1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18$ENV{GIT_DIR} ||= '.git';
19$Git::SVN::default_repo_id = 'svn';
20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21$Git::SVN::Ra::_log_window_size = 100;
22$Git::SVN::_minimize_url = 'unset';
23if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 }
30 }
31}
32$Git::SVN::Log::TZ = $ENV{TZ};
34$ENV{TZ} = 'UTC';
35$| = 1; # unbuffer STDOUT
36sub fatal (@) { print STDERR "@_\n"; exit 1 }
38require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39require SVN::Ra;
40require SVN::Delta;
41if ($SVN::Core::VERSION lt '1.1.0') {
42 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
43}
44my $can_compress = eval { require Compress::Zlib; 1};
45push @Git::SVN::Ra::ISA, 'SVN::Ra';
46push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
47push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
48use Carp qw/croak/;
49use Digest::MD5;
50use IO::File qw//;
51use File::Basename qw/dirname basename/;
52use File::Path qw/mkpath/;
53use File::Spec;
54use File::Find;
55use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
56use IPC::Open3;
57use Git;
58BEGIN {
60 # import functions from Git into our packages, en masse
61 no strict 'refs';
62 foreach (qw/command command_oneline command_noisy command_output_pipe
63 command_input_pipe command_close_pipe
64 command_bidi_pipe command_close_bidi_pipe/) {
65 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
66 Git::SVN::Migration Git::SVN::Log Git::SVN),
67 __PACKAGE__) {
68 *{"${package}::$_"} = \&{"Git::$_"};
69 }
70 }
71}
72my ($SVN);
74$sha1 = qr/[a-f\d]{40}/;
76$sha1_short = qr/[a-f\d]{4,40}/;
77my ($_stdin, $_help, $_edit,
78 $_message, $_file, $_branch_dest,
79 $_template, $_shared,
80 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
81 $_merge, $_strategy, $_dry_run, $_local,
82 $_prefix, $_no_checkout, $_url, $_verbose,
83 $_git_format, $_commit_url, $_tag);
84$Git::SVN::_follow_parent = 1;
85$_q ||= 0;
86my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
106my ($_trunk, @_tags, @_branches, $_stdlayout);
108my %icv;
109my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
110 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
111 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
112 'stdlayout|s' => \$_stdlayout,
113 'minimize-url|m!' => \$Git::SVN::_minimize_url,
114 'no-metadata' => sub { $icv{noMetadata} = 1 },
115 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
116 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
117 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
118 %remote_opts );
119my %cmt_opts = ( 'edit|e' => \$_edit,
120 'rmdir' => \$SVN::Git::Editor::_rmdir,
121 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
122 'l=i' => \$SVN::Git::Editor::_rename_limit,
123 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
124);
125my %cmd = (
127 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
128 { 'revision|r=s' => \$_revision,
129 'fetch-all|all' => \$_fetch_all,
130 'parent|p' => \$_fetch_parent,
131 %fc_opts } ],
132 clone => [ \&cmd_clone, "Initialize and fetch revisions",
133 { 'revision|r=s' => \$_revision,
134 %fc_opts, %init_opts } ],
135 init => [ \&cmd_init, "Initialize a repo for tracking" .
136 " (requires URL argument)",
137 \%init_opts ],
138 'multi-init' => [ \&cmd_multi_init,
139 "Deprecated alias for ".
140 "'$0 init -T<trunk> -b<branches> -t<tags>'",
141 \%init_opts ],
142 dcommit => [ \&cmd_dcommit,
143 'Commit several diffs to merge with upstream',
144 { 'merge|m|M' => \$_merge,
145 'strategy|s=s' => \$_strategy,
146 'verbose|v' => \$_verbose,
147 'dry-run|n' => \$_dry_run,
148 'fetch-all|all' => \$_fetch_all,
149 'commit-url=s' => \$_commit_url,
150 'revision|r=i' => \$_revision,
151 'no-rebase' => \$_no_rebase,
152 %cmt_opts, %fc_opts } ],
153 branch => [ \&cmd_branch,
154 'Create a branch in the SVN repository',
155 { 'message|m=s' => \$_message,
156 'destination|d=s' => \$_branch_dest,
157 'dry-run|n' => \$_dry_run,
158 'tag|t' => \$_tag } ],
159 tag => [ sub { $_tag = 1; cmd_branch(@_) },
160 'Create a tag in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run } ],
164 'set-tree' => [ \&cmd_set_tree,
165 "Set an SVN repository to a git tree-ish",
166 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
167 'create-ignore' => [ \&cmd_create_ignore,
168 'Create a .gitignore per svn:ignore',
169 { 'revision|r=i' => \$_revision
170 } ],
171 'mkdirs' => [ \&cmd_mkdirs ,
172 "recreate empty directories after a checkout",
173 { 'revision|r=i' => \$_revision } ],
174 'propget' => [ \&cmd_propget,
175 'Print the value of a property on a file or directory',
176 { 'revision|r=i' => \$_revision } ],
177 'proplist' => [ \&cmd_proplist,
178 'List all properties of a file or directory',
179 { 'revision|r=i' => \$_revision } ],
180 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
181 { 'revision|r=i' => \$_revision
182 } ],
183 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
184 { 'revision|r=i' => \$_revision
185 } ],
186 'multi-fetch' => [ \&cmd_multi_fetch,
187 "Deprecated alias for $0 fetch --all",
188 { 'revision|r=s' => \$_revision, %fc_opts } ],
189 'migrate' => [ sub { },
190 # no-op, we automatically run this anyways,
191 'Migrate configuration/metadata/layout from
192 previous versions of git-svn',
193 { 'minimize' => \$Git::SVN::Migration::_minimize,
194 %remote_opts } ],
195 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
196 { 'limit=i' => \$Git::SVN::Log::limit,
197 'revision|r=s' => \$_revision,
198 'verbose|v' => \$Git::SVN::Log::verbose,
199 'incremental' => \$Git::SVN::Log::incremental,
200 'oneline' => \$Git::SVN::Log::oneline,
201 'show-commit' => \$Git::SVN::Log::show_commit,
202 'non-recursive' => \$Git::SVN::Log::non_recursive,
203 'authors-file|A=s' => \$_authors,
204 'color' => \$Git::SVN::Log::color,
205 'pager=s' => \$Git::SVN::Log::pager
206 } ],
207 'find-rev' => [ \&cmd_find_rev,
208 "Translate between SVN revision numbers and tree-ish",
209 {} ],
210 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
211 { 'merge|m|M' => \$_merge,
212 'verbose|v' => \$_verbose,
213 'strategy|s=s' => \$_strategy,
214 'local|l' => \$_local,
215 'fetch-all|all' => \$_fetch_all,
216 'dry-run|n' => \$_dry_run,
217 %fc_opts } ],
218 'commit-diff' => [ \&cmd_commit_diff,
219 'Commit a diff between two trees',
220 { 'message|m=s' => \$_message,
221 'file|F=s' => \$_file,
222 'revision|r=s' => \$_revision,
223 %cmt_opts } ],
224 'info' => [ \&cmd_info,
225 "Show info about the latest SVN revision
226 on the current branch",
227 { 'url' => \$_url, } ],
228 'blame' => [ \&Git::SVN::Log::cmd_blame,
229 "Show what revision and author last modified each line of a file",
230 { 'git-format' => \$_git_format } ],
231 'reset' => [ \&cmd_reset,
232 "Undo fetches back to the specified SVN revision",
233 { 'revision|r=s' => \$_revision,
234 'parent|p' => \$_fetch_parent } ],
235 'gc' => [ \&cmd_gc,
236 "Compress unhandled.log files in .git/svn and remove " .
237 "index files in .git/svn",
238 {} ],
239);
240my $cmd;
242for (my $i = 0; $i < @ARGV; $i++) {
243 if (defined $cmd{$ARGV[$i]}) {
244 $cmd = $ARGV[$i];
245 splice @ARGV, $i, 1;
246 last;
247 } elsif ($ARGV[$i] eq 'help') {
248 $cmd = $ARGV[$i+1];
249 usage(0);
250 }
251};
252# make sure we're always running at the top-level working directory
254unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
255 unless (-d $ENV{GIT_DIR}) {
256 if ($git_dir_user_set) {
257 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
258 "but it is not a directory\n";
259 }
260 my $git_dir = delete $ENV{GIT_DIR};
261 my $cdup = undef;
262 git_cmd_try {
263 $cdup = command_oneline(qw/rev-parse --show-cdup/);
264 $git_dir = '.' unless ($cdup);
265 chomp $cdup if ($cdup);
266 $cdup = "." unless ($cdup && length $cdup);
267 } "Already at toplevel, but $git_dir not found\n";
268 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
269 unless (-d $git_dir) {
270 die "$git_dir still not found after going to ",
271 "'$cdup'\n";
272 }
273 $ENV{GIT_DIR} = $git_dir;
274 }
275 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
276}
277my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
279read_git_config(\%opts);
281if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
282 Getopt::Long::Configure('pass_through');
283}
284my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
285 'minimize-connections' => \$Git::SVN::Migration::_minimize,
286 'id|i=s' => \$Git::SVN::default_ref_id,
287 'svn-remote|remote|R=s' => sub {
288 $Git::SVN::no_reuse_existing = 1;
289 $Git::SVN::default_repo_id = $_[1] });
290exit 1 if (!$rv && $cmd && $cmd ne 'log');
291usage(0) if $_help;
293version() if $_version;
294usage(1) unless defined $cmd;
295load_authors() if $_authors;
296if (defined $_authors_prog) {
297 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
298}
299unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
301 Git::SVN::Migration::migration_check();
302}
303Git::SVN::init_vars();
304eval {
305 Git::SVN::verify_remotes_sanity();
306 $cmd{$cmd}->[0]->(@ARGV);
307};
308fatal $@ if $@;
309post_fetch_checkout();
310exit 0;
311####################### primary functions ######################
313sub usage {
314 my $exit = shift || 0;
315 my $fd = $exit ? \*STDERR : \*STDOUT;
316 print $fd <<"";
317git-svn - bidirectional operations between a single Subversion tree and git
318Usage: git svn <command> [options] [arguments]\n
319 print $fd "Available commands:\n" unless $cmd;
321 foreach (sort keys %cmd) {
323 next if $cmd && $cmd ne $_;
324 next if /^multi-/; # don't show deprecated commands
325 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
326 foreach (sort keys %{$cmd{$_}->[2]}) {
327 # mixed-case options are for .git/config only
328 next if /[A-Z]/ && /^[a-z]+$/i;
329 # prints out arguments as they should be passed:
330 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
331 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
332 "--$_" : "-$_" }
333 split /\|/,$_)," $x\n";
334 }
335 }
336 print $fd <<"";
337\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
338arbitrary identifier if you're tracking multiple SVN branches/repositories in
339one git repository and want to keep them separate. See git-svn(1) for more
340information.
341 exit $exit;
343}
344sub version {
346 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
347 exit 0;
348}
349sub do_git_init_db {
351 unless (-d $ENV{GIT_DIR}) {
352 my @init_db = ('init');
353 push @init_db, "--template=$_template" if defined $_template;
354 if (defined $_shared) {
355 if ($_shared =~ /[a-z]/) {
356 push @init_db, "--shared=$_shared";
357 } else {
358 push @init_db, "--shared";
359 }
360 }
361 command_noisy(@init_db);
362 $_repository = Git->repository(Repository => ".git");
363 }
364 command_noisy('config', 'core.autocrlf', 'false');
365 my $set;
366 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
367 foreach my $i (keys %icv) {
368 die "'$set' and '$i' cannot both be set\n" if $set;
369 next unless defined $icv{$i};
370 command_noisy('config', "$pfx.$i", $icv{$i});
371 $set = $i;
372 }
373 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
374 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
375 if defined $$ignore_regex;
376}
377sub init_subdir {
379 my $repo_path = shift or return;
380 mkpath([$repo_path]) unless -d $repo_path;
381 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
382 $ENV{GIT_DIR} = '.git';
383 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
384}
385sub cmd_clone {
387 my ($url, $path) = @_;
388 if (!defined $path &&
389 (defined $_trunk || @_branches || @_tags ||
390 defined $_stdlayout) &&
391 $url !~ m#^[a-z\+]+://#) {
392 $path = $url;
393 }
394 $path = basename($url) if !defined $path || !length $path;
395 cmd_init($url, $path);
396 Git::SVN::fetch_all($Git::SVN::default_repo_id);
397 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
398}
399sub cmd_init {
401 if (defined $_stdlayout) {
402 $_trunk = 'trunk' if (!defined $_trunk);
403 @_tags = 'tags' if (! @_tags);
404 @_branches = 'branches' if (! @_branches);
405 }
406 if (defined $_trunk || @_branches || @_tags) {
407 return cmd_multi_init(@_);
408 }
409 my $url = shift or die "SVN repository location required ",
410 "as a command-line argument\n";
411 $url = canonicalize_url($url);
412 init_subdir(@_);
413 do_git_init_db();
414 if ($Git::SVN::_minimize_url eq 'unset') {
416 $Git::SVN::_minimize_url = 0;
417 }
418 Git::SVN->init($url);
420}
421sub cmd_fetch {
423 if (grep /^\d+=./, @_) {
424 die "'<rev>=<commit>' fetch arguments are ",
425 "no longer supported.\n";
426 }
427 my ($remote) = @_;
428 if (@_ > 1) {
429 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
430 }
431 if ($_fetch_parent) {
432 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
433 unless ($gs) {
434 die "Unable to determine upstream SVN information from ",
435 "working tree history\n";
436 }
437 # just fetch, don't checkout.
438 $_no_checkout = 'true';
439 $_fetch_all ? $gs->fetch_all : $gs->fetch;
440 } elsif ($_fetch_all) {
441 cmd_multi_fetch();
442 } else {
443 $remote ||= $Git::SVN::default_repo_id;
444 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
445 }
446}
447sub cmd_set_tree {
449 my (@commits) = @_;
450 if ($_stdin || !@commits) {
451 print "Reading from stdin...\n";
452 @commits = ();
453 while (<STDIN>) {
454 if (/\b($sha1_short)\b/o) {
455 unshift @commits, $1;
456 }
457 }
458 }
459 my @revs;
460 foreach my $c (@commits) {
461 my @tmp = command('rev-parse',$c);
462 if (scalar @tmp == 1) {
463 push @revs, $tmp[0];
464 } elsif (scalar @tmp > 1) {
465 push @revs, reverse(command('rev-list',@tmp));
466 } else {
467 fatal "Failed to rev-parse $c";
468 }
469 }
470 my $gs = Git::SVN->new;
471 my ($r_last, $cmt_last) = $gs->last_rev_commit;
472 $gs->fetch;
473 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
474 fatal "There are new revisions that were fetched ",
475 "and need to be merged (or acknowledged) ",
476 "before committing.\nlast rev: $r_last\n",
477 " current: $gs->{last_rev}";
478 }
479 $gs->set_tree($_) foreach @revs;
480 print "Done committing ",scalar @revs," revisions to SVN\n";
481 unlink $gs->{index};
482}
483sub cmd_dcommit {
485 my $head = shift;
486 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
487 'Cannot dcommit with a dirty index. Commit your changes first, '
488 . "or stash them with `git stash'.\n";
489 $head ||= 'HEAD';
490 my $old_head;
492 if ($head ne 'HEAD') {
493 $old_head = eval {
494 command_oneline([qw/symbolic-ref -q HEAD/])
495 };
496 if ($old_head) {
497 $old_head =~ s{^refs/heads/}{};
498 } else {
499 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
500 }
501 command(['checkout', $head], STDERR => 0);
502 }
503 my @refs;
505 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
506 unless ($gs) {
507 die "Unable to determine upstream SVN information from ",
508 "$head history.\nPerhaps the repository is empty.";
509 }
510 if (defined $_commit_url) {
512 $url = $_commit_url;
513 } else {
514 $url = eval { command_oneline('config', '--get',
515 "svn-remote.$gs->{repo_id}.commiturl") };
516 if (!$url) {
517 $url = $gs->full_url
518 }
519 }
520 my $last_rev = $_revision if defined $_revision;
522 if ($url) {
523 print "Committing to $url ...\n";
524 }
525 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
526 if ($_no_rebase && scalar(@$linear_refs) > 1) {
527 warn "Attempting to commit more than one change while ",
528 "--no-rebase is enabled.\n",
529 "If these changes depend on each other, re-running ",
530 "without --no-rebase may be required."
531 }
532 my $expect_url = $url;
533 Git::SVN::remove_username($expect_url);
534 while (1) {
535 my $d = shift @$linear_refs or last;
536 unless (defined $last_rev) {
537 (undef, $last_rev, undef) = cmt_metadata("$d~1");
538 unless (defined $last_rev) {
539 fatal "Unable to extract revision information ",
540 "from commit $d~1";
541 }
542 }
543 if ($_dry_run) {
544 print "diff-tree $d~1 $d\n";
545 } else {
546 my $cmt_rev;
547 my %ed_opts = ( r => $last_rev,
548 log => get_commit_entry($d)->{log},
549 ra => Git::SVN::Ra->new($url),
550 config => SVN::Core::config_get_config(
551 $Git::SVN::Ra::config_dir
552 ),
553 tree_a => "$d~1",
554 tree_b => $d,
555 editor_cb => sub {
556 print "Committed r$_[0]\n";
557 $cmt_rev = $_[0];
558 },
559 svn_path => '');
560 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
561 print "No changes\n$d~1 == $d\n";
562 } elsif ($parents->{$d} && @{$parents->{$d}}) {
563 $gs->{inject_parents_dcommit}->{$cmt_rev} =
564 $parents->{$d};
565 }
566 $_fetch_all ? $gs->fetch_all : $gs->fetch;
567 $last_rev = $cmt_rev;
568 next if $_no_rebase;
569 # we always want to rebase against the current HEAD,
571 # not any head that was passed to us
572 my @diff = command('diff-tree', $d,
573 $gs->refname, '--');
574 my @finish;
575 if (@diff) {
576 @finish = rebase_cmd();
577 print STDERR "W: $d and ", $gs->refname,
578 " differ, using @finish:\n",
579 join("\n", @diff), "\n";
580 } else {
581 print "No changes between current HEAD and ",
582 $gs->refname,
583 "\nResetting to the latest ",
584 $gs->refname, "\n";
585 @finish = qw/reset --mixed/;
586 }
587 command_noisy(@finish, $gs->refname);
588 if (@diff) {
589 @refs = ();
590 my ($url_, $rev_, $uuid_, $gs_) =
591 working_head_info('HEAD', \@refs);
592 my ($linear_refs_, $parents_) =
593 linearize_history($gs_, \@refs);
594 if (scalar(@$linear_refs) !=
595 scalar(@$linear_refs_)) {
596 fatal "# of revisions changed ",
597 "\nbefore:\n",
598 join("\n", @$linear_refs),
599 "\n\nafter:\n",
600 join("\n", @$linear_refs_), "\n",
601 'If you are attempting to commit ',
602 "merges, try running:\n\t",
603 'git rebase --interactive',
604 '--preserve-merges ',
605 $gs->refname,
606 "\nBefore dcommitting";
607 }
608 if ($url_ ne $expect_url) {
609 if ($url_ eq $gs->metadata_url) {
610 print
611 "Accepting rewritten URL:",
612 " $url_\n";
613 } else {
614 fatal
615 "URL mismatch after rebase:",
616 " $url_ != $expect_url";
617 }
618 }
619 if ($uuid_ ne $uuid) {
620 fatal "uuid mismatch after rebase: ",
621 "$uuid_ != $uuid";
622 }
623 # remap parents
624 my (%p, @l, $i);
625 for ($i = 0; $i < scalar @$linear_refs; $i++) {
626 my $new = $linear_refs_->[$i] or next;
627 $p{$new} =
628 $parents->{$linear_refs->[$i]};
629 push @l, $new;
630 }
631 $parents = \%p;
632 $linear_refs = \@l;
633 }
634 }
635 }
636 if ($old_head) {
638 my $new_head = command_oneline(qw/rev-parse HEAD/);
639 my $new_is_symbolic = eval {
640 command_oneline(qw/symbolic-ref -q HEAD/);
641 };
642 if ($new_is_symbolic) {
643 print "dcommitted the branch ", $head, "\n";
644 } else {
645 print "dcommitted on a detached HEAD because you gave ",
646 "a revision argument.\n",
647 "The rewritten commit is: ", $new_head, "\n";
648 }
649 command(['checkout', $old_head], STDERR => 0);
650 }
651 unlink $gs->{index};
653}
654sub cmd_branch {
656 my ($branch_name, $head) = @_;
657 unless (defined $branch_name && length $branch_name) {
659 die(($_tag ? "tag" : "branch") . " name required\n");
660 }
661 $head ||= 'HEAD';
662 my ($src, $rev, undef, $gs) = working_head_info($head);
664 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
666 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
667 my $glob;
668 if ($#{$allglobs} == 0) {
669 $glob = $allglobs->[0];
670 } else {
671 unless(defined $_branch_dest) {
672 die "Multiple ",
673 $_tag ? "tag" : "branch",
674 " paths defined for Subversion repository.\n",
675 "You must specify where you want to create the ",
676 $_tag ? "tag" : "branch",
677 " with the --destination argument.\n";
678 }
679 foreach my $g (@{$allglobs}) {
680 # SVN::Git::Editor could probably be moved to Git.pm..
681 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
682 if ($_branch_dest =~ /$re/) {
683 $glob = $g;
684 last;
685 }
686 }
687 unless (defined $glob) {
688 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
689 foreach my $g (@{$allglobs}) {
690 $g->{path}->{left} =~ /$dest_re/ or next;
691 if (defined $glob) {
692 die "Ambiguous destination: ",
693 $_branch_dest, "\nmatches both '",
694 $glob->{path}->{left}, "' and '",
695 $g->{path}->{left}, "'\n";
696 }
697 $glob = $g;
698 }
699 unless (defined $glob) {
700 die "Unknown ",
701 $_tag ? "tag" : "branch",
702 " destination $_branch_dest\n";
703 }
704 }
705 }
706 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
707 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
708 my $ctx = SVN::Client->new(
710 auth => Git::SVN::Ra::_auth_providers(),
711 log_msg => sub {
712 ${ $_[0] } = defined $_message
713 ? $_message
714 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
715 . $branch_name;
716 },
717 );
718 eval {
720 $ctx->ls($dst, 'HEAD', 0);
721 } and die "branch ${branch_name} already exists\n";
722 print "Copying ${src} at r${rev} to ${dst}...\n";
724 $ctx->copy($src, $rev, $dst)
725 unless $_dry_run;
726 $gs->fetch_all;
728}
729sub cmd_find_rev {
731 my $revision_or_hash = shift or die "SVN or git revision required ",
732 "as a command-line argument\n";
733 my $result;
734 if ($revision_or_hash =~ /^r\d+$/) {
735 my $head = shift;
736 $head ||= 'HEAD';
737 my @refs;
738 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
739 unless ($gs) {
740 die "Unable to determine upstream SVN information from ",
741 "$head history\n";
742 }
743 my $desired_revision = substr($revision_or_hash, 1);
744 $result = $gs->rev_map_get($desired_revision, $uuid);
745 } else {
746 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
747 $result = $rev;
748 }
749 print "$result\n" if $result;
750}
751sub cmd_rebase {
753 command_noisy(qw/update-index --refresh/);
754 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
755 unless ($gs) {
756 die "Unable to determine upstream SVN information from ",
757 "working tree history\n";
758 }
759 if ($_dry_run) {
760 print "Remote Branch: " . $gs->refname . "\n";
761 print "SVN URL: " . $url . "\n";
762 return;
763 }
764 if (command(qw/diff-index HEAD --/)) {
765 print STDERR "Cannot rebase with uncommited changes:\n";
766 command_noisy('status');
767 exit 1;
768 }
769 unless ($_local) {
770 # rebase will checkout for us, so no need to do it explicitly
771 $_no_checkout = 'true';
772 $_fetch_all ? $gs->fetch_all : $gs->fetch;
773 }
774 command_noisy(rebase_cmd(), $gs->refname);
775 $gs->mkemptydirs;
776}
777sub cmd_show_ignore {
779 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
780 $gs ||= Git::SVN->new;
781 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
782 $gs->prop_walk($gs->{path}, $r, sub {
783 my ($gs, $path, $props) = @_;
784 print STDOUT "\n# $path\n";
785 my $s = $props->{'svn:ignore'} or return;
786 $s =~ s/[\r\n]+/\n/g;
787 $s =~ s/^\n+//;
788 chomp $s;
789 $s =~ s#^#$path#gm;
790 print STDOUT "$s\n";
791 });
792}
793sub cmd_show_externals {
795 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
796 $gs ||= Git::SVN->new;
797 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
798 $gs->prop_walk($gs->{path}, $r, sub {
799 my ($gs, $path, $props) = @_;
800 print STDOUT "\n# $path\n";
801 my $s = $props->{'svn:externals'} or return;
802 $s =~ s/[\r\n]+/\n/g;
803 chomp $s;
804 $s =~ s#^#$path#gm;
805 print STDOUT "$s\n";
806 });
807}
808sub cmd_create_ignore {
810 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
811 $gs ||= Git::SVN->new;
812 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
813 $gs->prop_walk($gs->{path}, $r, sub {
814 my ($gs, $path, $props) = @_;
815 # $path is of the form /path/to/dir/
816 $path = '.' . $path;
817 # SVN can have attributes on empty directories,
818 # which git won't track
819 mkpath([$path]) unless -d $path;
820 my $ignore = $path . '.gitignore';
821 my $s = $props->{'svn:ignore'} or return;
822 open(GITIGNORE, '>', $ignore)
823 or fatal("Failed to open `$ignore' for writing: $!");
824 $s =~ s/[\r\n]+/\n/g;
825 $s =~ s/^\n+//;
826 chomp $s;
827 # Prefix all patterns so that the ignore doesn't apply
828 # to sub-directories.
829 $s =~ s#^#/#gm;
830 print GITIGNORE "$s\n";
831 close(GITIGNORE)
832 or fatal("Failed to close `$ignore': $!");
833 command_noisy('add', '-f', $ignore);
834 });
835}
836sub cmd_mkdirs {
838 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
839 $gs ||= Git::SVN->new;
840 $gs->mkemptydirs($_revision);
841}
842sub canonicalize_path {
844 my ($path) = @_;
845 my $dot_slash_added = 0;
846 if (substr($path, 0, 1) ne "/") {
847 $path = "./" . $path;
848 $dot_slash_added = 1;
849 }
850 # File::Spec->canonpath doesn't collapse x/../y into y (for a
851 # good reason), so let's do this manually.
852 $path =~ s#/+#/#g;
853 $path =~ s#/\.(?:/|$)#/#g;
854 $path =~ s#/[^/]+/\.\.##g;
855 $path =~ s#/$##g;
856 $path =~ s#^\./## if $dot_slash_added;
857 $path =~ s#^/##;
858 $path =~ s#^\.$##;
859 return $path;
860}
861sub canonicalize_url {
863 my ($url) = @_;
864 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
865 return $url;
866}
867# get_svnprops(PATH)
869# ------------------
870# Helper for cmd_propget and cmd_proplist below.
871sub get_svnprops {
872 my $path = shift;
873 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
874 $gs ||= Git::SVN->new;
875 # prefix THE PATH by the sub-directory from which the user
877 # invoked us.
878 $path = $cmd_dir_prefix . $path;
879 fatal("No such file or directory: $path") unless -e $path;
880 my $is_dir = -d $path ? 1 : 0;
881 $path = $gs->{path} . '/' . $path;
882 # canonicalize the path (otherwise libsvn will abort or fail to
884 # find the file)
885 $path = canonicalize_path($path);
886 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
888 my $props;
889 if ($is_dir) {
890 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
891 }
892 else {
893 (undef, $props) = $gs->ra->get_file($path, $r, undef);
894 }
895 return $props;
896}
897# cmd_propget (PROP, PATH)
899# ------------------------
900# Print the SVN property PROP for PATH.
901sub cmd_propget {
902 my ($prop, $path) = @_;
903 $path = '.' if not defined $path;
904 usage(1) if not defined $prop;
905 my $props = get_svnprops($path);
906 if (not defined $props->{$prop}) {
907 fatal("`$path' does not have a `$prop' SVN property.");
908 }
909 print $props->{$prop} . "\n";
910}
911# cmd_proplist (PATH)
913# -------------------
914# Print the list of SVN properties for PATH.
915sub cmd_proplist {
916 my $path = shift;
917 $path = '.' if not defined $path;
918 my $props = get_svnprops($path);
919 print "Properties on '$path':\n";
920 foreach (sort keys %{$props}) {
921 print " $_\n";
922 }
923}
924sub cmd_multi_init {
926 my $url = shift;
927 unless (defined $_trunk || @_branches || @_tags) {
928 usage(1);
929 }
930 $_prefix = '' unless defined $_prefix;
932 if (defined $url) {
933 $url = canonicalize_url($url);
934 init_subdir(@_);
935 }
936 do_git_init_db();
937 if (defined $_trunk) {
938 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
939 # try both old-style and new-style lookups:
940 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
941 unless ($gs_trunk) {
942 my ($trunk_url, $trunk_path) =
943 complete_svn_url($url, $_trunk);
944 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
945 undef, $trunk_ref);
946 }
947 }
948 return unless @_branches || @_tags;
949 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
950 foreach my $path (@_branches) {
951 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
952 }
953 foreach my $path (@_tags) {
954 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
955 }
956}
957sub cmd_multi_fetch {
959 my $remotes = Git::SVN::read_all_remotes();
960 foreach my $repo_id (sort keys %$remotes) {
961 if ($remotes->{$repo_id}->{url}) {
962 Git::SVN::fetch_all($repo_id, $remotes);
963 }
964 }
965}
966# this command is special because it requires no metadata
968sub cmd_commit_diff {
969 my ($ta, $tb, $url) = @_;
970 my $usage = "Usage: $0 commit-diff -r<revision> ".
971 "<tree-ish> <tree-ish> [<URL>]";
972 fatal($usage) if (!defined $ta || !defined $tb);
973 my $svn_path = '';
974 if (!defined $url) {
975 my $gs = eval { Git::SVN->new };
976 if (!$gs) {
977 fatal("Needed URL or usable git-svn --id in ",
978 "the command-line\n", $usage);
979 }
980 $url = $gs->{url};
981 $svn_path = $gs->{path};
982 }
983 unless (defined $_revision) {
984 fatal("-r|--revision is a required argument\n", $usage);
985 }
986 if (defined $_message && defined $_file) {
987 fatal("Both --message/-m and --file/-F specified ",
988 "for the commit message.\n",
989 "I have no idea what you mean");
990 }
991 if (defined $_file) {
992 $_message = file_to_s($_file);
993 } else {
994 $_message ||= get_commit_entry($tb)->{log};
995 }
996 my $ra ||= Git::SVN::Ra->new($url);
997 my $r = $_revision;
998 if ($r eq 'HEAD') {
999 $r = $ra->get_latest_revnum;
1000 } elsif ($r !~ /^\d+$/) {
1001 die "revision argument: $r not understood by git-svn\n";
1002 }
1003 my %ed_opts = ( r => $r,
1004 log => $_message,
1005 ra => $ra,
1006 tree_a => $ta,
1007 tree_b => $tb,
1008 editor_cb => sub { print "Committed r$_[0]\n" },
1009 svn_path => $svn_path );
1010 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1011 print "No changes\n$ta == $tb\n";
1012 }
1013}
1014sub escape_uri_only {
1016 my ($uri) = @_;
1017 my @tmp;
1018 foreach (split m{/}, $uri) {
1019 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1020 push @tmp, $_;
1021 }
1022 join('/', @tmp);
1023}
1024sub escape_url {
1026 my ($url) = @_;
1027 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1028 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1029 $url = "$scheme://$domain$uri";
1030 }
1031 $url;
1032}
1033sub cmd_info {
1035 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1036 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1037 if (exists $_[1]) {
1038 die "Too many arguments specified\n";
1039 }
1040 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1042 if (!$file_type && !$diff_status) {
1044 print STDERR "svn: '$path' is not under version control\n";
1045 exit 1;
1046 }
1047 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1049 unless ($gs) {
1050 die "Unable to determine upstream SVN information from ",
1051 "working tree history\n";
1052 }
1053 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1055 $path = "." if $path eq "";
1056 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1058 if ($_url) {
1060 print escape_url($full_url), "\n";
1061 return;
1062 }
1063 my $result = "Path: $path\n";
1065 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1066 $result .= "URL: " . escape_url($full_url) . "\n";
1067 eval {
1069 my $repos_root = $gs->repos_root;
1070 Git::SVN::remove_username($repos_root);
1071 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1072 };
1073 if ($@) {
1074 $result .= "Repository Root: (offline)\n";
1075 }
1076 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1077 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1078 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1079 $result .= "Node Kind: " .
1081 ($file_type eq "dir" ? "directory" : "file") . "\n";
1082 my $schedule = $diff_status eq "A"
1084 ? "add"
1085 : ($diff_status eq "D" ? "delete" : "normal");
1086 $result .= "Schedule: $schedule\n";
1087 if ($diff_status eq "A") {
1089 print $result, "\n";
1090 return;
1091 }
1092 my ($lc_author, $lc_rev, $lc_date_utc);
1094 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1095 my $log = command_output_pipe(@args);
1096 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1097 while (<$log>) {
1098 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1099 $lc_author = $1;
1100 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1101 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1102 (undef, $lc_rev, undef) = ::extract_metadata($1);
1103 }
1104 }
1105 close $log;
1106 Git::SVN::Log::set_local_timezone();
1108 $result .= "Last Changed Author: $lc_author\n";
1110 $result .= "Last Changed Rev: $lc_rev\n";
1111 $result .= "Last Changed Date: " .
1112 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1113 if ($file_type ne "dir") {
1115 my $text_last_updated_date =
1116 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1117 $result .=
1118 "Text Last Updated: " .
1119 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1120 "\n";
1121 my $checksum;
1122 if ($diff_status eq "D") {
1123 my ($fh, $ctx) =
1124 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1125 if ($file_type eq "link") {
1126 my $file_name = <$fh>;
1127 $checksum = md5sum("link $file_name");
1128 } else {
1129 $checksum = md5sum($fh);
1130 }
1131 command_close_pipe($fh, $ctx);
1132 } elsif ($file_type eq "link") {
1133 my $file_name =
1134 command(qw(cat-file blob), "HEAD:$path");
1135 $checksum =
1136 md5sum("link " . $file_name);
1137 } else {
1138 open FILE, "<", $path or die $!;
1139 $checksum = md5sum(\*FILE);
1140 close FILE or die $!;
1141 }
1142 $result .= "Checksum: " . $checksum . "\n";
1143 }
1144 print $result, "\n";
1146}
1147sub cmd_reset {
1149 my $target = shift || $_revision or die "SVN revision required\n";
1150 $target = $1 if $target =~ /^r(\d+)$/;
1151 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1152 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1153 unless ($gs) {
1154 die "Unable to determine upstream SVN information from ".
1155 "history\n";
1156 }
1157 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1158 $gs->rev_map_set($r, $c, 'reset', $uuid);
1159 print "r$r = $c ($gs->{ref_id})\n";
1160}
1161sub cmd_gc {
1163 if (!$can_compress) {
1164 warn "Compress::Zlib could not be found; unhandled.log " .
1165 "files will not be compressed.\n";
1166 }
1167 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1168}
1169########################### utility functions #########################
1171sub rebase_cmd {
1173 my @cmd = qw/rebase/;
1174 push @cmd, '-v' if $_verbose;
1175 push @cmd, qw/--merge/ if $_merge;
1176 push @cmd, "--strategy=$_strategy" if $_strategy;
1177 @cmd;
1178}
1179sub post_fetch_checkout {
1181 return if $_no_checkout;
1182 my $gs = $Git::SVN::_head or return;
1183 return if verify_ref('refs/heads/master^0');
1184 # look for "trunk" ref if it exists
1186 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1187 my $fetch = $remote->{fetch};
1188 if ($fetch) {
1189 foreach my $p (keys %$fetch) {
1190 basename($fetch->{$p}) eq 'trunk' or next;
1191 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1192 last;
1193 }
1194 }
1195 my $valid_head = verify_ref('HEAD^0');
1197 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1198 return if ($valid_head || !verify_ref('HEAD^0'));
1199 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1201 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1202 return if -f $index;
1203 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1205 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1206 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1207 print STDERR "Checked out HEAD:\n ",
1208 $gs->full_url, " r", $gs->last_rev, "\n";
1209 $gs->mkemptydirs($gs->last_rev);
1210}
1211sub complete_svn_url {
1213 my ($url, $path) = @_;
1214 $path =~ s#/+$##;
1215 if ($path !~ m#^[a-z\+]+://#) {
1216 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1217 fatal("E: '$path' is not a complete URL ",
1218 "and a separate URL is not specified");
1219 }
1220 return ($url, $path);
1221 }
1222 return ($path, '');
1223}
1224sub complete_url_ls_init {
1226 my ($ra, $repo_path, $switch, $pfx) = @_;
1227 unless ($repo_path) {
1228 print STDERR "W: $switch not specified\n";
1229 return;
1230 }
1231 $repo_path =~ s#/+$##;
1232 if ($repo_path =~ m#^[a-z\+]+://#) {
1233 $ra = Git::SVN::Ra->new($repo_path);
1234 $repo_path = '';
1235 } else {
1236 $repo_path =~ s#^/+##;
1237 unless ($ra) {
1238 fatal("E: '$repo_path' is not a complete URL ",
1239 "and a separate URL is not specified");
1240 }
1241 }
1242 my $url = $ra->{url};
1243 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1244 my $k = "svn-remote.$gs->{repo_id}.url";
1245 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1246 if ($orig_url && ($orig_url ne $gs->{url})) {
1247 die "$k already set: $orig_url\n",
1248 "wanted to set to: $gs->{url}\n";
1249 }
1250 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1251 my $remote_path = "$gs->{path}/$repo_path";
1252 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1253 $remote_path =~ s#/+#/#g;
1254 $remote_path =~ s#^/##g;
1255 $remote_path .= "/*" if $remote_path !~ /\*/;
1256 my ($n) = ($switch =~ /^--(\w+)/);
1257 if (length $pfx && $pfx !~ m#/$#) {
1258 die "--prefix='$pfx' must have a trailing slash '/'\n";
1259 }
1260 command_noisy('config',
1261 '--add',
1262 "svn-remote.$gs->{repo_id}.$n",
1263 "$remote_path:refs/remotes/$pfx*" .
1264 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1265}
1266sub verify_ref {
1268 my ($ref) = @_;
1269 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1270 { STDERR => 0 }); };
1271}
1272sub get_tree_from_treeish {
1274 my ($treeish) = @_;
1275 # $treeish can be a symbolic ref, too:
1276 my $type = command_oneline(qw/cat-file -t/, $treeish);
1277 my $expected;
1278 while ($type eq 'tag') {
1279 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1280 }
1281 if ($type eq 'commit') {
1282 $expected = (grep /^tree /, command(qw/cat-file commit/,
1283 $treeish))[0];
1284 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1285 die "Unable to get tree from $treeish\n" unless $expected;
1286 } elsif ($type eq 'tree') {
1287 $expected = $treeish;
1288 } else {
1289 die "$treeish is a $type, expected tree, tag or commit\n";
1290 }
1291 return $expected;
1292}
1293sub get_commit_entry {
1295 my ($treeish) = shift;
1296 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1297 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1298 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1299 open my $log_fh, '>', $commit_editmsg or croak $!;
1300 my $type = command_oneline(qw/cat-file -t/, $treeish);
1302 if ($type eq 'commit' || $type eq 'tag') {
1303 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1304 $type, $treeish);
1305 my $in_msg = 0;
1306 my $author;
1307 my $saw_from = 0;
1308 my $msgbuf = "";
1309 while (<$msg_fh>) {
1310 if (!$in_msg) {
1311 $in_msg = 1 if (/^\s*$/);
1312 $author = $1 if (/^author (.*>)/);
1313 } elsif (/^git-svn-id: /) {
1314 # skip this for now, we regenerate the
1315 # correct one on re-fetch anyways
1316 # TODO: set *:merge properties or like...
1317 } else {
1318 if (/^From:/ || /^Signed-off-by:/) {
1319 $saw_from = 1;
1320 }
1321 $msgbuf .= $_;
1322 }
1323 }
1324 $msgbuf =~ s/\s+$//s;
1325 if ($Git::SVN::_add_author_from && defined($author)
1326 && !$saw_from) {
1327 $msgbuf .= "\n\nFrom: $author";
1328 }
1329 print $log_fh $msgbuf or croak $!;
1330 command_close_pipe($msg_fh, $ctx);
1331 }
1332 close $log_fh or croak $!;
1333 if ($_edit || ($type eq 'tree')) {
1335 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1336 # TODO: strip out spaces, comments, like git-commit.sh
1337 system($editor, $commit_editmsg);
1338 }
1339 rename $commit_editmsg, $commit_msg or croak $!;
1340 {
1341 require Encode;
1342 # SVN requires messages to be UTF-8 when entering the repo
1343 local $/;
1344 open $log_fh, '<', $commit_msg or croak $!;
1345 binmode $log_fh;
1346 chomp($log_entry{log} = <$log_fh>);
1347 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1349 my $msg = $log_entry{log};
1350 eval { $msg = Encode::decode($enc, $msg, 1) };
1352 if ($@) {
1353 die "Could not decode as $enc:\n", $msg,
1354 "\nPerhaps you need to set i18n.commitencoding\n";
1355 }
1356 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1358 die "Could not encode as UTF-8:\n$msg\n" if $@;
1359 $log_entry{log} = $msg;
1361 close $log_fh or croak $!;
1363 }
1364 unlink $commit_msg;
1365 \%log_entry;
1366}
1367sub s_to_file {
1369 my ($str, $file, $mode) = @_;
1370 open my $fd,'>',$file or croak $!;
1371 print $fd $str,"\n" or croak $!;
1372 close $fd or croak $!;
1373 chmod ($mode &~ umask, $file) if (defined $mode);
1374}
1375sub file_to_s {
1377 my $file = shift;
1378 open my $fd,'<',$file or croak "$!: file: $file\n";
1379 local $/;
1380 my $ret = <$fd>;
1381 close $fd or croak $!;
1382 $ret =~ s/\s*$//s;
1383 return $ret;
1384}
1385# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1387sub load_authors {
1388 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1389 my $log = $cmd eq 'log';
1390 while (<$authors>) {
1391 chomp;
1392 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1393 my ($user, $name, $email) = ($1, $2, $3);
1394 if ($log) {
1395 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1396 } else {
1397 $users{$user} = [$name, $email];
1398 }
1399 }
1400 close $authors or croak $!;
1401}
1402# convert GetOpt::Long specs for use by git-config
1404sub read_git_config {
1405 my $opts = shift;
1406 my @config_only;
1407 foreach my $o (keys %$opts) {
1408 # if we have mixedCase and a long option-only, then
1409 # it's a config-only variable that we don't need for
1410 # the command-line.
1411 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1412 my $v = $opts->{$o};
1413 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1414 $key =~ s/-//g;
1415 my $arg = 'git config';
1416 $arg .= ' --int' if ($o =~ /[:=]i$/);
1417 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1418 if (ref $v eq 'ARRAY') {
1419 chomp(my @tmp = `$arg --get-all svn.$key`);
1420 @$v = @tmp if @tmp;
1421 } else {
1422 chomp(my $tmp = `$arg --get svn.$key`);
1423 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1424 $$v = $tmp;
1425 }
1426 }
1427 }
1428 delete @$opts{@config_only} if @config_only;
1429}
1430sub extract_metadata {
1432 my $id = shift or return (undef, undef, undef);
1433 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1434 \s([a-f\d\-]+)$/ix);
1435 if (!defined $rev || !$uuid || !$url) {
1436 # some of the original repositories I made had
1437 # identifiers like this:
1438 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1439 }
1440 return ($url, $rev, $uuid);
1441}
1442sub cmt_metadata {
1444 return extract_metadata((grep(/^git-svn-id: /,
1445 command(qw/cat-file commit/, shift)))[-1]);
1446}
1447sub cmt_sha2rev_batch {
1449 my %s2r;
1450 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1451 my $list = shift;
1452 foreach my $sha (@{$list}) {
1454 my $first = 1;
1455 my $size = 0;
1456 print $out $sha, "\n";
1457 while (my $line = <$in>) {
1459 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1460 last;
1461 } elsif ($first &&
1462 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1463 $first = 0;
1464 $size = $1;
1465 next;
1466 } elsif ($line =~ /^(git-svn-id: )/) {
1467 my (undef, $rev, undef) =
1468 extract_metadata($line);
1469 $s2r{$sha} = $rev;
1470 }
1471 $size -= length($line);
1473 last if ($size == 0);
1474 }
1475 }
1476 command_close_bidi_pipe($pid, $in, $out, $ctx);
1478 return \%s2r;
1480}
1481sub working_head_info {
1483 my ($head, $refs) = @_;
1484 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1485 my ($fh, $ctx) = command_output_pipe(@args, $head);
1486 my $hash;
1487 my %max;
1488 while (<$fh>) {
1489 if ( m{^commit ($::sha1)$} ) {
1490 unshift @$refs, $hash if $hash and $refs;
1491 $hash = $1;
1492 next;
1493 }
1494 next unless s{^\s*(git-svn-id:)}{$1};
1495 my ($url, $rev, $uuid) = extract_metadata($_);
1496 if (defined $url && defined $rev) {
1497 next if $max{$url} and $max{$url} < $rev;
1498 if (my $gs = Git::SVN->find_by_url($url)) {
1499 my $c = $gs->rev_map_get($rev, $uuid);
1500 if ($c && $c eq $hash) {
1501 close $fh; # break the pipe
1502 return ($url, $rev, $uuid, $gs);
1503 } else {
1504 $max{$url} ||= $gs->rev_map_max;
1505 }
1506 }
1507 }
1508 }
1509 command_close_pipe($fh, $ctx);
1510 (undef, undef, undef, undef);
1511}
1512sub read_commit_parents {
1514 my ($parents, $c) = @_;
1515 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1516 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1517 @{$parents->{$c}} = split(/ /, $p);
1518}
1519sub linearize_history {
1521 my ($gs, $refs) = @_;
1522 my %parents;
1523 foreach my $c (@$refs) {
1524 read_commit_parents(\%parents, $c);
1525 }
1526 my @linear_refs;
1528 my %skip = ();
1529 my $last_svn_commit = $gs->last_commit;
1530 foreach my $c (reverse @$refs) {
1531 next if $c eq $last_svn_commit;
1532 last if $skip{$c};
1533 unshift @linear_refs, $c;
1535 $skip{$c} = 1;
1536 # we only want the first parent to diff against for linear
1538 # history, we save the rest to inject when we finalize the
1539 # svn commit
1540 my $fp_a = verify_ref("$c~1");
1541 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1542 if (!$fp_a || !$fp_b) {
1543 die "Commit $c\n",
1544 "has no parent commit, and therefore ",
1545 "nothing to diff against.\n",
1546 "You should be working from a repository ",
1547 "originally created by git-svn\n";
1548 }
1549 if ($fp_a ne $fp_b) {
1550 die "$c~1 = $fp_a, however parsing commit $c ",
1551 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1552 }
1553 foreach my $p (@{$parents{$c}}) {
1555 $skip{$p} = 1;
1556 }
1557 }
1558 (\@linear_refs, \%parents);
1559}
1560sub find_file_type_and_diff_status {
1562 my ($path) = @_;
1563 return ('dir', '') if $path eq '';
1564 my $diff_output =
1566 command_oneline(qw(diff --cached --name-status --), $path) || "";
1567 my $diff_status = (split(' ', $diff_output))[0] || "";
1568 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1570 return (undef, undef) if !$diff_status && !$ls_tree;
1572 if ($diff_status eq "A") {
1574 return ("link", $diff_status) if -l $path;
1575 return ("dir", $diff_status) if -d $path;
1576 return ("file", $diff_status);
1577 }
1578 my $mode = (split(' ', $ls_tree))[0] || "";
1580 return ("link", $diff_status) if $mode eq "120000";
1582 return ("dir", $diff_status) if $mode eq "040000";
1583 return ("file", $diff_status);
1584}
1585sub md5sum {
1587 my $arg = shift;
1588 my $ref = ref $arg;
1589 my $md5 = Digest::MD5->new();
1590 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1591 $md5->addfile($arg) or croak $!;
1592 } elsif ($ref eq 'SCALAR') {
1593 $md5->add($$arg) or croak $!;
1594 } elsif (!$ref) {
1595 $md5->add($arg) or croak $!;
1596 } else {
1597 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1598 }
1599 return $md5->hexdigest();
1600}
1601sub gc_directory {
1603 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1604 my $out_filename = $_ . ".gz";
1605 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1606 binmode $in_fh;
1607 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1608 die "Unable to open $out_filename: $!\n";
1609 my $res;
1611 while ($res = sysread($in_fh, my $str, 1024)) {
1612 $gz->gzwrite($str) or
1613 die "Unable to write: ".$gz->gzerror()."!\n";
1614 }
1615 unlink $_ or die "unlink $File::Find::name: $!\n";
1616 } elsif (-f $_ && basename($_) eq "index") {
1617 unlink $_ or die "unlink $_: $!\n";
1618 }
1619}
1620package Git::SVN;
1622use strict;
1623use warnings;
1624use Fcntl qw/:DEFAULT :seek/;
1625use constant rev_map_fmt => 'NH40';
1626use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1627 $_repack $_repack_flags $_use_svm_props $_head
1628 $_use_svnsync_props $no_reuse_existing $_minimize_url
1629 $_use_log_author $_add_author_from $_localtime/;
1630use Carp qw/croak/;
1631use File::Path qw/mkpath/;
1632use File::Copy qw/copy/;
1633use IPC::Open3;
1634my ($_gc_nr, $_gc_period);
1636# properties that we do not log:
1638my %SKIP_PROP;
1639BEGIN {
1640 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1641 svn:special svn:executable
1642 svn:entry:committed-rev
1643 svn:entry:last-author
1644 svn:entry:uuid
1645 svn:entry:committed-date/;
1646 # some options are read globally, but can be overridden locally
1648 # per [svn-remote "..."] section. Command-line options will *NOT*
1649 # override options set in an [svn-remote "..."] section
1650 no strict 'refs';
1651 for my $option (qw/follow_parent no_metadata use_svm_props
1652 use_svnsync_props/) {
1653 my $key = $option;
1654 $key =~ tr/_//d;
1655 my $prop = "-$option";
1656 *$option = sub {
1657 my ($self) = @_;
1658 return $self->{$prop} if exists $self->{$prop};
1659 my $k = "svn-remote.$self->{repo_id}.$key";
1660 eval { command_oneline(qw/config --get/, $k) };
1661 if ($@) {
1662 $self->{$prop} = ${"Git::SVN::_$option"};
1663 } else {
1664 my $v = command_oneline(qw/config --bool/,$k);
1665 $self->{$prop} = $v eq 'false' ? 0 : 1;
1666 }
1667 return $self->{$prop};
1668 }
1669 }
1670}
1671my (%LOCKFILES, %INDEX_FILES);
1674END {
1675 unlink keys %LOCKFILES if %LOCKFILES;
1676 unlink keys %INDEX_FILES if %INDEX_FILES;
1677}
1678sub resolve_local_globs {
1680 my ($url, $fetch, $glob_spec) = @_;
1681 return unless defined $glob_spec;
1682 my $ref = $glob_spec->{ref};
1683 my $path = $glob_spec->{path};
1684 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1685 next unless m#^$ref->{regex}$#;
1686 my $p = $1;
1687 my $pathname = desanitize_refname($path->full_path($p));
1688 my $refname = desanitize_refname($ref->full_path($p));
1689 if (my $existing = $fetch->{$pathname}) {
1690 if ($existing ne $refname) {
1691 die "Refspec conflict:\n",
1692 "existing: $existing\n",
1693 " globbed: $refname\n";
1694 }
1695 my $u = (::cmt_metadata("$refname"))[0];
1696 $u =~ s!^\Q$url\E(/|$)!! or die
1697 "$refname: '$url' not found in '$u'\n";
1698 if ($pathname ne $u) {
1699 warn "W: Refspec glob conflict ",
1700 "(ref: $refname):\n",
1701 "expected path: $pathname\n",
1702 " real path: $u\n",
1703 "Continuing ahead with $u\n";
1704 next;
1705 }
1706 } else {
1707 $fetch->{$pathname} = $refname;
1708 }
1709 }
1710}
1711sub parse_revision_argument {
1713 my ($base, $head) = @_;
1714 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1715 return ($base, $head);
1716 }
1717 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1718 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1719 return ($head, $head) if ($::_revision eq 'HEAD');
1720 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1721 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1722 die "revision argument: $::_revision not understood by git-svn\n";
1723}
1724sub fetch_all {
1726 my ($repo_id, $remotes) = @_;
1727 if (ref $repo_id) {
1728 my $gs = $repo_id;
1729 $repo_id = undef;
1730 $repo_id = $gs->{repo_id};
1731 }
1732 $remotes ||= read_all_remotes();
1733 my $remote = $remotes->{$repo_id} or
1734 die "[svn-remote \"$repo_id\"] unknown\n";
1735 my $fetch = $remote->{fetch};
1736 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1737 my (@gs, @globs);
1738 my $ra = Git::SVN::Ra->new($url);
1739 my $uuid = $ra->get_uuid;
1740 my $head = $ra->get_latest_revnum;
1741 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1742 my $base = defined $fetch ? $head : 0;
1743 # read the max revs for wildcard expansion (branches/*, tags/*)
1745 foreach my $t (qw/branches tags/) {
1746 defined $remote->{$t} or next;
1747 push @globs, @{$remote->{$t}};
1748 my $max_rev = eval { tmp_config(qw/--int --get/,
1750 "svn-remote.$repo_id.${t}-maxRev") };
1751 if (defined $max_rev && ($max_rev < $base)) {
1752 $base = $max_rev;
1753 } elsif (!defined $max_rev) {
1754 $base = 0;
1755 }
1756 }
1757 if ($fetch) {
1759 foreach my $p (sort keys %$fetch) {
1760 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1761 my $lr = $gs->rev_map_max;
1762 if (defined $lr) {
1763 $base = $lr if ($lr < $base);
1764 }
1765 push @gs, $gs;
1766 }
1767 }
1768 ($base, $head) = parse_revision_argument($base, $head);
1770 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1771}
1772sub read_all_remotes {
1774 my $r = {};
1775 my $use_svm_props = eval { command_oneline(qw/config --bool
1776 svn.useSvmProps/) };
1777 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1778 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1779 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1780 if (m!^(.+)\.fetch=$svn_refspec$!) {
1781 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1782 die("svn-remote.$remote: remote ref '$remote_ref' "
1783 . "must start with 'refs/'\n")
1784 unless $remote_ref =~ m{^refs/};
1785 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1786 $r->{$remote}->{svm} = {} if $use_svm_props;
1787 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1788 $r->{$1}->{svm} = {};
1789 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1790 $r->{$1}->{url} = $2;
1791 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1792 my ($remote, $t, $local_ref, $remote_ref) =
1793 ($1, $2, $3, $4);
1794 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1795 . "must start with 'refs/'\n")
1796 unless $remote_ref =~ m{^refs/};
1797 my $rs = {
1798 t => $t,
1799 remote => $remote,
1800 path => Git::SVN::GlobSpec->new($local_ref),
1801 ref => Git::SVN::GlobSpec->new($remote_ref) };
1802 if (length($rs->{ref}->{right}) != 0) {
1803 die "The '*' glob character must be the last ",
1804 "character of '$remote_ref'\n";
1805 }
1806 push @{ $r->{$remote}->{$t} }, $rs;
1807 }
1808 }
1809 map {
1811 if (defined $r->{$_}->{svm}) {
1812 my $svm;
1813 eval {
1814 my $section = "svn-remote.$_";
1815 $svm = {
1816 source => tmp_config('--get',
1817 "$section.svm-source"),
1818 replace => tmp_config('--get',
1819 "$section.svm-replace"),
1820 }
1821 };
1822 $r->{$_}->{svm} = $svm;
1823 }
1824 } keys %$r;
1825 $r;
1827}
1828sub init_vars {
1830 $_gc_nr = $_gc_period = 1000;
1831 if (defined $_repack || defined $_repack_flags) {
1832 warn "Repack options are obsolete; they have no effect.\n";
1833 }
1834}
1835sub verify_remotes_sanity {
1837 return unless -d $ENV{GIT_DIR};
1838 my %seen;
1839 foreach (command(qw/config -l/)) {
1840 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1841 if ($seen{$1}) {
1842 die "Remote ref refs/remote/$1 is tracked by",
1843 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1844 "Please resolve this ambiguity in ",
1845 "your git configuration file before ",
1846 "continuing\n";
1847 }
1848 $seen{$1} = $_;
1849 }
1850 }
1851}
1852sub find_existing_remote {
1854 my ($url, $remotes) = @_;
1855 return undef if $no_reuse_existing;
1856 my $existing;
1857 foreach my $repo_id (keys %$remotes) {
1858 my $u = $remotes->{$repo_id}->{url} or next;
1859 next if $u ne $url;
1860 $existing = $repo_id;
1861 last;
1862 }
1863 $existing;
1864}
1865sub init_remote_config {
1867 my ($self, $url, $no_write) = @_;
1868 $url =~ s!/+$!!; # strip trailing slash
1869 my $r = read_all_remotes();
1870 my $existing = find_existing_remote($url, $r);
1871 if ($existing) {
1872 unless ($no_write) {
1873 print STDERR "Using existing ",
1874 "[svn-remote \"$existing\"]\n";
1875 }
1876 $self->{repo_id} = $existing;
1877 } elsif ($_minimize_url) {
1878 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1879 $existing = find_existing_remote($min_url, $r);
1880 if ($existing) {
1881 unless ($no_write) {
1882 print STDERR "Using existing ",
1883 "[svn-remote \"$existing\"]\n";
1884 }
1885 $self->{repo_id} = $existing;
1886 }
1887 if ($min_url ne $url) {
1888 unless ($no_write) {
1889 print STDERR "Using higher level of URL: ",
1890 "$url => $min_url\n";
1891 }
1892 my $old_path = $self->{path};
1893 $self->{path} = $url;
1894 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1895 if (length $old_path) {
1896 $self->{path} .= "/$old_path";
1897 }
1898 $url = $min_url;
1899 }
1900 }
1901 my $orig_url;
1902 if (!$existing) {
1903 # verify that we aren't overwriting anything:
1904 $orig_url = eval {
1905 command_oneline('config', '--get',
1906 "svn-remote.$self->{repo_id}.url")
1907 };
1908 if ($orig_url && ($orig_url ne $url)) {
1909 die "svn-remote.$self->{repo_id}.url already set: ",
1910 "$orig_url\nwanted to set to: $url\n";
1911 }
1912 }
1913 my ($xrepo_id, $xpath) = find_ref($self->refname);
1914 if (!$no_write && defined $xpath) {
1915 die "svn-remote.$xrepo_id.fetch already set to track ",
1916 "$xpath:", $self->refname, "\n";
1917 }
1918 unless ($no_write) {
1919 command_noisy('config',
1920 "svn-remote.$self->{repo_id}.url", $url);
1921 $self->{path} =~ s{^/}{};
1922 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1923 command_noisy('config', '--add',
1924 "svn-remote.$self->{repo_id}.fetch",
1925 "$self->{path}:".$self->refname);
1926 }
1927 $self->{url} = $url;
1928}
1929sub find_by_url { # repos_root and, path are optional
1931 my ($class, $full_url, $repos_root, $path) = @_;
1932 return undef unless defined $full_url;
1934 remove_username($full_url);
1935 remove_username($repos_root) if defined $repos_root;
1936 my $remotes = read_all_remotes();
1937 if (defined $full_url && defined $repos_root && !defined $path) {
1938 $path = $full_url;
1939 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1940 }
1941 foreach my $repo_id (keys %$remotes) {
1942 my $u = $remotes->{$repo_id}->{url} or next;
1943 remove_username($u);
1944 next if defined $repos_root && $repos_root ne $u;
1945 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1947 foreach my $t (qw/branches tags/) {
1948 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1949 resolve_local_globs($u, $fetch, $globspec);
1950 }
1951 }
1952 my $p = $path;
1953 my $rwr = rewrite_root({repo_id => $repo_id});
1954 my $svm = $remotes->{$repo_id}->{svm}
1955 if defined $remotes->{$repo_id}->{svm};
1956 unless (defined $p) {
1957 $p = $full_url;
1958 my $z = $u;
1959 my $prefix = '';
1960 if ($rwr) {
1961 $z = $rwr;
1962 remove_username($z);
1963 } elsif (defined $svm) {
1964 $z = $svm->{source};
1965 $prefix = $svm->{replace};
1966 $prefix =~ s#^\Q$u\E(?:/|$)##;
1967 $prefix =~ s#/$##;
1968 }
1969 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1970 }
1971 foreach my $f (keys %$fetch) {
1972 next if $f ne $p;
1973 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1974 }
1975 }
1976 undef;
1977}
1978sub init {
1980 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1981 my $self = _new($class, $repo_id, $ref_id, $path);
1982 if (defined $url) {
1983 $self->init_remote_config($url, $no_write);
1984 }
1985 $self;
1986}
1987sub find_ref {
1989 my ($ref_id) = @_;
1990 foreach (command(qw/config -l/)) {
1991 next unless m!^svn-remote\.(.+)\.fetch=
1992 \s*(.*?)\s*:\s*(.+?)\s*$!x;
1993 my ($repo_id, $path, $ref) = ($1, $2, $3);
1994 if ($ref eq $ref_id) {
1995 $path = '' if ($path =~ m#^\./?#);
1996 return ($repo_id, $path);
1997 }
1998 }
1999 (undef, undef, undef);
2000}
2001sub new {
2003 my ($class, $ref_id, $repo_id, $path) = @_;
2004 if (defined $ref_id && !defined $repo_id && !defined $path) {
2005 ($repo_id, $path) = find_ref($ref_id);
2006 if (!defined $repo_id) {
2007 die "Could not find a \"svn-remote.*.fetch\" key ",
2008 "in the repository configuration matching: ",
2009 "$ref_id\n";
2010 }
2011 }
2012 my $self = _new($class, $repo_id, $ref_id, $path);
2013 if (!defined $self->{path} || !length $self->{path}) {
2014 my $fetch = command_oneline('config', '--get',
2015 "svn-remote.$repo_id.fetch",
2016 ":$ref_id\$") or
2017 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2018 "\":$ref_id\$\" in config\n";
2019 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2020 }
2021 $self->{url} = command_oneline('config', '--get',
2022 "svn-remote.$repo_id.url") or
2023 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2024 $self->rebuild;
2025 $self;
2026}
2027sub refname {
2029 my ($refname) = $_[0]->{ref_id} ;
2030 # It cannot end with a slash /, we'll throw up on this because
2032 # SVN can't have directories with a slash in their name, either:
2033 if ($refname =~ m{/$}) {
2034 die "ref: '$refname' ends with a trailing slash, this is ",
2035 "not permitted by git nor Subversion\n";
2036 }
2037 # It cannot have ASCII control character space, tilde ~, caret ^,
2039 # colon :, question-mark ?, asterisk *, space, or open bracket [
2040 # anywhere.
2041 #
2042 # Additionally, % must be escaped because it is used for escaping
2043 # and we want our escaped refname to be reversible
2044 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2045 # no slash-separated component can begin with a dot .
2047 # /.* becomes /%2E*
2048 $refname =~ s{/\.}{/%2E}g;
2049 # It cannot have two consecutive dots .. anywhere
2051 # .. becomes %2E%2E
2052 $refname =~ s{\.\.}{%2E%2E}g;
2053 return $refname;
2055}
2056sub desanitize_refname {
2058 my ($refname) = @_;
2059 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2060 return $refname;
2061}
2062sub svm_uuid {
2064 my ($self) = @_;
2065 return $self->{svm}->{uuid} if $self->svm;
2066 $self->ra;
2067 unless ($self->{svm}) {
2068 die "SVM UUID not cached, and reading remotely failed\n";
2069 }
2070 $self->{svm}->{uuid};
2071}
2072sub svm {
2074 my ($self) = @_;
2075 return $self->{svm} if $self->{svm};
2076 my $svm;
2077 # see if we have it in our config, first:
2078 eval {
2079 my $section = "svn-remote.$self->{repo_id}";
2080 $svm = {
2081 source => tmp_config('--get', "$section.svm-source"),
2082 uuid => tmp_config('--get', "$section.svm-uuid"),
2083 replace => tmp_config('--get', "$section.svm-replace"),
2084 }
2085 };
2086 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2087 $self->{svm} = $svm;
2088 }
2089 $self->{svm};
2090}
2091sub _set_svm_vars {
2093 my ($self, $ra) = @_;
2094 return $ra if $self->svm;
2095 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2097 "(svm:source, svm:uuid) ",
2098 "from the following URLs:\n" );
2099 sub read_svm_props {
2100 my ($self, $ra, $path, $r) = @_;
2101 my $props = ($ra->get_dir($path, $r))[2];
2102 my $src = $props->{'svm:source'};
2103 my $uuid = $props->{'svm:uuid'};
2104 return undef if (!$src || !$uuid);
2105 chomp($src, $uuid);
2107 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2109 or die "doesn't look right - svm:uuid is '$uuid'\n";
2110 # the '!' is used to mark the repos_root!/relative/path
2112 $src =~ s{/?!/?}{/};
2113 $src =~ s{/+$}{}; # no trailing slashes please
2114 # username is of no interest
2115 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2116 my $replace = $ra->{url};
2118 $replace .= "/$path" if length $path;
2119 my $section = "svn-remote.$self->{repo_id}";
2121 tmp_config("$section.svm-source", $src);
2122 tmp_config("$section.svm-replace", $replace);
2123 tmp_config("$section.svm-uuid", $uuid);
2124 $self->{svm} = {
2125 source => $src,
2126 uuid => $uuid,
2127 replace => $replace
2128 };
2129 }
2130 my $r = $ra->get_latest_revnum;
2132 my $path = $self->{path};
2133 my %tried;
2134 while (length $path) {
2135 unless ($tried{"$self->{url}/$path"}) {
2136 return $ra if $self->read_svm_props($ra, $path, $r);
2137 $tried{"$self->{url}/$path"} = 1;
2138 }
2139 $path =~ s#/?[^/]+$##;
2140 }
2141 die "Path: '$path' should be ''\n" if $path ne '';
2142 return $ra if $self->read_svm_props($ra, $path, $r);
2143 $tried{"$self->{url}/$path"} = 1;
2144 if ($ra->{repos_root} eq $self->{url}) {
2146 die @err, (map { " $_\n" } keys %tried), "\n";
2147 }
2148 # nope, make sure we're connected to the repository root:
2150 my $ok;
2151 my @tried_b;
2152 $path = $ra->{svn_path};
2153 $ra = Git::SVN::Ra->new($ra->{repos_root});
2154 while (length $path) {
2155 unless ($tried{"$ra->{url}/$path"}) {
2156 $ok = $self->read_svm_props($ra, $path, $r);
2157 last if $ok;
2158 $tried{"$ra->{url}/$path"} = 1;
2159 }
2160 $path =~ s#/?[^/]+$##;
2161 }
2162 die "Path: '$path' should be ''\n" if $path ne '';
2163 $ok ||= $self->read_svm_props($ra, $path, $r);
2164 $tried{"$ra->{url}/$path"} = 1;
2165 if (!$ok) {
2166 die @err, (map { " $_\n" } keys %tried), "\n";
2167 }
2168 Git::SVN::Ra->new($self->{url});
2169}
2170sub svnsync {
2172 my ($self) = @_;
2173 return $self->{svnsync} if $self->{svnsync};
2174 if ($self->no_metadata) {
2176 die "Can't have both 'noMetadata' and ",
2177 "'useSvnsyncProps' options set!\n";
2178 }
2179 if ($self->rewrite_root) {
2180 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2181 "options set!\n";
2182 }
2183 my $svnsync;
2185 # see if we have it in our config, first:
2186 eval {
2187 my $section = "svn-remote.$self->{repo_id}";
2188 my $url = tmp_config('--get', "$section.svnsync-url");
2190 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2191 die "doesn't look right - svn:sync-from-url is '$url'\n";
2192 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2194 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2195 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2196 $svnsync = { url => $url, uuid => $uuid }
2198 };
2199 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2200 return $self->{svnsync} = $svnsync;
2201 }
2202 my $err = "useSvnsyncProps set, but failed to read " .
2204 "svnsync property: svn:sync-from-";
2205 my $rp = $self->ra->rev_proplist(0);
2206 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2208 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2209 die "doesn't look right - svn:sync-from-url is '$url'\n";
2210 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2212 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2213 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2214 my $section = "svn-remote.$self->{repo_id}";
2216 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2217 tmp_config('--add', "$section.svnsync-url", $url);
2218 return $self->{svnsync} = { url => $url, uuid => $uuid };
2219}
2220# this allows us to memoize our SVN::Ra UUID locally and avoid a
2222# remote lookup (useful for 'git svn log').
2223sub ra_uuid {
2224 my ($self) = @_;
2225 unless ($self->{ra_uuid}) {
2226 my $key = "svn-remote.$self->{repo_id}.uuid";
2227 my $uuid = eval { tmp_config('--get', $key) };
2228 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2229 $self->{ra_uuid} = $uuid;
2230 } else {
2231 die "ra_uuid called without URL\n" unless $self->{url};
2232 $self->{ra_uuid} = $self->ra->get_uuid;
2233 tmp_config('--add', $key, $self->{ra_uuid});
2234 }
2235 }
2236 $self->{ra_uuid};
2237}
2238sub _set_repos_root {
2240 my ($self, $repos_root) = @_;
2241 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2242 $repos_root ||= $self->ra->{repos_root};
2243 tmp_config($k, $repos_root);
2244 $repos_root;
2245}
2246sub repos_root {
2248 my ($self) = @_;
2249 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2250 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2251}
2252sub ra {
2254 my ($self) = shift;
2255 my $ra = Git::SVN::Ra->new($self->{url});
2256 $self->_set_repos_root($ra->{repos_root});
2257 if ($self->use_svm_props && !$self->{svm}) {
2258 if ($self->no_metadata) {
2259 die "Can't have both 'noMetadata' and ",
2260 "'useSvmProps' options set!\n";
2261 } elsif ($self->use_svnsync_props) {
2262 die "Can't have both 'useSvnsyncProps' and ",
2263 "'useSvmProps' options set!\n";
2264 }
2265 $ra = $self->_set_svm_vars($ra);
2266 $self->{-want_revprops} = 1;
2267 }
2268 $ra;
2269}
2270# prop_walk(PATH, REV, SUB)
2272# -------------------------
2273# Recursively traverse PATH at revision REV and invoke SUB for each
2274# directory that contains a SVN property. SUB will be invoked as
2275# follows: &SUB(gs, path, props); where `gs' is this instance of
2276# Git::SVN, `path' the path to the directory where the properties
2277# `props' were found. The `path' will be relative to point of checkout,
2278# that is, if url://repo/trunk is the current Git branch, and that
2279# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2280# as `path' (note the trailing `/').
2281sub prop_walk {
2282 my ($self, $path, $rev, $sub) = @_;
2283 $path =~ s#^/##;
2285 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2286 $path =~ s#^/*#/#g;
2287 my $p = $path;
2288 # Strip the irrelevant part of the path.
2289 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2290 # Ensure the path is terminated by a `/'.
2291 $p =~ s#/*$#/#;
2292 # The properties contain all the internal SVN stuff nobody
2294 # (usually) cares about.
2295 my $interesting_props = 0;
2296 foreach (keys %{$props}) {
2297 # If it doesn't start with `svn:', it must be a
2298 # user-defined property.
2299 ++$interesting_props and next if $_ !~ /^svn:/;
2300 # FIXME: Fragile, if SVN adds new public properties,
2301 # this needs to be updated.
2302 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2303 |eol-style|mime-type
2304 |externals|needs-lock)$/x;
2305 }
2306 &$sub($self, $p, $props) if $interesting_props;
2307 foreach (sort keys %$dirent) {
2309 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2310 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2311 }
2312}
2313sub last_rev { ($_[0]->last_rev_commit)[0] }
2315sub last_commit { ($_[0]->last_rev_commit)[1] }
2316# returns the newest SVN revision number and newest commit SHA1
2318sub last_rev_commit {
2319 my ($self) = @_;
2320 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2321 return ($self->{last_rev}, $self->{last_commit});
2322 }
2323 my $c = ::verify_ref($self->refname.'^0');
2324 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2325 my $rev = (::cmt_metadata($c))[1];
2326 if (defined $rev) {
2327 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2328 return ($rev, $c);
2329 }
2330 }
2331 my $map_path = $self->map_path;
2332 unless (-e $map_path) {
2333 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2334 return (undef, undef);
2335 }
2336 my ($rev, $commit) = $self->rev_map_max(1);
2337 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2338 return ($rev, $commit);
2339}
2340sub get_fetch_range {
2342 my ($self, $min, $max) = @_;
2343 $max ||= $self->ra->get_latest_revnum;
2344 $min ||= $self->rev_map_max;
2345 (++$min, $max);
2346}
2347sub tmp_config {
2349 my (@args) = @_;
2350 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2351 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2352 if (! -f $config && -f $old_def_config) {
2353 rename $old_def_config, $config or
2354 die "Failed rename $old_def_config => $config: $!\n";
2355 }
2356 my $old_config = $ENV{GIT_CONFIG};
2357 $ENV{GIT_CONFIG} = $config;
2358 $@ = undef;
2359 my @ret = eval {
2360 unless (-f $config) {
2361 mkfile($config);
2362 open my $fh, '>', $config or
2363 die "Can't open $config: $!\n";
2364 print $fh "; This file is used internally by ",
2365 "git-svn\n" or die
2366 "Couldn't write to $config: $!\n";
2367 print $fh "; You should not have to edit it\n" or
2368 die "Couldn't write to $config: $!\n";
2369 close $fh or die "Couldn't close $config: $!\n";
2370 }
2371 command('config', @args);
2372 };
2373 my $err = $@;
2374 if (defined $old_config) {
2375 $ENV{GIT_CONFIG} = $old_config;
2376 } else {
2377 delete $ENV{GIT_CONFIG};
2378 }
2379 die $err if $err;
2380 wantarray ? @ret : $ret[0];
2381}
2382sub tmp_index_do {
2384 my ($self, $sub) = @_;
2385 my $old_index = $ENV{GIT_INDEX_FILE};
2386 $ENV{GIT_INDEX_FILE} = $self->{index};
2387 $@ = undef;
2388 my @ret = eval {
2389 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2390 mkpath([$dir]) unless -d $dir;
2391 &$sub;
2392 };
2393 my $err = $@;
2394 if (defined $old_index) {
2395 $ENV{GIT_INDEX_FILE} = $old_index;
2396 } else {
2397 delete $ENV{GIT_INDEX_FILE};
2398 }
2399 die $err if $err;
2400 wantarray ? @ret : $ret[0];
2401}
2402sub assert_index_clean {
2404 my ($self, $treeish) = @_;
2405 $self->tmp_index_do(sub {
2407 command_noisy('read-tree', $treeish) unless -e $self->{index};
2408 my $x = command_oneline('write-tree');
2409 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2410 /^tree ($::sha1)/mo);
2411 return if $y eq $x;
2412 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2414 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2415 command_noisy('read-tree', $treeish);
2416 $x = command_oneline('write-tree');
2417 if ($y ne $x) {
2418 ::fatal "trees ($treeish) $y != $x\n",
2419 "Something is seriously wrong...";
2420 }
2421 });
2422}
2423sub get_commit_parents {
2425 my ($self, $log_entry) = @_;
2426 my (%seen, @ret, @tmp);
2427 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2428 if (my $ip = $self->{inject_parents}) {
2429 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2430 push @tmp, $commit;
2431 }
2432 }
2433 if (my $cur = ::verify_ref($self->refname.'^0')) {
2434 push @tmp, $cur;
2435 }
2436 if (my $ipd = $self->{inject_parents_dcommit}) {
2437 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2438 push @tmp, @$commit;
2439 }
2440 }
2441 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2442 while (my $p = shift @tmp) {
2443 next if $seen{$p};
2444 $seen{$p} = 1;
2445 push @ret, $p;
2446 # MAXPARENT is defined to 16 in commit-tree.c:
2447 last if @ret >= 16;
2448 }
2449 if (@tmp) {
2450 die "r$log_entry->{revision}: No room for parents:\n\t",
2451 join("\n\t", @tmp), "\n";
2452 }
2453 @ret;
2454}
2455sub rewrite_root {
2457 my ($self) = @_;
2458 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2459 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2460 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2461 if ($rwr) {
2462 $rwr =~ s#/+$##;
2463 if ($rwr !~ m#^[a-z\+]+://#) {
2464 die "$rwr is not a valid URL (key: $k)\n";
2465 }
2466 }
2467 $self->{-rewrite_root} = $rwr;
2468}
2469sub metadata_url {
2471 my ($self) = @_;
2472 ($self->rewrite_root || $self->{url}) .
2473 (length $self->{path} ? '/' . $self->{path} : '');
2474}
2475sub full_url {
2477 my ($self) = @_;
2478 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2479}
2480sub set_commit_header_env {
2483 my ($log_entry) = @_;
2484 my %env;
2485 foreach my $ned (qw/NAME EMAIL DATE/) {
2486 foreach my $ac (qw/AUTHOR COMMITTER/) {
2487 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2488 }
2489 }
2490 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2492 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2493 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2494 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2496 ? $log_entry->{commit_name}
2497 : $log_entry->{name};
2498 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2499 ? $log_entry->{commit_email}
2500 : $log_entry->{email};
2501 \%env;
2502}
2503sub restore_commit_header_env {
2505 my ($env) = @_;
2506 foreach my $ned (qw/NAME EMAIL DATE/) {
2507 foreach my $ac (qw/AUTHOR COMMITTER/) {
2508 my $k = "GIT_${ac}_${ned}";
2509 if (defined $env->{$k}) {
2510 $ENV{$k} = $env->{$k};
2511 } else {
2512 delete $ENV{$k};
2513 }
2514 }
2515 }
2516}
2517sub gc {
2519 command_noisy('gc', '--auto');
2520};
2521sub do_git_commit {
2523 my ($self, $log_entry) = @_;
2524 my $lr = $self->last_rev;
2525 if (defined $lr && $lr >= $log_entry->{revision}) {
2526 die "Last fetched revision of ", $self->refname,
2527 " was r$lr, but we are about to fetch: ",
2528 "r$log_entry->{revision}!\n";
2529 }
2530 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2531 croak "$log_entry->{revision} = $c already exists! ",
2532 "Why are we refetching it?\n";
2533 }
2534 my $old_env = set_commit_header_env($log_entry);
2535 my $tree = $log_entry->{tree};
2536 if (!defined $tree) {
2537 $tree = $self->tmp_index_do(sub {
2538 command_oneline('write-tree') });
2539 }
2540 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2541 my @exec = ('git', 'commit-tree', $tree);
2543 foreach ($self->get_commit_parents($log_entry)) {
2544 push @exec, '-p', $_;
2545 }
2546 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2547 or croak $!;
2548 binmode $msg_fh;
2549 # we always get UTF-8 from SVN, but we may want our commits in
2551 # a different encoding.
2552 if (my $enc = Git::config('i18n.commitencoding')) {
2553 require Encode;
2554 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2555 }
2556 print $msg_fh $log_entry->{log} or croak $!;
2557 restore_commit_header_env($old_env);
2558 unless ($self->no_metadata) {
2559 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2560 or croak $!;
2561 }
2562 $msg_fh->flush == 0 or croak $!;
2563 close $msg_fh or croak $!;
2564 chomp(my $commit = do { local $/; <$out_fh> });
2565 close $out_fh or croak $!;
2566 waitpid $pid, 0;
2567 croak $? if $?;
2568 if ($commit !~ /^$::sha1$/o) {
2569 die "Failed to commit, invalid sha1: $commit\n";
2570 }
2571 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2573 $self->{last_rev} = $log_entry->{revision};
2575 $self->{last_commit} = $commit;
2576 print "r$log_entry->{revision}" unless $::_q > 1;
2577 if (defined $log_entry->{svm_revision}) {
2578 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2579 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2580 0, $self->svm_uuid);
2581 }
2582 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2583 if (--$_gc_nr == 0) {
2584 $_gc_nr = $_gc_period;
2585 gc();
2586 }
2587 return $commit;
2588}
2589sub match_paths {
2591 my ($self, $paths, $r) = @_;
2592 return 1 if $self->{path} eq '';
2593 if (my $path = $paths->{"/$self->{path}"}) {
2594 return ($path->{action} eq 'D') ? 0 : 1;
2595 }
2596 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2597 if (grep /$self->{path_regex}/, keys %$paths) {
2598 return 1;
2599 }
2600 my $c = '';
2601 foreach (split m#/#, $self->{path}) {
2602 $c .= "/$_";
2603 next unless ($paths->{$c} &&
2604 ($paths->{$c}->{action} =~ /^[AR]$/));
2605 if ($self->ra->check_path($self->{path}, $r) ==
2606 $SVN::Node::dir) {
2607 return 1;
2608 }
2609 }
2610 return 0;
2611}
2612sub find_parent_branch {
2614 my ($self, $paths, $rev) = @_;
2615 return undef unless $self->follow_parent;
2616 unless (defined $paths) {
2617 my $err_handler = $SVN::Error::handler;
2618 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2619 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2620 sub { $paths = $_[0] });
2621 $SVN::Error::handler = $err_handler;
2622 }
2623 return undef unless defined $paths;
2624 # look for a parent from another branch:
2626 my @b_path_components = split m#/#, $self->{path};
2627 my @a_path_components;
2628 my $i;
2629 while (@b_path_components) {
2630 $i = $paths->{'/'.join('/', @b_path_components)};
2631 last if $i && defined $i->{copyfrom_path};
2632 unshift(@a_path_components, pop(@b_path_components));
2633 }
2634 return undef unless defined $i && defined $i->{copyfrom_path};
2635 my $branch_from = $i->{copyfrom_path};
2636 if (@a_path_components) {
2637 print STDERR "branch_from: $branch_from => ";
2638 $branch_from .= '/'.join('/', @a_path_components);
2639 print STDERR $branch_from, "\n";
2640 }
2641 my $r = $i->{copyfrom_rev};
2642 my $repos_root = $self->ra->{repos_root};
2643 my $url = $self->ra->{url};
2644 my $new_url = $url . $branch_from;
2645 print STDERR "Found possible branch point: ",
2646 "$new_url => ", $self->full_url, ", $r\n"
2647 unless $::_q > 1;
2648 $branch_from =~ s#^/##;
2649 my $gs = $self->other_gs($new_url, $url,
2650 $branch_from, $r, $self->{ref_id});
2651 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2652 {
2653 my ($base, $head);
2654 if (!defined $r0 || !defined $parent) {
2655 ($base, $head) = parse_revision_argument(0, $r);
2656 } else {
2657 if ($r0 < $r) {
2658 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2659 0, 1, sub { $base = $_[1] - 1 });
2660 }
2661 }
2662 if (defined $base && $base <= $r) {
2663 $gs->fetch($base, $r);
2664 }
2665 ($r0, $parent) = $gs->find_rev_before($r, 1);
2666 }
2667 if (defined $r0 && defined $parent) {
2668 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2669 unless $::_q > 1;
2670 my $ed;
2671 if ($self->ra->can_do_switch) {
2672 $self->assert_index_clean($parent);
2673 print STDERR "Following parent with do_switch\n"
2674 unless $::_q > 1;
2675 # do_switch works with svn/trunk >= r22312, but that
2676 # is not included with SVN 1.4.3 (the latest version
2677 # at the moment), so we can't rely on it
2678 $self->{last_rev} = $r0;
2679 $self->{last_commit} = $parent;
2680 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2681 $gs->ra->gs_do_switch($r0, $rev, $gs,
2682 $self->full_url, $ed)
2683 or die "SVN connection failed somewhere...\n";
2684 } elsif ($self->ra->trees_match($new_url, $r0,
2685 $self->full_url, $rev)) {
2686 print STDERR "Trees match:\n",
2687 " $new_url\@$r0\n",
2688 " ${\$self->full_url}\@$rev\n",
2689 "Following parent with no changes\n"
2690 unless $::_q > 1;
2691 $self->tmp_index_do(sub {
2692 command_noisy('read-tree', $parent);
2693 });
2694 $self->{last_commit} = $parent;
2695 } else {
2696 print STDERR "Following parent with do_update\n"
2697 unless $::_q > 1;
2698 $ed = SVN::Git::Fetcher->new($self);
2699 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2700 or die "SVN connection failed somewhere...\n";
2701 }
2702 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2703 return $self->make_log_entry($rev, [$parent], $ed);
2704 }
2705 return undef;
2706}
2707sub do_fetch {
2709 my ($self, $paths, $rev) = @_;
2710 my $ed;
2711 my ($last_rev, @parents);
2712 if (my $lc = $self->last_commit) {
2713 # we can have a branch that was deleted, then re-added
2714 # under the same name but copied from another path, in
2715 # which case we'll have multiple parents (we don't
2716 # want to break the original ref, nor lose copypath info):
2717 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2718 push @{$log_entry->{parents}}, $lc;
2719 return $log_entry;
2720 }
2721 $ed = SVN::Git::Fetcher->new($self);
2722 $last_rev = $self->{last_rev};
2723 $ed->{c} = $lc;
2724 @parents = ($lc);
2725 } else {
2726 $last_rev = $rev;
2727 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2728 return $log_entry;
2729 }
2730 $ed = SVN::Git::Fetcher->new($self);
2731 }
2732 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2733 die "SVN connection failed somewhere...\n";
2734 }
2735 $self->make_log_entry($rev, \@parents, $ed);
2736}
2737sub mkemptydirs {
2739 my ($self, $r) = @_;
2740 my %empty_dirs = ();
2741 open my $fh, '<', "$self->{dir}/unhandled.log" or return;
2743 binmode $fh or croak "binmode: $!";
2744 while (<$fh>) {
2745 if (defined $r && /^r(\d+)$/) {
2746 last if $1 > $r;
2747 } elsif (/^ \+empty_dir: (.+)$/) {
2748 $empty_dirs{$1} = 1;
2749 } elsif (/^ \-empty_dir: (.+)$/) {
2750 delete $empty_dirs{$1};
2751 }
2752 }
2753 close $fh;
2754 foreach my $d (sort keys %empty_dirs) {
2755 $d = uri_decode($d);
2756 next if -d $d;
2757 if (-e _) {
2758 warn "$d exists but is not a directory\n";
2759 } else {
2760 print "creating empty directory: $d\n";
2761 mkpath([$d]);
2762 }
2763 }
2764}
2765sub get_untracked {
2767 my ($self, $ed) = @_;
2768 my @out;
2769 my $h = $ed->{empty};
2770 foreach (sort keys %$h) {
2771 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2772 push @out, " $act: " . uri_encode($_);
2773 warn "W: $act: $_\n";
2774 }
2775 foreach my $t (qw/dir_prop file_prop/) {
2776 $h = $ed->{$t} or next;
2777 foreach my $path (sort keys %$h) {
2778 my $ppath = $path eq '' ? '.' : $path;
2779 foreach my $prop (sort keys %{$h->{$path}}) {
2780 next if $SKIP_PROP{$prop};
2781 my $v = $h->{$path}->{$prop};
2782 my $t_ppath_prop = "$t: " .
2783 uri_encode($ppath) . ' ' .
2784 uri_encode($prop);
2785 if (defined $v) {
2786 push @out, " +$t_ppath_prop " .
2787 uri_encode($v);
2788 } else {
2789 push @out, " -$t_ppath_prop";
2790 }
2791 }
2792 }
2793 }
2794 foreach my $t (qw/absent_file absent_directory/) {
2795 $h = $ed->{$t} or next;
2796 foreach my $parent (sort keys %$h) {
2797 foreach my $path (sort @{$h->{$parent}}) {
2798 push @out, " $t: " .
2799 uri_encode("$parent/$path");
2800 warn "W: $t: $parent/$path ",
2801 "Insufficient permissions?\n";
2802 }
2803 }
2804 }
2805 \@out;
2806}
2807# parse_svn_date(DATE)
2809# --------------------
2810# Given a date (in UTC) from Subversion, return a string in the format
2811# "<TZ Offset> <local date/time>" that Git will use.
2812#
2813# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2814# is true we'll convert it to the local timezone instead.
2815sub parse_svn_date {
2816 my $date = shift || return '+0000 1970-01-01 00:00:00';
2817 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2818 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2819 croak "Unable to parse date: $date\n";
2820 my $parsed_date; # Set next.
2821 if ($Git::SVN::_localtime) {
2823 # Translate the Subversion datetime to an epoch time.
2824 # Begin by switching ourselves to $date's timezone, UTC.
2825 my $old_env_TZ = $ENV{TZ};
2826 $ENV{TZ} = 'UTC';
2827 my $epoch_in_UTC =
2829 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2830 # Determine our local timezone (including DST) at the
2832 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2833 # value of TZ, if any, at the time we were run.
2834 if (defined $Git::SVN::Log::TZ) {
2835 $ENV{TZ} = $Git::SVN::Log::TZ;
2836 } else {
2837 delete $ENV{TZ};
2838 }
2839 my $our_TZ =
2841 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2842 # This converts $epoch_in_UTC into our local timezone.
2844 my ($sec, $min, $hour, $mday, $mon, $year,
2845 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2846 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2848 $our_TZ, $year + 1900, $mon + 1,
2849 $mday, $hour, $min, $sec);
2850 # Reset us to the timezone in effect when we entered
2852 # this routine.
2853 if (defined $old_env_TZ) {
2854 $ENV{TZ} = $old_env_TZ;
2855 } else {
2856 delete $ENV{TZ};
2857 }
2858 } else {
2859 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2860 }
2861 return $parsed_date;
2863}
2864sub other_gs {
2866 my ($self, $new_url, $url,
2867 $branch_from, $r, $old_ref_id) = @_;
2868 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2869 unless ($gs) {
2870 my $ref_id = $old_ref_id;
2871 $ref_id =~ s/\@\d+$//;
2872 $ref_id .= "\@$r";
2873 # just grow a tail if we're not unique enough :x
2874 $ref_id .= '-' while find_ref($ref_id);
2875 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2876 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2877 if ($u =~ s#^\Q$url\E(/|$)##) {
2878 $p = $u;
2879 $u = $url;
2880 $repo_id = $self->{repo_id};
2881 }
2882 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2883 }
2884 $gs
2885}
2886sub call_authors_prog {
2888 my ($orig_author) = @_;
2889 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2890 my $author = `$::_authors_prog $orig_author`;
2891 if ($? != 0) {
2892 die "$::_authors_prog failed with exit code $?\n"
2893 }
2894 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2895 my ($name, $email) = ($1, $2);
2896 $email = undef if length $2 == 0;
2897 return [$name, $email];
2898 } else {
2899 die "Author: $orig_author: $::_authors_prog returned "
2900 . "invalid author format: $author\n";
2901 }
2902}
2903sub check_author {
2905 my ($author) = @_;
2906 if (!defined $author || length $author == 0) {
2907 $author = '(no author)';
2908 }
2909 if (!defined $::users{$author}) {
2910 if (defined $::_authors_prog) {
2911 $::users{$author} = call_authors_prog($author);
2912 } elsif (defined $::_authors) {
2913 die "Author: $author not defined in $::_authors file\n";
2914 }
2915 }
2916 $author;
2917}
2918sub find_extra_svk_parents {
2920 my ($self, $ed, $tickets, $parents) = @_;
2921 # aha! svk:merge property changed...
2922 my @tickets = split "\n", $tickets;
2923 my @known_parents;
2924 for my $ticket ( @tickets ) {
2925 my ($uuid, $path, $rev) = split /:/, $ticket;
2926 if ( $uuid eq $self->ra_uuid ) {
2927 my $url = $self->rewrite_root || $self->{url};
2928 my $repos_root = $url;
2929 my $branch_from = $path;
2930 $branch_from =~ s{^/}{};
2931 my $gs = $self->other_gs($repos_root."/".$branch_from,
2932 $url,
2933 $branch_from,
2934 $rev,
2935 $self->{ref_id});
2936 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
2937 # wahey! we found it, but it might be
2938 # an old one (!)
2939 push @known_parents, $commit;
2940 }
2941 }
2942 }
2943 for my $parent ( @known_parents ) {
2944 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
2945 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2946 my $new;
2947 while ( <$msg_fh> ) {
2948 $new=1;last;
2949 }
2950 command_close_pipe($msg_fh, $ctx);
2951 if ( $new ) {
2952 print STDERR
2953 "Found merge parent (svk:merge ticket): $parent\n";
2954 push @$parents, $parent;
2955 }
2956 }
2957}
2958# note: this function should only be called if the various dirprops
2960# have actually changed
2961sub find_extra_svn_parents {
2962 my ($self, $ed, $mergeinfo, $parents) = @_;
2963 # aha! svk:merge property changed...
2964 # We first search for merged tips which are not in our
2966 # history. Then, we figure out which git revisions are in
2967 # that tip, but not this revision. If all of those revisions
2968 # are now marked as merge, we can add the tip as a parent.
2969 my @merges = split "\n", $mergeinfo;
2970 my @merge_tips;
2971 my @merged_commit_ranges;
2972 my $url = $self->rewrite_root || $self->{url};
2973 for my $merge ( @merges ) {
2974 my ($source, $revs) = split ":", $merge;
2975 my $path = $source;
2976 $path =~ s{^/}{};
2977 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
2978 if ( !$gs ) {
2979 warn "Couldn't find revmap for $url$source\n";
2980 next;
2981 }
2982 my @ranges = split ",", $revs;
2983 my ($tip, $tip_commit);
2984 # find the tip
2985 for my $range ( @ranges ) {
2986 my ($bottom, $top) = split "-", $range;
2987 $top ||= $bottom;
2988 my $bottom_commit =
2989 $gs->rev_map_get($bottom, $self->ra_uuid) ||
2990 $gs->rev_map_get($bottom+1, $self->ra_uuid);
2991 my $top_commit;
2992 for (; !$top_commit && $top >= $bottom; --$top) {
2993 $top_commit =
2994 $gs->rev_map_get($top, $self->ra_uuid);
2995 }
2996 unless ($top_commit and $bottom_commit) {
2998 warn "W:unknown path/rev in svn:mergeinfo "
2999 ."dirprop: $source:$range\n";
3000 next;
3001 }
3002 push @merged_commit_ranges,
3004 "$bottom_commit..$top_commit";
3005 if ( !defined $tip or $top > $tip ) {
3007 $tip = $top;
3008 $tip_commit = $top_commit;
3009 }
3010 }
3011 unless (!$tip_commit or
3012 grep { $_ eq $tip_commit } @$parents ) {
3013 push @merge_tips, $tip_commit;
3014 } else {
3015 push @merge_tips, undef;
3016 }
3017 }
3018 for my $merge_tip ( @merge_tips ) {
3019 my $spec = shift @merges;
3020 next unless $merge_tip;
3021 my @cmd = ('rev-list', "-1", $merge_tip,
3022 "--not", @$parents );
3023 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3024 my $new;
3025 while ( <$msg_fh> ) {
3026 $new=1;last;
3027 }
3028 command_close_pipe($msg_fh, $ctx);
3029 if ( $new ) {
3030 push @cmd, @merged_commit_ranges;
3031 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3032 my $unmerged;
3033 while ( <$msg_fh> ) {
3034 $unmerged=1;last;
3035 }
3036 command_close_pipe($msg_fh, $ctx);
3037 if ( $unmerged ) {
3038 warn "W:svn cherry-pick ignored ($spec)\n";
3039 } else {
3040 warn
3041 "Found merge parent (svn:mergeinfo prop): ",
3042 $merge_tip, "\n";
3043 push @$parents, $merge_tip;
3044 }
3045 }
3046 }
3047}
3048sub make_log_entry {
3050 my ($self, $rev, $parents, $ed) = @_;
3051 my $untracked = $self->get_untracked($ed);
3052 my @parents = @$parents;
3054 my $ps = $ed->{path_strip} || "";
3055 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3056 my $props = $ed->{dir_prop}{$path};
3057 if ( $props->{"svk:merge"} ) {
3058 $self->find_extra_svk_parents
3059 ($ed, $props->{"svk:merge"}, \@parents);
3060 }
3061 if ( $props->{"svn:mergeinfo"} ) {
3062 $self->find_extra_svn_parents
3063 ($ed,
3064 $props->{"svn:mergeinfo"},
3065 \@parents);
3066 }
3067 }
3068 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3070 print $un "r$rev\n" or croak $!;
3071 print $un $_, "\n" foreach @$untracked;
3072 my %log_entry = ( parents => \@parents, revision => $rev,
3073 log => '');
3074 my $headrev;
3076 my $logged = delete $self->{logged_rev_props};
3077 if (!$logged || $self->{-want_revprops}) {
3078 my $rp = $self->ra->rev_proplist($rev);
3079 foreach (sort keys %$rp) {
3080 my $v = $rp->{$_};
3081 if (/^svn:(author|date|log)$/) {
3082 $log_entry{$1} = $v;
3083 } elsif ($_ eq 'svm:headrev') {
3084 $headrev = $v;
3085 } else {
3086 print $un " rev_prop: ", uri_encode($_), ' ',
3087 uri_encode($v), "\n";
3088 }
3089 }
3090 } else {
3091 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3092 }
3093 close $un or croak $!;
3094 $log_entry{date} = parse_svn_date($log_entry{date});
3096 $log_entry{log} .= "\n";
3097 my $author = $log_entry{author} = check_author($log_entry{author});
3098 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3099 : ($author, undef);
3100 my ($commit_name, $commit_email) = ($name, $email);
3102 if ($_use_log_author) {
3103 my $name_field;
3104 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3105 $name_field = $1;
3106 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3107 $name_field = $1;
3108 }
3109 if (!defined $name_field) {
3110 if (!defined $email) {
3111 $email = $name;
3112 }
3113 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3114 ($name, $email) = ($1, $2);
3115 } elsif ($name_field =~ /(.*)@/) {
3116 ($name, $email) = ($1, $name_field);
3117 } else {
3118 ($name, $email) = ($name_field, $name_field);
3119 }
3120 }
3121 if (defined $headrev && $self->use_svm_props) {
3122 if ($self->rewrite_root) {
3123 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3124 "options set!\n";
3125 }
3126 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3127 # we don't want "SVM: initializing mirror for junk" ...
3128 return undef if $r == 0;
3129 my $svm = $self->svm;
3130 if ($uuid ne $svm->{uuid}) {
3131 die "UUID mismatch on SVM path:\n",
3132 "expected: $svm->{uuid}\n",
3133 " got: $uuid\n";
3134 }
3135 my $full_url = $self->full_url;
3136 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3137 die "Failed to replace '$svm->{replace}' with ",
3138 "'$svm->{source}' in $full_url\n";
3139 # throw away username for storing in records
3140 remove_username($full_url);
3141 $log_entry{metadata} = "$full_url\@$r $uuid";
3142 $log_entry{svm_revision} = $r;
3143 $email ||= "$author\@$uuid";
3144 $commit_email ||= "$author\@$uuid";
3145 } elsif ($self->use_svnsync_props) {
3146 my $full_url = $self->svnsync->{url};
3147 $full_url .= "/$self->{path}" if length $self->{path};
3148 remove_username($full_url);
3149 my $uuid = $self->svnsync->{uuid};
3150 $log_entry{metadata} = "$full_url\@$rev $uuid";
3151 $email ||= "$author\@$uuid";
3152 $commit_email ||= "$author\@$uuid";
3153 } else {
3154 my $url = $self->metadata_url;
3155 remove_username($url);
3156 $log_entry{metadata} = "$url\@$rev " .
3157 $self->ra->get_uuid;
3158 $email ||= "$author\@" . $self->ra->get_uuid;
3159 $commit_email ||= "$author\@" . $self->ra->get_uuid;
3160 }
3161 $log_entry{name} = $name;
3162 $log_entry{email} = $email;
3163 $log_entry{commit_name} = $commit_name;
3164 $log_entry{commit_email} = $commit_email;
3165 \%log_entry;
3166}
3167sub fetch {
3169 my ($self, $min_rev, $max_rev, @parents) = @_;
3170 my ($last_rev, $last_commit) = $self->last_rev_commit;
3171 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3172 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3173}
3174sub set_tree_cb {
3176 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3177 $self->{inject_parents} = { $rev => $tree };
3178 $self->fetch(undef, undef);
3179}
3180sub set_tree {
3182 my ($self, $tree) = (shift, shift);
3183 my $log_entry = ::get_commit_entry($tree);
3184 unless ($self->{last_rev}) {
3185 ::fatal("Must have an existing revision to commit");
3186 }
3187 my %ed_opts = ( r => $self->{last_rev},
3188 log => $log_entry->{log},
3189 ra => $self->ra,
3190 tree_a => $self->{last_commit},
3191 tree_b => $tree,
3192 editor_cb => sub {
3193 $self->set_tree_cb($log_entry, $tree, @_) },
3194 svn_path => $self->{path} );
3195 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3196 print "No changes\nr$self->{last_rev} = $tree\n";
3197 }
3198}
3199sub rebuild_from_rev_db {
3201 my ($self, $path) = @_;
3202 my $r = -1;
3203 open my $fh, '<', $path or croak "open: $!";
3204 binmode $fh or croak "binmode: $!";
3205 while (<$fh>) {
3206 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3207 chomp($_);
3208 ++$r;
3209 next if $_ eq ('0' x 40);
3210 $self->rev_map_set($r, $_);
3211 print "r$r = $_\n";
3212 }
3213 close $fh or croak "close: $!";
3214 unlink $path or croak "unlink: $!";
3215}
3216sub rebuild {
3218 my ($self) = @_;
3219 my $map_path = $self->map_path;
3220 my $partial = (-e $map_path && ! -z $map_path);
3221 return unless ::verify_ref($self->refname.'^0');
3222 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3223 my $rev_db = $self->rev_db_path;
3224 $self->rebuild_from_rev_db($rev_db);
3225 if ($self->use_svm_props) {
3226 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3227 $self->rebuild_from_rev_db($svm_rev_db);
3228 }
3229 $self->unlink_rev_db_symlink;
3230 return;
3231 }
3232 print "Rebuilding $map_path ...\n" if (!$partial);
3233 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3234 (undef, undef));
3235 my ($log, $ctx) =
3236 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3237 ($head ? "$head.." : "") . $self->refname,
3238 '--');
3239 my $metadata_url = $self->metadata_url;
3240 remove_username($metadata_url);
3241 my $svn_uuid = $self->ra_uuid;
3242 my $c;
3243 while (<$log>) {
3244 if ( m{^commit ($::sha1)$} ) {
3245 $c = $1;
3246 next;
3247 }
3248 next unless s{^\s*(git-svn-id:)}{$1};
3249 my ($url, $rev, $uuid) = ::extract_metadata($_);
3250 remove_username($url);
3251 # ignore merges (from set-tree)
3253 next if (!defined $rev || !$uuid);
3254 # if we merged or otherwise started elsewhere, this is
3256 # how we break out of it
3257 if (($uuid ne $svn_uuid) ||
3258 ($metadata_url && $url && ($url ne $metadata_url))) {
3259 next;
3260 }
3261 if ($partial && $head) {
3262 print "Partial-rebuilding $map_path ...\n";
3263 print "Currently at $base_rev = $head\n";
3264 $head = undef;
3265 }
3266 $self->rev_map_set($rev, $c);
3268 print "r$rev = $c\n";
3269 }
3270 command_close_pipe($log, $ctx);
3271 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3272 my $rev_db_path = $self->rev_db_path;
3273 if (-f $self->rev_db_path) {
3274 unlink $self->rev_db_path or croak "unlink: $!";
3275 }
3276 $self->unlink_rev_db_symlink;
3277}
3278# rev_map:
3280# Tie::File seems to be prone to offset errors if revisions get sparse,
3281# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3282# one of my favorite modules is out :< Next up would be one of the DBM
3283# modules, but I'm not sure which is most portable...
3284#
3285# This is the replacement for the rev_db format, which was too big
3286# and inefficient for large repositories with a lot of sparse history
3287# (mainly tags)
3288#
3289# The format is this:
3290# - 24 bytes for every record,
3291# * 4 bytes for the integer representing an SVN revision number
3292# * 20 bytes representing the sha1 of a git commit
3293# - No empty padding records like the old format
3294# (except the last record, which can be overwritten)
3295# - new records are written append-only since SVN revision numbers
3296# increase monotonically
3297# - lookups on SVN revision number are done via a binary search
3298# - Piping the file to xxd -c24 is a good way of dumping it for
3299# viewing or editing (piped back through xxd -r), should the need
3300# ever arise.
3301# - The last record can be padding revision with an all-zero sha1
3302# This is used to optimize fetch performance when using multiple
3303# "fetch" directives in .git/config
3304#
3305# These files are disposable unless noMetadata or useSvmProps is set
3306sub _rev_map_set {
3308 my ($fh, $rev, $commit) = @_;
3309 binmode $fh or croak "binmode: $!";
3311 my $size = (stat($fh))[7];
3312 ($size % 24) == 0 or croak "inconsistent size: $size";
3313 my $wr_offset = 0;
3315 if ($size > 0) {
3316 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3317 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3318 $read == 24 or croak "read only $read bytes (!= 24)";
3319 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3320 if ($last_commit eq ('0' x40)) {
3321 if ($size >= 48) {
3322 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3323 $read = sysread($fh, $buf, 24) or
3324 croak "read: $!";
3325 $read == 24 or
3326 croak "read only $read bytes (!= 24)";
3327 ($last_rev, $last_commit) =
3328 unpack(rev_map_fmt, $buf);
3329 if ($last_commit eq ('0' x40)) {
3330 croak "inconsistent .rev_map\n";
3331 }
3332 }
3333 if ($last_rev >= $rev) {
3334 croak "last_rev is higher!: $last_rev >= $rev";
3335 }
3336 $wr_offset = -24;
3337 }
3338 }
3339 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3340 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3341 croak "write: $!";
3342}
3343sub _rev_map_reset {
3345 my ($fh, $rev, $commit) = @_;
3346 my $c = _rev_map_get($fh, $rev);
3347 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3348 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3349 truncate $fh, $offset or croak "truncate: $!";
3350}
3351sub mkfile {
3353 my ($path) = @_;
3354 unless (-e $path) {
3355 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3356 mkpath([$dir]) unless -d $dir;
3357 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3358 close $fh or die "Couldn't close (create) $path: $!\n";
3359 }
3360}
3361sub rev_map_set {
3363 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3364 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3365 my $db = $self->map_path($uuid);
3366 my $db_lock = "$db.lock";
3367 my $sig;
3368 $update_ref ||= 0;
3369 if ($update_ref) {
3370 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3371 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3372 }
3373 mkfile($db);
3374 $LOCKFILES{$db_lock} = 1;
3376 my $sync;
3377 # both of these options make our .rev_db file very, very important
3378 # and we can't afford to lose it because rebuild() won't work
3379 if ($self->use_svm_props || $self->no_metadata) {
3380 $sync = 1;
3381 copy($db, $db_lock) or die "rev_map_set(@_): ",
3382 "Failed to copy: ",
3383 "$db => $db_lock ($!)\n";
3384 } else {
3385 rename $db, $db_lock or die "rev_map_set(@_): ",
3386 "Failed to rename: ",
3387 "$db => $db_lock ($!)\n";
3388 }
3389 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3391 or croak "Couldn't open $db_lock: $!\n";
3392 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3393 _rev_map_set($fh, $rev, $commit);
3394 if ($sync) {
3395 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3396 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3397 }
3398 close $fh or croak $!;
3399 if ($update_ref) {
3400 $_head = $self;
3401 my $note = "";
3402 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3403 command_noisy('update-ref', '-m', "r$rev$note",
3404 $self->refname, $commit);
3405 }
3406 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3407 "$db_lock => $db ($!)\n";
3408 delete $LOCKFILES{$db_lock};
3409 if ($update_ref) {
3410 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3411 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3412 kill $sig, $$ if defined $sig;
3413 }
3414}
3415# If want_commit, this will return an array of (rev, commit) where
3417# commit _must_ be a valid commit in the archive.
3418# Otherwise, it'll return the max revision (whether or not the
3419# commit is valid or just a 0x40 placeholder).
3420sub rev_map_max {
3421 my ($self, $want_commit) = @_;
3422 $self->rebuild;
3423 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3424 $want_commit ? ($r, $c) : $r;
3425}
3426sub rev_map_max_norebuild {
3428 my ($self, $want_commit) = @_;
3429 my $map_path = $self->map_path;
3430 stat $map_path or return $want_commit ? (0, undef) : 0;
3431 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3432 binmode $fh or croak "binmode: $!";
3433 my $size = (stat($fh))[7];
3434 ($size % 24) == 0 or croak "inconsistent size: $size";
3435 if ($size == 0) {
3437 close $fh or croak "close: $!";
3438 return $want_commit ? (0, undef) : 0;
3439 }
3440 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3442 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3443 my ($r, $c) = unpack(rev_map_fmt, $buf);
3444 if ($want_commit && $c eq ('0' x40)) {
3445 if ($size < 48) {
3446 return $want_commit ? (0, undef) : 0;
3447 }
3448 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3449 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3450 ($r, $c) = unpack(rev_map_fmt, $buf);
3451 if ($c eq ('0'x40)) {
3452 croak "Penultimate record is all-zeroes in $map_path";
3453 }
3454 }
3455 close $fh or croak "close: $!";
3456 $want_commit ? ($r, $c) : $r;
3457}
3458sub rev_map_get {
3460 my ($self, $rev, $uuid) = @_;
3461 my $map_path = $self->map_path($uuid);
3462 return undef unless -e $map_path;
3463 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3465 my $c = _rev_map_get($fh, $rev);
3466 close($fh) or croak "close: $!";
3467 $c
3468}
3469sub _rev_map_get {
3471 my ($fh, $rev) = @_;
3472 binmode $fh or croak "binmode: $!";
3474 my $size = (stat($fh))[7];
3475 ($size % 24) == 0 or croak "inconsistent size: $size";
3476 if ($size == 0) {
3478 return undef;
3479 }
3480 my ($l, $u) = (0, $size - 24);
3482 my ($r, $c, $buf);
3483 while ($l <= $u) {
3485 my $i = int(($l/24 + $u/24) / 2) * 24;
3486 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3487 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3488 my ($r, $c) = unpack(rev_map_fmt, $buf);
3489 if ($r < $rev) {
3491 $l = $i + 24;
3492 } elsif ($r > $rev) {
3493 $u = $i - 24;
3494 } else { # $r == $rev
3495 return $c eq ('0' x 40) ? undef : $c;
3496 }
3497 }
3498 undef;
3499}
3500# Finds the first svn revision that exists on (if $eq_ok is true) or
3502# before $rev for the current branch. It will not search any lower
3503# than $min_rev. Returns the git commit hash and svn revision number
3504# if found, else (undef, undef).
3505sub find_rev_before {
3506 my ($self, $rev, $eq_ok, $min_rev) = @_;
3507 --$rev unless $eq_ok;
3508 $min_rev ||= 1;
3509 my $max_rev = $self->rev_map_max;
3510 $rev = $max_rev if ($rev > $max_rev);
3511 while ($rev >= $min_rev) {
3512 if (my $c = $self->rev_map_get($rev)) {
3513 return ($rev, $c);
3514 }
3515 --$rev;
3516 }
3517 return (undef, undef);
3518}
3519# Finds the first svn revision that exists on (if $eq_ok is true) or
3521# after $rev for the current branch. It will not search any higher
3522# than $max_rev. Returns the git commit hash and svn revision number
3523# if found, else (undef, undef).
3524sub find_rev_after {
3525 my ($self, $rev, $eq_ok, $max_rev) = @_;
3526 ++$rev unless $eq_ok;
3527 $max_rev ||= $self->rev_map_max;
3528 while ($rev <= $max_rev) {
3529 if (my $c = $self->rev_map_get($rev)) {
3530 return ($rev, $c);
3531 }
3532 ++$rev;
3533 }
3534 return (undef, undef);
3535}
3536sub _new {
3538 my ($class, $repo_id, $ref_id, $path) = @_;
3539 unless (defined $repo_id && length $repo_id) {
3540 $repo_id = $Git::SVN::default_repo_id;
3541 }
3542 unless (defined $ref_id && length $ref_id) {
3543 $_prefix = '' unless defined($_prefix);
3544 $_[2] = $ref_id =
3545 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3546 }
3547 $_[1] = $repo_id;
3548 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3549 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3551 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3552 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3553 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3554 if (-d $old_dir && ! -d $dir) {
3555 $dir = $old_dir;
3556 }
3557 }
3558 $_[3] = $path = '' unless (defined $path);
3560 mkpath([$dir]);
3561 bless {
3562 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3563 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3564 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3565}
3566# for read-only access of old .rev_db formats
3568sub unlink_rev_db_symlink {
3569 my ($self) = @_;
3570 my $link = $self->rev_db_path;
3571 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3572 if (-l $link) {
3573 unlink $link or croak "unlink: $link failed!";
3574 }
3575}
3576sub rev_db_path {
3578 my ($self, $uuid) = @_;
3579 my $db_path = $self->map_path($uuid);
3580 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3581 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3582 $db_path;
3583}
3584# the new replacement for .rev_db
3586sub map_path {
3587 my ($self, $uuid) = @_;
3588 $uuid ||= $self->ra_uuid;
3589 "$self->{map_root}.$uuid";
3590}
3591sub uri_encode {
3593 my ($f) = @_;
3594 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3595 $f
3596}
3597sub uri_decode {
3599 my ($f) = @_;
3600 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3601 $f
3602}
3603sub remove_username {
3605 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3606}
3607package Git::SVN::Prompt;
3609use strict;
3610use warnings;
3611require SVN::Core;
3612use vars qw/$_no_auth_cache $_username/;
3613sub simple {
3615 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3616 $may_save = undef if $_no_auth_cache;
3617 $default_username = $_username if defined $_username;
3618 if (defined $default_username && length $default_username) {
3619 if (defined $realm && length $realm) {
3620 print STDERR "Authentication realm: $realm\n";
3621 STDERR->flush;
3622 }
3623 $cred->username($default_username);
3624 } else {
3625 username($cred, $realm, $may_save, $pool);
3626 }
3627 $cred->password(_read_password("Password for '" .
3628 $cred->username . "': ", $realm));
3629 $cred->may_save($may_save);
3630 $SVN::_Core::SVN_NO_ERROR;
3631}
3632sub ssl_server_trust {
3634 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3635 $may_save = undef if $_no_auth_cache;
3636 print STDERR "Error validating server certificate for '$realm':\n";
3637 {
3638 no warnings 'once';
3639 # All variables SVN::Auth::SSL::* are used only once,
3640 # so we're shutting up Perl warnings about this.
3641 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3642 print STDERR " - The certificate is not issued ",
3643 "by a trusted authority. Use the\n",
3644 " fingerprint to validate ",
3645 "the certificate manually!\n";
3646 }
3647 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3648 print STDERR " - The certificate hostname ",
3649 "does not match.\n";
3650 }
3651 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3652 print STDERR " - The certificate is not yet valid.\n";
3653 }
3654 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3655 print STDERR " - The certificate has expired.\n";
3656 }
3657 if ($failures & $SVN::Auth::SSL::OTHER) {
3658 print STDERR " - The certificate has ",
3659 "an unknown error.\n";
3660 }
3661 } # no warnings 'once'
3662 printf STDERR
3663 "Certificate information:\n".
3664 " - Hostname: %s\n".
3665 " - Valid: from %s until %s\n".
3666 " - Issuer: %s\n".
3667 " - Fingerprint: %s\n",
3668 map $cert_info->$_, qw(hostname valid_from valid_until
3669 issuer_dname fingerprint);
3670 my $choice;
3671prompt:
3672 print STDERR $may_save ?
3673 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3674 "(R)eject or accept (t)emporarily? ";
3675 STDERR->flush;
3676 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3677 if ($choice =~ /^t$/i) {
3678 $cred->may_save(undef);
3679 } elsif ($choice =~ /^r$/i) {
3680 return -1;
3681 } elsif ($may_save && $choice =~ /^p$/i) {
3682 $cred->may_save($may_save);
3683 } else {
3684 goto prompt;
3685 }
3686 $cred->accepted_failures($failures);
3687 $SVN::_Core::SVN_NO_ERROR;
3688}
3689sub ssl_client_cert {
3691 my ($cred, $realm, $may_save, $pool) = @_;
3692 $may_save = undef if $_no_auth_cache;
3693 print STDERR "Client certificate filename: ";
3694 STDERR->flush;
3695 chomp(my $filename = <STDIN>);
3696 $cred->cert_file($filename);
3697 $cred->may_save($may_save);
3698 $SVN::_Core::SVN_NO_ERROR;
3699}
3700sub ssl_client_cert_pw {
3702 my ($cred, $realm, $may_save, $pool) = @_;
3703 $may_save = undef if $_no_auth_cache;
3704 $cred->password(_read_password("Password: ", $realm));
3705 $cred->may_save($may_save);
3706 $SVN::_Core::SVN_NO_ERROR;
3707}
3708sub username {
3710 my ($cred, $realm, $may_save, $pool) = @_;
3711 $may_save = undef if $_no_auth_cache;
3712 if (defined $realm && length $realm) {
3713 print STDERR "Authentication realm: $realm\n";
3714 }
3715 my $username;
3716 if (defined $_username) {
3717 $username = $_username;
3718 } else {
3719 print STDERR "Username: ";
3720 STDERR->flush;
3721 chomp($username = <STDIN>);
3722 }
3723 $cred->username($username);
3724 $cred->may_save($may_save);
3725 $SVN::_Core::SVN_NO_ERROR;
3726}
3727sub _read_password {
3729 my ($prompt, $realm) = @_;
3730 print STDERR $prompt;
3731 STDERR->flush;
3732 require Term::ReadKey;
3733 Term::ReadKey::ReadMode('noecho');
3734 my $password = '';
3735 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3736 last if $key =~ /[\012\015]/; # \n\r
3737 $password .= $key;
3738 }
3739 Term::ReadKey::ReadMode('restore');
3740 print STDERR "\n";
3741 STDERR->flush;
3742 $password;
3743}
3744package SVN::Git::Fetcher;
3746use vars qw/@ISA/;
3747use strict;
3748use warnings;
3749use Carp qw/croak/;
3750use File::Temp qw/tempfile/;
3751use IO::File qw//;
3752use vars qw/$_ignore_regex/;
3753# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3755sub new {
3756 my ($class, $git_svn, $switch_path) = @_;
3757 my $self = SVN::Delta::Editor->new;
3758 bless $self, $class;
3759 if (exists $git_svn->{last_commit}) {
3760 $self->{c} = $git_svn->{last_commit};
3761 $self->{empty_symlinks} =
3762 _mark_empty_symlinks($git_svn, $switch_path);
3763 }
3764 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3765 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3766 $self->{empty} = {};
3767 $self->{dir_prop} = {};
3768 $self->{file_prop} = {};
3769 $self->{absent_dir} = {};
3770 $self->{absent_file} = {};
3771 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3772 $self;
3773}
3774# this uses the Ra object, so it must be called before do_{switch,update},
3776# not inside them (when the Git::SVN::Fetcher object is passed) to
3777# do_{switch,update}
3778sub _mark_empty_symlinks {
3779 my ($git_svn, $switch_path) = @_;
3780 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3781 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3782 my %ret;
3784 my ($rev, $cmt) = $git_svn->last_rev_commit;
3785 return {} unless ($rev && $cmt);
3786 # allow the warning to be printed for each revision we fetch to
3788 # ensure the user sees it. The user can also disable the workaround
3789 # on the repository even while git svn is running and the next
3790 # revision fetched will skip this expensive function.
3791 my $printed_warning;
3792 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3793 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3794 local $/ = "\0";
3795 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3796 $pfx .= '/' if length($pfx);
3797 while (<$ls>) {
3798 chomp;
3799 s/\A100644 blob $empty_blob\t//o or next;
3800 unless ($printed_warning) {
3801 print STDERR "Scanning for empty symlinks, ",
3802 "this may take a while if you have ",
3803 "many empty files\n",
3804 "You may disable this with `",
3805 "git config svn.brokenSymlinkWorkaround ",
3806 "false'.\n",
3807 "This may be done in a different ",
3808 "terminal without restarting ",
3809 "git svn\n";
3810 $printed_warning = 1;
3811 }
3812 my $path = $_;
3813 my (undef, $props) =
3814 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3815 if ($props->{'svn:special'}) {
3816 $ret{$path} = 1;
3817 }
3818 }
3819 command_close_pipe($ls, $ctx);
3820 \%ret;
3821}
3822# returns true if a given path is inside a ".git" directory
3824sub in_dot_git {
3825 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3826}
3827# return value: 0 -- don't ignore, 1 -- ignore
3829sub is_path_ignored {
3830 my ($self, $path) = @_;
3831 return 1 if in_dot_git($path);
3832 return 1 if defined($self->{ignore_regex}) &&
3833 $path =~ m!$self->{ignore_regex}!;
3834 return 0 unless defined($_ignore_regex);
3835 return 1 if $path =~ m!$_ignore_regex!o;
3836 return 0;
3837}
3838sub set_path_strip {
3840 my ($self, $path) = @_;
3841 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3842}
3843sub open_root {
3845 { path => '' };
3846}
3847sub open_directory {
3849 my ($self, $path, $pb, $rev) = @_;
3850 { path => $path };
3851}
3852sub git_path {
3854 my ($self, $path) = @_;
3855 if ($self->{path_strip}) {
3856 $path =~ s!$self->{path_strip}!! or
3857 die "Failed to strip path '$path' ($self->{path_strip})\n";
3858 }
3859 $path;
3860}
3861sub delete_entry {
3863 my ($self, $path, $rev, $pb) = @_;
3864 return undef if $self->is_path_ignored($path);
3865 my $gpath = $self->git_path($path);
3867 return undef if ($gpath eq '');
3868 # remove entire directories.
3870 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3871 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3872 if ($tree) {
3873 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3874 -r --name-only -z/,
3875 $tree);
3876 local $/ = "\0";
3877 while (<$ls>) {
3878 chomp;
3879 my $rmpath = "$gpath/$_";
3880 $self->{gii}->remove($rmpath);
3881 print "\tD\t$rmpath\n" unless $::_q;
3882 }
3883 print "\tD\t$gpath/\n" unless $::_q;
3884 command_close_pipe($ls, $ctx);
3885 $self->{empty}->{$path} = 0
3886 } else {
3887 $self->{gii}->remove($gpath);
3888 print "\tD\t$gpath\n" unless $::_q;
3889 }
3890 undef;
3891}
3892sub open_file {
3894 my ($self, $path, $pb, $rev) = @_;
3895 my ($mode, $blob);
3896 goto out if $self->is_path_ignored($path);
3898 my $gpath = $self->git_path($path);
3900 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3901 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3902 unless (defined $mode && defined $blob) {
3903 die "$path was not found in commit $self->{c} (r$rev)\n";
3904 }
3905 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3906 $mode = '120000';
3907 }
3908out:
3909 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3910 pool => SVN::Pool->new, action => 'M' };
3911}
3912sub add_file {
3914 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3915 my $mode;
3916 if (!$self->is_path_ignored($path)) {
3918 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3919 delete $self->{empty}->{$dir};
3920 $mode = '100644';
3921 }
3922 { path => $path, mode_a => $mode, mode_b => $mode,
3923 pool => SVN::Pool->new, action => 'A' };
3924}
3925sub add_directory {
3927 my ($self, $path, $cp_path, $cp_rev) = @_;
3928 goto out if $self->is_path_ignored($path);
3929 my $gpath = $self->git_path($path);
3930 if ($gpath eq '') {
3931 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3932 -r --name-only -z/,
3933 $self->{c});
3934 local $/ = "\0";
3935 while (<$ls>) {
3936 chomp;
3937 $self->{gii}->remove($_);
3938 print "\tD\t$_\n" unless $::_q;
3939 }
3940 command_close_pipe($ls, $ctx);
3941 $self->{empty}->{$path} = 0;
3942 }
3943 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3944 delete $self->{empty}->{$dir};
3945 $self->{empty}->{$path} = 1;
3946out:
3947 { path => $path };
3948}
3949sub change_dir_prop {
3951 my ($self, $db, $prop, $value) = @_;
3952 return undef if $self->is_path_ignored($db->{path});
3953 $self->{dir_prop}->{$db->{path}} ||= {};
3954 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3955 undef;
3956}
3957sub absent_directory {
3959 my ($self, $path, $pb) = @_;
3960 return undef if $self->is_path_ignored($path);
3961 $self->{absent_dir}->{$pb->{path}} ||= [];
3962 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3963 undef;
3964}
3965sub absent_file {
3967 my ($self, $path, $pb) = @_;
3968 return undef if $self->is_path_ignored($path);
3969 $self->{absent_file}->{$pb->{path}} ||= [];
3970 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3971 undef;
3972}
3973sub change_file_prop {
3975 my ($self, $fb, $prop, $value) = @_;
3976 return undef if $self->is_path_ignored($fb->{path});
3977 if ($prop eq 'svn:executable') {
3978 if ($fb->{mode_b} != 120000) {
3979 $fb->{mode_b} = defined $value ? 100755 : 100644;
3980 }
3981 } elsif ($prop eq 'svn:special') {
3982 $fb->{mode_b} = defined $value ? 120000 : 100644;
3983 } else {
3984 $self->{file_prop}->{$fb->{path}} ||= {};
3985 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3986 }
3987 undef;
3988}
3989sub apply_textdelta {
3991 my ($self, $fb, $exp) = @_;
3992 return undef if $self->is_path_ignored($fb->{path});
3993 my $fh = $::_repository->temp_acquire('svn_delta');
3994 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3995 # (but $base does not,) so dup() it for reading in close_file
3996 open my $dup, '<&', $fh or croak $!;
3997 my $base = $::_repository->temp_acquire('git_blob');
3998 if ($fb->{blob}) {
4000 my ($base_is_link, $size);
4001 if ($fb->{mode_a} eq '120000' &&
4003 ! $self->{empty_symlinks}->{$fb->{path}}) {
4004 print $base 'link ' or die "print $!\n";
4005 $base_is_link = 1;
4006 }
4007 retry:
4008 $size = $::_repository->cat_blob($fb->{blob}, $base);
4009 die "Failed to read object $fb->{blob}" if ($size < 0);
4010 if (defined $exp) {
4012 seek $base, 0, 0 or croak $!;
4013 my $got = ::md5sum($base);
4014 if ($got ne $exp) {
4015 my $err = "Checksum mismatch: ".
4016 "$fb->{path} $fb->{blob}\n" .
4017 "expected: $exp\n" .
4018 " got: $got\n";
4019 if ($base_is_link) {
4020 warn $err,
4021 "Retrying... (possibly ",
4022 "a bad symlink from SVN)\n";
4023 $::_repository->temp_reset($base);
4024 $base_is_link = 0;
4025 goto retry;
4026 }
4027 die $err;
4028 }
4029 }
4030 }
4031 seek $base, 0, 0 or croak $!;
4032 $fb->{fh} = $fh;
4033 $fb->{base} = $base;
4034 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4035}
4036sub close_file {
4038 my ($self, $fb, $exp) = @_;
4039 return undef if $self->is_path_ignored($fb->{path});
4040 my $hash;
4042 my $path = $self->git_path($fb->{path});
4043 if (my $fh = $fb->{fh}) {
4044 if (defined $exp) {
4045 seek($fh, 0, 0) or croak $!;
4046 my $got = ::md5sum($fh);
4047 if ($got ne $exp) {
4048 die "Checksum mismatch: $path\n",
4049 "expected: $exp\n got: $got\n";
4050 }
4051 }
4052 if ($fb->{mode_b} == 120000) {
4053 sysseek($fh, 0, 0) or croak $!;
4054 my $rd = sysread($fh, my $buf, 5);
4055 if (!defined $rd) {
4057 croak "sysread: $!\n";
4058 } elsif ($rd == 0) {
4059 warn "$path has mode 120000",
4060 " but it points to nothing\n",
4061 "converting to an empty file with mode",
4062 " 100644\n";
4063 $fb->{mode_b} = '100644';
4064 } elsif ($buf ne 'link ') {
4065 warn "$path has mode 120000",
4066 " but is not a link\n";
4067 } else {
4068 my $tmp_fh = $::_repository->temp_acquire(
4069 'svn_hash');
4070 my $res;
4071 while ($res = sysread($fh, my $str, 1024)) {
4072 my $out = syswrite($tmp_fh, $str, $res);
4073 defined($out) && $out == $res
4074 or croak("write ",
4075 Git::temp_path($tmp_fh),
4076 ": $!\n");
4077 }
4078 defined $res or croak $!;
4079 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4081 Git::temp_release($tmp_fh, 1);
4082 }
4083 }
4084 $hash = $::_repository->hash_and_insert_object(
4086 Git::temp_path($fh));
4087 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4088 Git::temp_release($fb->{base}, 1);
4090 Git::temp_release($fh, 1);
4091 } else {
4092 $hash = $fb->{blob} or die "no blob information\n";
4093 }
4094 $fb->{pool}->clear;
4095 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4096 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4097 undef;
4098}
4099sub abort_edit {
4101 my $self = shift;
4102 $self->{nr} = $self->{gii}->{nr};
4103 delete $self->{gii};
4104 $self->SUPER::abort_edit(@_);
4105}
4106sub close_edit {
4108 my $self = shift;
4109 $self->{git_commit_ok} = 1;
4110 $self->{nr} = $self->{gii}->{nr};
4111 delete $self->{gii};
4112 $self->SUPER::close_edit(@_);
4113}
4114package SVN::Git::Editor;
4116use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4117use strict;
4118use warnings;
4119use Carp qw/croak/;
4120use IO::File;
4121sub new {
4123 my ($class, $opts) = @_;
4124 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4125 die "$_ required!\n" unless (defined $opts->{$_});
4126 }
4127 my $pool = SVN::Pool->new;
4129 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4130 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4131 $opts->{r}, $mods);
4132 # $opts->{ra} functions should not be used after this:
4134 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4135 $opts->{editor_cb}, $pool);
4136 my $self = SVN::Delta::Editor->new(@ce, $pool);
4137 bless $self, $class;
4138 foreach (qw/svn_path r tree_a tree_b/) {
4139 $self->{$_} = $opts->{$_};
4140 }
4141 $self->{url} = $opts->{ra}->{url};
4142 $self->{mods} = $mods;
4143 $self->{types} = $types;
4144 $self->{pool} = $pool;
4145 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4146 $self->{rm} = { };
4147 $self->{path_prefix} = length $self->{svn_path} ?
4148 "$self->{svn_path}/" : '';
4149 $self->{config} = $opts->{config};
4150 return $self;
4151}
4152sub generate_diff {
4154 my ($tree_a, $tree_b) = @_;
4155 my @diff_tree = qw(diff-tree -z -r);
4156 if ($_cp_similarity) {
4157 push @diff_tree, "-C$_cp_similarity";
4158 } else {
4159 push @diff_tree, '-C';
4160 }
4161 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4162 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4163 push @diff_tree, $tree_a, $tree_b;
4164 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4165 local $/ = "\0";
4166 my $state = 'meta';
4167 my @mods;
4168 while (<$diff_fh>) {
4169 chomp $_; # this gets rid of the trailing "\0"
4170 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4171 ($::sha1)\s($::sha1)\s
4172 ([MTCRAD])\d*$/xo) {
4173 push @mods, { mode_a => $1, mode_b => $2,
4174 sha1_a => $3, sha1_b => $4,
4175 chg => $5 };
4176 if ($5 =~ /^(?:C|R)$/) {
4177 $state = 'file_a';
4178 } else {
4179 $state = 'file_b';
4180 }
4181 } elsif ($state eq 'file_a') {
4182 my $x = $mods[$#mods] or croak "Empty array\n";
4183 if ($x->{chg} !~ /^(?:C|R)$/) {
4184 croak "Error parsing $_, $x->{chg}\n";
4185 }
4186 $x->{file_a} = $_;
4187 $state = 'file_b';
4188 } elsif ($state eq 'file_b') {
4189 my $x = $mods[$#mods] or croak "Empty array\n";
4190 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4191 croak "Error parsing $_, $x->{chg}\n";
4192 }
4193 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4194 croak "Error parsing $_, $x->{chg}\n";
4195 }
4196 $x->{file_b} = $_;
4197 $state = 'meta';
4198 } else {
4199 croak "Error parsing $_\n";
4200 }
4201 }
4202 command_close_pipe($diff_fh, $ctx);
4203 \@mods;
4204}
4205sub check_diff_paths {
4207 my ($ra, $pfx, $rev, $mods) = @_;
4208 my %types;
4209 $pfx .= '/' if length $pfx;
4210 sub type_diff_paths {
4212 my ($ra, $types, $path, $rev) = @_;
4213 my @p = split m#/+#, $path;
4214 my $c = shift @p;
4215 unless (defined $types->{$c}) {
4216 $types->{$c} = $ra->check_path($c, $rev);
4217 }
4218 while (@p) {
4219 $c .= '/' . shift @p;
4220 next if defined $types->{$c};
4221 $types->{$c} = $ra->check_path($c, $rev);
4222 }
4223 }
4224 foreach my $m (@$mods) {
4226 foreach my $f (qw/file_a file_b/) {
4227 next unless defined $m->{$f};
4228 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4229 if (length $pfx.$dir && ! defined $types{$dir}) {
4230 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4231 }
4232 }
4233 }
4234 \%types;
4235}
4236sub split_path {
4238 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4239}
4240sub repo_path {
4242 my ($self, $path) = @_;
4243 $self->{path_prefix}.(defined $path ? $path : '');
4244}
4245sub url_path {
4247 my ($self, $path) = @_;
4248 if ($self->{url} =~ m#^https?://#) {
4249 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4250 }
4251 $self->{url} . '/' . $self->repo_path($path);
4252}
4253sub rmdirs {
4255 my ($self) = @_;
4256 my $rm = $self->{rm};
4257 delete $rm->{''}; # we never delete the url we're tracking
4258 return unless %$rm;
4259 foreach (keys %$rm) {
4261 my @d = split m#/#, $_;
4262 my $c = shift @d;
4263 $rm->{$c} = 1;
4264 while (@d) {
4265 $c .= '/' . shift @d;
4266 $rm->{$c} = 1;
4267 }
4268 }
4269 delete $rm->{$self->{svn_path}};
4270 delete $rm->{''}; # we never delete the url we're tracking
4271 return unless %$rm;
4272 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4274 $self->{tree_b});
4275 local $/ = "\0";
4276 while (<$fh>) {
4277 chomp;
4278 my @dn = split m#/#, $_;
4279 while (pop @dn) {
4280 delete $rm->{join '/', @dn};
4281 }
4282 unless (%$rm) {
4283 close $fh;
4284 return;
4285 }
4286 }
4287 command_close_pipe($fh, $ctx);
4288 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4290 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4291 $self->close_directory($bat->{$d}, $p);
4292 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4293 print "\tD+\t$d/\n" unless $::_q;
4294 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4295 delete $bat->{$d};
4296 }
4297}
4298sub open_or_add_dir {
4300 my ($self, $full_path, $baton) = @_;
4301 my $t = $self->{types}->{$full_path};
4302 if (!defined $t) {
4303 die "$full_path not known in r$self->{r} or we have a bug!\n";
4304 }
4305 {
4306 no warnings 'once';
4307 # SVN::Node::none and SVN::Node::file are used only once,
4308 # so we're shutting up Perl's warnings about them.
4309 if ($t == $SVN::Node::none) {
4310 return $self->add_directory($full_path, $baton,
4311 undef, -1, $self->{pool});
4312 } elsif ($t == $SVN::Node::dir) {
4313 return $self->open_directory($full_path, $baton,
4314 $self->{r}, $self->{pool});
4315 } # no warnings 'once'
4316 print STDERR "$full_path already exists in repository at ",
4317 "r$self->{r} and it is not a directory (",
4318 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4319 } # no warnings 'once'
4320 exit 1;
4321}
4322sub ensure_path {
4324 my ($self, $path) = @_;
4325 my $bat = $self->{bat};
4326 my $repo_path = $self->repo_path($path);
4327 return $bat->{''} unless (length $repo_path);
4328 my @p = split m#/+#, $repo_path;
4329 my $c = shift @p;
4330 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4331 while (@p) {
4332 my $c0 = $c;
4333 $c .= '/' . shift @p;
4334 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4335 }
4336 return $bat->{$c};
4337}
4338# Subroutine to convert a globbing pattern to a regular expression.
4340# From perl cookbook.
4341sub glob2pat {
4342 my $globstr = shift;
4343 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4344 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4345 return '^' . $globstr . '$';
4346}
4347sub check_autoprop {
4349 my ($self, $pattern, $properties, $file, $fbat) = @_;
4350 # Convert the globbing pattern to a regular expression.
4351 my $regex = glob2pat($pattern);
4352 # Check if the pattern matches the file name.
4353 if($file =~ m/($regex)/) {
4354 # Parse the list of properties to set.
4355 my @props = split(/;/, $properties);
4356 foreach my $prop (@props) {
4357 # Parse 'name=value' syntax and set the property.
4358 if ($prop =~ /([^=]+)=(.*)/) {
4359 my ($n,$v) = ($1,$2);
4360 for ($n, $v) {
4361 s/^\s+//; s/\s+$//;
4362 }
4363 $self->change_file_prop($fbat, $n, $v);
4364 }
4365 }
4366 }
4367}
4368sub apply_autoprops {
4370 my ($self, $file, $fbat) = @_;
4371 my $conf_t = ${$self->{config}}{'config'};
4372 no warnings 'once';
4373 # Check [miscellany]/enable-auto-props in svn configuration.
4374 if (SVN::_Core::svn_config_get_bool(
4375 $conf_t,
4376 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4377 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4378 0)) {
4379 # Auto-props are enabled. Enumerate them to look for matches.
4380 my $callback = sub {
4381 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4382 };
4383 SVN::_Core::svn_config_enumerate(
4384 $conf_t,
4385 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4386 $callback);
4387 }
4388}
4389sub A {
4391 my ($self, $m) = @_;
4392 my ($dir, $file) = split_path($m->{file_b});
4393 my $pbat = $self->ensure_path($dir);
4394 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4395 undef, -1);
4396 print "\tA\t$m->{file_b}\n" unless $::_q;
4397 $self->apply_autoprops($file, $fbat);
4398 $self->chg_file($fbat, $m);
4399 $self->close_file($fbat,undef,$self->{pool});
4400}
4401sub C {
4403 my ($self, $m) = @_;
4404 my ($dir, $file) = split_path($m->{file_b});
4405 my $pbat = $self->ensure_path($dir);
4406 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4407 $self->url_path($m->{file_a}), $self->{r});
4408 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4409 $self->chg_file($fbat, $m);
4410 $self->close_file($fbat,undef,$self->{pool});
4411}
4412sub delete_entry {
4414 my ($self, $path, $pbat) = @_;
4415 my $rpath = $self->repo_path($path);
4416 my ($dir, $file) = split_path($rpath);
4417 $self->{rm}->{$dir} = 1;
4418 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4419}
4420sub R {
4422 my ($self, $m) = @_;
4423 my ($dir, $file) = split_path($m->{file_b});
4424 my $pbat = $self->ensure_path($dir);
4425 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4426 $self->url_path($m->{file_a}), $self->{r});
4427 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4428 $self->apply_autoprops($file, $fbat);
4429 $self->chg_file($fbat, $m);
4430 $self->close_file($fbat,undef,$self->{pool});
4431 ($dir, $file) = split_path($m->{file_a});
4433 $pbat = $self->ensure_path($dir);
4434 $self->delete_entry($m->{file_a}, $pbat);
4435}
4436sub M {
4438 my ($self, $m) = @_;
4439 my ($dir, $file) = split_path($m->{file_b});
4440 my $pbat = $self->ensure_path($dir);
4441 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4442 $pbat,$self->{r},$self->{pool});
4443 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4444 $self->chg_file($fbat, $m);
4445 $self->close_file($fbat,undef,$self->{pool});
4446}
4447sub T { shift->M(@_) }
4449sub change_file_prop {
4451 my ($self, $fbat, $pname, $pval) = @_;
4452 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4453}
4454sub _chg_file_get_blob ($$$$) {
4456 my ($self, $fbat, $m, $which) = @_;
4457 my $fh = $::_repository->temp_acquire("git_blob_$which");
4458 if ($m->{"mode_$which"} =~ /^120/) {
4459 print $fh 'link ' or croak $!;
4460 $self->change_file_prop($fbat,'svn:special','*');
4461 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4462 $self->change_file_prop($fbat,'svn:special',undef);
4463 }
4464 my $blob = $m->{"sha1_$which"};
4465 return ($fh,) if ($blob =~ /^0{40}$/);
4466 my $size = $::_repository->cat_blob($blob, $fh);
4467 croak "Failed to read object $blob" if ($size < 0);
4468 $fh->flush == 0 or croak $!;
4469 seek $fh, 0, 0 or croak $!;
4470 my $exp = ::md5sum($fh);
4472 seek $fh, 0, 0 or croak $!;
4473 return ($fh, $exp);
4474}
4475sub chg_file {
4477 my ($self, $fbat, $m) = @_;
4478 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4479 $self->change_file_prop($fbat,'svn:executable','*');
4480 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4481 $self->change_file_prop($fbat,'svn:executable',undef);
4482 }
4483 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4484 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4485 my $pool = SVN::Pool->new;
4486 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4487 if (-s $fh_a) {
4488 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4489 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4490 if (defined $res) {
4491 die "Unexpected result from send_txstream: $res\n",
4492 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4493 }
4494 } else {
4495 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4496 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4497 if ($got ne $exp_b);
4498 }
4499 Git::temp_release($fh_b, 1);
4500 Git::temp_release($fh_a, 1);
4501 $pool->clear;
4502}
4503sub D {
4505 my ($self, $m) = @_;
4506 my ($dir, $file) = split_path($m->{file_b});
4507 my $pbat = $self->ensure_path($dir);
4508 print "\tD\t$m->{file_b}\n" unless $::_q;
4509 $self->delete_entry($m->{file_b}, $pbat);
4510}
4511sub close_edit {
4513 my ($self) = @_;
4514 my ($p,$bat) = ($self->{pool}, $self->{bat});
4515 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4516 next if $_ eq '';
4517 $self->close_directory($bat->{$_}, $p);
4518 }
4519 $self->close_directory($bat->{''}, $p);
4520 $self->SUPER::close_edit($p);
4521 $p->clear;
4522}
4523sub abort_edit {
4525 my ($self) = @_;
4526 $self->SUPER::abort_edit($self->{pool});
4527}
4528sub DESTROY {
4530 my $self = shift;
4531 $self->SUPER::DESTROY(@_);
4532 $self->{pool}->clear;
4533}
4534# this drives the editor
4536sub apply_diff {
4537 my ($self) = @_;
4538 my $mods = $self->{mods};
4539 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4540 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4541 my $f = $m->{chg};
4542 if (defined $o{$f}) {
4543 $self->$f($m);
4544 } else {
4545 fatal("Invalid change type: $f");
4546 }
4547 }
4548 $self->rmdirs if $_rmdir;
4549 if (@$mods == 0) {
4550 $self->abort_edit;
4551 } else {
4552 $self->close_edit;
4553 }
4554 return scalar @$mods;
4555}
4556package Git::SVN::Ra;
4558use vars qw/@ISA $config_dir $_log_window_size/;
4559use strict;
4560use warnings;
4561my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4562BEGIN {
4564 # enforce temporary pool usage for some simple functions
4565 no strict 'refs';
4566 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4567 get_file/) {
4568 my $SUPER = "SUPER::$f";
4569 *$f = sub {
4570 my $self = shift;
4571 my $pool = SVN::Pool->new;
4572 my @ret = $self->$SUPER(@_,$pool);
4573 $pool->clear;
4574 wantarray ? @ret : $ret[0];
4575 };
4576 }
4577}
4578sub _auth_providers () {
4580 [
4581 SVN::Client::get_simple_provider(),
4582 SVN::Client::get_ssl_server_trust_file_provider(),
4583 SVN::Client::get_simple_prompt_provider(
4584 \&Git::SVN::Prompt::simple, 2),
4585 SVN::Client::get_ssl_client_cert_file_provider(),
4586 SVN::Client::get_ssl_client_cert_prompt_provider(
4587 \&Git::SVN::Prompt::ssl_client_cert, 2),
4588 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4589 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4590 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4591 SVN::Client::get_username_provider(),
4592 SVN::Client::get_ssl_server_trust_prompt_provider(
4593 \&Git::SVN::Prompt::ssl_server_trust),
4594 SVN::Client::get_username_prompt_provider(
4595 \&Git::SVN::Prompt::username, 2)
4596 ]
4597}
4598sub escape_uri_only {
4600 my ($uri) = @_;
4601 my @tmp;
4602 foreach (split m{/}, $uri) {
4603 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4604 push @tmp, $_;
4605 }
4606 join('/', @tmp);
4607}
4608sub escape_url {
4610 my ($url) = @_;
4611 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4612 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4613 $url = "$scheme://$domain$uri";
4614 }
4615 $url;
4616}
4617sub new {
4619 my ($class, $url) = @_;
4620 $url =~ s!/+$!!;
4621 return $RA if ($RA && $RA->{url} eq $url);
4622 SVN::_Core::svn_config_ensure($config_dir, undef);
4624 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4625 my $config = SVN::Core::config_get_config($config_dir);
4626 $RA = undef;
4627 my $dont_store_passwords = 1;
4628 my $conf_t = ${$config}{'config'};
4629 {
4630 no warnings 'once';
4631 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4632 # produces warnings that variables are used only once.
4633 # I had not found the better way to shut them up, so
4634 # the warnings of type 'once' are disabled in this block.
4635 if (SVN::_Core::svn_config_get_bool($conf_t,
4636 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4637 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4638 1) == 0) {
4639 SVN::_Core::svn_auth_set_parameter($baton,
4640 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4641 bless (\$dont_store_passwords, "_p_void"));
4642 }
4643 if (SVN::_Core::svn_config_get_bool($conf_t,
4644 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4645 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4646 1) == 0) {
4647 $Git::SVN::Prompt::_no_auth_cache = 1;
4648 }
4649 } # no warnings 'once'
4650 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4651 config => $config,
4652 pool => SVN::Pool->new,
4653 auth_provider_callbacks => $callbacks);
4654 $self->{url} = $url;
4655 $self->{svn_path} = $url;
4656 $self->{repos_root} = $self->get_repos_root;
4657 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4658 $self->{cache} = { check_path => { r => 0, data => {} },
4659 get_dir => { r => 0, data => {} } };
4660 $RA = bless $self, $class;
4661}
4662sub check_path {
4664 my ($self, $path, $r) = @_;
4665 my $cache = $self->{cache}->{check_path};
4666 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4667 return $cache->{data}->{$path};
4668 }
4669 my $pool = SVN::Pool->new;
4670 my $t = $self->SUPER::check_path($path, $r, $pool);
4671 $pool->clear;
4672 if ($r != $cache->{r}) {
4673 %{$cache->{data}} = ();
4674 $cache->{r} = $r;
4675 }
4676 $cache->{data}->{$path} = $t;
4677}
4678sub get_dir {
4680 my ($self, $dir, $r) = @_;
4681 my $cache = $self->{cache}->{get_dir};
4682 if ($r == $cache->{r}) {
4683 if (my $x = $cache->{data}->{$dir}) {
4684 return wantarray ? @$x : $x->[0];
4685 }
4686 }
4687 my $pool = SVN::Pool->new;
4688 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4689 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4690 $pool->clear;
4691 if ($r != $cache->{r}) {
4692 %{$cache->{data}} = ();
4693 $cache->{r} = $r;
4694 }
4695 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4696 wantarray ? (\%dirents, $r, $props) : \%dirents;
4697}
4698sub DESTROY {
4700 # do not call the real DESTROY since we store ourselves in $RA
4701}
4702# get_log(paths, start, end, limit,
4704# discover_changed_paths, strict_node_history, receiver)
4705sub get_log {
4706 my ($self, @args) = @_;
4707 my $pool = SVN::Pool->new;
4708 # svn_log_changed_path_t objects passed to get_log are likely to be
4710 # overwritten even if only the refs are copied to an external variable,
4711 # so we should dup the structures in their entirety. Using an
4712 # externally passed pool (instead of our temporary and quickly cleared
4713 # pool in Git::SVN::Ra) does not help matters at all...
4714 my $receiver = pop @args;
4715 my $prefix = "/".$self->{svn_path};
4716 $prefix =~ s#/+($)##;
4717 my $prefix_regex = qr#^\Q$prefix\E#;
4718 push(@args, sub {
4719 my ($paths) = $_[0];
4720 return &$receiver(@_) unless $paths;
4721 $_[0] = ();
4722 foreach my $p (keys %$paths) {
4723 my $i = $paths->{$p};
4724 # Make path relative to our url, not repos_root
4725 $p =~ s/$prefix_regex//;
4726 my %s = map { $_ => $i->$_; }
4727 qw/copyfrom_path copyfrom_rev action/;
4728 if ($s{'copyfrom_path'}) {
4729 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4730 }
4731 $_[0]{$p} = \%s;
4732 }
4733 &$receiver(@_);
4734 });
4735 # the limit parameter was not supported in SVN 1.1.x, so we
4738 # drop it. Therefore, the receiver callback passed to it
4739 # is made aware of this limitation by being wrapped if
4740 # the limit passed to is being wrapped.
4741 if ($SVN::Core::VERSION le '1.2.0') {
4742 my $limit = splice(@args, 3, 1);
4743 if ($limit > 0) {
4744 my $receiver = pop @args;
4745 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4746 }
4747 }
4748 my $ret = $self->SUPER::get_log(@args, $pool);
4749 $pool->clear;
4750 $ret;
4751}
4752sub trees_match {
4754 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4755 my $ctx = SVN::Client->new(auth => _auth_providers);
4756 my $out = IO::File->new_tmpfile;
4757 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4759 # $ctx->diff(), so we'll create a default one
4760 my $pool = SVN::Pool->new_default_sub;
4761 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4763 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4764 $out->flush;
4765 my $ret = (($out->stat)[7] == 0);
4766 close $out or croak $!;
4767 $ret;
4769}
4770sub get_commit_editor {
4772 my ($self, $log, $cb, $pool) = @_;
4773 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4774 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4775}
4776sub gs_do_update {
4778 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4779 my $new = ($rev_a == $rev_b);
4780 my $path = $gs->{path};
4781 if ($new && -e $gs->{index}) {
4783 unlink $gs->{index} or die
4784 "Couldn't unlink index: $gs->{index}: $!\n";
4785 }
4786 my $pool = SVN::Pool->new;
4787 $editor->set_path_strip($path);
4788 my (@pc) = split m#/#, $path;
4789 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4790 1, $editor, $pool);
4791 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4792 # Since we can't rely on svn_ra_reparent being available, we'll
4794 # just have to do some magic with set_path to make it so
4795 # we only want a partial path.
4796 my $sp = '';
4797 my $final = join('/', @pc);
4798 while (@pc) {
4799 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4800 $sp .= '/' if length $sp;
4801 $sp .= shift @pc;
4802 }
4803 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4804 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4806 $reporter->finish_report($pool);
4808 $pool->clear;
4809 $editor->{git_commit_ok};
4810}
4811# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4813# svn_ra_reparent didn't work before 1.4)
4814sub gs_do_switch {
4815 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4816 my $path = $gs->{path};
4817 my $pool = SVN::Pool->new;
4818 my $full_url = $self->{url};
4820 my $old_url = $full_url;
4821 $full_url .= '/' . $path if length $path;
4822 my ($ra, $reparented);
4823 if ($old_url =~ m#^svn(\+ssh)?://# ||
4825 ($full_url =~ m#^https?://# &&
4826 escape_url($full_url) ne $full_url)) {
4827 $_[0] = undef;
4828 $self = undef;
4829 $RA = undef;
4830 $ra = Git::SVN::Ra->new($full_url);
4831 $ra_invalid = 1;
4832 } elsif ($old_url ne $full_url) {
4833 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4834 $self->{url} = $full_url;
4835 $reparented = 1;
4836 }
4837 $ra ||= $self;
4839 $url_b = escape_url($url_b);
4840 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4841 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4842 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4843 $reporter->finish_report($pool);
4844 if ($reparented) {
4846 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4847 $self->{url} = $old_url;
4848 }
4849 $pool->clear;
4851 $editor->{git_commit_ok};
4852}
4853sub longest_common_path {
4855 my ($gsv, $globs) = @_;
4856 my %common;
4857 my $common_max = scalar @$gsv;
4858 foreach my $gs (@$gsv) {
4860 my @tmp = split m#/#, $gs->{path};
4861 my $p = '';
4862 foreach (@tmp) {
4863 $p .= length($p) ? "/$_" : $_;
4864 $common{$p} ||= 0;
4865 $common{$p}++;
4866 }
4867 }
4868 $globs ||= [];
4869 $common_max += scalar @$globs;
4870 foreach my $glob (@$globs) {
4871 my @tmp = split m#/#, $glob->{path}->{left};
4872 my $p = '';
4873 foreach (@tmp) {
4874 $p .= length($p) ? "/$_" : $_;
4875 $common{$p} ||= 0;
4876 $common{$p}++;
4877 }
4878 }
4879 my $longest_path = '';
4881 foreach (sort {length $b <=> length $a} keys %common) {
4882 if ($common{$_} == $common_max) {
4883 $longest_path = $_;
4884 last;
4885 }
4886 }
4887 $longest_path;
4888}
4889sub gs_fetch_loop_common {
4891 my ($self, $base, $head, $gsv, $globs) = @_;
4892 return if ($base > $head);
4893 my $inc = $_log_window_size;
4894 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4895 my $longest_path = longest_common_path($gsv, $globs);
4896 my $ra_url = $self->{url};
4897 my $find_trailing_edge;
4898 while (1) {
4899 my %revs;
4900 my $err;
4901 my $err_handler = $SVN::Error::handler;
4902 $SVN::Error::handler = sub {
4903 ($err) = @_;
4904 skip_unknown_revs($err);
4905 };
4906 sub _cb {
4907 my ($paths, $r, $author, $date, $log) = @_;
4908 [ $paths,
4909 { author => $author, date => $date, log => $log } ];
4910 }
4911 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4912 sub { $revs{$_[1]} = _cb(@_) });
4913 if ($err) {
4914 print "Checked through r$max\r";
4915 } else {
4916 $find_trailing_edge = 1;
4917 }
4918 if ($err and $find_trailing_edge) {
4919 print STDERR "Path '$longest_path' ",
4920 "was probably deleted:\n",
4921 $err->expanded_message,
4922 "\nWill attempt to follow ",
4923 "revisions r$min .. r$max ",
4924 "committed before the deletion\n";
4925 my $hi = $max;
4926 while (--$hi >= $min) {
4927 my $ok;
4928 $self->get_log([$longest_path], $min, $hi,
4929 0, 1, 1, sub {
4930 $ok = $_[1];
4931 $revs{$_[1]} = _cb(@_) });
4932 if ($ok) {
4933 print STDERR "r$min .. r$ok OK\n";
4934 last;
4935 }
4936 }
4937 $find_trailing_edge = 0;
4938 }
4939 $SVN::Error::handler = $err_handler;
4940 my %exists = map { $_->{path} => $_ } @$gsv;
4942 foreach my $r (sort {$a <=> $b} keys %revs) {
4943 my ($paths, $logged) = @{$revs{$r}};
4944 foreach my $gs ($self->match_globs(\%exists, $paths,
4946 $globs, $r)) {
4947 if ($gs->rev_map_max >= $r) {
4948 next;
4949 }
4950 next unless $gs->match_paths($paths, $r);
4951 $gs->{logged_rev_props} = $logged;
4952 if (my $last_commit = $gs->last_commit) {
4953 $gs->assert_index_clean($last_commit);
4954 }
4955 my $log_entry = $gs->do_fetch($paths, $r);
4956 if ($log_entry) {
4957 $gs->do_git_commit($log_entry);
4958 }
4959 $INDEX_FILES{$gs->{index}} = 1;
4960 }
4961 foreach my $g (@$globs) {
4962 my $k = "svn-remote.$g->{remote}." .
4963 "$g->{t}-maxRev";
4964 Git::SVN::tmp_config($k, $r);
4965 }
4966 if ($ra_invalid) {
4967 $_[0] = undef;
4968 $self = undef;
4969 $RA = undef;
4970 $self = Git::SVN::Ra->new($ra_url);
4971 $ra_invalid = undef;
4972 }
4973 }
4974 # pre-fill the .rev_db since it'll eventually get filled in
4975 # with '0' x40 if something new gets committed
4976 foreach my $gs (@$gsv) {
4977 next if $gs->rev_map_max >= $max;
4978 next if defined $gs->rev_map_get($max);
4979 $gs->rev_map_set($max, 0 x40);
4980 }
4981 foreach my $g (@$globs) {
4982 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4983 Git::SVN::tmp_config($k, $max);
4984 }
4985 last if $max >= $head;
4986 $min = $max + 1;
4987 $max += $inc;
4988 $max = $head if ($max > $head);
4989 }
4990 Git::SVN::gc();
4991}
4992sub get_dir_globbed {
4994 my ($self, $left, $depth, $r) = @_;
4995 my @x = eval { $self->get_dir($left, $r) };
4997 return unless scalar @x == 3;
4998 my $dirents = $x[0];
4999 my @finalents;
5000 foreach my $de (keys %$dirents) {
5001 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5002 if ($depth > 1) {
5003 my @args = ("$left/$de", $depth - 1, $r);
5004 foreach my $dir ($self->get_dir_globbed(@args)) {
5005 push @finalents, "$de/$dir";
5006 }
5007 } else {
5008 push @finalents, $de;
5009 }
5010 }
5011 @finalents;
5012}
5013sub match_globs {
5015 my ($self, $exists, $paths, $globs, $r) = @_;
5016 sub get_dir_check {
5018 my ($self, $exists, $g, $r) = @_;
5019 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5021 $g->{path}->{depth},
5022 $r);
5023 foreach my $de (@dirs) {
5025 my $p = $g->{path}->full_path($de);
5026 next if $exists->{$p};
5027 next if (length $g->{path}->{right} &&
5028 ($self->check_path($p, $r) !=
5029 $SVN::Node::dir));
5030 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5031 $g->{ref}->full_path($de), 1);
5032 }
5033 }
5034 foreach my $g (@$globs) {
5035 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5036 if ($path->{action} =~ /^[AR]$/) {
5037 get_dir_check($self, $exists, $g, $r);
5038 }
5039 }
5040 foreach (keys %$paths) {
5041 if (/$g->{path}->{left_regex}/ &&
5042 !/$g->{path}->{regex}/) {
5043 next if $paths->{$_}->{action} !~ /^[AR]$/;
5044 get_dir_check($self, $exists, $g, $r);
5045 }
5046 next unless /$g->{path}->{regex}/;
5047 my $p = $1;
5048 my $pathname = $g->{path}->full_path($p);
5049 next if $exists->{$pathname};
5050 next if ($self->check_path($pathname, $r) !=
5051 $SVN::Node::dir);
5052 $exists->{$pathname} = Git::SVN->init(
5053 $self->{url}, $pathname, undef,
5054 $g->{ref}->full_path($p), 1);
5055 }
5056 my $c = '';
5057 foreach (split m#/#, $g->{path}->{left}) {
5058 $c .= "/$_";
5059 next unless ($paths->{$c} &&
5060 ($paths->{$c}->{action} =~ /^[AR]$/));
5061 get_dir_check($self, $exists, $g, $r);
5062 }
5063 }
5064 values %$exists;
5065}
5066sub minimize_url {
5068 my ($self) = @_;
5069 return $self->{url} if ($self->{url} eq $self->{repos_root});
5070 my $url = $self->{repos_root};
5071 my @components = split(m!/!, $self->{svn_path});
5072 my $c = '';
5073 do {
5074 $url .= "/$c" if length $c;
5075 eval {
5076 my $ra = (ref $self)->new($url);
5077 my $latest = $ra->get_latest_revnum;
5078 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5079 };
5080 } while ($@ && ($c = shift @components));
5081 $url;
5082}
5083sub can_do_switch {
5085 my $self = shift;
5086 unless (defined $can_do_switch) {
5087 my $pool = SVN::Pool->new;
5088 my $rep = eval {
5089 $self->do_switch(1, '', 0, $self->{url},
5090 SVN::Delta::Editor->new, $pool);
5091 };
5092 if ($@) {
5093 $can_do_switch = 0;
5094 } else {
5095 $rep->abort_report($pool);
5096 $can_do_switch = 1;
5097 }
5098 $pool->clear;
5099 }
5100 $can_do_switch;
5101}
5102sub skip_unknown_revs {
5104 my ($err) = @_;
5105 my $errno = $err->apr_err();
5106 # Maybe the branch we're tracking didn't
5107 # exist when the repo started, so it's
5108 # not an error if it doesn't, just continue
5109 #
5110 # Wonderfully consistent library, eh?
5111 # 160013 - svn:// and file://
5112 # 175002 - http(s)://
5113 # 175007 - http(s):// (this repo required authorization, too...)
5114 # More codes may be discovered later...
5115 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5116 my $err_key = $err->expanded_message;
5117 # revision numbers change every time, filter them out
5118 $err_key =~ s/\d+/\0/g;
5119 $err_key = "$errno\0$err_key";
5120 unless ($ignored_err{$err_key}) {
5121 warn "W: Ignoring error from SVN, path probably ",
5122 "does not exist: ($errno): ",
5123 $err->expanded_message,"\n";
5124 warn "W: Do not be alarmed at the above message ",
5125 "git-svn is just searching aggressively for ",
5126 "old history.\n",
5127 "This may take a while on large repositories\n";
5128 $ignored_err{$err_key} = 1;
5129 }
5130 return;
5131 }
5132 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5133}
5134package Git::SVN::Log;
5136use strict;
5137use warnings;
5138use POSIX qw/strftime/;
5139use Time::Local;
5140use constant commit_log_separator => ('-' x 72) . "\n";
5141use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5142 %rusers $show_commit $incremental/;
5143my $l_fmt;
5144sub cmt_showable {
5146 my ($c) = @_;
5147 return 1 if defined $c->{r};
5148 # big commit message got truncated by the 16k pretty buffer in rev-list
5150 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5151 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5152 @{$c->{l}} = ();
5153 my @log = command(qw/cat-file commit/, $c->{c});
5154 # shift off the headers
5156 shift @log while ($log[0] ne '');
5157 shift @log;
5158 # TODO: make $c->{l} not have a trailing newline in the future
5160 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5161 (undef, $c->{r}, undef) = ::extract_metadata(
5163 (grep(/^git-svn-id: /, @log))[-1]);
5164 }
5165 return defined $c->{r};
5166}
5167sub log_use_color {
5169 return $color || Git->repository->get_colorbool('color.diff');
5170}
5171sub git_svn_log_cmd {
5173 my ($r_min, $r_max, @args) = @_;
5174 my $head = 'HEAD';
5175 my (@files, @log_opts);
5176 foreach my $x (@args) {
5177 if ($x eq '--' || @files) {
5178 push @files, $x;
5179 } else {
5180 if (::verify_ref("$x^0")) {
5181 $head = $x;
5182 } else {
5183 push @log_opts, $x;
5184 }
5185 }
5186 }
5187 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5189 $gs ||= Git::SVN->_new;
5190 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5191 $gs->refname);
5192 push @cmd, '-r' unless $non_recursive;
5193 push @cmd, qw/--raw --name-status/ if $verbose;
5194 push @cmd, '--color' if log_use_color();
5195 push @cmd, @log_opts;
5196 if (defined $r_max && $r_max == $r_min) {
5197 push @cmd, '--max-count=1';
5198 if (my $c = $gs->rev_map_get($r_max)) {
5199 push @cmd, $c;
5200 }
5201 } elsif (defined $r_max) {
5202 if ($r_max < $r_min) {
5203 ($r_min, $r_max) = ($r_max, $r_min);
5204 }
5205 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5206 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5207 # If there are no commits in the range, both $c_max and $c_min
5208 # will be undefined. If there is at least 1 commit in the
5209 # range, both will be defined.
5210 return () if !defined $c_min || !defined $c_max;
5211 if ($c_min eq $c_max) {
5212 push @cmd, '--max-count=1', $c_min;
5213 } else {
5214 push @cmd, '--boundary', "$c_min..$c_max";
5215 }
5216 }
5217 return (@cmd, @files);
5218}
5219# adapted from pager.c
5221sub config_pager {
5222 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5223 if (!defined $pager) {
5224 $pager = 'less';
5225 } elsif (length $pager == 0 || $pager eq 'cat') {
5226 $pager = undef;
5227 }
5228 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5229}
5230sub run_pager {
5232 return unless -t *STDOUT && defined $pager;
5233 pipe my ($rfd, $wfd) or return;
5234 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5235 if (!$pid) {
5236 open STDOUT, '>&', $wfd or
5237 ::fatal "Can't redirect to stdout: $!";
5238 return;
5239 }
5240 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5241 $ENV{LESS} ||= 'FRSX';
5242 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5243}
5244sub format_svn_date {
5246 # some systmes don't handle or mishandle %z, so be creative.
5247 my $t = shift || time;
5248 my $gm = timelocal(gmtime($t));
5249 my $sign = qw( + + - )[ $t <=> $gm ];
5250 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5251 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5252}
5253sub parse_git_date {
5255 my ($t, $tz) = @_;
5256 # Date::Parse isn't in the standard Perl distro :(
5257 if ($tz =~ s/^\+//) {
5258 $t += tz_to_s_offset($tz);
5259 } elsif ($tz =~ s/^\-//) {
5260 $t -= tz_to_s_offset($tz);
5261 }
5262 return $t;
5263}
5264sub set_local_timezone {
5266 if (defined $TZ) {
5267 $ENV{TZ} = $TZ;
5268 } else {
5269 delete $ENV{TZ};
5270 }
5271}
5272sub tz_to_s_offset {
5274 my ($tz) = @_;
5275 $tz =~ s/(\d\d)$//;
5276 return ($1 * 60) + ($tz * 3600);
5277}
5278sub get_author_info {
5280 my ($dest, $author, $t, $tz) = @_;
5281 $author =~ s/(?:^\s*|\s*$)//g;
5282 $dest->{a_raw} = $author;
5283 my $au;
5284 if ($::_authors) {
5285 $au = $rusers{$author} || undef;
5286 }
5287 if (!$au) {
5288 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5289 }
5290 $dest->{t} = $t;
5291 $dest->{tz} = $tz;
5292 $dest->{a} = $au;
5293 $dest->{t_utc} = parse_git_date($t, $tz);
5294}
5295sub process_commit {
5297 my ($c, $r_min, $r_max, $defer) = @_;
5298 if (defined $r_min && defined $r_max) {
5299 if ($r_min == $c->{r} && $r_min == $r_max) {
5300 show_commit($c);
5301 return 0;
5302 }
5303 return 1 if $r_min == $r_max;
5304 if ($r_min < $r_max) {
5305 # we need to reverse the print order
5306 return 0 if (defined $limit && --$limit < 0);
5307 push @$defer, $c;
5308 return 1;
5309 }
5310 if ($r_min != $r_max) {
5311 return 1 if ($r_min < $c->{r});
5312 return 1 if ($r_max > $c->{r});
5313 }
5314 }
5315 return 0 if (defined $limit && --$limit < 0);
5316 show_commit($c);
5317 return 1;
5318}
5319sub show_commit {
5321 my $c = shift;
5322 if ($oneline) {
5323 my $x = "\n";
5324 if (my $l = $c->{l}) {
5325 while ($l->[0] =~ /^\s*$/) { shift @$l }
5326 $x = $l->[0];
5327 }
5328 $l_fmt ||= 'A' . length($c->{r});
5329 print 'r',pack($l_fmt, $c->{r}),' | ';
5330 print "$c->{c} | " if $show_commit;
5331 print $x;
5332 } else {
5333 show_commit_normal($c);
5334 }
5335}
5336sub show_commit_changed_paths {
5338 my ($c) = @_;
5339 return unless $c->{changed};
5340 print "Changed paths:\n", @{$c->{changed}};
5341}
5342sub show_commit_normal {
5344 my ($c) = @_;
5345 print commit_log_separator, "r$c->{r} | ";
5346 print "$c->{c} | " if $show_commit;
5347 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5348 my $nr_line = 0;
5349 if (my $l = $c->{l}) {
5351 while ($l->[$#$l] eq "\n" && $#$l > 0
5352 && $l->[($#$l - 1)] eq "\n") {
5353 pop @$l;
5354 }
5355 $nr_line = scalar @$l;
5356 if (!$nr_line) {
5357 print "1 line\n\n\n";
5358 } else {
5359 if ($nr_line == 1) {
5360 $nr_line = '1 line';
5361 } else {
5362 $nr_line .= ' lines';
5363 }
5364 print $nr_line, "\n";
5365 show_commit_changed_paths($c);
5366 print "\n";
5367 print $_ foreach @$l;
5368 }
5369 } else {
5370 print "1 line\n";
5371 show_commit_changed_paths($c);
5372 print "\n";
5373 }
5375 foreach my $x (qw/raw stat diff/) {
5376 if ($c->{$x}) {
5377 print "\n";
5378 print $_ foreach @{$c->{$x}}
5379 }
5380 }
5381}
5382sub cmd_show_log {
5384 my (@args) = @_;
5385 my ($r_min, $r_max);
5386 my $r_last = -1; # prevent dupes
5387 set_local_timezone();
5388 if (defined $::_revision) {
5389 if ($::_revision =~ /^(\d+):(\d+)$/) {
5390 ($r_min, $r_max) = ($1, $2);
5391 } elsif ($::_revision =~ /^\d+$/) {
5392 $r_min = $r_max = $::_revision;
5393 } else {
5394 ::fatal "-r$::_revision is not supported, use ",
5395 "standard 'git log' arguments instead";
5396 }
5397 }
5398 config_pager();
5400 @args = git_svn_log_cmd($r_min, $r_max, @args);
5401 if (!@args) {
5402 print commit_log_separator unless $incremental || $oneline;
5403 return;
5404 }
5405 my $log = command_output_pipe(@args);
5406 run_pager();
5407 my (@k, $c, $d, $stat);
5408 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5409 while (<$log>) {
5410 if (/^${esc_color}commit -?($::sha1_short)/o) {
5411 my $cmt = $1;
5412 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5413 $r_last = $c->{r};
5414 process_commit($c, $r_min, $r_max, \@k) or
5415 goto out;
5416 }
5417 $d = undef;
5418 $c = { c => $cmt };
5419 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5420 get_author_info($c, $1, $2, $3);
5421 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5422 # ignore
5423 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5424 push @{$c->{raw}}, $_;
5425 } elsif (/^${esc_color}[ACRMDT]\t/) {
5426 # we could add $SVN->{svn_path} here, but that requires
5427 # remote access at the moment (repo_path_split)...
5428 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5429 push @{$c->{changed}}, $_;
5430 } elsif (/^${esc_color}diff /o) {
5431 $d = 1;
5432 push @{$c->{diff}}, $_;
5433 } elsif ($d) {
5434 push @{$c->{diff}}, $_;
5435 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5436 $esc_color*[\+\-]*$esc_color$/x) {
5437 $stat = 1;
5438 push @{$c->{stat}}, $_;
5439 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5440 push @{$c->{stat}}, $_;
5441 $stat = undef;
5442 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5443 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5444 } elsif (s/^${esc_color} //o) {
5445 push @{$c->{l}}, $_;
5446 }
5447 }
5448 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5449 $r_last = $c->{r};
5450 process_commit($c, $r_min, $r_max, \@k);
5451 }
5452 if (@k) {
5453 ($r_min, $r_max) = ($r_max, $r_min);
5454 process_commit($_, $r_min, $r_max) foreach reverse @k;
5455 }
5456out:
5457 close $log;
5458 print commit_log_separator unless $incremental || $oneline;
5459}
5460sub cmd_blame {
5462 my $path = pop;
5463 config_pager();
5465 run_pager();
5466 my ($fh, $ctx, $rev);
5468 if ($_git_format) {
5470 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5471 while (my $line = <$fh>) {
5472 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5473 # Uncommitted edits show up as a rev ID of
5474 # all zeros, which we can't look up with
5475 # cmt_metadata
5476 if ($1 !~ /^0+$/) {
5477 (undef, $rev, undef) =
5478 ::cmt_metadata($1);
5479 $rev = '0' if (!$rev);
5480 } else {
5481 $rev = '0';
5482 }
5483 $rev = sprintf('%-10s', $rev);
5484 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5485 }
5486 print $line;
5487 }
5488 } else {
5489 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5490 '--', $path);
5491 my ($sha1);
5492 my %authors;
5493 my @buffer;
5494 my %dsha; #distinct sha keys
5495 while (my $line = <$fh>) {
5497 push @buffer, $line;
5498 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5499 $dsha{$1} = 1;
5500 }
5501 }
5502 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5504 foreach my $line (@buffer) {
5506 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5507 $rev = $s2r->{$1};
5508 $rev = '0' if (!$rev)
5509 }
5510 elsif ($line =~ /^author (.*)/) {
5511 $authors{$rev} = $1;
5512 $authors{$rev} =~ s/\s/_/g;
5513 }
5514 elsif ($line =~ /^\t(.*)$/) {
5515 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5516 }
5517 }
5518 }
5519 command_close_pipe($fh, $ctx);
5520}
5521package Git::SVN::Migration;
5523# these version numbers do NOT correspond to actual version numbers
5524# of git nor git-svn. They are just relative.
5525#
5526# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5527#
5528# v1 layout: .git/$id/info/url, refs/remotes/$id
5529#
5530# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5531#
5532# v3 layout: .git/svn/$id, refs/remotes/$id
5533# - info/url may remain for backwards compatibility
5534# - this is what we migrate up to this layout automatically,
5535# - this will be used by git svn init on single branches
5536# v3.1 layout (auto migrated):
5537# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5538# for backwards compatibility
5539#
5540# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5541# - this is only created for newly multi-init-ed
5542# repositories. Similar in spirit to the
5543# --use-separate-remotes option in git-clone (now default)
5544# - we do not automatically migrate to this (following
5545# the example set by core git)
5546#
5547# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5548# - newer, more-efficient format that uses 24-bytes per record
5549# with no filler space.
5550# - use xxd -c24 < .rev_map.$UUID to view and debug
5551# - This is a one-way migration, repositories updated to the
5552# new format will not be able to use old git-svn without
5553# rebuilding the .rev_db. Rebuilding the rev_db is not
5554# possible if noMetadata or useSvmProps are set; but should
5555# be no problem for users that use the (sensible) defaults.
5556use strict;
5557use warnings;
5558use Carp qw/croak/;
5559use File::Path qw/mkpath/;
5560use File::Basename qw/dirname basename/;
5561use vars qw/$_minimize/;
5562sub migrate_from_v0 {
5564 my $git_dir = $ENV{GIT_DIR};
5565 return undef unless -d $git_dir;
5566 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5567 my $migrated = 0;
5568 while (<$fh>) {
5569 chomp;
5570 my ($id, $orig_ref) = ($_, $_);
5571 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5572 next unless -f "$git_dir/$id/info/url";
5573 my $new_ref = "refs/remotes/$id";
5574 if (::verify_ref("$new_ref^0")) {
5575 print STDERR "W: $orig_ref is probably an old ",
5576 "branch used by an ancient version of ",
5577 "git-svn.\n",
5578 "However, $new_ref also exists.\n",
5579 "We will not be able ",
5580 "to use this branch until this ",
5581 "ambiguity is resolved.\n";
5582 next;
5583 }
5584 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5585 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5586 command_noisy('update-ref', $new_ref, $orig_ref);
5587 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5588 $migrated++;
5589 }
5590 command_close_pipe($fh, $ctx);
5591 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5592 $migrated;
5593}
5594sub migrate_from_v1 {
5596 my $git_dir = $ENV{GIT_DIR};
5597 my $migrated = 0;
5598 return $migrated unless -d $git_dir;
5599 my $svn_dir = "$git_dir/svn";
5600 # just in case somebody used 'svn' as their $id at some point...
5602 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5603 print STDERR "Migrating from a git-svn v1 layout...\n";
5605 mkpath([$svn_dir]);
5606 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5607 "$svn_dir\n\t(required for this version ",
5608 "($::VERSION) of git-svn) does not exist.\n";
5609 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5610 while (<$fh>) {
5611 my $x = $_;
5612 next unless $x =~ s#^refs/remotes/##;
5613 chomp $x;
5614 next unless -f "$git_dir/$x/info/url";
5615 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5616 next unless $u;
5617 my $dn = dirname("$git_dir/svn/$x");
5618 mkpath([$dn]) unless -d $dn;
5619 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5620 mkpath(["$git_dir/svn/svn"]);
5621 print STDERR " - $git_dir/$x/info => ",
5622 "$git_dir/svn/$x/info\n";
5623 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5624 croak "$!: $x";
5625 # don't worry too much about these, they probably
5626 # don't exist with repos this old (save for index,
5627 # and we can easily regenerate that)
5628 foreach my $f (qw/unhandled.log index .rev_db/) {
5629 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5630 }
5631 } else {
5632 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5633 rename "$git_dir/$x", "$git_dir/svn/$x" or
5634 croak "$!: $x";
5635 }
5636 $migrated++;
5637 }
5638 command_close_pipe($fh, $ctx);
5639 print STDERR "Done migrating from a git-svn v1 layout\n";
5640 $migrated;
5641}
5642sub read_old_urls {
5644 my ($l_map, $pfx, $path) = @_;
5645 my @dir;
5646 foreach (<$path/*>) {
5647 if (-r "$_/info/url") {
5648 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5649 my $ref_id = $pfx . basename $_;
5650 my $url = ::file_to_s("$_/info/url");
5651 $l_map->{$ref_id} = $url;
5652 } elsif (-d $_) {
5653 push @dir, $_;
5654 }
5655 }
5656 foreach (@dir) {
5657 my $x = $_;
5658 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5659 read_old_urls($l_map, $x, $_);
5660 }
5661}
5662sub migrate_from_v2 {
5664 my @cfg = command(qw/config -l/);
5665 return if grep /^svn-remote\..+\.url=/, @cfg;
5666 my %l_map;
5667 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5668 my $migrated = 0;
5669 foreach my $ref_id (sort keys %l_map) {
5671 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5672 if ($@) {
5673 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5674 }
5675 $migrated++;
5676 }
5677 $migrated;
5678}
5679sub minimize_connections {
5681 my $r = Git::SVN::read_all_remotes();
5682 my $new_urls = {};
5683 my $root_repos = {};
5684 foreach my $repo_id (keys %$r) {
5685 my $url = $r->{$repo_id}->{url} or next;
5686 my $fetch = $r->{$repo_id}->{fetch} or next;
5687 my $ra = Git::SVN::Ra->new($url);
5688 # skip existing cases where we already connect to the root
5690 if (($ra->{url} eq $ra->{repos_root}) ||
5691 ($ra->{repos_root} eq $repo_id)) {
5692 $root_repos->{$ra->{url}} = $repo_id;
5693 next;
5694 }
5695 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5697 my $root_path = $ra->{url};
5698 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5699 foreach my $path (keys %$fetch) {
5700 my $ref_id = $fetch->{$path};
5701 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5702 # make sure we can read when connecting to
5704 # a higher level of a repository
5705 my ($last_rev, undef) = $gs->last_rev_commit;
5706 if (!defined $last_rev) {
5707 $last_rev = eval {
5708 $root_ra->get_latest_revnum;
5709 };
5710 next if $@;
5711 }
5712 my $new = $root_path;
5713 $new .= length $path ? "/$path" : '';
5714 eval {
5715 $root_ra->get_log([$new], $last_rev, $last_rev,
5716 0, 0, 1, sub { });
5717 };
5718 next if $@;
5719 $new_urls->{$ra->{repos_root}}->{$new} =
5720 { ref_id => $ref_id,
5721 old_repo_id => $repo_id,
5722 old_path => $path };
5723 }
5724 }
5725 my @emptied;
5727 foreach my $url (keys %$new_urls) {
5728 # see if we can re-use an existing [svn-remote "repo_id"]
5729 # instead of creating a(n ugly) new section:
5730 my $repo_id = $root_repos->{$url} || $url;
5731 my $fetch = $new_urls->{$url};
5733 foreach my $path (keys %$fetch) {
5734 my $x = $fetch->{$path};
5735 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5736 my $pfx = "svn-remote.$x->{old_repo_id}";
5737 my $old_fetch = quotemeta("$x->{old_path}:".
5739 "$x->{ref_id}");
5740 command_noisy(qw/config --unset/,
5741 "$pfx.fetch", '^'. $old_fetch . '$');
5742 delete $r->{$x->{old_repo_id}}->
5743 {fetch}->{$x->{old_path}};
5744 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5745 command_noisy(qw/config --unset/,
5746 "$pfx.url");
5747 push @emptied, $x->{old_repo_id}
5748 }
5749 }
5750 }
5751 if (@emptied) {
5752 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5753 print STDERR <<EOF;
5754The following [svn-remote] sections in your config file ($file) are empty
5755and can be safely removed:
5756EOF
5757 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5758 }
5759}
5760sub migration_check {
5762 migrate_from_v0();
5763 migrate_from_v1();
5764 migrate_from_v2();
5765 minimize_connections() if $_minimize;
5766}
5767package Git::IndexInfo;
5769use strict;
5770use warnings;
5771use Git qw/command_input_pipe command_close_pipe/;
5772sub new {
5774 my ($class) = @_;
5775 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5776 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5777}
5778sub remove {
5780 my ($self, $path) = @_;
5781 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5782 return ++$self->{nr};
5783 }
5784 undef;
5785}
5786sub update {
5788 my ($self, $mode, $hash, $path) = @_;
5789 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5790 return ++$self->{nr};
5791 }
5792 undef;
5793}
5794sub DESTROY {
5796 my ($self) = @_;
5797 command_close_pipe($self->{gui}, $self->{ctx});
5798}
5799package Git::SVN::GlobSpec;
5801use strict;
5802use warnings;
5803sub new {
5805 my ($class, $glob) = @_;
5806 my $re = $glob;
5807 $re =~ s!/+$!!g; # no need for trailing slashes
5808 $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5809 my $temp = $re;
5810 my ($left, $right) = ($1, $3);
5811 $re = $2;
5812 my $depth = $re =~ tr/*/*/;
5813 if ($depth != $temp =~ tr/*/*/) {
5814 die "Only one set of wildcard directories " .
5815 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5816 }
5817 if ($depth == 0) {
5818 die "One '*' is needed for glob: '$glob'\n";
5819 }
5820 $re =~ s!\*!\[^/\]*!g;
5821 $re = quotemeta($left) . "($re)" . quotemeta($right);
5822 if (length $left && !($left =~ s!/+$!!g)) {
5823 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5824 }
5825 if (length $right && !($right =~ s!^/+!!g)) {
5826 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5827 }
5828 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5829 bless { left => $left, right => $right, left_regex => $left_re,
5830 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5831}
5832sub full_path {
5834 my ($self, $path) = @_;
5835 return (length $self->{left} ? "$self->{left}/" : '') .
5836 $path . (length $self->{right} ? "/$self->{right}" : '');
5837}
5838__END__
5840Data structures:
5842$remotes = { # returned by read_all_remotes()
5845 'svn' => {
5846 # svn-remote.svn.url=https://svn.musicpd.org
5847 url => 'https://svn.musicpd.org',
5848 # svn-remote.svn.fetch=mpd/trunk:trunk
5849 fetch => {
5850 'mpd/trunk' => 'trunk',
5851 },
5852 # svn-remote.svn.tags=mpd/tags/*:tags/*
5853 tags => {
5854 path => {
5855 left => 'mpd/tags',
5856 right => '',
5857 regex => qr!mpd/tags/([^/]+)$!,
5858 glob => 'tags/*',
5859 },
5860 ref => {
5861 left => 'tags',
5862 right => '',
5863 regex => qr!tags/([^/]+)$!,
5864 glob => 'tags/*',
5865 },
5866 }
5867 }
5868};
5869$log_entry hashref as returned by libsvn_log_entry()
5871{
5872 log => 'whitespace-formatted log entry
5873', # trailing newline is preserved
5874 revision => '8', # integer
5875 date => '2004-02-24T17:01:44.108345Z', # commit date
5876 author => 'committer name'
5877};
5878# this is generated by generate_diff();
5881@mods = array of diff-index line hashes, each element represents one line
5882 of diff-index output
5883diff-index line ($m hash)
5885{
5886 mode_a => first column of diff-index output, no leading ':',
5887 mode_b => second column of diff-index output,
5888 sha1_b => sha1sum of the final blob,
5889 chg => change type [MCRADT],
5890 file_a => original file name of a file (iff chg is 'C' or 'R')
5891 file_b => new/current file name of a file (any chg)
5892}
5893;
5894# retval of read_url_paths{,_all}();
5896$l_map = {
5897 # repository root url
5898 'https://svn.musicpd.org' => {
5899 # repository path # GIT_SVN_ID
5900 'mpd/trunk' => 'trunk',
5901 'mpd/tags/0.11.5' => 'tags/0.11.5',
5902 },
5903}
5904Notes:
5906 I don't trust the each() function on unless I created %hash myself
5907 because the internal iterator may not have started at base.