8d2ef3d16f32d89a0d110a4539f9460436e0ce3b
1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18$ENV{GIT_DIR} ||= '.git';
19$Git::SVN::default_repo_id = 'svn';
20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21$Git::SVN::Ra::_log_window_size = 100;
22$Git::SVN::_minimize_url = 'unset';
23if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
30 }
31 }
32}
33$Git::SVN::Log::TZ = $ENV{TZ};
35$ENV{TZ} = 'UTC';
36$| = 1; # unbuffer STDOUT
37sub fatal (@) { print STDERR "@_\n"; exit 1 }
39sub _req_svn {
40 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
41 require SVN::Ra;
42 require SVN::Delta;
43 if ($SVN::Core::VERSION lt '1.1.0') {
44 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
45 }
46}
47my $can_compress = eval { require Compress::Zlib; 1};
48push @Git::SVN::Ra::ISA, 'SVN::Ra';
49push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
50push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
51use Carp qw/croak/;
52use Digest::MD5;
53use IO::File qw//;
54use File::Basename qw/dirname basename/;
55use File::Path qw/mkpath/;
56use File::Spec;
57use File::Find;
58use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
59use IPC::Open3;
60use Git;
61BEGIN {
63 # import functions from Git into our packages, en masse
64 no strict 'refs';
65 foreach (qw/command command_oneline command_noisy command_output_pipe
66 command_input_pipe command_close_pipe
67 command_bidi_pipe command_close_bidi_pipe/) {
68 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
69 Git::SVN::Migration Git::SVN::Log Git::SVN),
70 __PACKAGE__) {
71 *{"${package}::$_"} = \&{"Git::$_"};
72 }
73 }
74}
75my ($SVN);
77$sha1 = qr/[a-f\d]{40}/;
79$sha1_short = qr/[a-f\d]{4,40}/;
80my ($_stdin, $_help, $_edit,
81 $_message, $_file, $_branch_dest,
82 $_template, $_shared,
83 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
84 $_merge, $_strategy, $_dry_run, $_local,
85 $_prefix, $_no_checkout, $_url, $_verbose,
86 $_git_format, $_commit_url, $_tag);
87$Git::SVN::_follow_parent = 1;
88$_q ||= 0;
89my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
90 'config-dir=s' => \$Git::SVN::Ra::config_dir,
91 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
92 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
93my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
94 'authors-file|A=s' => \$_authors,
95 'authors-prog=s' => \$_authors_prog,
96 'repack:i' => \$Git::SVN::_repack,
97 'noMetadata' => \$Git::SVN::_no_metadata,
98 'useSvmProps' => \$Git::SVN::_use_svm_props,
99 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
100 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
101 'no-checkout' => \$_no_checkout,
102 'quiet|q+' => \$_q,
103 'repack-flags|repack-args|repack-opts=s' =>
104 \$Git::SVN::_repack_flags,
105 'use-log-author' => \$Git::SVN::_use_log_author,
106 'add-author-from' => \$Git::SVN::_add_author_from,
107 'localtime' => \$Git::SVN::_localtime,
108 %remote_opts );
109my ($_trunk, @_tags, @_branches, $_stdlayout);
111my %icv;
112my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
113 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
114 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
115 'stdlayout|s' => \$_stdlayout,
116 'minimize-url|m!' => \$Git::SVN::_minimize_url,
117 'no-metadata' => sub { $icv{noMetadata} = 1 },
118 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
119 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
120 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
121 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
122 %remote_opts );
123my %cmt_opts = ( 'edit|e' => \$_edit,
124 'rmdir' => \$SVN::Git::Editor::_rmdir,
125 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
126 'l=i' => \$SVN::Git::Editor::_rename_limit,
127 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
128);
129my %cmd = (
131 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
132 { 'revision|r=s' => \$_revision,
133 'fetch-all|all' => \$_fetch_all,
134 'parent|p' => \$_fetch_parent,
135 %fc_opts } ],
136 clone => [ \&cmd_clone, "Initialize and fetch revisions",
137 { 'revision|r=s' => \$_revision,
138 %fc_opts, %init_opts } ],
139 init => [ \&cmd_init, "Initialize a repo for tracking" .
140 " (requires URL argument)",
141 \%init_opts ],
142 'multi-init' => [ \&cmd_multi_init,
143 "Deprecated alias for ".
144 "'$0 init -T<trunk> -b<branches> -t<tags>'",
145 \%init_opts ],
146 dcommit => [ \&cmd_dcommit,
147 'Commit several diffs to merge with upstream',
148 { 'merge|m|M' => \$_merge,
149 'strategy|s=s' => \$_strategy,
150 'verbose|v' => \$_verbose,
151 'dry-run|n' => \$_dry_run,
152 'fetch-all|all' => \$_fetch_all,
153 'commit-url=s' => \$_commit_url,
154 'revision|r=i' => \$_revision,
155 'no-rebase' => \$_no_rebase,
156 %cmt_opts, %fc_opts } ],
157 branch => [ \&cmd_branch,
158 'Create a branch in the SVN repository',
159 { 'message|m=s' => \$_message,
160 'destination|d=s' => \$_branch_dest,
161 'dry-run|n' => \$_dry_run,
162 'tag|t' => \$_tag,
163 'username=s' => \$Git::SVN::Prompt::_username,
164 'commit-url=s' => \$_commit_url } ],
165 tag => [ sub { $_tag = 1; cmd_branch(@_) },
166 'Create a tag in the SVN repository',
167 { 'message|m=s' => \$_message,
168 'destination|d=s' => \$_branch_dest,
169 'dry-run|n' => \$_dry_run,
170 'username=s' => \$Git::SVN::Prompt::_username,
171 'commit-url=s' => \$_commit_url } ],
172 'set-tree' => [ \&cmd_set_tree,
173 "Set an SVN repository to a git tree-ish",
174 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
175 'create-ignore' => [ \&cmd_create_ignore,
176 'Create a .gitignore per svn:ignore',
177 { 'revision|r=i' => \$_revision
178 } ],
179 'mkdirs' => [ \&cmd_mkdirs ,
180 "recreate empty directories after a checkout",
181 { 'revision|r=i' => \$_revision } ],
182 'propget' => [ \&cmd_propget,
183 'Print the value of a property on a file or directory',
184 { 'revision|r=i' => \$_revision } ],
185 'proplist' => [ \&cmd_proplist,
186 'List all properties of a file or directory',
187 { 'revision|r=i' => \$_revision } ],
188 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
189 { 'revision|r=i' => \$_revision
190 } ],
191 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
192 { 'revision|r=i' => \$_revision
193 } ],
194 'multi-fetch' => [ \&cmd_multi_fetch,
195 "Deprecated alias for $0 fetch --all",
196 { 'revision|r=s' => \$_revision, %fc_opts } ],
197 'migrate' => [ sub { },
198 # no-op, we automatically run this anyways,
199 'Migrate configuration/metadata/layout from
200 previous versions of git-svn',
201 { 'minimize' => \$Git::SVN::Migration::_minimize,
202 %remote_opts } ],
203 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
204 { 'limit=i' => \$Git::SVN::Log::limit,
205 'revision|r=s' => \$_revision,
206 'verbose|v' => \$Git::SVN::Log::verbose,
207 'incremental' => \$Git::SVN::Log::incremental,
208 'oneline' => \$Git::SVN::Log::oneline,
209 'show-commit' => \$Git::SVN::Log::show_commit,
210 'non-recursive' => \$Git::SVN::Log::non_recursive,
211 'authors-file|A=s' => \$_authors,
212 'color' => \$Git::SVN::Log::color,
213 'pager=s' => \$Git::SVN::Log::pager
214 } ],
215 'find-rev' => [ \&cmd_find_rev,
216 "Translate between SVN revision numbers and tree-ish",
217 {} ],
218 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
219 { 'merge|m|M' => \$_merge,
220 'verbose|v' => \$_verbose,
221 'strategy|s=s' => \$_strategy,
222 'local|l' => \$_local,
223 'fetch-all|all' => \$_fetch_all,
224 'dry-run|n' => \$_dry_run,
225 %fc_opts } ],
226 'commit-diff' => [ \&cmd_commit_diff,
227 'Commit a diff between two trees',
228 { 'message|m=s' => \$_message,
229 'file|F=s' => \$_file,
230 'revision|r=s' => \$_revision,
231 %cmt_opts } ],
232 'info' => [ \&cmd_info,
233 "Show info about the latest SVN revision
234 on the current branch",
235 { 'url' => \$_url, } ],
236 'blame' => [ \&Git::SVN::Log::cmd_blame,
237 "Show what revision and author last modified each line of a file",
238 { 'git-format' => \$_git_format } ],
239 'reset' => [ \&cmd_reset,
240 "Undo fetches back to the specified SVN revision",
241 { 'revision|r=s' => \$_revision,
242 'parent|p' => \$_fetch_parent } ],
243 'gc' => [ \&cmd_gc,
244 "Compress unhandled.log files in .git/svn and remove " .
245 "index files in .git/svn",
246 {} ],
247);
248my $cmd;
250for (my $i = 0; $i < @ARGV; $i++) {
251 if (defined $cmd{$ARGV[$i]}) {
252 $cmd = $ARGV[$i];
253 splice @ARGV, $i, 1;
254 last;
255 } elsif ($ARGV[$i] eq 'help') {
256 $cmd = $ARGV[$i+1];
257 usage(0);
258 }
259};
260# make sure we're always running at the top-level working directory
262unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
263 unless (-d $ENV{GIT_DIR}) {
264 if ($git_dir_user_set) {
265 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
266 "but it is not a directory\n";
267 }
268 my $git_dir = delete $ENV{GIT_DIR};
269 my $cdup = undef;
270 git_cmd_try {
271 $cdup = command_oneline(qw/rev-parse --show-cdup/);
272 $git_dir = '.' unless ($cdup);
273 chomp $cdup if ($cdup);
274 $cdup = "." unless ($cdup && length $cdup);
275 } "Already at toplevel, but $git_dir not found\n";
276 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
277 unless (-d $git_dir) {
278 die "$git_dir still not found after going to ",
279 "'$cdup'\n";
280 }
281 $ENV{GIT_DIR} = $git_dir;
282 }
283 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
284}
285my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
287read_git_config(\%opts);
289if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
290 Getopt::Long::Configure('pass_through');
291}
292my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
293 'minimize-connections' => \$Git::SVN::Migration::_minimize,
294 'id|i=s' => \$Git::SVN::default_ref_id,
295 'svn-remote|remote|R=s' => sub {
296 $Git::SVN::no_reuse_existing = 1;
297 $Git::SVN::default_repo_id = $_[1] });
298exit 1 if (!$rv && $cmd && $cmd ne 'log');
299usage(0) if $_help;
301version() if $_version;
302usage(1) unless defined $cmd;
303load_authors() if $_authors;
304if (defined $_authors_prog) {
305 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
306}
307unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
309 Git::SVN::Migration::migration_check();
310}
311Git::SVN::init_vars();
312eval {
313 Git::SVN::verify_remotes_sanity();
314 $cmd{$cmd}->[0]->(@ARGV);
315};
316fatal $@ if $@;
317post_fetch_checkout();
318exit 0;
319####################### primary functions ######################
321sub usage {
322 my $exit = shift || 0;
323 my $fd = $exit ? \*STDERR : \*STDOUT;
324 print $fd <<"";
325git-svn - bidirectional operations between a single Subversion tree and git
326Usage: git svn <command> [options] [arguments]\n
327 print $fd "Available commands:\n" unless $cmd;
329 foreach (sort keys %cmd) {
331 next if $cmd && $cmd ne $_;
332 next if /^multi-/; # don't show deprecated commands
333 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
334 foreach (sort keys %{$cmd{$_}->[2]}) {
335 # mixed-case options are for .git/config only
336 next if /[A-Z]/ && /^[a-z]+$/i;
337 # prints out arguments as they should be passed:
338 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
339 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
340 "--$_" : "-$_" }
341 split /\|/,$_)," $x\n";
342 }
343 }
344 print $fd <<"";
345\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
346arbitrary identifier if you're tracking multiple SVN branches/repositories in
347one git repository and want to keep them separate. See git-svn(1) for more
348information.
349 exit $exit;
351}
352sub version {
354 ::_req_svn();
355 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
356 exit 0;
357}
358sub do_git_init_db {
360 unless (-d $ENV{GIT_DIR}) {
361 my @init_db = ('init');
362 push @init_db, "--template=$_template" if defined $_template;
363 if (defined $_shared) {
364 if ($_shared =~ /[a-z]/) {
365 push @init_db, "--shared=$_shared";
366 } else {
367 push @init_db, "--shared";
368 }
369 }
370 command_noisy(@init_db);
371 $_repository = Git->repository(Repository => ".git");
372 }
373 my $set;
374 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
375 foreach my $i (keys %icv) {
376 die "'$set' and '$i' cannot both be set\n" if $set;
377 next unless defined $icv{$i};
378 command_noisy('config', "$pfx.$i", $icv{$i});
379 $set = $i;
380 }
381 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
382 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
383 if defined $$ignore_regex;
384}
385sub init_subdir {
387 my $repo_path = shift or return;
388 mkpath([$repo_path]) unless -d $repo_path;
389 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
390 $ENV{GIT_DIR} = '.git';
391 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
392}
393sub cmd_clone {
395 my ($url, $path) = @_;
396 if (!defined $path &&
397 (defined $_trunk || @_branches || @_tags ||
398 defined $_stdlayout) &&
399 $url !~ m#^[a-z\+]+://#) {
400 $path = $url;
401 }
402 $path = basename($url) if !defined $path || !length $path;
403 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
404 cmd_init($url, $path);
405 command_oneline('config', 'svn.authorsfile', $authors_absolute)
406 if $_authors;
407 Git::SVN::fetch_all($Git::SVN::default_repo_id);
408}
409sub cmd_init {
411 if (defined $_stdlayout) {
412 $_trunk = 'trunk' if (!defined $_trunk);
413 @_tags = 'tags' if (! @_tags);
414 @_branches = 'branches' if (! @_branches);
415 }
416 if (defined $_trunk || @_branches || @_tags) {
417 return cmd_multi_init(@_);
418 }
419 my $url = shift or die "SVN repository location required ",
420 "as a command-line argument\n";
421 $url = canonicalize_url($url);
422 init_subdir(@_);
423 do_git_init_db();
424 if ($Git::SVN::_minimize_url eq 'unset') {
426 $Git::SVN::_minimize_url = 0;
427 }
428 Git::SVN->init($url);
430}
431sub cmd_fetch {
433 if (grep /^\d+=./, @_) {
434 die "'<rev>=<commit>' fetch arguments are ",
435 "no longer supported.\n";
436 }
437 my ($remote) = @_;
438 if (@_ > 1) {
439 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
440 }
441 $Git::SVN::no_reuse_existing = undef;
442 if ($_fetch_parent) {
443 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
444 unless ($gs) {
445 die "Unable to determine upstream SVN information from ",
446 "working tree history\n";
447 }
448 # just fetch, don't checkout.
449 $_no_checkout = 'true';
450 $_fetch_all ? $gs->fetch_all : $gs->fetch;
451 } elsif ($_fetch_all) {
452 cmd_multi_fetch();
453 } else {
454 $remote ||= $Git::SVN::default_repo_id;
455 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
456 }
457}
458sub cmd_set_tree {
460 my (@commits) = @_;
461 if ($_stdin || !@commits) {
462 print "Reading from stdin...\n";
463 @commits = ();
464 while (<STDIN>) {
465 if (/\b($sha1_short)\b/o) {
466 unshift @commits, $1;
467 }
468 }
469 }
470 my @revs;
471 foreach my $c (@commits) {
472 my @tmp = command('rev-parse',$c);
473 if (scalar @tmp == 1) {
474 push @revs, $tmp[0];
475 } elsif (scalar @tmp > 1) {
476 push @revs, reverse(command('rev-list',@tmp));
477 } else {
478 fatal "Failed to rev-parse $c";
479 }
480 }
481 my $gs = Git::SVN->new;
482 my ($r_last, $cmt_last) = $gs->last_rev_commit;
483 $gs->fetch;
484 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
485 fatal "There are new revisions that were fetched ",
486 "and need to be merged (or acknowledged) ",
487 "before committing.\nlast rev: $r_last\n",
488 " current: $gs->{last_rev}";
489 }
490 $gs->set_tree($_) foreach @revs;
491 print "Done committing ",scalar @revs," revisions to SVN\n";
492 unlink $gs->{index};
493}
494sub cmd_dcommit {
496 my $head = shift;
497 command_noisy(qw/update-index --refresh/);
498 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
499 'Cannot dcommit with a dirty index. Commit your changes first, '
500 . "or stash them with `git stash'.\n";
501 $head ||= 'HEAD';
502 my $old_head;
504 if ($head ne 'HEAD') {
505 $old_head = eval {
506 command_oneline([qw/symbolic-ref -q HEAD/])
507 };
508 if ($old_head) {
509 $old_head =~ s{^refs/heads/}{};
510 } else {
511 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
512 }
513 command(['checkout', $head], STDERR => 0);
514 }
515 my @refs;
517 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
518 unless ($gs) {
519 die "Unable to determine upstream SVN information from ",
520 "$head history.\nPerhaps the repository is empty.";
521 }
522 if (defined $_commit_url) {
524 $url = $_commit_url;
525 } else {
526 $url = eval { command_oneline('config', '--get',
527 "svn-remote.$gs->{repo_id}.commiturl") };
528 if (!$url) {
529 $url = $gs->full_url
530 }
531 }
532 my $last_rev = $_revision if defined $_revision;
534 if ($url) {
535 print "Committing to $url ...\n";
536 }
537 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
538 if ($_no_rebase && scalar(@$linear_refs) > 1) {
539 warn "Attempting to commit more than one change while ",
540 "--no-rebase is enabled.\n",
541 "If these changes depend on each other, re-running ",
542 "without --no-rebase may be required."
543 }
544 my $expect_url = $url;
545 Git::SVN::remove_username($expect_url);
546 while (1) {
547 my $d = shift @$linear_refs or last;
548 unless (defined $last_rev) {
549 (undef, $last_rev, undef) = cmt_metadata("$d~1");
550 unless (defined $last_rev) {
551 fatal "Unable to extract revision information ",
552 "from commit $d~1";
553 }
554 }
555 if ($_dry_run) {
556 print "diff-tree $d~1 $d\n";
557 } else {
558 my $cmt_rev;
559 my %ed_opts = ( r => $last_rev,
560 log => get_commit_entry($d)->{log},
561 ra => Git::SVN::Ra->new($url),
562 config => SVN::Core::config_get_config(
563 $Git::SVN::Ra::config_dir
564 ),
565 tree_a => "$d~1",
566 tree_b => $d,
567 editor_cb => sub {
568 print "Committed r$_[0]\n";
569 $cmt_rev = $_[0];
570 },
571 svn_path => '');
572 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
573 print "No changes\n$d~1 == $d\n";
574 } elsif ($parents->{$d} && @{$parents->{$d}}) {
575 $gs->{inject_parents_dcommit}->{$cmt_rev} =
576 $parents->{$d};
577 }
578 $_fetch_all ? $gs->fetch_all : $gs->fetch;
579 $last_rev = $cmt_rev;
580 next if $_no_rebase;
581 # we always want to rebase against the current HEAD,
583 # not any head that was passed to us
584 my @diff = command('diff-tree', $d,
585 $gs->refname, '--');
586 my @finish;
587 if (@diff) {
588 @finish = rebase_cmd();
589 print STDERR "W: $d and ", $gs->refname,
590 " differ, using @finish:\n",
591 join("\n", @diff), "\n";
592 } else {
593 print "No changes between current HEAD and ",
594 $gs->refname,
595 "\nResetting to the latest ",
596 $gs->refname, "\n";
597 @finish = qw/reset --mixed/;
598 }
599 command_noisy(@finish, $gs->refname);
600 if (@diff) {
601 @refs = ();
602 my ($url_, $rev_, $uuid_, $gs_) =
603 working_head_info('HEAD', \@refs);
604 my ($linear_refs_, $parents_) =
605 linearize_history($gs_, \@refs);
606 if (scalar(@$linear_refs) !=
607 scalar(@$linear_refs_)) {
608 fatal "# of revisions changed ",
609 "\nbefore:\n",
610 join("\n", @$linear_refs),
611 "\n\nafter:\n",
612 join("\n", @$linear_refs_), "\n",
613 'If you are attempting to commit ',
614 "merges, try running:\n\t",
615 'git rebase --interactive',
616 '--preserve-merges ',
617 $gs->refname,
618 "\nBefore dcommitting";
619 }
620 if ($url_ ne $expect_url) {
621 if ($url_ eq $gs->metadata_url) {
622 print
623 "Accepting rewritten URL:",
624 " $url_\n";
625 } else {
626 fatal
627 "URL mismatch after rebase:",
628 " $url_ != $expect_url";
629 }
630 }
631 if ($uuid_ ne $uuid) {
632 fatal "uuid mismatch after rebase: ",
633 "$uuid_ != $uuid";
634 }
635 # remap parents
636 my (%p, @l, $i);
637 for ($i = 0; $i < scalar @$linear_refs; $i++) {
638 my $new = $linear_refs_->[$i] or next;
639 $p{$new} =
640 $parents->{$linear_refs->[$i]};
641 push @l, $new;
642 }
643 $parents = \%p;
644 $linear_refs = \@l;
645 }
646 }
647 }
648 if ($old_head) {
650 my $new_head = command_oneline(qw/rev-parse HEAD/);
651 my $new_is_symbolic = eval {
652 command_oneline(qw/symbolic-ref -q HEAD/);
653 };
654 if ($new_is_symbolic) {
655 print "dcommitted the branch ", $head, "\n";
656 } else {
657 print "dcommitted on a detached HEAD because you gave ",
658 "a revision argument.\n",
659 "The rewritten commit is: ", $new_head, "\n";
660 }
661 command(['checkout', $old_head], STDERR => 0);
662 }
663 unlink $gs->{index};
665}
666sub cmd_branch {
668 my ($branch_name, $head) = @_;
669 unless (defined $branch_name && length $branch_name) {
671 die(($_tag ? "tag" : "branch") . " name required\n");
672 }
673 $head ||= 'HEAD';
674 my (undef, $rev, undef, $gs) = working_head_info($head);
676 my $src = $gs->full_url;
677 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
679 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
680 my $glob;
681 if ($#{$allglobs} == 0) {
682 $glob = $allglobs->[0];
683 } else {
684 unless(defined $_branch_dest) {
685 die "Multiple ",
686 $_tag ? "tag" : "branch",
687 " paths defined for Subversion repository.\n",
688 "You must specify where you want to create the ",
689 $_tag ? "tag" : "branch",
690 " with the --destination argument.\n";
691 }
692 foreach my $g (@{$allglobs}) {
693 # SVN::Git::Editor could probably be moved to Git.pm..
694 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
695 if ($_branch_dest =~ /$re/) {
696 $glob = $g;
697 last;
698 }
699 }
700 unless (defined $glob) {
701 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
702 foreach my $g (@{$allglobs}) {
703 $g->{path}->{left} =~ /$dest_re/ or next;
704 if (defined $glob) {
705 die "Ambiguous destination: ",
706 $_branch_dest, "\nmatches both '",
707 $glob->{path}->{left}, "' and '",
708 $g->{path}->{left}, "'\n";
709 }
710 $glob = $g;
711 }
712 unless (defined $glob) {
713 die "Unknown ",
714 $_tag ? "tag" : "branch",
715 " destination $_branch_dest\n";
716 }
717 }
718 }
719 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
720 my $url;
721 if (defined $_commit_url) {
722 $url = $_commit_url;
723 } else {
724 $url = eval { command_oneline('config', '--get',
725 "svn-remote.$gs->{repo_id}.commiturl") };
726 if (!$url) {
727 $url = $remote->{url};
728 }
729 }
730 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
731 if ($dst =~ /^https:/ && $src =~ /^http:/) {
733 $src=~s/^http:/https:/;
734 }
735 ::_req_svn();
737 my $ctx = SVN::Client->new(
739 auth => Git::SVN::Ra::_auth_providers(),
740 log_msg => sub {
741 ${ $_[0] } = defined $_message
742 ? $_message
743 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
744 . $branch_name;
745 },
746 );
747 eval {
749 $ctx->ls($dst, 'HEAD', 0);
750 } and die "branch ${branch_name} already exists\n";
751 print "Copying ${src} at r${rev} to ${dst}...\n";
753 $ctx->copy($src, $rev, $dst)
754 unless $_dry_run;
755 $gs->fetch_all;
757}
758sub cmd_find_rev {
760 my $revision_or_hash = shift or die "SVN or git revision required ",
761 "as a command-line argument\n";
762 my $result;
763 if ($revision_or_hash =~ /^r\d+$/) {
764 my $head = shift;
765 $head ||= 'HEAD';
766 my @refs;
767 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
768 unless ($gs) {
769 die "Unable to determine upstream SVN information from ",
770 "$head history\n";
771 }
772 my $desired_revision = substr($revision_or_hash, 1);
773 $result = $gs->rev_map_get($desired_revision, $uuid);
774 } else {
775 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
776 $result = $rev;
777 }
778 print "$result\n" if $result;
779}
780sub cmd_rebase {
782 command_noisy(qw/update-index --refresh/);
783 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
784 unless ($gs) {
785 die "Unable to determine upstream SVN information from ",
786 "working tree history\n";
787 }
788 if ($_dry_run) {
789 print "Remote Branch: " . $gs->refname . "\n";
790 print "SVN URL: " . $url . "\n";
791 return;
792 }
793 if (command(qw/diff-index HEAD --/)) {
794 print STDERR "Cannot rebase with uncommited changes:\n";
795 command_noisy('status');
796 exit 1;
797 }
798 unless ($_local) {
799 # rebase will checkout for us, so no need to do it explicitly
800 $_no_checkout = 'true';
801 $_fetch_all ? $gs->fetch_all : $gs->fetch;
802 }
803 command_noisy(rebase_cmd(), $gs->refname);
804 $gs->mkemptydirs;
805}
806sub cmd_show_ignore {
808 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
809 $gs ||= Git::SVN->new;
810 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
811 $gs->prop_walk($gs->{path}, $r, sub {
812 my ($gs, $path, $props) = @_;
813 print STDOUT "\n# $path\n";
814 my $s = $props->{'svn:ignore'} or return;
815 $s =~ s/[\r\n]+/\n/g;
816 $s =~ s/^\n+//;
817 chomp $s;
818 $s =~ s#^#$path#gm;
819 print STDOUT "$s\n";
820 });
821}
822sub cmd_show_externals {
824 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
825 $gs ||= Git::SVN->new;
826 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
827 $gs->prop_walk($gs->{path}, $r, sub {
828 my ($gs, $path, $props) = @_;
829 print STDOUT "\n# $path\n";
830 my $s = $props->{'svn:externals'} or return;
831 $s =~ s/[\r\n]+/\n/g;
832 chomp $s;
833 $s =~ s#^#$path#gm;
834 print STDOUT "$s\n";
835 });
836}
837sub cmd_create_ignore {
839 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
840 $gs ||= Git::SVN->new;
841 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
842 $gs->prop_walk($gs->{path}, $r, sub {
843 my ($gs, $path, $props) = @_;
844 # $path is of the form /path/to/dir/
845 $path = '.' . $path;
846 # SVN can have attributes on empty directories,
847 # which git won't track
848 mkpath([$path]) unless -d $path;
849 my $ignore = $path . '.gitignore';
850 my $s = $props->{'svn:ignore'} or return;
851 open(GITIGNORE, '>', $ignore)
852 or fatal("Failed to open `$ignore' for writing: $!");
853 $s =~ s/[\r\n]+/\n/g;
854 $s =~ s/^\n+//;
855 chomp $s;
856 # Prefix all patterns so that the ignore doesn't apply
857 # to sub-directories.
858 $s =~ s#^#/#gm;
859 print GITIGNORE "$s\n";
860 close(GITIGNORE)
861 or fatal("Failed to close `$ignore': $!");
862 command_noisy('add', '-f', $ignore);
863 });
864}
865sub cmd_mkdirs {
867 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
868 $gs ||= Git::SVN->new;
869 $gs->mkemptydirs($_revision);
870}
871sub canonicalize_path {
873 my ($path) = @_;
874 my $dot_slash_added = 0;
875 if (substr($path, 0, 1) ne "/") {
876 $path = "./" . $path;
877 $dot_slash_added = 1;
878 }
879 # File::Spec->canonpath doesn't collapse x/../y into y (for a
880 # good reason), so let's do this manually.
881 $path =~ s#/+#/#g;
882 $path =~ s#/\.(?:/|$)#/#g;
883 $path =~ s#/[^/]+/\.\.##g;
884 $path =~ s#/$##g;
885 $path =~ s#^\./## if $dot_slash_added;
886 $path =~ s#^/##;
887 $path =~ s#^\.$##;
888 return $path;
889}
890sub canonicalize_url {
892 my ($url) = @_;
893 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
894 return $url;
895}
896# get_svnprops(PATH)
898# ------------------
899# Helper for cmd_propget and cmd_proplist below.
900sub get_svnprops {
901 my $path = shift;
902 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
903 $gs ||= Git::SVN->new;
904 # prefix THE PATH by the sub-directory from which the user
906 # invoked us.
907 $path = $cmd_dir_prefix . $path;
908 fatal("No such file or directory: $path") unless -e $path;
909 my $is_dir = -d $path ? 1 : 0;
910 $path = $gs->{path} . '/' . $path;
911 # canonicalize the path (otherwise libsvn will abort or fail to
913 # find the file)
914 $path = canonicalize_path($path);
915 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
917 my $props;
918 if ($is_dir) {
919 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
920 }
921 else {
922 (undef, $props) = $gs->ra->get_file($path, $r, undef);
923 }
924 return $props;
925}
926# cmd_propget (PROP, PATH)
928# ------------------------
929# Print the SVN property PROP for PATH.
930sub cmd_propget {
931 my ($prop, $path) = @_;
932 $path = '.' if not defined $path;
933 usage(1) if not defined $prop;
934 my $props = get_svnprops($path);
935 if (not defined $props->{$prop}) {
936 fatal("`$path' does not have a `$prop' SVN property.");
937 }
938 print $props->{$prop} . "\n";
939}
940# cmd_proplist (PATH)
942# -------------------
943# Print the list of SVN properties for PATH.
944sub cmd_proplist {
945 my $path = shift;
946 $path = '.' if not defined $path;
947 my $props = get_svnprops($path);
948 print "Properties on '$path':\n";
949 foreach (sort keys %{$props}) {
950 print " $_\n";
951 }
952}
953sub cmd_multi_init {
955 my $url = shift;
956 unless (defined $_trunk || @_branches || @_tags) {
957 usage(1);
958 }
959 $_prefix = '' unless defined $_prefix;
961 if (defined $url) {
962 $url = canonicalize_url($url);
963 init_subdir(@_);
964 }
965 do_git_init_db();
966 if (defined $_trunk) {
967 $_trunk =~ s#^/+##;
968 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
969 # try both old-style and new-style lookups:
970 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
971 unless ($gs_trunk) {
972 my ($trunk_url, $trunk_path) =
973 complete_svn_url($url, $_trunk);
974 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
975 undef, $trunk_ref);
976 }
977 }
978 return unless @_branches || @_tags;
979 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
980 foreach my $path (@_branches) {
981 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
982 }
983 foreach my $path (@_tags) {
984 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
985 }
986}
987sub cmd_multi_fetch {
989 $Git::SVN::no_reuse_existing = undef;
990 my $remotes = Git::SVN::read_all_remotes();
991 foreach my $repo_id (sort keys %$remotes) {
992 if ($remotes->{$repo_id}->{url}) {
993 Git::SVN::fetch_all($repo_id, $remotes);
994 }
995 }
996}
997# this command is special because it requires no metadata
999sub cmd_commit_diff {
1000 my ($ta, $tb, $url) = @_;
1001 my $usage = "Usage: $0 commit-diff -r<revision> ".
1002 "<tree-ish> <tree-ish> [<URL>]";
1003 fatal($usage) if (!defined $ta || !defined $tb);
1004 my $svn_path = '';
1005 if (!defined $url) {
1006 my $gs = eval { Git::SVN->new };
1007 if (!$gs) {
1008 fatal("Needed URL or usable git-svn --id in ",
1009 "the command-line\n", $usage);
1010 }
1011 $url = $gs->{url};
1012 $svn_path = $gs->{path};
1013 }
1014 unless (defined $_revision) {
1015 fatal("-r|--revision is a required argument\n", $usage);
1016 }
1017 if (defined $_message && defined $_file) {
1018 fatal("Both --message/-m and --file/-F specified ",
1019 "for the commit message.\n",
1020 "I have no idea what you mean");
1021 }
1022 if (defined $_file) {
1023 $_message = file_to_s($_file);
1024 } else {
1025 $_message ||= get_commit_entry($tb)->{log};
1026 }
1027 my $ra ||= Git::SVN::Ra->new($url);
1028 my $r = $_revision;
1029 if ($r eq 'HEAD') {
1030 $r = $ra->get_latest_revnum;
1031 } elsif ($r !~ /^\d+$/) {
1032 die "revision argument: $r not understood by git-svn\n";
1033 }
1034 my %ed_opts = ( r => $r,
1035 log => $_message,
1036 ra => $ra,
1037 tree_a => $ta,
1038 tree_b => $tb,
1039 editor_cb => sub { print "Committed r$_[0]\n" },
1040 svn_path => $svn_path );
1041 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1042 print "No changes\n$ta == $tb\n";
1043 }
1044}
1045sub escape_uri_only {
1047 my ($uri) = @_;
1048 my @tmp;
1049 foreach (split m{/}, $uri) {
1050 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1051 push @tmp, $_;
1052 }
1053 join('/', @tmp);
1054}
1055sub escape_url {
1057 my ($url) = @_;
1058 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1059 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1060 $url = "$scheme://$domain$uri";
1061 }
1062 $url;
1063}
1064sub cmd_info {
1066 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1067 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1068 if (exists $_[1]) {
1069 die "Too many arguments specified\n";
1070 }
1071 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1073 if (!$file_type && !$diff_status) {
1075 print STDERR "svn: '$path' is not under version control\n";
1076 exit 1;
1077 }
1078 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1080 unless ($gs) {
1081 die "Unable to determine upstream SVN information from ",
1082 "working tree history\n";
1083 }
1084 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1086 $path = "." if $path eq "";
1087 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1089 if ($_url) {
1091 print escape_url($full_url), "\n";
1092 return;
1093 }
1094 my $result = "Path: $path\n";
1096 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1097 $result .= "URL: " . escape_url($full_url) . "\n";
1098 eval {
1100 my $repos_root = $gs->repos_root;
1101 Git::SVN::remove_username($repos_root);
1102 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1103 };
1104 if ($@) {
1105 $result .= "Repository Root: (offline)\n";
1106 }
1107 ::_req_svn();
1108 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1109 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1110 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1111 $result .= "Node Kind: " .
1113 ($file_type eq "dir" ? "directory" : "file") . "\n";
1114 my $schedule = $diff_status eq "A"
1116 ? "add"
1117 : ($diff_status eq "D" ? "delete" : "normal");
1118 $result .= "Schedule: $schedule\n";
1119 if ($diff_status eq "A") {
1121 print $result, "\n";
1122 return;
1123 }
1124 my ($lc_author, $lc_rev, $lc_date_utc);
1126 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1127 my $log = command_output_pipe(@args);
1128 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1129 while (<$log>) {
1130 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1131 $lc_author = $1;
1132 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1133 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1134 (undef, $lc_rev, undef) = ::extract_metadata($1);
1135 }
1136 }
1137 close $log;
1138 Git::SVN::Log::set_local_timezone();
1140 $result .= "Last Changed Author: $lc_author\n";
1142 $result .= "Last Changed Rev: $lc_rev\n";
1143 $result .= "Last Changed Date: " .
1144 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1145 if ($file_type ne "dir") {
1147 my $text_last_updated_date =
1148 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1149 $result .=
1150 "Text Last Updated: " .
1151 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1152 "\n";
1153 my $checksum;
1154 if ($diff_status eq "D") {
1155 my ($fh, $ctx) =
1156 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1157 if ($file_type eq "link") {
1158 my $file_name = <$fh>;
1159 $checksum = md5sum("link $file_name");
1160 } else {
1161 $checksum = md5sum($fh);
1162 }
1163 command_close_pipe($fh, $ctx);
1164 } elsif ($file_type eq "link") {
1165 my $file_name =
1166 command(qw(cat-file blob), "HEAD:$path");
1167 $checksum =
1168 md5sum("link " . $file_name);
1169 } else {
1170 open FILE, "<", $path or die $!;
1171 $checksum = md5sum(\*FILE);
1172 close FILE or die $!;
1173 }
1174 $result .= "Checksum: " . $checksum . "\n";
1175 }
1176 print $result, "\n";
1178}
1179sub cmd_reset {
1181 my $target = shift || $_revision or die "SVN revision required\n";
1182 $target = $1 if $target =~ /^r(\d+)$/;
1183 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1184 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1185 unless ($gs) {
1186 die "Unable to determine upstream SVN information from ".
1187 "history\n";
1188 }
1189 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1190 die "Cannot find SVN revision $target\n" unless defined($c);
1191 $gs->rev_map_set($r, $c, 'reset', $uuid);
1192 print "r$r = $c ($gs->{ref_id})\n";
1193}
1194sub cmd_gc {
1196 if (!$can_compress) {
1197 warn "Compress::Zlib could not be found; unhandled.log " .
1198 "files will not be compressed.\n";
1199 }
1200 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1201}
1202########################### utility functions #########################
1204sub rebase_cmd {
1206 my @cmd = qw/rebase/;
1207 push @cmd, '-v' if $_verbose;
1208 push @cmd, qw/--merge/ if $_merge;
1209 push @cmd, "--strategy=$_strategy" if $_strategy;
1210 @cmd;
1211}
1212sub post_fetch_checkout {
1214 return if $_no_checkout;
1215 my $gs = $Git::SVN::_head or return;
1216 return if verify_ref('refs/heads/master^0');
1217 # look for "trunk" ref if it exists
1219 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1220 my $fetch = $remote->{fetch};
1221 if ($fetch) {
1222 foreach my $p (keys %$fetch) {
1223 basename($fetch->{$p}) eq 'trunk' or next;
1224 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1225 last;
1226 }
1227 }
1228 my $valid_head = verify_ref('HEAD^0');
1230 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1231 return if ($valid_head || !verify_ref('HEAD^0'));
1232 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1234 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1235 return if -f $index;
1236 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1238 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1239 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1240 print STDERR "Checked out HEAD:\n ",
1241 $gs->full_url, " r", $gs->last_rev, "\n";
1242 $gs->mkemptydirs($gs->last_rev);
1243}
1244sub complete_svn_url {
1246 my ($url, $path) = @_;
1247 $path =~ s#/+$##;
1248 if ($path !~ m#^[a-z\+]+://#) {
1249 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1250 fatal("E: '$path' is not a complete URL ",
1251 "and a separate URL is not specified");
1252 }
1253 return ($url, $path);
1254 }
1255 return ($path, '');
1256}
1257sub complete_url_ls_init {
1259 my ($ra, $repo_path, $switch, $pfx) = @_;
1260 unless ($repo_path) {
1261 print STDERR "W: $switch not specified\n";
1262 return;
1263 }
1264 $repo_path =~ s#/+$##;
1265 if ($repo_path =~ m#^[a-z\+]+://#) {
1266 $ra = Git::SVN::Ra->new($repo_path);
1267 $repo_path = '';
1268 } else {
1269 $repo_path =~ s#^/+##;
1270 unless ($ra) {
1271 fatal("E: '$repo_path' is not a complete URL ",
1272 "and a separate URL is not specified");
1273 }
1274 }
1275 my $url = $ra->{url};
1276 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1277 my $k = "svn-remote.$gs->{repo_id}.url";
1278 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1279 if ($orig_url && ($orig_url ne $gs->{url})) {
1280 die "$k already set: $orig_url\n",
1281 "wanted to set to: $gs->{url}\n";
1282 }
1283 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1284 my $remote_path = "$gs->{path}/$repo_path";
1285 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1286 $remote_path =~ s#/+#/#g;
1287 $remote_path =~ s#^/##g;
1288 $remote_path .= "/*" if $remote_path !~ /\*/;
1289 my ($n) = ($switch =~ /^--(\w+)/);
1290 if (length $pfx && $pfx !~ m#/$#) {
1291 die "--prefix='$pfx' must have a trailing slash '/'\n";
1292 }
1293 command_noisy('config',
1294 '--add',
1295 "svn-remote.$gs->{repo_id}.$n",
1296 "$remote_path:refs/remotes/$pfx*" .
1297 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1298}
1299sub verify_ref {
1301 my ($ref) = @_;
1302 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1303 { STDERR => 0 }); };
1304}
1305sub get_tree_from_treeish {
1307 my ($treeish) = @_;
1308 # $treeish can be a symbolic ref, too:
1309 my $type = command_oneline(qw/cat-file -t/, $treeish);
1310 my $expected;
1311 while ($type eq 'tag') {
1312 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1313 }
1314 if ($type eq 'commit') {
1315 $expected = (grep /^tree /, command(qw/cat-file commit/,
1316 $treeish))[0];
1317 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1318 die "Unable to get tree from $treeish\n" unless $expected;
1319 } elsif ($type eq 'tree') {
1320 $expected = $treeish;
1321 } else {
1322 die "$treeish is a $type, expected tree, tag or commit\n";
1323 }
1324 return $expected;
1325}
1326sub get_commit_entry {
1328 my ($treeish) = shift;
1329 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1330 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1331 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1332 open my $log_fh, '>', $commit_editmsg or croak $!;
1333 my $type = command_oneline(qw/cat-file -t/, $treeish);
1335 if ($type eq 'commit' || $type eq 'tag') {
1336 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1337 $type, $treeish);
1338 my $in_msg = 0;
1339 my $author;
1340 my $saw_from = 0;
1341 my $msgbuf = "";
1342 while (<$msg_fh>) {
1343 if (!$in_msg) {
1344 $in_msg = 1 if (/^\s*$/);
1345 $author = $1 if (/^author (.*>)/);
1346 } elsif (/^git-svn-id: /) {
1347 # skip this for now, we regenerate the
1348 # correct one on re-fetch anyways
1349 # TODO: set *:merge properties or like...
1350 } else {
1351 if (/^From:/ || /^Signed-off-by:/) {
1352 $saw_from = 1;
1353 }
1354 $msgbuf .= $_;
1355 }
1356 }
1357 $msgbuf =~ s/\s+$//s;
1358 if ($Git::SVN::_add_author_from && defined($author)
1359 && !$saw_from) {
1360 $msgbuf .= "\n\nFrom: $author";
1361 }
1362 print $log_fh $msgbuf or croak $!;
1363 command_close_pipe($msg_fh, $ctx);
1364 }
1365 close $log_fh or croak $!;
1366 if ($_edit || ($type eq 'tree')) {
1368 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1369 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1370 }
1371 rename $commit_editmsg, $commit_msg or croak $!;
1372 {
1373 require Encode;
1374 # SVN requires messages to be UTF-8 when entering the repo
1375 local $/;
1376 open $log_fh, '<', $commit_msg or croak $!;
1377 binmode $log_fh;
1378 chomp($log_entry{log} = <$log_fh>);
1379 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1381 my $msg = $log_entry{log};
1382 eval { $msg = Encode::decode($enc, $msg, 1) };
1384 if ($@) {
1385 die "Could not decode as $enc:\n", $msg,
1386 "\nPerhaps you need to set i18n.commitencoding\n";
1387 }
1388 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1390 die "Could not encode as UTF-8:\n$msg\n" if $@;
1391 $log_entry{log} = $msg;
1393 close $log_fh or croak $!;
1395 }
1396 unlink $commit_msg;
1397 \%log_entry;
1398}
1399sub s_to_file {
1401 my ($str, $file, $mode) = @_;
1402 open my $fd,'>',$file or croak $!;
1403 print $fd $str,"\n" or croak $!;
1404 close $fd or croak $!;
1405 chmod ($mode &~ umask, $file) if (defined $mode);
1406}
1407sub file_to_s {
1409 my $file = shift;
1410 open my $fd,'<',$file or croak "$!: file: $file\n";
1411 local $/;
1412 my $ret = <$fd>;
1413 close $fd or croak $!;
1414 $ret =~ s/\s*$//s;
1415 return $ret;
1416}
1417# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1419sub load_authors {
1420 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1421 my $log = $cmd eq 'log';
1422 while (<$authors>) {
1423 chomp;
1424 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1425 my ($user, $name, $email) = ($1, $2, $3);
1426 if ($log) {
1427 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1428 } else {
1429 $users{$user} = [$name, $email];
1430 }
1431 }
1432 close $authors or croak $!;
1433}
1434# convert GetOpt::Long specs for use by git-config
1436sub read_git_config {
1437 my $opts = shift;
1438 my @config_only;
1439 foreach my $o (keys %$opts) {
1440 # if we have mixedCase and a long option-only, then
1441 # it's a config-only variable that we don't need for
1442 # the command-line.
1443 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1444 my $v = $opts->{$o};
1445 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1446 $key =~ s/-//g;
1447 my $arg = 'git config';
1448 $arg .= ' --int' if ($o =~ /[:=]i$/);
1449 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1450 if (ref $v eq 'ARRAY') {
1451 chomp(my @tmp = `$arg --get-all svn.$key`);
1452 @$v = @tmp if @tmp;
1453 } else {
1454 chomp(my $tmp = `$arg --get svn.$key`);
1455 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1456 $$v = $tmp;
1457 }
1458 }
1459 }
1460 delete @$opts{@config_only} if @config_only;
1461}
1462sub extract_metadata {
1464 my $id = shift or return (undef, undef, undef);
1465 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1466 \s([a-f\d\-]+)$/ix);
1467 if (!defined $rev || !$uuid || !$url) {
1468 # some of the original repositories I made had
1469 # identifiers like this:
1470 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1471 }
1472 return ($url, $rev, $uuid);
1473}
1474sub cmt_metadata {
1476 return extract_metadata((grep(/^git-svn-id: /,
1477 command(qw/cat-file commit/, shift)))[-1]);
1478}
1479sub cmt_sha2rev_batch {
1481 my %s2r;
1482 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1483 my $list = shift;
1484 foreach my $sha (@{$list}) {
1486 my $first = 1;
1487 my $size = 0;
1488 print $out $sha, "\n";
1489 while (my $line = <$in>) {
1491 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1492 last;
1493 } elsif ($first &&
1494 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1495 $first = 0;
1496 $size = $1;
1497 next;
1498 } elsif ($line =~ /^(git-svn-id: )/) {
1499 my (undef, $rev, undef) =
1500 extract_metadata($line);
1501 $s2r{$sha} = $rev;
1502 }
1503 $size -= length($line);
1505 last if ($size == 0);
1506 }
1507 }
1508 command_close_bidi_pipe($pid, $in, $out, $ctx);
1510 return \%s2r;
1512}
1513sub working_head_info {
1515 my ($head, $refs) = @_;
1516 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1517 my ($fh, $ctx) = command_output_pipe(@args, $head);
1518 my $hash;
1519 my %max;
1520 while (<$fh>) {
1521 if ( m{^commit ($::sha1)$} ) {
1522 unshift @$refs, $hash if $hash and $refs;
1523 $hash = $1;
1524 next;
1525 }
1526 next unless s{^\s*(git-svn-id:)}{$1};
1527 my ($url, $rev, $uuid) = extract_metadata($_);
1528 if (defined $url && defined $rev) {
1529 next if $max{$url} and $max{$url} < $rev;
1530 if (my $gs = Git::SVN->find_by_url($url)) {
1531 my $c = $gs->rev_map_get($rev, $uuid);
1532 if ($c && $c eq $hash) {
1533 close $fh; # break the pipe
1534 return ($url, $rev, $uuid, $gs);
1535 } else {
1536 $max{$url} ||= $gs->rev_map_max;
1537 }
1538 }
1539 }
1540 }
1541 command_close_pipe($fh, $ctx);
1542 (undef, undef, undef, undef);
1543}
1544sub read_commit_parents {
1546 my ($parents, $c) = @_;
1547 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1548 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1549 @{$parents->{$c}} = split(/ /, $p);
1550}
1551sub linearize_history {
1553 my ($gs, $refs) = @_;
1554 my %parents;
1555 foreach my $c (@$refs) {
1556 read_commit_parents(\%parents, $c);
1557 }
1558 my @linear_refs;
1560 my %skip = ();
1561 my $last_svn_commit = $gs->last_commit;
1562 foreach my $c (reverse @$refs) {
1563 next if $c eq $last_svn_commit;
1564 last if $skip{$c};
1565 unshift @linear_refs, $c;
1567 $skip{$c} = 1;
1568 # we only want the first parent to diff against for linear
1570 # history, we save the rest to inject when we finalize the
1571 # svn commit
1572 my $fp_a = verify_ref("$c~1");
1573 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1574 if (!$fp_a || !$fp_b) {
1575 die "Commit $c\n",
1576 "has no parent commit, and therefore ",
1577 "nothing to diff against.\n",
1578 "You should be working from a repository ",
1579 "originally created by git-svn\n";
1580 }
1581 if ($fp_a ne $fp_b) {
1582 die "$c~1 = $fp_a, however parsing commit $c ",
1583 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1584 }
1585 foreach my $p (@{$parents{$c}}) {
1587 $skip{$p} = 1;
1588 }
1589 }
1590 (\@linear_refs, \%parents);
1591}
1592sub find_file_type_and_diff_status {
1594 my ($path) = @_;
1595 return ('dir', '') if $path eq '';
1596 my $diff_output =
1598 command_oneline(qw(diff --cached --name-status --), $path) || "";
1599 my $diff_status = (split(' ', $diff_output))[0] || "";
1600 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1602 return (undef, undef) if !$diff_status && !$ls_tree;
1604 if ($diff_status eq "A") {
1606 return ("link", $diff_status) if -l $path;
1607 return ("dir", $diff_status) if -d $path;
1608 return ("file", $diff_status);
1609 }
1610 my $mode = (split(' ', $ls_tree))[0] || "";
1612 return ("link", $diff_status) if $mode eq "120000";
1614 return ("dir", $diff_status) if $mode eq "040000";
1615 return ("file", $diff_status);
1616}
1617sub md5sum {
1619 my $arg = shift;
1620 my $ref = ref $arg;
1621 my $md5 = Digest::MD5->new();
1622 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1623 $md5->addfile($arg) or croak $!;
1624 } elsif ($ref eq 'SCALAR') {
1625 $md5->add($$arg) or croak $!;
1626 } elsif (!$ref) {
1627 $md5->add($arg) or croak $!;
1628 } else {
1629 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1630 }
1631 return $md5->hexdigest();
1632}
1633sub gc_directory {
1635 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1636 my $out_filename = $_ . ".gz";
1637 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1638 binmode $in_fh;
1639 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1640 die "Unable to open $out_filename: $!\n";
1641 my $res;
1643 while ($res = sysread($in_fh, my $str, 1024)) {
1644 $gz->gzwrite($str) or
1645 die "Unable to write: ".$gz->gzerror()."!\n";
1646 }
1647 unlink $_ or die "unlink $File::Find::name: $!\n";
1648 } elsif (-f $_ && basename($_) eq "index") {
1649 unlink $_ or die "unlink $_: $!\n";
1650 }
1651}
1652package Git::SVN;
1654use strict;
1655use warnings;
1656use Fcntl qw/:DEFAULT :seek/;
1657use constant rev_map_fmt => 'NH40';
1658use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1659 $_repack $_repack_flags $_use_svm_props $_head
1660 $_use_svnsync_props $no_reuse_existing $_minimize_url
1661 $_use_log_author $_add_author_from $_localtime/;
1662use Carp qw/croak/;
1663use File::Path qw/mkpath/;
1664use File::Copy qw/copy/;
1665use IPC::Open3;
1666use Memoize; # core since 5.8.0, Jul 2002
1667use Memoize::Storable;
1668my ($_gc_nr, $_gc_period);
1670# properties that we do not log:
1672my %SKIP_PROP;
1673BEGIN {
1674 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1675 svn:special svn:executable
1676 svn:entry:committed-rev
1677 svn:entry:last-author
1678 svn:entry:uuid
1679 svn:entry:committed-date/;
1680 # some options are read globally, but can be overridden locally
1682 # per [svn-remote "..."] section. Command-line options will *NOT*
1683 # override options set in an [svn-remote "..."] section
1684 no strict 'refs';
1685 for my $option (qw/follow_parent no_metadata use_svm_props
1686 use_svnsync_props/) {
1687 my $key = $option;
1688 $key =~ tr/_//d;
1689 my $prop = "-$option";
1690 *$option = sub {
1691 my ($self) = @_;
1692 return $self->{$prop} if exists $self->{$prop};
1693 my $k = "svn-remote.$self->{repo_id}.$key";
1694 eval { command_oneline(qw/config --get/, $k) };
1695 if ($@) {
1696 $self->{$prop} = ${"Git::SVN::_$option"};
1697 } else {
1698 my $v = command_oneline(qw/config --bool/,$k);
1699 $self->{$prop} = $v eq 'false' ? 0 : 1;
1700 }
1701 return $self->{$prop};
1702 }
1703 }
1704}
1705my (%LOCKFILES, %INDEX_FILES);
1708END {
1709 unlink keys %LOCKFILES if %LOCKFILES;
1710 unlink keys %INDEX_FILES if %INDEX_FILES;
1711}
1712sub resolve_local_globs {
1714 my ($url, $fetch, $glob_spec) = @_;
1715 return unless defined $glob_spec;
1716 my $ref = $glob_spec->{ref};
1717 my $path = $glob_spec->{path};
1718 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1719 next unless m#^$ref->{regex}$#;
1720 my $p = $1;
1721 my $pathname = desanitize_refname($path->full_path($p));
1722 my $refname = desanitize_refname($ref->full_path($p));
1723 if (my $existing = $fetch->{$pathname}) {
1724 if ($existing ne $refname) {
1725 die "Refspec conflict:\n",
1726 "existing: $existing\n",
1727 " globbed: $refname\n";
1728 }
1729 my $u = (::cmt_metadata("$refname"))[0];
1730 $u =~ s!^\Q$url\E(/|$)!! or die
1731 "$refname: '$url' not found in '$u'\n";
1732 if ($pathname ne $u) {
1733 warn "W: Refspec glob conflict ",
1734 "(ref: $refname):\n",
1735 "expected path: $pathname\n",
1736 " real path: $u\n",
1737 "Continuing ahead with $u\n";
1738 next;
1739 }
1740 } else {
1741 $fetch->{$pathname} = $refname;
1742 }
1743 }
1744}
1745sub parse_revision_argument {
1747 my ($base, $head) = @_;
1748 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1749 return ($base, $head);
1750 }
1751 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1752 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1753 return ($head, $head) if ($::_revision eq 'HEAD');
1754 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1755 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1756 die "revision argument: $::_revision not understood by git-svn\n";
1757}
1758sub fetch_all {
1760 my ($repo_id, $remotes) = @_;
1761 if (ref $repo_id) {
1762 my $gs = $repo_id;
1763 $repo_id = undef;
1764 $repo_id = $gs->{repo_id};
1765 }
1766 $remotes ||= read_all_remotes();
1767 my $remote = $remotes->{$repo_id} or
1768 die "[svn-remote \"$repo_id\"] unknown\n";
1769 my $fetch = $remote->{fetch};
1770 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1771 my (@gs, @globs);
1772 my $ra = Git::SVN::Ra->new($url);
1773 my $uuid = $ra->get_uuid;
1774 my $head = $ra->get_latest_revnum;
1775 # ignore errors, $head revision may not even exist anymore
1777 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1778 warn "W: $@\n" if $@;
1779 my $base = defined $fetch ? $head : 0;
1781 # read the max revs for wildcard expansion (branches/*, tags/*)
1783 foreach my $t (qw/branches tags/) {
1784 defined $remote->{$t} or next;
1785 push @globs, @{$remote->{$t}};
1786 my $max_rev = eval { tmp_config(qw/--int --get/,
1788 "svn-remote.$repo_id.${t}-maxRev") };
1789 if (defined $max_rev && ($max_rev < $base)) {
1790 $base = $max_rev;
1791 } elsif (!defined $max_rev) {
1792 $base = 0;
1793 }
1794 }
1795 if ($fetch) {
1797 foreach my $p (sort keys %$fetch) {
1798 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1799 my $lr = $gs->rev_map_max;
1800 if (defined $lr) {
1801 $base = $lr if ($lr < $base);
1802 }
1803 push @gs, $gs;
1804 }
1805 }
1806 ($base, $head) = parse_revision_argument($base, $head);
1808 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1809}
1810sub read_all_remotes {
1812 my $r = {};
1813 my $use_svm_props = eval { command_oneline(qw/config --bool
1814 svn.useSvmProps/) };
1815 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1816 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1817 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1818 if (m!^(.+)\.fetch=$svn_refspec$!) {
1819 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1820 die("svn-remote.$remote: remote ref '$remote_ref' "
1821 . "must start with 'refs/'\n")
1822 unless $remote_ref =~ m{^refs/};
1823 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1824 $r->{$remote}->{svm} = {} if $use_svm_props;
1825 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1826 $r->{$1}->{svm} = {};
1827 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1828 $r->{$1}->{url} = $2;
1829 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1830 my ($remote, $t, $local_ref, $remote_ref) =
1831 ($1, $2, $3, $4);
1832 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1833 . "must start with 'refs/'\n")
1834 unless $remote_ref =~ m{^refs/};
1835 my $rs = {
1836 t => $t,
1837 remote => $remote,
1838 path => Git::SVN::GlobSpec->new($local_ref, 1),
1839 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1840 if (length($rs->{ref}->{right}) != 0) {
1841 die "The '*' glob character must be the last ",
1842 "character of '$remote_ref'\n";
1843 }
1844 push @{ $r->{$remote}->{$t} }, $rs;
1845 }
1846 }
1847 map {
1849 if (defined $r->{$_}->{svm}) {
1850 my $svm;
1851 eval {
1852 my $section = "svn-remote.$_";
1853 $svm = {
1854 source => tmp_config('--get',
1855 "$section.svm-source"),
1856 replace => tmp_config('--get',
1857 "$section.svm-replace"),
1858 }
1859 };
1860 $r->{$_}->{svm} = $svm;
1861 }
1862 } keys %$r;
1863 $r;
1865}
1866sub init_vars {
1868 $_gc_nr = $_gc_period = 1000;
1869 if (defined $_repack || defined $_repack_flags) {
1870 warn "Repack options are obsolete; they have no effect.\n";
1871 }
1872}
1873sub verify_remotes_sanity {
1875 return unless -d $ENV{GIT_DIR};
1876 my %seen;
1877 foreach (command(qw/config -l/)) {
1878 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1879 if ($seen{$1}) {
1880 die "Remote ref refs/remote/$1 is tracked by",
1881 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1882 "Please resolve this ambiguity in ",
1883 "your git configuration file before ",
1884 "continuing\n";
1885 }
1886 $seen{$1} = $_;
1887 }
1888 }
1889}
1890sub find_existing_remote {
1892 my ($url, $remotes) = @_;
1893 return undef if $no_reuse_existing;
1894 my $existing;
1895 foreach my $repo_id (keys %$remotes) {
1896 my $u = $remotes->{$repo_id}->{url} or next;
1897 next if $u ne $url;
1898 $existing = $repo_id;
1899 last;
1900 }
1901 $existing;
1902}
1903sub init_remote_config {
1905 my ($self, $url, $no_write) = @_;
1906 $url =~ s!/+$!!; # strip trailing slash
1907 my $r = read_all_remotes();
1908 my $existing = find_existing_remote($url, $r);
1909 if ($existing) {
1910 unless ($no_write) {
1911 print STDERR "Using existing ",
1912 "[svn-remote \"$existing\"]\n";
1913 }
1914 $self->{repo_id} = $existing;
1915 } elsif ($_minimize_url) {
1916 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1917 $existing = find_existing_remote($min_url, $r);
1918 if ($existing) {
1919 unless ($no_write) {
1920 print STDERR "Using existing ",
1921 "[svn-remote \"$existing\"]\n";
1922 }
1923 $self->{repo_id} = $existing;
1924 }
1925 if ($min_url ne $url) {
1926 unless ($no_write) {
1927 print STDERR "Using higher level of URL: ",
1928 "$url => $min_url\n";
1929 }
1930 my $old_path = $self->{path};
1931 $self->{path} = $url;
1932 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1933 if (length $old_path) {
1934 $self->{path} .= "/$old_path";
1935 }
1936 $url = $min_url;
1937 }
1938 }
1939 my $orig_url;
1940 if (!$existing) {
1941 # verify that we aren't overwriting anything:
1942 $orig_url = eval {
1943 command_oneline('config', '--get',
1944 "svn-remote.$self->{repo_id}.url")
1945 };
1946 if ($orig_url && ($orig_url ne $url)) {
1947 die "svn-remote.$self->{repo_id}.url already set: ",
1948 "$orig_url\nwanted to set to: $url\n";
1949 }
1950 }
1951 my ($xrepo_id, $xpath) = find_ref($self->refname);
1952 if (!$no_write && defined $xpath) {
1953 die "svn-remote.$xrepo_id.fetch already set to track ",
1954 "$xpath:", $self->refname, "\n";
1955 }
1956 unless ($no_write) {
1957 command_noisy('config',
1958 "svn-remote.$self->{repo_id}.url", $url);
1959 $self->{path} =~ s{^/}{};
1960 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1961 command_noisy('config', '--add',
1962 "svn-remote.$self->{repo_id}.fetch",
1963 "$self->{path}:".$self->refname);
1964 }
1965 $self->{url} = $url;
1966}
1967sub find_by_url { # repos_root and, path are optional
1969 my ($class, $full_url, $repos_root, $path) = @_;
1970 return undef unless defined $full_url;
1972 remove_username($full_url);
1973 remove_username($repos_root) if defined $repos_root;
1974 my $remotes = read_all_remotes();
1975 if (defined $full_url && defined $repos_root && !defined $path) {
1976 $path = $full_url;
1977 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1978 }
1979 foreach my $repo_id (keys %$remotes) {
1980 my $u = $remotes->{$repo_id}->{url} or next;
1981 remove_username($u);
1982 next if defined $repos_root && $repos_root ne $u;
1983 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1985 foreach my $t (qw/branches tags/) {
1986 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1987 resolve_local_globs($u, $fetch, $globspec);
1988 }
1989 }
1990 my $p = $path;
1991 my $rwr = rewrite_root({repo_id => $repo_id});
1992 my $svm = $remotes->{$repo_id}->{svm}
1993 if defined $remotes->{$repo_id}->{svm};
1994 unless (defined $p) {
1995 $p = $full_url;
1996 my $z = $u;
1997 my $prefix = '';
1998 if ($rwr) {
1999 $z = $rwr;
2000 remove_username($z);
2001 } elsif (defined $svm) {
2002 $z = $svm->{source};
2003 $prefix = $svm->{replace};
2004 $prefix =~ s#^\Q$u\E(?:/|$)##;
2005 $prefix =~ s#/$##;
2006 }
2007 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2008 }
2009 foreach my $f (keys %$fetch) {
2010 next if $f ne $p;
2011 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2012 }
2013 }
2014 undef;
2015}
2016sub init {
2018 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2019 my $self = _new($class, $repo_id, $ref_id, $path);
2020 if (defined $url) {
2021 $self->init_remote_config($url, $no_write);
2022 }
2023 $self;
2024}
2025sub find_ref {
2027 my ($ref_id) = @_;
2028 foreach (command(qw/config -l/)) {
2029 next unless m!^svn-remote\.(.+)\.fetch=
2030 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2031 my ($repo_id, $path, $ref) = ($1, $2, $3);
2032 if ($ref eq $ref_id) {
2033 $path = '' if ($path =~ m#^\./?#);
2034 return ($repo_id, $path);
2035 }
2036 }
2037 (undef, undef, undef);
2038}
2039sub new {
2041 my ($class, $ref_id, $repo_id, $path) = @_;
2042 if (defined $ref_id && !defined $repo_id && !defined $path) {
2043 ($repo_id, $path) = find_ref($ref_id);
2044 if (!defined $repo_id) {
2045 die "Could not find a \"svn-remote.*.fetch\" key ",
2046 "in the repository configuration matching: ",
2047 "$ref_id\n";
2048 }
2049 }
2050 my $self = _new($class, $repo_id, $ref_id, $path);
2051 if (!defined $self->{path} || !length $self->{path}) {
2052 my $fetch = command_oneline('config', '--get',
2053 "svn-remote.$repo_id.fetch",
2054 ":$ref_id\$") or
2055 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2056 "\":$ref_id\$\" in config\n";
2057 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2058 }
2059 $self->{path} =~ s{/+}{/}g;
2060 $self->{path} =~ s{\A/}{};
2061 $self->{path} =~ s{/\z}{};
2062 $self->{url} = command_oneline('config', '--get',
2063 "svn-remote.$repo_id.url") or
2064 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2065 $self->rebuild;
2066 $self;
2067}
2068sub refname {
2070 my ($refname) = $_[0]->{ref_id} ;
2071 # It cannot end with a slash /, we'll throw up on this because
2073 # SVN can't have directories with a slash in their name, either:
2074 if ($refname =~ m{/$}) {
2075 die "ref: '$refname' ends with a trailing slash, this is ",
2076 "not permitted by git nor Subversion\n";
2077 }
2078 # It cannot have ASCII control character space, tilde ~, caret ^,
2080 # colon :, question-mark ?, asterisk *, space, or open bracket [
2081 # anywhere.
2082 #
2083 # Additionally, % must be escaped because it is used for escaping
2084 # and we want our escaped refname to be reversible
2085 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2086 # no slash-separated component can begin with a dot .
2088 # /.* becomes /%2E*
2089 $refname =~ s{/\.}{/%2E}g;
2090 # It cannot have two consecutive dots .. anywhere
2092 # .. becomes %2E%2E
2093 $refname =~ s{\.\.}{%2E%2E}g;
2094 # trailing dots and .lock are not allowed
2096 # .$ becomes %2E and .lock becomes %2Elock
2097 $refname =~ s{\.(?=$|lock$)}{%2E};
2098 # the sequence @{ is used to access the reflog
2100 # @{ becomes %40{
2101 $refname =~ s{\@\{}{%40\{}g;
2102 return $refname;
2104}
2105sub desanitize_refname {
2107 my ($refname) = @_;
2108 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2109 return $refname;
2110}
2111sub svm_uuid {
2113 my ($self) = @_;
2114 return $self->{svm}->{uuid} if $self->svm;
2115 $self->ra;
2116 unless ($self->{svm}) {
2117 die "SVM UUID not cached, and reading remotely failed\n";
2118 }
2119 $self->{svm}->{uuid};
2120}
2121sub svm {
2123 my ($self) = @_;
2124 return $self->{svm} if $self->{svm};
2125 my $svm;
2126 # see if we have it in our config, first:
2127 eval {
2128 my $section = "svn-remote.$self->{repo_id}";
2129 $svm = {
2130 source => tmp_config('--get', "$section.svm-source"),
2131 uuid => tmp_config('--get', "$section.svm-uuid"),
2132 replace => tmp_config('--get', "$section.svm-replace"),
2133 }
2134 };
2135 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2136 $self->{svm} = $svm;
2137 }
2138 $self->{svm};
2139}
2140sub _set_svm_vars {
2142 my ($self, $ra) = @_;
2143 return $ra if $self->svm;
2144 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2146 "(svm:source, svm:uuid) ",
2147 "from the following URLs:\n" );
2148 sub read_svm_props {
2149 my ($self, $ra, $path, $r) = @_;
2150 my $props = ($ra->get_dir($path, $r))[2];
2151 my $src = $props->{'svm:source'};
2152 my $uuid = $props->{'svm:uuid'};
2153 return undef if (!$src || !$uuid);
2154 chomp($src, $uuid);
2156 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2158 or die "doesn't look right - svm:uuid is '$uuid'\n";
2159 # the '!' is used to mark the repos_root!/relative/path
2161 $src =~ s{/?!/?}{/};
2162 $src =~ s{/+$}{}; # no trailing slashes please
2163 # username is of no interest
2164 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2165 my $replace = $ra->{url};
2167 $replace .= "/$path" if length $path;
2168 my $section = "svn-remote.$self->{repo_id}";
2170 tmp_config("$section.svm-source", $src);
2171 tmp_config("$section.svm-replace", $replace);
2172 tmp_config("$section.svm-uuid", $uuid);
2173 $self->{svm} = {
2174 source => $src,
2175 uuid => $uuid,
2176 replace => $replace
2177 };
2178 }
2179 my $r = $ra->get_latest_revnum;
2181 my $path = $self->{path};
2182 my %tried;
2183 while (length $path) {
2184 unless ($tried{"$self->{url}/$path"}) {
2185 return $ra if $self->read_svm_props($ra, $path, $r);
2186 $tried{"$self->{url}/$path"} = 1;
2187 }
2188 $path =~ s#/?[^/]+$##;
2189 }
2190 die "Path: '$path' should be ''\n" if $path ne '';
2191 return $ra if $self->read_svm_props($ra, $path, $r);
2192 $tried{"$self->{url}/$path"} = 1;
2193 if ($ra->{repos_root} eq $self->{url}) {
2195 die @err, (map { " $_\n" } keys %tried), "\n";
2196 }
2197 # nope, make sure we're connected to the repository root:
2199 my $ok;
2200 my @tried_b;
2201 $path = $ra->{svn_path};
2202 $ra = Git::SVN::Ra->new($ra->{repos_root});
2203 while (length $path) {
2204 unless ($tried{"$ra->{url}/$path"}) {
2205 $ok = $self->read_svm_props($ra, $path, $r);
2206 last if $ok;
2207 $tried{"$ra->{url}/$path"} = 1;
2208 }
2209 $path =~ s#/?[^/]+$##;
2210 }
2211 die "Path: '$path' should be ''\n" if $path ne '';
2212 $ok ||= $self->read_svm_props($ra, $path, $r);
2213 $tried{"$ra->{url}/$path"} = 1;
2214 if (!$ok) {
2215 die @err, (map { " $_\n" } keys %tried), "\n";
2216 }
2217 Git::SVN::Ra->new($self->{url});
2218}
2219sub svnsync {
2221 my ($self) = @_;
2222 return $self->{svnsync} if $self->{svnsync};
2223 if ($self->no_metadata) {
2225 die "Can't have both 'noMetadata' and ",
2226 "'useSvnsyncProps' options set!\n";
2227 }
2228 if ($self->rewrite_root) {
2229 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2230 "options set!\n";
2231 }
2232 if ($self->rewrite_uuid) {
2233 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2234 "options set!\n";
2235 }
2236 my $svnsync;
2238 # see if we have it in our config, first:
2239 eval {
2240 my $section = "svn-remote.$self->{repo_id}";
2241 my $url = tmp_config('--get', "$section.svnsync-url");
2243 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2244 die "doesn't look right - svn:sync-from-url is '$url'\n";
2245 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2247 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2248 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2249 $svnsync = { url => $url, uuid => $uuid }
2251 };
2252 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2253 return $self->{svnsync} = $svnsync;
2254 }
2255 my $err = "useSvnsyncProps set, but failed to read " .
2257 "svnsync property: svn:sync-from-";
2258 my $rp = $self->ra->rev_proplist(0);
2259 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2261 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2262 die "doesn't look right - svn:sync-from-url is '$url'\n";
2263 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2265 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2266 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2267 my $section = "svn-remote.$self->{repo_id}";
2269 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2270 tmp_config('--add', "$section.svnsync-url", $url);
2271 return $self->{svnsync} = { url => $url, uuid => $uuid };
2272}
2273# this allows us to memoize our SVN::Ra UUID locally and avoid a
2275# remote lookup (useful for 'git svn log').
2276sub ra_uuid {
2277 my ($self) = @_;
2278 unless ($self->{ra_uuid}) {
2279 my $key = "svn-remote.$self->{repo_id}.uuid";
2280 my $uuid = eval { tmp_config('--get', $key) };
2281 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2282 $self->{ra_uuid} = $uuid;
2283 } else {
2284 die "ra_uuid called without URL\n" unless $self->{url};
2285 $self->{ra_uuid} = $self->ra->get_uuid;
2286 tmp_config('--add', $key, $self->{ra_uuid});
2287 }
2288 }
2289 $self->{ra_uuid};
2290}
2291sub _set_repos_root {
2293 my ($self, $repos_root) = @_;
2294 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2295 $repos_root ||= $self->ra->{repos_root};
2296 tmp_config($k, $repos_root);
2297 $repos_root;
2298}
2299sub repos_root {
2301 my ($self) = @_;
2302 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2303 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2304}
2305sub ra {
2307 my ($self) = shift;
2308 my $ra = Git::SVN::Ra->new($self->{url});
2309 $self->_set_repos_root($ra->{repos_root});
2310 if ($self->use_svm_props && !$self->{svm}) {
2311 if ($self->no_metadata) {
2312 die "Can't have both 'noMetadata' and ",
2313 "'useSvmProps' options set!\n";
2314 } elsif ($self->use_svnsync_props) {
2315 die "Can't have both 'useSvnsyncProps' and ",
2316 "'useSvmProps' options set!\n";
2317 }
2318 $ra = $self->_set_svm_vars($ra);
2319 $self->{-want_revprops} = 1;
2320 }
2321 $ra;
2322}
2323# prop_walk(PATH, REV, SUB)
2325# -------------------------
2326# Recursively traverse PATH at revision REV and invoke SUB for each
2327# directory that contains a SVN property. SUB will be invoked as
2328# follows: &SUB(gs, path, props); where `gs' is this instance of
2329# Git::SVN, `path' the path to the directory where the properties
2330# `props' were found. The `path' will be relative to point of checkout,
2331# that is, if url://repo/trunk is the current Git branch, and that
2332# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2333# as `path' (note the trailing `/').
2334sub prop_walk {
2335 my ($self, $path, $rev, $sub) = @_;
2336 $path =~ s#^/##;
2338 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2339 $path =~ s#^/*#/#g;
2340 my $p = $path;
2341 # Strip the irrelevant part of the path.
2342 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2343 # Ensure the path is terminated by a `/'.
2344 $p =~ s#/*$#/#;
2345 # The properties contain all the internal SVN stuff nobody
2347 # (usually) cares about.
2348 my $interesting_props = 0;
2349 foreach (keys %{$props}) {
2350 # If it doesn't start with `svn:', it must be a
2351 # user-defined property.
2352 ++$interesting_props and next if $_ !~ /^svn:/;
2353 # FIXME: Fragile, if SVN adds new public properties,
2354 # this needs to be updated.
2355 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2356 |eol-style|mime-type
2357 |externals|needs-lock)$/x;
2358 }
2359 &$sub($self, $p, $props) if $interesting_props;
2360 foreach (sort keys %$dirent) {
2362 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2363 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2364 }
2365}
2366sub last_rev { ($_[0]->last_rev_commit)[0] }
2368sub last_commit { ($_[0]->last_rev_commit)[1] }
2369# returns the newest SVN revision number and newest commit SHA1
2371sub last_rev_commit {
2372 my ($self) = @_;
2373 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2374 return ($self->{last_rev}, $self->{last_commit});
2375 }
2376 my $c = ::verify_ref($self->refname.'^0');
2377 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2378 my $rev = (::cmt_metadata($c))[1];
2379 if (defined $rev) {
2380 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2381 return ($rev, $c);
2382 }
2383 }
2384 my $map_path = $self->map_path;
2385 unless (-e $map_path) {
2386 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2387 return (undef, undef);
2388 }
2389 my ($rev, $commit) = $self->rev_map_max(1);
2390 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2391 return ($rev, $commit);
2392}
2393sub get_fetch_range {
2395 my ($self, $min, $max) = @_;
2396 $max ||= $self->ra->get_latest_revnum;
2397 $min ||= $self->rev_map_max;
2398 (++$min, $max);
2399}
2400sub tmp_config {
2402 my (@args) = @_;
2403 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2404 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2405 if (! -f $config && -f $old_def_config) {
2406 rename $old_def_config, $config or
2407 die "Failed rename $old_def_config => $config: $!\n";
2408 }
2409 my $old_config = $ENV{GIT_CONFIG};
2410 $ENV{GIT_CONFIG} = $config;
2411 $@ = undef;
2412 my @ret = eval {
2413 unless (-f $config) {
2414 mkfile($config);
2415 open my $fh, '>', $config or
2416 die "Can't open $config: $!\n";
2417 print $fh "; This file is used internally by ",
2418 "git-svn\n" or die
2419 "Couldn't write to $config: $!\n";
2420 print $fh "; You should not have to edit it\n" or
2421 die "Couldn't write to $config: $!\n";
2422 close $fh or die "Couldn't close $config: $!\n";
2423 }
2424 command('config', @args);
2425 };
2426 my $err = $@;
2427 if (defined $old_config) {
2428 $ENV{GIT_CONFIG} = $old_config;
2429 } else {
2430 delete $ENV{GIT_CONFIG};
2431 }
2432 die $err if $err;
2433 wantarray ? @ret : $ret[0];
2434}
2435sub tmp_index_do {
2437 my ($self, $sub) = @_;
2438 my $old_index = $ENV{GIT_INDEX_FILE};
2439 $ENV{GIT_INDEX_FILE} = $self->{index};
2440 $@ = undef;
2441 my @ret = eval {
2442 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2443 mkpath([$dir]) unless -d $dir;
2444 &$sub;
2445 };
2446 my $err = $@;
2447 if (defined $old_index) {
2448 $ENV{GIT_INDEX_FILE} = $old_index;
2449 } else {
2450 delete $ENV{GIT_INDEX_FILE};
2451 }
2452 die $err if $err;
2453 wantarray ? @ret : $ret[0];
2454}
2455sub assert_index_clean {
2457 my ($self, $treeish) = @_;
2458 $self->tmp_index_do(sub {
2460 command_noisy('read-tree', $treeish) unless -e $self->{index};
2461 my $x = command_oneline('write-tree');
2462 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2463 /^tree ($::sha1)/mo);
2464 return if $y eq $x;
2465 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2467 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2468 command_noisy('read-tree', $treeish);
2469 $x = command_oneline('write-tree');
2470 if ($y ne $x) {
2471 ::fatal "trees ($treeish) $y != $x\n",
2472 "Something is seriously wrong...";
2473 }
2474 });
2475}
2476sub get_commit_parents {
2478 my ($self, $log_entry) = @_;
2479 my (%seen, @ret, @tmp);
2480 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2481 if (my $ip = $self->{inject_parents}) {
2482 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2483 push @tmp, $commit;
2484 }
2485 }
2486 if (my $cur = ::verify_ref($self->refname.'^0')) {
2487 push @tmp, $cur;
2488 }
2489 if (my $ipd = $self->{inject_parents_dcommit}) {
2490 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2491 push @tmp, @$commit;
2492 }
2493 }
2494 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2495 while (my $p = shift @tmp) {
2496 next if $seen{$p};
2497 $seen{$p} = 1;
2498 push @ret, $p;
2499 }
2500 @ret;
2501}
2502sub rewrite_root {
2504 my ($self) = @_;
2505 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2506 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2507 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2508 if ($rwr) {
2509 $rwr =~ s#/+$##;
2510 if ($rwr !~ m#^[a-z\+]+://#) {
2511 die "$rwr is not a valid URL (key: $k)\n";
2512 }
2513 }
2514 $self->{-rewrite_root} = $rwr;
2515}
2516sub rewrite_uuid {
2518 my ($self) = @_;
2519 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2520 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2521 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2522 if ($rwid) {
2523 $rwid =~ s#/+$##;
2524 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2525 die "$rwid is not a valid UUID (key: $k)\n";
2526 }
2527 }
2528 $self->{-rewrite_uuid} = $rwid;
2529}
2530sub metadata_url {
2532 my ($self) = @_;
2533 ($self->rewrite_root || $self->{url}) .
2534 (length $self->{path} ? '/' . $self->{path} : '');
2535}
2536sub full_url {
2538 my ($self) = @_;
2539 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2540}
2541sub set_commit_header_env {
2544 my ($log_entry) = @_;
2545 my %env;
2546 foreach my $ned (qw/NAME EMAIL DATE/) {
2547 foreach my $ac (qw/AUTHOR COMMITTER/) {
2548 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2549 }
2550 }
2551 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2553 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2554 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2555 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2557 ? $log_entry->{commit_name}
2558 : $log_entry->{name};
2559 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2560 ? $log_entry->{commit_email}
2561 : $log_entry->{email};
2562 \%env;
2563}
2564sub restore_commit_header_env {
2566 my ($env) = @_;
2567 foreach my $ned (qw/NAME EMAIL DATE/) {
2568 foreach my $ac (qw/AUTHOR COMMITTER/) {
2569 my $k = "GIT_${ac}_${ned}";
2570 if (defined $env->{$k}) {
2571 $ENV{$k} = $env->{$k};
2572 } else {
2573 delete $ENV{$k};
2574 }
2575 }
2576 }
2577}
2578sub gc {
2580 command_noisy('gc', '--auto');
2581};
2582sub do_git_commit {
2584 my ($self, $log_entry) = @_;
2585 my $lr = $self->last_rev;
2586 if (defined $lr && $lr >= $log_entry->{revision}) {
2587 die "Last fetched revision of ", $self->refname,
2588 " was r$lr, but we are about to fetch: ",
2589 "r$log_entry->{revision}!\n";
2590 }
2591 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2592 croak "$log_entry->{revision} = $c already exists! ",
2593 "Why are we refetching it?\n";
2594 }
2595 my $old_env = set_commit_header_env($log_entry);
2596 my $tree = $log_entry->{tree};
2597 if (!defined $tree) {
2598 $tree = $self->tmp_index_do(sub {
2599 command_oneline('write-tree') });
2600 }
2601 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2602 my @exec = ('git', 'commit-tree', $tree);
2604 foreach ($self->get_commit_parents($log_entry)) {
2605 push @exec, '-p', $_;
2606 }
2607 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2608 or croak $!;
2609 binmode $msg_fh;
2610 # we always get UTF-8 from SVN, but we may want our commits in
2612 # a different encoding.
2613 if (my $enc = Git::config('i18n.commitencoding')) {
2614 require Encode;
2615 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2616 }
2617 print $msg_fh $log_entry->{log} or croak $!;
2618 restore_commit_header_env($old_env);
2619 unless ($self->no_metadata) {
2620 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2621 or croak $!;
2622 }
2623 $msg_fh->flush == 0 or croak $!;
2624 close $msg_fh or croak $!;
2625 chomp(my $commit = do { local $/; <$out_fh> });
2626 close $out_fh or croak $!;
2627 waitpid $pid, 0;
2628 croak $? if $?;
2629 if ($commit !~ /^$::sha1$/o) {
2630 die "Failed to commit, invalid sha1: $commit\n";
2631 }
2632 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2634 $self->{last_rev} = $log_entry->{revision};
2636 $self->{last_commit} = $commit;
2637 print "r$log_entry->{revision}" unless $::_q > 1;
2638 if (defined $log_entry->{svm_revision}) {
2639 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2640 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2641 0, $self->svm_uuid);
2642 }
2643 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2644 if (--$_gc_nr == 0) {
2645 $_gc_nr = $_gc_period;
2646 gc();
2647 }
2648 return $commit;
2649}
2650sub match_paths {
2652 my ($self, $paths, $r) = @_;
2653 return 1 if $self->{path} eq '';
2654 if (my $path = $paths->{"/$self->{path}"}) {
2655 return ($path->{action} eq 'D') ? 0 : 1;
2656 }
2657 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2658 if (grep /$self->{path_regex}/, keys %$paths) {
2659 return 1;
2660 }
2661 my $c = '';
2662 foreach (split m#/#, $self->{path}) {
2663 $c .= "/$_";
2664 next unless ($paths->{$c} &&
2665 ($paths->{$c}->{action} =~ /^[AR]$/));
2666 if ($self->ra->check_path($self->{path}, $r) ==
2667 $SVN::Node::dir) {
2668 return 1;
2669 }
2670 }
2671 return 0;
2672}
2673sub find_parent_branch {
2675 my ($self, $paths, $rev) = @_;
2676 return undef unless $self->follow_parent;
2677 unless (defined $paths) {
2678 my $err_handler = $SVN::Error::handler;
2679 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2680 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2681 sub { $paths = $_[0] });
2682 $SVN::Error::handler = $err_handler;
2683 }
2684 return undef unless defined $paths;
2685 # look for a parent from another branch:
2687 my @b_path_components = split m#/#, $self->{path};
2688 my @a_path_components;
2689 my $i;
2690 while (@b_path_components) {
2691 $i = $paths->{'/'.join('/', @b_path_components)};
2692 last if $i && defined $i->{copyfrom_path};
2693 unshift(@a_path_components, pop(@b_path_components));
2694 }
2695 return undef unless defined $i && defined $i->{copyfrom_path};
2696 my $branch_from = $i->{copyfrom_path};
2697 if (@a_path_components) {
2698 print STDERR "branch_from: $branch_from => ";
2699 $branch_from .= '/'.join('/', @a_path_components);
2700 print STDERR $branch_from, "\n";
2701 }
2702 my $r = $i->{copyfrom_rev};
2703 my $repos_root = $self->ra->{repos_root};
2704 my $url = $self->ra->{url};
2705 my $new_url = $url . $branch_from;
2706 print STDERR "Found possible branch point: ",
2707 "$new_url => ", $self->full_url, ", $r\n"
2708 unless $::_q > 1;
2709 $branch_from =~ s#^/##;
2710 my $gs = $self->other_gs($new_url, $url,
2711 $branch_from, $r, $self->{ref_id});
2712 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2713 {
2714 my ($base, $head);
2715 if (!defined $r0 || !defined $parent) {
2716 ($base, $head) = parse_revision_argument(0, $r);
2717 } else {
2718 if ($r0 < $r) {
2719 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2720 0, 1, sub { $base = $_[1] - 1 });
2721 }
2722 }
2723 if (defined $base && $base <= $r) {
2724 $gs->fetch($base, $r);
2725 }
2726 ($r0, $parent) = $gs->find_rev_before($r, 1);
2727 }
2728 if (defined $r0 && defined $parent) {
2729 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2730 unless $::_q > 1;
2731 my $ed;
2732 if ($self->ra->can_do_switch) {
2733 $self->assert_index_clean($parent);
2734 print STDERR "Following parent with do_switch\n"
2735 unless $::_q > 1;
2736 # do_switch works with svn/trunk >= r22312, but that
2737 # is not included with SVN 1.4.3 (the latest version
2738 # at the moment), so we can't rely on it
2739 $self->{last_rev} = $r0;
2740 $self->{last_commit} = $parent;
2741 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2742 $gs->ra->gs_do_switch($r0, $rev, $gs,
2743 $self->full_url, $ed)
2744 or die "SVN connection failed somewhere...\n";
2745 } elsif ($self->ra->trees_match($new_url, $r0,
2746 $self->full_url, $rev)) {
2747 print STDERR "Trees match:\n",
2748 " $new_url\@$r0\n",
2749 " ${\$self->full_url}\@$rev\n",
2750 "Following parent with no changes\n"
2751 unless $::_q > 1;
2752 $self->tmp_index_do(sub {
2753 command_noisy('read-tree', $parent);
2754 });
2755 $self->{last_commit} = $parent;
2756 } else {
2757 print STDERR "Following parent with do_update\n"
2758 unless $::_q > 1;
2759 $ed = SVN::Git::Fetcher->new($self);
2760 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2761 or die "SVN connection failed somewhere...\n";
2762 }
2763 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2764 return $self->make_log_entry($rev, [$parent], $ed);
2765 }
2766 return undef;
2767}
2768sub do_fetch {
2770 my ($self, $paths, $rev) = @_;
2771 my $ed;
2772 my ($last_rev, @parents);
2773 if (my $lc = $self->last_commit) {
2774 # we can have a branch that was deleted, then re-added
2775 # under the same name but copied from another path, in
2776 # which case we'll have multiple parents (we don't
2777 # want to break the original ref, nor lose copypath info):
2778 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2779 push @{$log_entry->{parents}}, $lc;
2780 return $log_entry;
2781 }
2782 $ed = SVN::Git::Fetcher->new($self);
2783 $last_rev = $self->{last_rev};
2784 $ed->{c} = $lc;
2785 @parents = ($lc);
2786 } else {
2787 $last_rev = $rev;
2788 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2789 return $log_entry;
2790 }
2791 $ed = SVN::Git::Fetcher->new($self);
2792 }
2793 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2794 die "SVN connection failed somewhere...\n";
2795 }
2796 $self->make_log_entry($rev, \@parents, $ed);
2797}
2798sub mkemptydirs {
2800 my ($self, $r) = @_;
2801 sub scan {
2803 my ($r, $empty_dirs, $line) = @_;
2804 if (defined $r && $line =~ /^r(\d+)$/) {
2805 return 0 if $1 > $r;
2806 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2807 $empty_dirs->{$1} = 1;
2808 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2809 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2810 delete @$empty_dirs{@d};
2811 }
2812 1; # continue
2813 };
2814 my %empty_dirs = ();
2816 my $gz_file = "$self->{dir}/unhandled.log.gz";
2817 if (-f $gz_file) {
2818 if (!$can_compress) {
2819 warn "Compress::Zlib could not be found; ",
2820 "empty directories in $gz_file will not be read\n";
2821 } else {
2822 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2823 die "Unable to open $gz_file: $!\n";
2824 my $line;
2825 while ($gz->gzreadline($line) > 0) {
2826 scan($r, \%empty_dirs, $line) or last;
2827 }
2828 $gz->gzclose;
2829 }
2830 }
2831 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2833 binmode $fh or croak "binmode: $!";
2834 while (<$fh>) {
2835 scan($r, \%empty_dirs, $_) or last;
2836 }
2837 close $fh;
2838 }
2839 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2841 foreach my $d (sort keys %empty_dirs) {
2842 $d = uri_decode($d);
2843 $d =~ s/$strip//;
2844 next unless length($d);
2845 next if -d $d;
2846 if (-e $d) {
2847 warn "$d exists but is not a directory\n";
2848 } else {
2849 print "creating empty directory: $d\n";
2850 mkpath([$d]);
2851 }
2852 }
2853}
2854sub get_untracked {
2856 my ($self, $ed) = @_;
2857 my @out;
2858 my $h = $ed->{empty};
2859 foreach (sort keys %$h) {
2860 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2861 push @out, " $act: " . uri_encode($_);
2862 warn "W: $act: $_\n";
2863 }
2864 foreach my $t (qw/dir_prop file_prop/) {
2865 $h = $ed->{$t} or next;
2866 foreach my $path (sort keys %$h) {
2867 my $ppath = $path eq '' ? '.' : $path;
2868 foreach my $prop (sort keys %{$h->{$path}}) {
2869 next if $SKIP_PROP{$prop};
2870 my $v = $h->{$path}->{$prop};
2871 my $t_ppath_prop = "$t: " .
2872 uri_encode($ppath) . ' ' .
2873 uri_encode($prop);
2874 if (defined $v) {
2875 push @out, " +$t_ppath_prop " .
2876 uri_encode($v);
2877 } else {
2878 push @out, " -$t_ppath_prop";
2879 }
2880 }
2881 }
2882 }
2883 foreach my $t (qw/absent_file absent_directory/) {
2884 $h = $ed->{$t} or next;
2885 foreach my $parent (sort keys %$h) {
2886 foreach my $path (sort @{$h->{$parent}}) {
2887 push @out, " $t: " .
2888 uri_encode("$parent/$path");
2889 warn "W: $t: $parent/$path ",
2890 "Insufficient permissions?\n";
2891 }
2892 }
2893 }
2894 \@out;
2895}
2896# parse_svn_date(DATE)
2898# --------------------
2899# Given a date (in UTC) from Subversion, return a string in the format
2900# "<TZ Offset> <local date/time>" that Git will use.
2901#
2902# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2903# is true we'll convert it to the local timezone instead.
2904sub parse_svn_date {
2905 my $date = shift || return '+0000 1970-01-01 00:00:00';
2906 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2907 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2908 croak "Unable to parse date: $date\n";
2909 my $parsed_date; # Set next.
2910 if ($Git::SVN::_localtime) {
2912 # Translate the Subversion datetime to an epoch time.
2913 # Begin by switching ourselves to $date's timezone, UTC.
2914 my $old_env_TZ = $ENV{TZ};
2915 $ENV{TZ} = 'UTC';
2916 my $epoch_in_UTC =
2918 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2919 # Determine our local timezone (including DST) at the
2921 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2922 # value of TZ, if any, at the time we were run.
2923 if (defined $Git::SVN::Log::TZ) {
2924 $ENV{TZ} = $Git::SVN::Log::TZ;
2925 } else {
2926 delete $ENV{TZ};
2927 }
2928 my $our_TZ =
2930 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2931 # This converts $epoch_in_UTC into our local timezone.
2933 my ($sec, $min, $hour, $mday, $mon, $year,
2934 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2935 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2937 $our_TZ, $year + 1900, $mon + 1,
2938 $mday, $hour, $min, $sec);
2939 # Reset us to the timezone in effect when we entered
2941 # this routine.
2942 if (defined $old_env_TZ) {
2943 $ENV{TZ} = $old_env_TZ;
2944 } else {
2945 delete $ENV{TZ};
2946 }
2947 } else {
2948 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2949 }
2950 return $parsed_date;
2952}
2953sub other_gs {
2955 my ($self, $new_url, $url,
2956 $branch_from, $r, $old_ref_id) = @_;
2957 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2958 unless ($gs) {
2959 my $ref_id = $old_ref_id;
2960 $ref_id =~ s/\@\d+$//;
2961 $ref_id .= "\@$r";
2962 # just grow a tail if we're not unique enough :x
2963 $ref_id .= '-' while find_ref($ref_id);
2964 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2965 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2966 if ($u =~ s#^\Q$url\E(/|$)##) {
2967 $p = $u;
2968 $u = $url;
2969 $repo_id = $self->{repo_id};
2970 }
2971 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2972 }
2973 $gs
2974}
2975sub call_authors_prog {
2977 my ($orig_author) = @_;
2978 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2979 my $author = `$::_authors_prog $orig_author`;
2980 if ($? != 0) {
2981 die "$::_authors_prog failed with exit code $?\n"
2982 }
2983 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2984 my ($name, $email) = ($1, $2);
2985 $email = undef if length $2 == 0;
2986 return [$name, $email];
2987 } else {
2988 die "Author: $orig_author: $::_authors_prog returned "
2989 . "invalid author format: $author\n";
2990 }
2991}
2992sub check_author {
2994 my ($author) = @_;
2995 if (!defined $author || length $author == 0) {
2996 $author = '(no author)';
2997 }
2998 if (!defined $::users{$author}) {
2999 if (defined $::_authors_prog) {
3000 $::users{$author} = call_authors_prog($author);
3001 } elsif (defined $::_authors) {
3002 die "Author: $author not defined in $::_authors file\n";
3003 }
3004 }
3005 $author;
3006}
3007sub find_extra_svk_parents {
3009 my ($self, $ed, $tickets, $parents) = @_;
3010 # aha! svk:merge property changed...
3011 my @tickets = split "\n", $tickets;
3012 my @known_parents;
3013 for my $ticket ( @tickets ) {
3014 my ($uuid, $path, $rev) = split /:/, $ticket;
3015 if ( $uuid eq $self->ra_uuid ) {
3016 my $url = $self->{url};
3017 my $repos_root = $url;
3018 my $branch_from = $path;
3019 $branch_from =~ s{^/}{};
3020 my $gs = $self->other_gs($repos_root."/".$branch_from,
3021 $url,
3022 $branch_from,
3023 $rev,
3024 $self->{ref_id});
3025 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3026 # wahey! we found it, but it might be
3027 # an old one (!)
3028 push @known_parents, [ $rev, $commit ];
3029 }
3030 }
3031 }
3032 # Ordering matters; highest-numbered commit merge tickets
3033 # first, as they may account for later merge ticket additions
3034 # or changes.
3035 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3036 for my $parent ( @known_parents ) {
3037 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3038 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3039 my $new;
3040 while ( <$msg_fh> ) {
3041 $new=1;last;
3042 }
3043 command_close_pipe($msg_fh, $ctx);
3044 if ( $new ) {
3045 print STDERR
3046 "Found merge parent (svk:merge ticket): $parent\n";
3047 push @$parents, $parent;
3048 }
3049 }
3050}
3051sub lookup_svn_merge {
3053 my $uuid = shift;
3054 my $url = shift;
3055 my $merge = shift;
3056 my ($source, $revs) = split ":", $merge;
3058 my $path = $source;
3059 $path =~ s{^/}{};
3060 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3061 if ( !$gs ) {
3062 warn "Couldn't find revmap for $url$source\n";
3063 return;
3064 }
3065 my @ranges = split ",", $revs;
3066 my ($tip, $tip_commit);
3067 my @merged_commit_ranges;
3068 # find the tip
3069 for my $range ( @ranges ) {
3070 my ($bottom, $top) = split "-", $range;
3071 $top ||= $bottom;
3072 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3073 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3074 unless ($top_commit and $bottom_commit) {
3076 warn "W:unknown path/rev in svn:mergeinfo "
3077 ."dirprop: $source:$range\n";
3078 next;
3079 }
3080 push @merged_commit_ranges,
3082 "$bottom_commit^..$top_commit";
3083 if ( !defined $tip or $top > $tip ) {
3085 $tip = $top;
3086 $tip_commit = $top_commit;
3087 }
3088 }
3089 return ($tip_commit, @merged_commit_ranges);
3090}
3091sub _rev_list {
3093 my ($msg_fh, $ctx) = command_output_pipe(
3094 "rev-list", @_,
3095 );
3096 my @rv;
3097 while ( <$msg_fh> ) {
3098 chomp;
3099 push @rv, $_;
3100 }
3101 command_close_pipe($msg_fh, $ctx);
3102 @rv;
3103}
3104sub check_cherry_pick {
3106 my $base = shift;
3107 my $tip = shift;
3108 my @ranges = @_;
3109 my %commits = map { $_ => 1 }
3110 _rev_list("--no-merges", $tip, "--not", $base);
3111 for my $range ( @ranges ) {
3112 delete @commits{_rev_list($range)};
3113 }
3114 for my $commit (keys %commits) {
3115 if (has_no_changes($commit)) {
3116 delete $commits{$commit};
3117 }
3118 }
3119 return (keys %commits);
3120}
3121sub has_no_changes {
3123 my $commit = shift;
3124 my @revs = split / /, command_oneline(
3126 qw(rev-list --parents -1 -m), $commit);
3127 # Commits with no parents, e.g. the start of a partial branch,
3129 # have changes by definition.
3130 return 1 if (@revs < 2);
3131 # Commits with multiple parents, e.g a merge, have no changes
3133 # by definition.
3134 return 0 if (@revs > 2);
3135 return (command_oneline("rev-parse", "$commit^{tree}") eq
3137 command_oneline("rev-parse", "$commit~1^{tree}"));
3138}
3139# The GIT_DIR environment variable is not always set until after the command
3141# line arguments are processed, so we can't memoize in a BEGIN block.
3142{
3143 my $memoized = 0;
3144 sub memoize_svn_mergeinfo_functions {
3146 return if $memoized;
3147 $memoized = 1;
3148 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3150 mkpath([$cache_path]) unless -d $cache_path;
3151 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3153 "$cache_path/lookup_svn_merge.db", 'nstore';
3154 memoize 'lookup_svn_merge',
3155 SCALAR_CACHE => 'FAULT',
3156 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3157 ;
3158 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3160 "$cache_path/check_cherry_pick.db", 'nstore';
3161 memoize 'check_cherry_pick',
3162 SCALAR_CACHE => 'FAULT',
3163 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3164 ;
3165 tie my %has_no_changes_cache => 'Memoize::Storable',
3167 "$cache_path/has_no_changes.db", 'nstore';
3168 memoize 'has_no_changes',
3169 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3170 LIST_CACHE => 'FAULT',
3171 ;
3172 }
3173 sub unmemoize_svn_mergeinfo_functions {
3175 return if not $memoized;
3176 $memoized = 0;
3177 Memoize::unmemoize 'lookup_svn_merge';
3179 Memoize::unmemoize 'check_cherry_pick';
3180 Memoize::unmemoize 'has_no_changes';
3181 }
3182}
3183END {
3185 # Force cache writeout explicitly instead of waiting for
3186 # global destruction to avoid segfault in Storable:
3187 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3188 unmemoize_svn_mergeinfo_functions();
3189}
3190sub parents_exclude {
3192 my $parents = shift;
3193 my @commits = @_;
3194 return unless @commits;
3195 my @excluded;
3197 my $excluded;
3198 do {
3199 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3200 $excluded = command_oneline(@cmd);
3201 if ( $excluded ) {
3202 my @new;
3203 my $found;
3204 for my $commit ( @commits ) {
3205 if ( $commit eq $excluded ) {
3206 push @excluded, $commit;
3207 $found++;
3208 last;
3209 }
3210 else {
3211 push @new, $commit;
3212 }
3213 }
3214 die "saw commit '$excluded' in rev-list output, "
3215 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3216 unless $found;
3217 @commits = @new;
3218 }
3219 }
3220 while ($excluded and @commits);
3221 return @excluded;
3223}
3224# note: this function should only be called if the various dirprops
3227# have actually changed
3228sub find_extra_svn_parents {
3229 my ($self, $ed, $mergeinfo, $parents) = @_;
3230 # aha! svk:merge property changed...
3231 memoize_svn_mergeinfo_functions();
3233 # We first search for merged tips which are not in our
3235 # history. Then, we figure out which git revisions are in
3236 # that tip, but not this revision. If all of those revisions
3237 # are now marked as merge, we can add the tip as a parent.
3238 my @merges = split "\n", $mergeinfo;
3239 my @merge_tips;
3240 my $url = $self->{url};
3241 my $uuid = $self->ra_uuid;
3242 my %ranges;
3243 for my $merge ( @merges ) {
3244 my ($tip_commit, @ranges) =
3245 lookup_svn_merge( $uuid, $url, $merge );
3246 unless (!$tip_commit or
3247 grep { $_ eq $tip_commit } @$parents ) {
3248 push @merge_tips, $tip_commit;
3249 $ranges{$tip_commit} = \@ranges;
3250 } else {
3251 push @merge_tips, undef;
3252 }
3253 }
3254 my %excluded = map { $_ => 1 }
3256 parents_exclude($parents, grep { defined } @merge_tips);
3257 # check merge tips for new parents
3259 my @new_parents;
3260 for my $merge_tip ( @merge_tips ) {
3261 my $spec = shift @merges;
3262 next unless $merge_tip and $excluded{$merge_tip};
3263 my $ranges = $ranges{$merge_tip};
3265 # check out 'new' tips
3267 my $merge_base;
3268 eval {
3269 $merge_base = command_oneline(
3270 "merge-base",
3271 @$parents, $merge_tip,
3272 );
3273 };
3274 if ($@) {
3275 die "An error occurred during merge-base"
3276 unless $@->isa("Git::Error::Command");
3277 warn "W: Cannot find common ancestor between ".
3279 "@$parents and $merge_tip. Ignoring merge info.\n";
3280 next;
3281 }
3282 # double check that there are no missing non-merge commits
3284 my (@incomplete) = check_cherry_pick(
3285 $merge_base, $merge_tip,
3286 @$ranges,
3287 );
3288 if ( @incomplete ) {
3290 warn "W:svn cherry-pick ignored ($spec) - missing "
3291 .@incomplete." commit(s) (eg $incomplete[0])\n";
3292 } else {
3293 warn
3294 "Found merge parent (svn:mergeinfo prop): ",
3295 $merge_tip, "\n";
3296 push @new_parents, $merge_tip;
3297 }
3298 }
3299 # cater for merges which merge commits from multiple branches
3301 if ( @new_parents > 1 ) {
3302 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3303 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3304 next if $i == $j;
3305 next unless $new_parents[$i];
3306 next unless $new_parents[$j];
3307 my $revs = command_oneline(
3308 "rev-list", "-1",
3309 "$new_parents[$i]..$new_parents[$j]",
3310 );
3311 if ( !$revs ) {
3312 undef($new_parents[$j]);
3313 }
3314 }
3315 }
3316 }
3317 push @$parents, grep { defined } @new_parents;
3318}
3319sub make_log_entry {
3321 my ($self, $rev, $parents, $ed) = @_;
3322 my $untracked = $self->get_untracked($ed);
3323 my @parents = @$parents;
3325 my $ps = $ed->{path_strip} || "";
3326 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3327 my $props = $ed->{dir_prop}{$path};
3328 if ( $props->{"svk:merge"} ) {
3329 $self->find_extra_svk_parents
3330 ($ed, $props->{"svk:merge"}, \@parents);
3331 }
3332 if ( $props->{"svn:mergeinfo"} ) {
3333 $self->find_extra_svn_parents
3334 ($ed,
3335 $props->{"svn:mergeinfo"},
3336 \@parents);
3337 }
3338 }
3339 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3341 print $un "r$rev\n" or croak $!;
3342 print $un $_, "\n" foreach @$untracked;
3343 my %log_entry = ( parents => \@parents, revision => $rev,
3344 log => '');
3345 my $headrev;
3347 my $logged = delete $self->{logged_rev_props};
3348 if (!$logged || $self->{-want_revprops}) {
3349 my $rp = $self->ra->rev_proplist($rev);
3350 foreach (sort keys %$rp) {
3351 my $v = $rp->{$_};
3352 if (/^svn:(author|date|log)$/) {
3353 $log_entry{$1} = $v;
3354 } elsif ($_ eq 'svm:headrev') {
3355 $headrev = $v;
3356 } else {
3357 print $un " rev_prop: ", uri_encode($_), ' ',
3358 uri_encode($v), "\n";
3359 }
3360 }
3361 } else {
3362 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3363 }
3364 close $un or croak $!;
3365 $log_entry{date} = parse_svn_date($log_entry{date});
3367 $log_entry{log} .= "\n";
3368 my $author = $log_entry{author} = check_author($log_entry{author});
3369 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3370 : ($author, undef);
3371 my ($commit_name, $commit_email) = ($name, $email);
3373 if ($_use_log_author) {
3374 my $name_field;
3375 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3376 $name_field = $1;
3377 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3378 $name_field = $1;
3379 }
3380 if (!defined $name_field) {
3381 if (!defined $email) {
3382 $email = $name;
3383 }
3384 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3385 ($name, $email) = ($1, $2);
3386 } elsif ($name_field =~ /(.*)@/) {
3387 ($name, $email) = ($1, $name_field);
3388 } else {
3389 ($name, $email) = ($name_field, $name_field);
3390 }
3391 }
3392 if (defined $headrev && $self->use_svm_props) {
3393 if ($self->rewrite_root) {
3394 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3395 "options set!\n";
3396 }
3397 if ($self->rewrite_uuid) {
3398 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3399 "options set!\n";
3400 }
3401 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3402 # we don't want "SVM: initializing mirror for junk" ...
3403 return undef if $r == 0;
3404 my $svm = $self->svm;
3405 if ($uuid ne $svm->{uuid}) {
3406 die "UUID mismatch on SVM path:\n",
3407 "expected: $svm->{uuid}\n",
3408 " got: $uuid\n";
3409 }
3410 my $full_url = $self->full_url;
3411 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3412 die "Failed to replace '$svm->{replace}' with ",
3413 "'$svm->{source}' in $full_url\n";
3414 # throw away username for storing in records
3415 remove_username($full_url);
3416 $log_entry{metadata} = "$full_url\@$r $uuid";
3417 $log_entry{svm_revision} = $r;
3418 $email ||= "$author\@$uuid";
3419 $commit_email ||= "$author\@$uuid";
3420 } elsif ($self->use_svnsync_props) {
3421 my $full_url = $self->svnsync->{url};
3422 $full_url .= "/$self->{path}" if length $self->{path};
3423 remove_username($full_url);
3424 my $uuid = $self->svnsync->{uuid};
3425 $log_entry{metadata} = "$full_url\@$rev $uuid";
3426 $email ||= "$author\@$uuid";
3427 $commit_email ||= "$author\@$uuid";
3428 } else {
3429 my $url = $self->metadata_url;
3430 remove_username($url);
3431 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3432 $log_entry{metadata} = "$url\@$rev " . $uuid;
3433 $email ||= "$author\@" . $uuid;
3434 $commit_email ||= "$author\@" . $uuid;
3435 }
3436 $log_entry{name} = $name;
3437 $log_entry{email} = $email;
3438 $log_entry{commit_name} = $commit_name;
3439 $log_entry{commit_email} = $commit_email;
3440 \%log_entry;
3441}
3442sub fetch {
3444 my ($self, $min_rev, $max_rev, @parents) = @_;
3445 my ($last_rev, $last_commit) = $self->last_rev_commit;
3446 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3447 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3448}
3449sub set_tree_cb {
3451 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3452 $self->{inject_parents} = { $rev => $tree };
3453 $self->fetch(undef, undef);
3454}
3455sub set_tree {
3457 my ($self, $tree) = (shift, shift);
3458 my $log_entry = ::get_commit_entry($tree);
3459 unless ($self->{last_rev}) {
3460 ::fatal("Must have an existing revision to commit");
3461 }
3462 my %ed_opts = ( r => $self->{last_rev},
3463 log => $log_entry->{log},
3464 ra => $self->ra,
3465 tree_a => $self->{last_commit},
3466 tree_b => $tree,
3467 editor_cb => sub {
3468 $self->set_tree_cb($log_entry, $tree, @_) },
3469 svn_path => $self->{path} );
3470 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3471 print "No changes\nr$self->{last_rev} = $tree\n";
3472 }
3473}
3474sub rebuild_from_rev_db {
3476 my ($self, $path) = @_;
3477 my $r = -1;
3478 open my $fh, '<', $path or croak "open: $!";
3479 binmode $fh or croak "binmode: $!";
3480 while (<$fh>) {
3481 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3482 chomp($_);
3483 ++$r;
3484 next if $_ eq ('0' x 40);
3485 $self->rev_map_set($r, $_);
3486 print "r$r = $_\n";
3487 }
3488 close $fh or croak "close: $!";
3489 unlink $path or croak "unlink: $!";
3490}
3491sub rebuild {
3493 my ($self) = @_;
3494 my $map_path = $self->map_path;
3495 my $partial = (-e $map_path && ! -z $map_path);
3496 return unless ::verify_ref($self->refname.'^0');
3497 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3498 my $rev_db = $self->rev_db_path;
3499 $self->rebuild_from_rev_db($rev_db);
3500 if ($self->use_svm_props) {
3501 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3502 $self->rebuild_from_rev_db($svm_rev_db);
3503 }
3504 $self->unlink_rev_db_symlink;
3505 return;
3506 }
3507 print "Rebuilding $map_path ...\n" if (!$partial);
3508 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3509 (undef, undef));
3510 my ($log, $ctx) =
3511 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3512 ($head ? "$head.." : "") . $self->refname,
3513 '--');
3514 my $metadata_url = $self->metadata_url;
3515 remove_username($metadata_url);
3516 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3517 my $c;
3518 while (<$log>) {
3519 if ( m{^commit ($::sha1)$} ) {
3520 $c = $1;
3521 next;
3522 }
3523 next unless s{^\s*(git-svn-id:)}{$1};
3524 my ($url, $rev, $uuid) = ::extract_metadata($_);
3525 remove_username($url);
3526 # ignore merges (from set-tree)
3528 next if (!defined $rev || !$uuid);
3529 # if we merged or otherwise started elsewhere, this is
3531 # how we break out of it
3532 if (($uuid ne $svn_uuid) ||
3533 ($metadata_url && $url && ($url ne $metadata_url))) {
3534 next;
3535 }
3536 if ($partial && $head) {
3537 print "Partial-rebuilding $map_path ...\n";
3538 print "Currently at $base_rev = $head\n";
3539 $head = undef;
3540 }
3541 $self->rev_map_set($rev, $c);
3543 print "r$rev = $c\n";
3544 }
3545 command_close_pipe($log, $ctx);
3546 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3547 my $rev_db_path = $self->rev_db_path;
3548 if (-f $self->rev_db_path) {
3549 unlink $self->rev_db_path or croak "unlink: $!";
3550 }
3551 $self->unlink_rev_db_symlink;
3552}
3553# rev_map:
3555# Tie::File seems to be prone to offset errors if revisions get sparse,
3556# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3557# one of my favorite modules is out :< Next up would be one of the DBM
3558# modules, but I'm not sure which is most portable...
3559#
3560# This is the replacement for the rev_db format, which was too big
3561# and inefficient for large repositories with a lot of sparse history
3562# (mainly tags)
3563#
3564# The format is this:
3565# - 24 bytes for every record,
3566# * 4 bytes for the integer representing an SVN revision number
3567# * 20 bytes representing the sha1 of a git commit
3568# - No empty padding records like the old format
3569# (except the last record, which can be overwritten)
3570# - new records are written append-only since SVN revision numbers
3571# increase monotonically
3572# - lookups on SVN revision number are done via a binary search
3573# - Piping the file to xxd -c24 is a good way of dumping it for
3574# viewing or editing (piped back through xxd -r), should the need
3575# ever arise.
3576# - The last record can be padding revision with an all-zero sha1
3577# This is used to optimize fetch performance when using multiple
3578# "fetch" directives in .git/config
3579#
3580# These files are disposable unless noMetadata or useSvmProps is set
3581sub _rev_map_set {
3583 my ($fh, $rev, $commit) = @_;
3584 binmode $fh or croak "binmode: $!";
3586 my $size = (stat($fh))[7];
3587 ($size % 24) == 0 or croak "inconsistent size: $size";
3588 my $wr_offset = 0;
3590 if ($size > 0) {
3591 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3592 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3593 $read == 24 or croak "read only $read bytes (!= 24)";
3594 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3595 if ($last_commit eq ('0' x40)) {
3596 if ($size >= 48) {
3597 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3598 $read = sysread($fh, $buf, 24) or
3599 croak "read: $!";
3600 $read == 24 or
3601 croak "read only $read bytes (!= 24)";
3602 ($last_rev, $last_commit) =
3603 unpack(rev_map_fmt, $buf);
3604 if ($last_commit eq ('0' x40)) {
3605 croak "inconsistent .rev_map\n";
3606 }
3607 }
3608 if ($last_rev >= $rev) {
3609 croak "last_rev is higher!: $last_rev >= $rev";
3610 }
3611 $wr_offset = -24;
3612 }
3613 }
3614 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3615 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3616 croak "write: $!";
3617}
3618sub _rev_map_reset {
3620 my ($fh, $rev, $commit) = @_;
3621 my $c = _rev_map_get($fh, $rev);
3622 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3623 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3624 truncate $fh, $offset or croak "truncate: $!";
3625}
3626sub mkfile {
3628 my ($path) = @_;
3629 unless (-e $path) {
3630 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3631 mkpath([$dir]) unless -d $dir;
3632 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3633 close $fh or die "Couldn't close (create) $path: $!\n";
3634 }
3635}
3636sub rev_map_set {
3638 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3639 defined $commit or die "missing arg3\n";
3640 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3641 my $db = $self->map_path($uuid);
3642 my $db_lock = "$db.lock";
3643 my $sig;
3644 $update_ref ||= 0;
3645 if ($update_ref) {
3646 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3647 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3648 }
3649 mkfile($db);
3650 $LOCKFILES{$db_lock} = 1;
3652 my $sync;
3653 # both of these options make our .rev_db file very, very important
3654 # and we can't afford to lose it because rebuild() won't work
3655 if ($self->use_svm_props || $self->no_metadata) {
3656 $sync = 1;
3657 copy($db, $db_lock) or die "rev_map_set(@_): ",
3658 "Failed to copy: ",
3659 "$db => $db_lock ($!)\n";
3660 } else {
3661 rename $db, $db_lock or die "rev_map_set(@_): ",
3662 "Failed to rename: ",
3663 "$db => $db_lock ($!)\n";
3664 }
3665 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3667 or croak "Couldn't open $db_lock: $!\n";
3668 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3669 _rev_map_set($fh, $rev, $commit);
3670 if ($sync) {
3671 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3672 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3673 }
3674 close $fh or croak $!;
3675 if ($update_ref) {
3676 $_head = $self;
3677 my $note = "";
3678 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3679 command_noisy('update-ref', '-m', "r$rev$note",
3680 $self->refname, $commit);
3681 }
3682 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3683 "$db_lock => $db ($!)\n";
3684 delete $LOCKFILES{$db_lock};
3685 if ($update_ref) {
3686 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3687 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3688 kill $sig, $$ if defined $sig;
3689 }
3690}
3691# If want_commit, this will return an array of (rev, commit) where
3693# commit _must_ be a valid commit in the archive.
3694# Otherwise, it'll return the max revision (whether or not the
3695# commit is valid or just a 0x40 placeholder).
3696sub rev_map_max {
3697 my ($self, $want_commit) = @_;
3698 $self->rebuild;
3699 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3700 $want_commit ? ($r, $c) : $r;
3701}
3702sub rev_map_max_norebuild {
3704 my ($self, $want_commit) = @_;
3705 my $map_path = $self->map_path;
3706 stat $map_path or return $want_commit ? (0, undef) : 0;
3707 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3708 binmode $fh or croak "binmode: $!";
3709 my $size = (stat($fh))[7];
3710 ($size % 24) == 0 or croak "inconsistent size: $size";
3711 if ($size == 0) {
3713 close $fh or croak "close: $!";
3714 return $want_commit ? (0, undef) : 0;
3715 }
3716 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3718 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3719 my ($r, $c) = unpack(rev_map_fmt, $buf);
3720 if ($want_commit && $c eq ('0' x40)) {
3721 if ($size < 48) {
3722 return $want_commit ? (0, undef) : 0;
3723 }
3724 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3725 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3726 ($r, $c) = unpack(rev_map_fmt, $buf);
3727 if ($c eq ('0'x40)) {
3728 croak "Penultimate record is all-zeroes in $map_path";
3729 }
3730 }
3731 close $fh or croak "close: $!";
3732 $want_commit ? ($r, $c) : $r;
3733}
3734sub rev_map_get {
3736 my ($self, $rev, $uuid) = @_;
3737 my $map_path = $self->map_path($uuid);
3738 return undef unless -e $map_path;
3739 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3741 my $c = _rev_map_get($fh, $rev);
3742 close($fh) or croak "close: $!";
3743 $c
3744}
3745sub _rev_map_get {
3747 my ($fh, $rev) = @_;
3748 binmode $fh or croak "binmode: $!";
3750 my $size = (stat($fh))[7];
3751 ($size % 24) == 0 or croak "inconsistent size: $size";
3752 if ($size == 0) {
3754 return undef;
3755 }
3756 my ($l, $u) = (0, $size - 24);
3758 my ($r, $c, $buf);
3759 while ($l <= $u) {
3761 my $i = int(($l/24 + $u/24) / 2) * 24;
3762 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3763 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3764 my ($r, $c) = unpack(rev_map_fmt, $buf);
3765 if ($r < $rev) {
3767 $l = $i + 24;
3768 } elsif ($r > $rev) {
3769 $u = $i - 24;
3770 } else { # $r == $rev
3771 return $c eq ('0' x 40) ? undef : $c;
3772 }
3773 }
3774 undef;
3775}
3776# Finds the first svn revision that exists on (if $eq_ok is true) or
3778# before $rev for the current branch. It will not search any lower
3779# than $min_rev. Returns the git commit hash and svn revision number
3780# if found, else (undef, undef).
3781sub find_rev_before {
3782 my ($self, $rev, $eq_ok, $min_rev) = @_;
3783 --$rev unless $eq_ok;
3784 $min_rev ||= 1;
3785 my $max_rev = $self->rev_map_max;
3786 $rev = $max_rev if ($rev > $max_rev);
3787 while ($rev >= $min_rev) {
3788 if (my $c = $self->rev_map_get($rev)) {
3789 return ($rev, $c);
3790 }
3791 --$rev;
3792 }
3793 return (undef, undef);
3794}
3795# Finds the first svn revision that exists on (if $eq_ok is true) or
3797# after $rev for the current branch. It will not search any higher
3798# than $max_rev. Returns the git commit hash and svn revision number
3799# if found, else (undef, undef).
3800sub find_rev_after {
3801 my ($self, $rev, $eq_ok, $max_rev) = @_;
3802 ++$rev unless $eq_ok;
3803 $max_rev ||= $self->rev_map_max;
3804 while ($rev <= $max_rev) {
3805 if (my $c = $self->rev_map_get($rev)) {
3806 return ($rev, $c);
3807 }
3808 ++$rev;
3809 }
3810 return (undef, undef);
3811}
3812sub _new {
3814 my ($class, $repo_id, $ref_id, $path) = @_;
3815 unless (defined $repo_id && length $repo_id) {
3816 $repo_id = $Git::SVN::default_repo_id;
3817 }
3818 unless (defined $ref_id && length $ref_id) {
3819 $_prefix = '' unless defined($_prefix);
3820 $_[2] = $ref_id =
3821 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3822 }
3823 $_[1] = $repo_id;
3824 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3825 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3827 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3828 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3829 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3830 if (-d $old_dir && ! -d $dir) {
3831 $dir = $old_dir;
3832 }
3833 }
3834 $_[3] = $path = '' unless (defined $path);
3836 mkpath([$dir]);
3837 bless {
3838 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3839 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3840 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3841}
3842# for read-only access of old .rev_db formats
3844sub unlink_rev_db_symlink {
3845 my ($self) = @_;
3846 my $link = $self->rev_db_path;
3847 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3848 if (-l $link) {
3849 unlink $link or croak "unlink: $link failed!";
3850 }
3851}
3852sub rev_db_path {
3854 my ($self, $uuid) = @_;
3855 my $db_path = $self->map_path($uuid);
3856 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3857 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3858 $db_path;
3859}
3860# the new replacement for .rev_db
3862sub map_path {
3863 my ($self, $uuid) = @_;
3864 $uuid ||= $self->ra_uuid;
3865 "$self->{map_root}.$uuid";
3866}
3867sub uri_encode {
3869 my ($f) = @_;
3870 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3871 $f
3872}
3873sub uri_decode {
3875 my ($f) = @_;
3876 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3877 $f
3878}
3879sub remove_username {
3881 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3882}
3883package Git::SVN::Prompt;
3885use strict;
3886use warnings;
3887require SVN::Core;
3888use vars qw/$_no_auth_cache $_username/;
3889sub simple {
3891 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3892 $may_save = undef if $_no_auth_cache;
3893 $default_username = $_username if defined $_username;
3894 if (defined $default_username && length $default_username) {
3895 if (defined $realm && length $realm) {
3896 print STDERR "Authentication realm: $realm\n";
3897 STDERR->flush;
3898 }
3899 $cred->username($default_username);
3900 } else {
3901 username($cred, $realm, $may_save, $pool);
3902 }
3903 $cred->password(_read_password("Password for '" .
3904 $cred->username . "': ", $realm));
3905 $cred->may_save($may_save);
3906 $SVN::_Core::SVN_NO_ERROR;
3907}
3908sub ssl_server_trust {
3910 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3911 $may_save = undef if $_no_auth_cache;
3912 print STDERR "Error validating server certificate for '$realm':\n";
3913 {
3914 no warnings 'once';
3915 # All variables SVN::Auth::SSL::* are used only once,
3916 # so we're shutting up Perl warnings about this.
3917 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3918 print STDERR " - The certificate is not issued ",
3919 "by a trusted authority. Use the\n",
3920 " fingerprint to validate ",
3921 "the certificate manually!\n";
3922 }
3923 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3924 print STDERR " - The certificate hostname ",
3925 "does not match.\n";
3926 }
3927 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3928 print STDERR " - The certificate is not yet valid.\n";
3929 }
3930 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3931 print STDERR " - The certificate has expired.\n";
3932 }
3933 if ($failures & $SVN::Auth::SSL::OTHER) {
3934 print STDERR " - The certificate has ",
3935 "an unknown error.\n";
3936 }
3937 } # no warnings 'once'
3938 printf STDERR
3939 "Certificate information:\n".
3940 " - Hostname: %s\n".
3941 " - Valid: from %s until %s\n".
3942 " - Issuer: %s\n".
3943 " - Fingerprint: %s\n",
3944 map $cert_info->$_, qw(hostname valid_from valid_until
3945 issuer_dname fingerprint);
3946 my $choice;
3947prompt:
3948 print STDERR $may_save ?
3949 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3950 "(R)eject or accept (t)emporarily? ";
3951 STDERR->flush;
3952 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3953 if ($choice =~ /^t$/i) {
3954 $cred->may_save(undef);
3955 } elsif ($choice =~ /^r$/i) {
3956 return -1;
3957 } elsif ($may_save && $choice =~ /^p$/i) {
3958 $cred->may_save($may_save);
3959 } else {
3960 goto prompt;
3961 }
3962 $cred->accepted_failures($failures);
3963 $SVN::_Core::SVN_NO_ERROR;
3964}
3965sub ssl_client_cert {
3967 my ($cred, $realm, $may_save, $pool) = @_;
3968 $may_save = undef if $_no_auth_cache;
3969 print STDERR "Client certificate filename: ";
3970 STDERR->flush;
3971 chomp(my $filename = <STDIN>);
3972 $cred->cert_file($filename);
3973 $cred->may_save($may_save);
3974 $SVN::_Core::SVN_NO_ERROR;
3975}
3976sub ssl_client_cert_pw {
3978 my ($cred, $realm, $may_save, $pool) = @_;
3979 $may_save = undef if $_no_auth_cache;
3980 $cred->password(_read_password("Password: ", $realm));
3981 $cred->may_save($may_save);
3982 $SVN::_Core::SVN_NO_ERROR;
3983}
3984sub username {
3986 my ($cred, $realm, $may_save, $pool) = @_;
3987 $may_save = undef if $_no_auth_cache;
3988 if (defined $realm && length $realm) {
3989 print STDERR "Authentication realm: $realm\n";
3990 }
3991 my $username;
3992 if (defined $_username) {
3993 $username = $_username;
3994 } else {
3995 print STDERR "Username: ";
3996 STDERR->flush;
3997 chomp($username = <STDIN>);
3998 }
3999 $cred->username($username);
4000 $cred->may_save($may_save);
4001 $SVN::_Core::SVN_NO_ERROR;
4002}
4003sub _read_password {
4005 my ($prompt, $realm) = @_;
4006 my $password = '';
4007 if (exists $ENV{GIT_ASKPASS}) {
4008 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4009 $password = <PH>;
4010 $password =~ s/[\012\015]//; # \n\r
4011 close(PH);
4012 } else {
4013 print STDERR $prompt;
4014 STDERR->flush;
4015 require Term::ReadKey;
4016 Term::ReadKey::ReadMode('noecho');
4017 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4018 last if $key =~ /[\012\015]/; # \n\r
4019 $password .= $key;
4020 }
4021 Term::ReadKey::ReadMode('restore');
4022 print STDERR "\n";
4023 STDERR->flush;
4024 }
4025 $password;
4026}
4027package SVN::Git::Fetcher;
4029use vars qw/@ISA/;
4030use strict;
4031use warnings;
4032use Carp qw/croak/;
4033use IO::File qw//;
4034use vars qw/$_ignore_regex/;
4035# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4037sub new {
4038 my ($class, $git_svn, $switch_path) = @_;
4039 my $self = SVN::Delta::Editor->new;
4040 bless $self, $class;
4041 if (exists $git_svn->{last_commit}) {
4042 $self->{c} = $git_svn->{last_commit};
4043 $self->{empty_symlinks} =
4044 _mark_empty_symlinks($git_svn, $switch_path);
4045 }
4046 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4047 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4048 $self->{empty} = {};
4049 $self->{dir_prop} = {};
4050 $self->{file_prop} = {};
4051 $self->{absent_dir} = {};
4052 $self->{absent_file} = {};
4053 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4054 $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4055 $self;
4056}
4057# this uses the Ra object, so it must be called before do_{switch,update},
4059# not inside them (when the Git::SVN::Fetcher object is passed) to
4060# do_{switch,update}
4061sub _mark_empty_symlinks {
4062 my ($git_svn, $switch_path) = @_;
4063 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4064 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4065 my %ret;
4067 my ($rev, $cmt) = $git_svn->last_rev_commit;
4068 return {} unless ($rev && $cmt);
4069 # allow the warning to be printed for each revision we fetch to
4071 # ensure the user sees it. The user can also disable the workaround
4072 # on the repository even while git svn is running and the next
4073 # revision fetched will skip this expensive function.
4074 my $printed_warning;
4075 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4076 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4077 local $/ = "\0";
4078 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4079 $pfx .= '/' if length($pfx);
4080 while (<$ls>) {
4081 chomp;
4082 s/\A100644 blob $empty_blob\t//o or next;
4083 unless ($printed_warning) {
4084 print STDERR "Scanning for empty symlinks, ",
4085 "this may take a while if you have ",
4086 "many empty files\n",
4087 "You may disable this with `",
4088 "git config svn.brokenSymlinkWorkaround ",
4089 "false'.\n",
4090 "This may be done in a different ",
4091 "terminal without restarting ",
4092 "git svn\n";
4093 $printed_warning = 1;
4094 }
4095 my $path = $_;
4096 my (undef, $props) =
4097 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4098 if ($props->{'svn:special'}) {
4099 $ret{$path} = 1;
4100 }
4101 }
4102 command_close_pipe($ls, $ctx);
4103 \%ret;
4104}
4105# returns true if a given path is inside a ".git" directory
4107sub in_dot_git {
4108 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4109}
4110# return value: 0 -- don't ignore, 1 -- ignore
4112sub is_path_ignored {
4113 my ($self, $path) = @_;
4114 return 1 if in_dot_git($path);
4115 return 1 if defined($self->{ignore_regex}) &&
4116 $path =~ m!$self->{ignore_regex}!;
4117 return 0 unless defined($_ignore_regex);
4118 return 1 if $path =~ m!$_ignore_regex!o;
4119 return 0;
4120}
4121sub set_path_strip {
4123 my ($self, $path) = @_;
4124 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4125}
4126sub open_root {
4128 { path => '' };
4129}
4130sub open_directory {
4132 my ($self, $path, $pb, $rev) = @_;
4133 { path => $path };
4134}
4135sub git_path {
4137 my ($self, $path) = @_;
4138 if (my $enc = $self->{pathnameencoding}) {
4139 require Encode;
4140 Encode::from_to($path, 'UTF-8', $enc);
4141 }
4142 if ($self->{path_strip}) {
4143 $path =~ s!$self->{path_strip}!! or
4144 die "Failed to strip path '$path' ($self->{path_strip})\n";
4145 }
4146 $path;
4147}
4148sub delete_entry {
4150 my ($self, $path, $rev, $pb) = @_;
4151 return undef if $self->is_path_ignored($path);
4152 my $gpath = $self->git_path($path);
4154 return undef if ($gpath eq '');
4155 # remove entire directories.
4157 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4158 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4159 if ($tree) {
4160 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4161 -r --name-only -z/,
4162 $tree);
4163 local $/ = "\0";
4164 while (<$ls>) {
4165 chomp;
4166 my $rmpath = "$gpath/$_";
4167 $self->{gii}->remove($rmpath);
4168 print "\tD\t$rmpath\n" unless $::_q;
4169 }
4170 print "\tD\t$gpath/\n" unless $::_q;
4171 command_close_pipe($ls, $ctx);
4172 } else {
4173 $self->{gii}->remove($gpath);
4174 print "\tD\t$gpath\n" unless $::_q;
4175 }
4176 $self->{empty}->{$path} = 0;
4177 undef;
4178}
4179sub open_file {
4181 my ($self, $path, $pb, $rev) = @_;
4182 my ($mode, $blob);
4183 goto out if $self->is_path_ignored($path);
4185 my $gpath = $self->git_path($path);
4187 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4188 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4189 unless (defined $mode && defined $blob) {
4190 die "$path was not found in commit $self->{c} (r$rev)\n";
4191 }
4192 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4193 $mode = '120000';
4194 }
4195out:
4196 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4197 pool => SVN::Pool->new, action => 'M' };
4198}
4199sub add_file {
4201 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4202 my $mode;
4203 if (!$self->is_path_ignored($path)) {
4205 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4206 delete $self->{empty}->{$dir};
4207 $mode = '100644';
4208 }
4209 { path => $path, mode_a => $mode, mode_b => $mode,
4210 pool => SVN::Pool->new, action => 'A' };
4211}
4212sub add_directory {
4214 my ($self, $path, $cp_path, $cp_rev) = @_;
4215 goto out if $self->is_path_ignored($path);
4216 my $gpath = $self->git_path($path);
4217 if ($gpath eq '') {
4218 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4219 -r --name-only -z/,
4220 $self->{c});
4221 local $/ = "\0";
4222 while (<$ls>) {
4223 chomp;
4224 $self->{gii}->remove($_);
4225 print "\tD\t$_\n" unless $::_q;
4226 }
4227 command_close_pipe($ls, $ctx);
4228 $self->{empty}->{$path} = 0;
4229 }
4230 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4231 delete $self->{empty}->{$dir};
4232 $self->{empty}->{$path} = 1;
4233out:
4234 { path => $path };
4235}
4236sub change_dir_prop {
4238 my ($self, $db, $prop, $value) = @_;
4239 return undef if $self->is_path_ignored($db->{path});
4240 $self->{dir_prop}->{$db->{path}} ||= {};
4241 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4242 undef;
4243}
4244sub absent_directory {
4246 my ($self, $path, $pb) = @_;
4247 return undef if $self->is_path_ignored($path);
4248 $self->{absent_dir}->{$pb->{path}} ||= [];
4249 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4250 undef;
4251}
4252sub absent_file {
4254 my ($self, $path, $pb) = @_;
4255 return undef if $self->is_path_ignored($path);
4256 $self->{absent_file}->{$pb->{path}} ||= [];
4257 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4258 undef;
4259}
4260sub change_file_prop {
4262 my ($self, $fb, $prop, $value) = @_;
4263 return undef if $self->is_path_ignored($fb->{path});
4264 if ($prop eq 'svn:executable') {
4265 if ($fb->{mode_b} != 120000) {
4266 $fb->{mode_b} = defined $value ? 100755 : 100644;
4267 }
4268 } elsif ($prop eq 'svn:special') {
4269 $fb->{mode_b} = defined $value ? 120000 : 100644;
4270 } else {
4271 $self->{file_prop}->{$fb->{path}} ||= {};
4272 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4273 }
4274 undef;
4275}
4276sub apply_textdelta {
4278 my ($self, $fb, $exp) = @_;
4279 return undef if $self->is_path_ignored($fb->{path});
4280 my $fh = $::_repository->temp_acquire('svn_delta');
4281 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4282 # (but $base does not,) so dup() it for reading in close_file
4283 open my $dup, '<&', $fh or croak $!;
4284 my $base = $::_repository->temp_acquire('git_blob');
4285 if ($fb->{blob}) {
4287 my ($base_is_link, $size);
4288 if ($fb->{mode_a} eq '120000' &&
4290 ! $self->{empty_symlinks}->{$fb->{path}}) {
4291 print $base 'link ' or die "print $!\n";
4292 $base_is_link = 1;
4293 }
4294 retry:
4295 $size = $::_repository->cat_blob($fb->{blob}, $base);
4296 die "Failed to read object $fb->{blob}" if ($size < 0);
4297 if (defined $exp) {
4299 seek $base, 0, 0 or croak $!;
4300 my $got = ::md5sum($base);
4301 if ($got ne $exp) {
4302 my $err = "Checksum mismatch: ".
4303 "$fb->{path} $fb->{blob}\n" .
4304 "expected: $exp\n" .
4305 " got: $got\n";
4306 if ($base_is_link) {
4307 warn $err,
4308 "Retrying... (possibly ",
4309 "a bad symlink from SVN)\n";
4310 $::_repository->temp_reset($base);
4311 $base_is_link = 0;
4312 goto retry;
4313 }
4314 die $err;
4315 }
4316 }
4317 }
4318 seek $base, 0, 0 or croak $!;
4319 $fb->{fh} = $fh;
4320 $fb->{base} = $base;
4321 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4322}
4323sub close_file {
4325 my ($self, $fb, $exp) = @_;
4326 return undef if $self->is_path_ignored($fb->{path});
4327 my $hash;
4329 my $path = $self->git_path($fb->{path});
4330 if (my $fh = $fb->{fh}) {
4331 if (defined $exp) {
4332 seek($fh, 0, 0) or croak $!;
4333 my $got = ::md5sum($fh);
4334 if ($got ne $exp) {
4335 die "Checksum mismatch: $path\n",
4336 "expected: $exp\n got: $got\n";
4337 }
4338 }
4339 if ($fb->{mode_b} == 120000) {
4340 sysseek($fh, 0, 0) or croak $!;
4341 my $rd = sysread($fh, my $buf, 5);
4342 if (!defined $rd) {
4344 croak "sysread: $!\n";
4345 } elsif ($rd == 0) {
4346 warn "$path has mode 120000",
4347 " but it points to nothing\n",
4348 "converting to an empty file with mode",
4349 " 100644\n";
4350 $fb->{mode_b} = '100644';
4351 } elsif ($buf ne 'link ') {
4352 warn "$path has mode 120000",
4353 " but is not a link\n";
4354 } else {
4355 my $tmp_fh = $::_repository->temp_acquire(
4356 'svn_hash');
4357 my $res;
4358 while ($res = sysread($fh, my $str, 1024)) {
4359 my $out = syswrite($tmp_fh, $str, $res);
4360 defined($out) && $out == $res
4361 or croak("write ",
4362 Git::temp_path($tmp_fh),
4363 ": $!\n");
4364 }
4365 defined $res or croak $!;
4366 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4368 Git::temp_release($tmp_fh, 1);
4369 }
4370 }
4371 $hash = $::_repository->hash_and_insert_object(
4373 Git::temp_path($fh));
4374 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4375 Git::temp_release($fb->{base}, 1);
4377 Git::temp_release($fh, 1);
4378 } else {
4379 $hash = $fb->{blob} or die "no blob information\n";
4380 }
4381 $fb->{pool}->clear;
4382 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4383 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4384 undef;
4385}
4386sub abort_edit {
4388 my $self = shift;
4389 $self->{nr} = $self->{gii}->{nr};
4390 delete $self->{gii};
4391 $self->SUPER::abort_edit(@_);
4392}
4393sub close_edit {
4395 my $self = shift;
4396 $self->{git_commit_ok} = 1;
4397 $self->{nr} = $self->{gii}->{nr};
4398 delete $self->{gii};
4399 $self->SUPER::close_edit(@_);
4400}
4401package SVN::Git::Editor;
4403use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4404use strict;
4405use warnings;
4406use Carp qw/croak/;
4407use IO::File;
4408sub new {
4410 my ($class, $opts) = @_;
4411 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4412 die "$_ required!\n" unless (defined $opts->{$_});
4413 }
4414 my $pool = SVN::Pool->new;
4416 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4417 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4418 $opts->{r}, $mods);
4419 # $opts->{ra} functions should not be used after this:
4421 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4422 $opts->{editor_cb}, $pool);
4423 my $self = SVN::Delta::Editor->new(@ce, $pool);
4424 bless $self, $class;
4425 foreach (qw/svn_path r tree_a tree_b/) {
4426 $self->{$_} = $opts->{$_};
4427 }
4428 $self->{url} = $opts->{ra}->{url};
4429 $self->{mods} = $mods;
4430 $self->{types} = $types;
4431 $self->{pool} = $pool;
4432 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4433 $self->{rm} = { };
4434 $self->{path_prefix} = length $self->{svn_path} ?
4435 "$self->{svn_path}/" : '';
4436 $self->{config} = $opts->{config};
4437 return $self;
4438}
4439sub generate_diff {
4441 my ($tree_a, $tree_b) = @_;
4442 my @diff_tree = qw(diff-tree -z -r);
4443 if ($_cp_similarity) {
4444 push @diff_tree, "-C$_cp_similarity";
4445 } else {
4446 push @diff_tree, '-C';
4447 }
4448 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4449 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4450 push @diff_tree, $tree_a, $tree_b;
4451 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4452 local $/ = "\0";
4453 my $state = 'meta';
4454 my @mods;
4455 while (<$diff_fh>) {
4456 chomp $_; # this gets rid of the trailing "\0"
4457 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4458 ($::sha1)\s($::sha1)\s
4459 ([MTCRAD])\d*$/xo) {
4460 push @mods, { mode_a => $1, mode_b => $2,
4461 sha1_a => $3, sha1_b => $4,
4462 chg => $5 };
4463 if ($5 =~ /^(?:C|R)$/) {
4464 $state = 'file_a';
4465 } else {
4466 $state = 'file_b';
4467 }
4468 } elsif ($state eq 'file_a') {
4469 my $x = $mods[$#mods] or croak "Empty array\n";
4470 if ($x->{chg} !~ /^(?:C|R)$/) {
4471 croak "Error parsing $_, $x->{chg}\n";
4472 }
4473 $x->{file_a} = $_;
4474 $state = 'file_b';
4475 } elsif ($state eq 'file_b') {
4476 my $x = $mods[$#mods] or croak "Empty array\n";
4477 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4478 croak "Error parsing $_, $x->{chg}\n";
4479 }
4480 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4481 croak "Error parsing $_, $x->{chg}\n";
4482 }
4483 $x->{file_b} = $_;
4484 $state = 'meta';
4485 } else {
4486 croak "Error parsing $_\n";
4487 }
4488 }
4489 command_close_pipe($diff_fh, $ctx);
4490 \@mods;
4491}
4492sub check_diff_paths {
4494 my ($ra, $pfx, $rev, $mods) = @_;
4495 my %types;
4496 $pfx .= '/' if length $pfx;
4497 sub type_diff_paths {
4499 my ($ra, $types, $path, $rev) = @_;
4500 my @p = split m#/+#, $path;
4501 my $c = shift @p;
4502 unless (defined $types->{$c}) {
4503 $types->{$c} = $ra->check_path($c, $rev);
4504 }
4505 while (@p) {
4506 $c .= '/' . shift @p;
4507 next if defined $types->{$c};
4508 $types->{$c} = $ra->check_path($c, $rev);
4509 }
4510 }
4511 foreach my $m (@$mods) {
4513 foreach my $f (qw/file_a file_b/) {
4514 next unless defined $m->{$f};
4515 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4516 if (length $pfx.$dir && ! defined $types{$dir}) {
4517 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4518 }
4519 }
4520 }
4521 \%types;
4522}
4523sub split_path {
4525 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4526}
4527sub repo_path {
4529 my ($self, $path) = @_;
4530 if (my $enc = $self->{pathnameencoding}) {
4531 require Encode;
4532 Encode::from_to($path, $enc, 'UTF-8');
4533 }
4534 $self->{path_prefix}.(defined $path ? $path : '');
4535}
4536sub url_path {
4538 my ($self, $path) = @_;
4539 if ($self->{url} =~ m#^https?://#) {
4540 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4541 }
4542 $self->{url} . '/' . $self->repo_path($path);
4543}
4544sub rmdirs {
4546 my ($self) = @_;
4547 my $rm = $self->{rm};
4548 delete $rm->{''}; # we never delete the url we're tracking
4549 return unless %$rm;
4550 foreach (keys %$rm) {
4552 my @d = split m#/#, $_;
4553 my $c = shift @d;
4554 $rm->{$c} = 1;
4555 while (@d) {
4556 $c .= '/' . shift @d;
4557 $rm->{$c} = 1;
4558 }
4559 }
4560 delete $rm->{$self->{svn_path}};
4561 delete $rm->{''}; # we never delete the url we're tracking
4562 return unless %$rm;
4563 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4565 $self->{tree_b});
4566 local $/ = "\0";
4567 while (<$fh>) {
4568 chomp;
4569 my @dn = split m#/#, $_;
4570 while (pop @dn) {
4571 delete $rm->{join '/', @dn};
4572 }
4573 unless (%$rm) {
4574 close $fh;
4575 return;
4576 }
4577 }
4578 command_close_pipe($fh, $ctx);
4579 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4581 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4582 $self->close_directory($bat->{$d}, $p);
4583 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4584 print "\tD+\t$d/\n" unless $::_q;
4585 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4586 delete $bat->{$d};
4587 }
4588}
4589sub open_or_add_dir {
4591 my ($self, $full_path, $baton) = @_;
4592 my $t = $self->{types}->{$full_path};
4593 if (!defined $t) {
4594 die "$full_path not known in r$self->{r} or we have a bug!\n";
4595 }
4596 {
4597 no warnings 'once';
4598 # SVN::Node::none and SVN::Node::file are used only once,
4599 # so we're shutting up Perl's warnings about them.
4600 if ($t == $SVN::Node::none) {
4601 return $self->add_directory($full_path, $baton,
4602 undef, -1, $self->{pool});
4603 } elsif ($t == $SVN::Node::dir) {
4604 return $self->open_directory($full_path, $baton,
4605 $self->{r}, $self->{pool});
4606 } # no warnings 'once'
4607 print STDERR "$full_path already exists in repository at ",
4608 "r$self->{r} and it is not a directory (",
4609 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4610 } # no warnings 'once'
4611 exit 1;
4612}
4613sub ensure_path {
4615 my ($self, $path) = @_;
4616 my $bat = $self->{bat};
4617 my $repo_path = $self->repo_path($path);
4618 return $bat->{''} unless (length $repo_path);
4619 my @p = split m#/+#, $repo_path;
4620 my $c = shift @p;
4621 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4622 while (@p) {
4623 my $c0 = $c;
4624 $c .= '/' . shift @p;
4625 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4626 }
4627 return $bat->{$c};
4628}
4629# Subroutine to convert a globbing pattern to a regular expression.
4631# From perl cookbook.
4632sub glob2pat {
4633 my $globstr = shift;
4634 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4635 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4636 return '^' . $globstr . '$';
4637}
4638sub check_autoprop {
4640 my ($self, $pattern, $properties, $file, $fbat) = @_;
4641 # Convert the globbing pattern to a regular expression.
4642 my $regex = glob2pat($pattern);
4643 # Check if the pattern matches the file name.
4644 if($file =~ m/($regex)/) {
4645 # Parse the list of properties to set.
4646 my @props = split(/;/, $properties);
4647 foreach my $prop (@props) {
4648 # Parse 'name=value' syntax and set the property.
4649 if ($prop =~ /([^=]+)=(.*)/) {
4650 my ($n,$v) = ($1,$2);
4651 for ($n, $v) {
4652 s/^\s+//; s/\s+$//;
4653 }
4654 $self->change_file_prop($fbat, $n, $v);
4655 }
4656 }
4657 }
4658}
4659sub apply_autoprops {
4661 my ($self, $file, $fbat) = @_;
4662 my $conf_t = ${$self->{config}}{'config'};
4663 no warnings 'once';
4664 # Check [miscellany]/enable-auto-props in svn configuration.
4665 if (SVN::_Core::svn_config_get_bool(
4666 $conf_t,
4667 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4668 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4669 0)) {
4670 # Auto-props are enabled. Enumerate them to look for matches.
4671 my $callback = sub {
4672 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4673 };
4674 SVN::_Core::svn_config_enumerate(
4675 $conf_t,
4676 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4677 $callback);
4678 }
4679}
4680sub A {
4682 my ($self, $m) = @_;
4683 my ($dir, $file) = split_path($m->{file_b});
4684 my $pbat = $self->ensure_path($dir);
4685 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4686 undef, -1);
4687 print "\tA\t$m->{file_b}\n" unless $::_q;
4688 $self->apply_autoprops($file, $fbat);
4689 $self->chg_file($fbat, $m);
4690 $self->close_file($fbat,undef,$self->{pool});
4691}
4692sub C {
4694 my ($self, $m) = @_;
4695 my ($dir, $file) = split_path($m->{file_b});
4696 my $pbat = $self->ensure_path($dir);
4697 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4698 $self->url_path($m->{file_a}), $self->{r});
4699 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4700 $self->chg_file($fbat, $m);
4701 $self->close_file($fbat,undef,$self->{pool});
4702}
4703sub delete_entry {
4705 my ($self, $path, $pbat) = @_;
4706 my $rpath = $self->repo_path($path);
4707 my ($dir, $file) = split_path($rpath);
4708 $self->{rm}->{$dir} = 1;
4709 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4710}
4711sub R {
4713 my ($self, $m) = @_;
4714 my ($dir, $file) = split_path($m->{file_b});
4715 my $pbat = $self->ensure_path($dir);
4716 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4717 $self->url_path($m->{file_a}), $self->{r});
4718 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4719 $self->apply_autoprops($file, $fbat);
4720 $self->chg_file($fbat, $m);
4721 $self->close_file($fbat,undef,$self->{pool});
4722 ($dir, $file) = split_path($m->{file_a});
4724 $pbat = $self->ensure_path($dir);
4725 $self->delete_entry($m->{file_a}, $pbat);
4726}
4727sub M {
4729 my ($self, $m) = @_;
4730 my ($dir, $file) = split_path($m->{file_b});
4731 my $pbat = $self->ensure_path($dir);
4732 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4733 $pbat,$self->{r},$self->{pool});
4734 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4735 $self->chg_file($fbat, $m);
4736 $self->close_file($fbat,undef,$self->{pool});
4737}
4738sub T { shift->M(@_) }
4740sub change_file_prop {
4742 my ($self, $fbat, $pname, $pval) = @_;
4743 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4744}
4745sub _chg_file_get_blob ($$$$) {
4747 my ($self, $fbat, $m, $which) = @_;
4748 my $fh = $::_repository->temp_acquire("git_blob_$which");
4749 if ($m->{"mode_$which"} =~ /^120/) {
4750 print $fh 'link ' or croak $!;
4751 $self->change_file_prop($fbat,'svn:special','*');
4752 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4753 $self->change_file_prop($fbat,'svn:special',undef);
4754 }
4755 my $blob = $m->{"sha1_$which"};
4756 return ($fh,) if ($blob =~ /^0{40}$/);
4757 my $size = $::_repository->cat_blob($blob, $fh);
4758 croak "Failed to read object $blob" if ($size < 0);
4759 $fh->flush == 0 or croak $!;
4760 seek $fh, 0, 0 or croak $!;
4761 my $exp = ::md5sum($fh);
4763 seek $fh, 0, 0 or croak $!;
4764 return ($fh, $exp);
4765}
4766sub chg_file {
4768 my ($self, $fbat, $m) = @_;
4769 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4770 $self->change_file_prop($fbat,'svn:executable','*');
4771 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4772 $self->change_file_prop($fbat,'svn:executable',undef);
4773 }
4774 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4775 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4776 my $pool = SVN::Pool->new;
4777 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4778 if (-s $fh_a) {
4779 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4780 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4781 if (defined $res) {
4782 die "Unexpected result from send_txstream: $res\n",
4783 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4784 }
4785 } else {
4786 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4787 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4788 if ($got ne $exp_b);
4789 }
4790 Git::temp_release($fh_b, 1);
4791 Git::temp_release($fh_a, 1);
4792 $pool->clear;
4793}
4794sub D {
4796 my ($self, $m) = @_;
4797 my ($dir, $file) = split_path($m->{file_b});
4798 my $pbat = $self->ensure_path($dir);
4799 print "\tD\t$m->{file_b}\n" unless $::_q;
4800 $self->delete_entry($m->{file_b}, $pbat);
4801}
4802sub close_edit {
4804 my ($self) = @_;
4805 my ($p,$bat) = ($self->{pool}, $self->{bat});
4806 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4807 next if $_ eq '';
4808 $self->close_directory($bat->{$_}, $p);
4809 }
4810 $self->close_directory($bat->{''}, $p);
4811 $self->SUPER::close_edit($p);
4812 $p->clear;
4813}
4814sub abort_edit {
4816 my ($self) = @_;
4817 $self->SUPER::abort_edit($self->{pool});
4818}
4819sub DESTROY {
4821 my $self = shift;
4822 $self->SUPER::DESTROY(@_);
4823 $self->{pool}->clear;
4824}
4825# this drives the editor
4827sub apply_diff {
4828 my ($self) = @_;
4829 my $mods = $self->{mods};
4830 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4831 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4832 my $f = $m->{chg};
4833 if (defined $o{$f}) {
4834 $self->$f($m);
4835 } else {
4836 fatal("Invalid change type: $f");
4837 }
4838 }
4839 $self->rmdirs if $_rmdir;
4840 if (@$mods == 0) {
4841 $self->abort_edit;
4842 } else {
4843 $self->close_edit;
4844 }
4845 return scalar @$mods;
4846}
4847package Git::SVN::Ra;
4849use vars qw/@ISA $config_dir $_log_window_size/;
4850use strict;
4851use warnings;
4852my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4853BEGIN {
4855 # enforce temporary pool usage for some simple functions
4856 no strict 'refs';
4857 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4858 get_file/) {
4859 my $SUPER = "SUPER::$f";
4860 *$f = sub {
4861 my $self = shift;
4862 my $pool = SVN::Pool->new;
4863 my @ret = $self->$SUPER(@_,$pool);
4864 $pool->clear;
4865 wantarray ? @ret : $ret[0];
4866 };
4867 }
4868}
4869sub _auth_providers () {
4871 [
4872 SVN::Client::get_simple_provider(),
4873 SVN::Client::get_ssl_server_trust_file_provider(),
4874 SVN::Client::get_simple_prompt_provider(
4875 \&Git::SVN::Prompt::simple, 2),
4876 SVN::Client::get_ssl_client_cert_file_provider(),
4877 SVN::Client::get_ssl_client_cert_prompt_provider(
4878 \&Git::SVN::Prompt::ssl_client_cert, 2),
4879 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4880 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4881 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4882 SVN::Client::get_username_provider(),
4883 SVN::Client::get_ssl_server_trust_prompt_provider(
4884 \&Git::SVN::Prompt::ssl_server_trust),
4885 SVN::Client::get_username_prompt_provider(
4886 \&Git::SVN::Prompt::username, 2)
4887 ]
4888}
4889sub escape_uri_only {
4891 my ($uri) = @_;
4892 my @tmp;
4893 foreach (split m{/}, $uri) {
4894 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4895 push @tmp, $_;
4896 }
4897 join('/', @tmp);
4898}
4899sub escape_url {
4901 my ($url) = @_;
4902 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4903 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4904 $url = "$scheme://$domain$uri";
4905 }
4906 $url;
4907}
4908sub new {
4910 my ($class, $url) = @_;
4911 $url =~ s!/+$!!;
4912 return $RA if ($RA && $RA->{url} eq $url);
4913 ::_req_svn();
4915 SVN::_Core::svn_config_ensure($config_dir, undef);
4917 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4918 my $config = SVN::Core::config_get_config($config_dir);
4919 $RA = undef;
4920 my $dont_store_passwords = 1;
4921 my $conf_t = ${$config}{'config'};
4922 {
4923 no warnings 'once';
4924 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4925 # produces warnings that variables are used only once.
4926 # I had not found the better way to shut them up, so
4927 # the warnings of type 'once' are disabled in this block.
4928 if (SVN::_Core::svn_config_get_bool($conf_t,
4929 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4930 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4931 1) == 0) {
4932 SVN::_Core::svn_auth_set_parameter($baton,
4933 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4934 bless (\$dont_store_passwords, "_p_void"));
4935 }
4936 if (SVN::_Core::svn_config_get_bool($conf_t,
4937 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4938 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4939 1) == 0) {
4940 $Git::SVN::Prompt::_no_auth_cache = 1;
4941 }
4942 } # no warnings 'once'
4943 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4944 config => $config,
4945 pool => SVN::Pool->new,
4946 auth_provider_callbacks => $callbacks);
4947 $self->{url} = $url;
4948 $self->{svn_path} = $url;
4949 $self->{repos_root} = $self->get_repos_root;
4950 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4951 $self->{cache} = { check_path => { r => 0, data => {} },
4952 get_dir => { r => 0, data => {} } };
4953 $RA = bless $self, $class;
4954}
4955sub check_path {
4957 my ($self, $path, $r) = @_;
4958 my $cache = $self->{cache}->{check_path};
4959 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4960 return $cache->{data}->{$path};
4961 }
4962 my $pool = SVN::Pool->new;
4963 my $t = $self->SUPER::check_path($path, $r, $pool);
4964 $pool->clear;
4965 if ($r != $cache->{r}) {
4966 %{$cache->{data}} = ();
4967 $cache->{r} = $r;
4968 }
4969 $cache->{data}->{$path} = $t;
4970}
4971sub get_dir {
4973 my ($self, $dir, $r) = @_;
4974 my $cache = $self->{cache}->{get_dir};
4975 if ($r == $cache->{r}) {
4976 if (my $x = $cache->{data}->{$dir}) {
4977 return wantarray ? @$x : $x->[0];
4978 }
4979 }
4980 my $pool = SVN::Pool->new;
4981 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4982 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4983 $pool->clear;
4984 if ($r != $cache->{r}) {
4985 %{$cache->{data}} = ();
4986 $cache->{r} = $r;
4987 }
4988 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4989 wantarray ? (\%dirents, $r, $props) : \%dirents;
4990}
4991sub DESTROY {
4993 # do not call the real DESTROY since we store ourselves in $RA
4994}
4995# get_log(paths, start, end, limit,
4997# discover_changed_paths, strict_node_history, receiver)
4998sub get_log {
4999 my ($self, @args) = @_;
5000 my $pool = SVN::Pool->new;
5001 # svn_log_changed_path_t objects passed to get_log are likely to be
5003 # overwritten even if only the refs are copied to an external variable,
5004 # so we should dup the structures in their entirety. Using an
5005 # externally passed pool (instead of our temporary and quickly cleared
5006 # pool in Git::SVN::Ra) does not help matters at all...
5007 my $receiver = pop @args;
5008 my $prefix = "/".$self->{svn_path};
5009 $prefix =~ s#/+($)##;
5010 my $prefix_regex = qr#^\Q$prefix\E#;
5011 push(@args, sub {
5012 my ($paths) = $_[0];
5013 return &$receiver(@_) unless $paths;
5014 $_[0] = ();
5015 foreach my $p (keys %$paths) {
5016 my $i = $paths->{$p};
5017 # Make path relative to our url, not repos_root
5018 $p =~ s/$prefix_regex//;
5019 my %s = map { $_ => $i->$_; }
5020 qw/copyfrom_path copyfrom_rev action/;
5021 if ($s{'copyfrom_path'}) {
5022 $s{'copyfrom_path'} =~ s/$prefix_regex//;
5023 }
5024 $_[0]{$p} = \%s;
5025 }
5026 &$receiver(@_);
5027 });
5028 # the limit parameter was not supported in SVN 1.1.x, so we
5031 # drop it. Therefore, the receiver callback passed to it
5032 # is made aware of this limitation by being wrapped if
5033 # the limit passed to is being wrapped.
5034 if ($SVN::Core::VERSION le '1.2.0') {
5035 my $limit = splice(@args, 3, 1);
5036 if ($limit > 0) {
5037 my $receiver = pop @args;
5038 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5039 }
5040 }
5041 my $ret = $self->SUPER::get_log(@args, $pool);
5042 $pool->clear;
5043 $ret;
5044}
5045sub trees_match {
5047 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5048 my $ctx = SVN::Client->new(auth => _auth_providers);
5049 my $out = IO::File->new_tmpfile;
5050 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5052 # $ctx->diff(), so we'll create a default one
5053 my $pool = SVN::Pool->new_default_sub;
5054 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5056 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5057 $out->flush;
5058 my $ret = (($out->stat)[7] == 0);
5059 close $out or croak $!;
5060 $ret;
5062}
5063sub get_commit_editor {
5065 my ($self, $log, $cb, $pool) = @_;
5066 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5067 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5068}
5069sub gs_do_update {
5071 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5072 my $new = ($rev_a == $rev_b);
5073 my $path = $gs->{path};
5074 if ($new && -e $gs->{index}) {
5076 unlink $gs->{index} or die
5077 "Couldn't unlink index: $gs->{index}: $!\n";
5078 }
5079 my $pool = SVN::Pool->new;
5080 $editor->set_path_strip($path);
5081 my (@pc) = split m#/#, $path;
5082 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5083 1, $editor, $pool);
5084 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5085 # Since we can't rely on svn_ra_reparent being available, we'll
5087 # just have to do some magic with set_path to make it so
5088 # we only want a partial path.
5089 my $sp = '';
5090 my $final = join('/', @pc);
5091 while (@pc) {
5092 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5093 $sp .= '/' if length $sp;
5094 $sp .= shift @pc;
5095 }
5096 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5097 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5099 $reporter->finish_report($pool);
5101 $pool->clear;
5102 $editor->{git_commit_ok};
5103}
5104# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5106# svn_ra_reparent didn't work before 1.4)
5107sub gs_do_switch {
5108 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5109 my $path = $gs->{path};
5110 my $pool = SVN::Pool->new;
5111 my $full_url = $self->{url};
5113 my $old_url = $full_url;
5114 $full_url .= '/' . $path if length $path;
5115 my ($ra, $reparented);
5116 if ($old_url =~ m#^svn(\+ssh)?://# ||
5118 ($full_url =~ m#^https?://# &&
5119 escape_url($full_url) ne $full_url)) {
5120 $_[0] = undef;
5121 $self = undef;
5122 $RA = undef;
5123 $ra = Git::SVN::Ra->new($full_url);
5124 $ra_invalid = 1;
5125 } elsif ($old_url ne $full_url) {
5126 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5127 $self->{url} = $full_url;
5128 $reparented = 1;
5129 }
5130 $ra ||= $self;
5132 $url_b = escape_url($url_b);
5133 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5134 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5135 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5136 $reporter->finish_report($pool);
5137 if ($reparented) {
5139 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5140 $self->{url} = $old_url;
5141 }
5142 $pool->clear;
5144 $editor->{git_commit_ok};
5145}
5146sub longest_common_path {
5148 my ($gsv, $globs) = @_;
5149 my %common;
5150 my $common_max = scalar @$gsv;
5151 foreach my $gs (@$gsv) {
5153 my @tmp = split m#/#, $gs->{path};
5154 my $p = '';
5155 foreach (@tmp) {
5156 $p .= length($p) ? "/$_" : $_;
5157 $common{$p} ||= 0;
5158 $common{$p}++;
5159 }
5160 }
5161 $globs ||= [];
5162 $common_max += scalar @$globs;
5163 foreach my $glob (@$globs) {
5164 my @tmp = split m#/#, $glob->{path}->{left};
5165 my $p = '';
5166 foreach (@tmp) {
5167 $p .= length($p) ? "/$_" : $_;
5168 $common{$p} ||= 0;
5169 $common{$p}++;
5170 }
5171 }
5172 my $longest_path = '';
5174 foreach (sort {length $b <=> length $a} keys %common) {
5175 if ($common{$_} == $common_max) {
5176 $longest_path = $_;
5177 last;
5178 }
5179 }
5180 $longest_path;
5181}
5182sub gs_fetch_loop_common {
5184 my ($self, $base, $head, $gsv, $globs) = @_;
5185 return if ($base > $head);
5186 my $inc = $_log_window_size;
5187 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5188 my $longest_path = longest_common_path($gsv, $globs);
5189 my $ra_url = $self->{url};
5190 my $find_trailing_edge;
5191 while (1) {
5192 my %revs;
5193 my $err;
5194 my $err_handler = $SVN::Error::handler;
5195 $SVN::Error::handler = sub {
5196 ($err) = @_;
5197 skip_unknown_revs($err);
5198 };
5199 sub _cb {
5200 my ($paths, $r, $author, $date, $log) = @_;
5201 [ $paths,
5202 { author => $author, date => $date, log => $log } ];
5203 }
5204 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5205 sub { $revs{$_[1]} = _cb(@_) });
5206 if ($err) {
5207 print "Checked through r$max\r";
5208 } else {
5209 $find_trailing_edge = 1;
5210 }
5211 if ($err and $find_trailing_edge) {
5212 print STDERR "Path '$longest_path' ",
5213 "was probably deleted:\n",
5214 $err->expanded_message,
5215 "\nWill attempt to follow ",
5216 "revisions r$min .. r$max ",
5217 "committed before the deletion\n";
5218 my $hi = $max;
5219 while (--$hi >= $min) {
5220 my $ok;
5221 $self->get_log([$longest_path], $min, $hi,
5222 0, 1, 1, sub {
5223 $ok = $_[1];
5224 $revs{$_[1]} = _cb(@_) });
5225 if ($ok) {
5226 print STDERR "r$min .. r$ok OK\n";
5227 last;
5228 }
5229 }
5230 $find_trailing_edge = 0;
5231 }
5232 $SVN::Error::handler = $err_handler;
5233 my %exists = map { $_->{path} => $_ } @$gsv;
5235 foreach my $r (sort {$a <=> $b} keys %revs) {
5236 my ($paths, $logged) = @{$revs{$r}};
5237 foreach my $gs ($self->match_globs(\%exists, $paths,
5239 $globs, $r)) {
5240 if ($gs->rev_map_max >= $r) {
5241 next;
5242 }
5243 next unless $gs->match_paths($paths, $r);
5244 $gs->{logged_rev_props} = $logged;
5245 if (my $last_commit = $gs->last_commit) {
5246 $gs->assert_index_clean($last_commit);
5247 }
5248 my $log_entry = $gs->do_fetch($paths, $r);
5249 if ($log_entry) {
5250 $gs->do_git_commit($log_entry);
5251 }
5252 $INDEX_FILES{$gs->{index}} = 1;
5253 }
5254 foreach my $g (@$globs) {
5255 my $k = "svn-remote.$g->{remote}." .
5256 "$g->{t}-maxRev";
5257 Git::SVN::tmp_config($k, $r);
5258 }
5259 if ($ra_invalid) {
5260 $_[0] = undef;
5261 $self = undef;
5262 $RA = undef;
5263 $self = Git::SVN::Ra->new($ra_url);
5264 $ra_invalid = undef;
5265 }
5266 }
5267 # pre-fill the .rev_db since it'll eventually get filled in
5268 # with '0' x40 if something new gets committed
5269 foreach my $gs (@$gsv) {
5270 next if $gs->rev_map_max >= $max;
5271 next if defined $gs->rev_map_get($max);
5272 $gs->rev_map_set($max, 0 x40);
5273 }
5274 foreach my $g (@$globs) {
5275 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5276 Git::SVN::tmp_config($k, $max);
5277 }
5278 last if $max >= $head;
5279 $min = $max + 1;
5280 $max += $inc;
5281 $max = $head if ($max > $head);
5282 }
5283 Git::SVN::gc();
5284}
5285sub get_dir_globbed {
5287 my ($self, $left, $depth, $r) = @_;
5288 my @x = eval { $self->get_dir($left, $r) };
5290 return unless scalar @x == 3;
5291 my $dirents = $x[0];
5292 my @finalents;
5293 foreach my $de (keys %$dirents) {
5294 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5295 if ($depth > 1) {
5296 my @args = ("$left/$de", $depth - 1, $r);
5297 foreach my $dir ($self->get_dir_globbed(@args)) {
5298 push @finalents, "$de/$dir";
5299 }
5300 } else {
5301 push @finalents, $de;
5302 }
5303 }
5304 @finalents;
5305}
5306sub match_globs {
5308 my ($self, $exists, $paths, $globs, $r) = @_;
5309 sub get_dir_check {
5311 my ($self, $exists, $g, $r) = @_;
5312 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5314 $g->{path}->{depth},
5315 $r);
5316 foreach my $de (@dirs) {
5318 my $p = $g->{path}->full_path($de);
5319 next if $exists->{$p};
5320 next if (length $g->{path}->{right} &&
5321 ($self->check_path($p, $r) !=
5322 $SVN::Node::dir));
5323 next unless $p =~ /$g->{path}->{regex}/;
5324 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5325 $g->{ref}->full_path($de), 1);
5326 }
5327 }
5328 foreach my $g (@$globs) {
5329 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5330 if ($path->{action} =~ /^[AR]$/) {
5331 get_dir_check($self, $exists, $g, $r);
5332 }
5333 }
5334 foreach (keys %$paths) {
5335 if (/$g->{path}->{left_regex}/ &&
5336 !/$g->{path}->{regex}/) {
5337 next if $paths->{$_}->{action} !~ /^[AR]$/;
5338 get_dir_check($self, $exists, $g, $r);
5339 }
5340 next unless /$g->{path}->{regex}/;
5341 my $p = $1;
5342 my $pathname = $g->{path}->full_path($p);
5343 next if $exists->{$pathname};
5344 next if ($self->check_path($pathname, $r) !=
5345 $SVN::Node::dir);
5346 $exists->{$pathname} = Git::SVN->init(
5347 $self->{url}, $pathname, undef,
5348 $g->{ref}->full_path($p), 1);
5349 }
5350 my $c = '';
5351 foreach (split m#/#, $g->{path}->{left}) {
5352 $c .= "/$_";
5353 next unless ($paths->{$c} &&
5354 ($paths->{$c}->{action} =~ /^[AR]$/));
5355 get_dir_check($self, $exists, $g, $r);
5356 }
5357 }
5358 values %$exists;
5359}
5360sub minimize_url {
5362 my ($self) = @_;
5363 return $self->{url} if ($self->{url} eq $self->{repos_root});
5364 my $url = $self->{repos_root};
5365 my @components = split(m!/!, $self->{svn_path});
5366 my $c = '';
5367 do {
5368 $url .= "/$c" if length $c;
5369 eval {
5370 my $ra = (ref $self)->new($url);
5371 my $latest = $ra->get_latest_revnum;
5372 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5373 };
5374 } while ($@ && ($c = shift @components));
5375 $url;
5376}
5377sub can_do_switch {
5379 my $self = shift;
5380 unless (defined $can_do_switch) {
5381 my $pool = SVN::Pool->new;
5382 my $rep = eval {
5383 $self->do_switch(1, '', 0, $self->{url},
5384 SVN::Delta::Editor->new, $pool);
5385 };
5386 if ($@) {
5387 $can_do_switch = 0;
5388 } else {
5389 $rep->abort_report($pool);
5390 $can_do_switch = 1;
5391 }
5392 $pool->clear;
5393 }
5394 $can_do_switch;
5395}
5396sub skip_unknown_revs {
5398 my ($err) = @_;
5399 my $errno = $err->apr_err();
5400 # Maybe the branch we're tracking didn't
5401 # exist when the repo started, so it's
5402 # not an error if it doesn't, just continue
5403 #
5404 # Wonderfully consistent library, eh?
5405 # 160013 - svn:// and file://
5406 # 175002 - http(s)://
5407 # 175007 - http(s):// (this repo required authorization, too...)
5408 # More codes may be discovered later...
5409 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5410 my $err_key = $err->expanded_message;
5411 # revision numbers change every time, filter them out
5412 $err_key =~ s/\d+/\0/g;
5413 $err_key = "$errno\0$err_key";
5414 unless ($ignored_err{$err_key}) {
5415 warn "W: Ignoring error from SVN, path probably ",
5416 "does not exist: ($errno): ",
5417 $err->expanded_message,"\n";
5418 warn "W: Do not be alarmed at the above message ",
5419 "git-svn is just searching aggressively for ",
5420 "old history.\n",
5421 "This may take a while on large repositories\n";
5422 $ignored_err{$err_key} = 1;
5423 }
5424 return;
5425 }
5426 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5427}
5428package Git::SVN::Log;
5430use strict;
5431use warnings;
5432use POSIX qw/strftime/;
5433use Time::Local;
5434use constant commit_log_separator => ('-' x 72) . "\n";
5435use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5436 %rusers $show_commit $incremental/;
5437my $l_fmt;
5438sub cmt_showable {
5440 my ($c) = @_;
5441 return 1 if defined $c->{r};
5442 # big commit message got truncated by the 16k pretty buffer in rev-list
5444 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5445 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5446 @{$c->{l}} = ();
5447 my @log = command(qw/cat-file commit/, $c->{c});
5448 # shift off the headers
5450 shift @log while ($log[0] ne '');
5451 shift @log;
5452 # TODO: make $c->{l} not have a trailing newline in the future
5454 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5455 (undef, $c->{r}, undef) = ::extract_metadata(
5457 (grep(/^git-svn-id: /, @log))[-1]);
5458 }
5459 return defined $c->{r};
5460}
5461sub log_use_color {
5463 return $color || Git->repository->get_colorbool('color.diff');
5464}
5465sub git_svn_log_cmd {
5467 my ($r_min, $r_max, @args) = @_;
5468 my $head = 'HEAD';
5469 my (@files, @log_opts);
5470 foreach my $x (@args) {
5471 if ($x eq '--' || @files) {
5472 push @files, $x;
5473 } else {
5474 if (::verify_ref("$x^0")) {
5475 $head = $x;
5476 } else {
5477 push @log_opts, $x;
5478 }
5479 }
5480 }
5481 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5483 $gs ||= Git::SVN->_new;
5484 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5485 $gs->refname);
5486 push @cmd, '-r' unless $non_recursive;
5487 push @cmd, qw/--raw --name-status/ if $verbose;
5488 push @cmd, '--color' if log_use_color();
5489 push @cmd, @log_opts;
5490 if (defined $r_max && $r_max == $r_min) {
5491 push @cmd, '--max-count=1';
5492 if (my $c = $gs->rev_map_get($r_max)) {
5493 push @cmd, $c;
5494 }
5495 } elsif (defined $r_max) {
5496 if ($r_max < $r_min) {
5497 ($r_min, $r_max) = ($r_max, $r_min);
5498 }
5499 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5500 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5501 # If there are no commits in the range, both $c_max and $c_min
5502 # will be undefined. If there is at least 1 commit in the
5503 # range, both will be defined.
5504 return () if !defined $c_min || !defined $c_max;
5505 if ($c_min eq $c_max) {
5506 push @cmd, '--max-count=1', $c_min;
5507 } else {
5508 push @cmd, '--boundary', "$c_min..$c_max";
5509 }
5510 }
5511 return (@cmd, @files);
5512}
5513# adapted from pager.c
5515sub config_pager {
5516 if (! -t *STDOUT) {
5517 $ENV{GIT_PAGER_IN_USE} = 'false';
5518 $pager = undef;
5519 return;
5520 }
5521 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5522 if ($pager eq 'cat') {
5523 $pager = undef;
5524 }
5525 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5526}
5527sub run_pager {
5529 return unless defined $pager;
5530 pipe my ($rfd, $wfd) or return;
5531 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5532 if (!$pid) {
5533 open STDOUT, '>&', $wfd or
5534 ::fatal "Can't redirect to stdout: $!";
5535 return;
5536 }
5537 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5538 $ENV{LESS} ||= 'FRSX';
5539 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5540}
5541sub format_svn_date {
5543 # some systmes don't handle or mishandle %z, so be creative.
5544 my $t = shift || time;
5545 my $gm = timelocal(gmtime($t));
5546 my $sign = qw( + + - )[ $t <=> $gm ];
5547 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5548 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5549}
5550sub parse_git_date {
5552 my ($t, $tz) = @_;
5553 # Date::Parse isn't in the standard Perl distro :(
5554 if ($tz =~ s/^\+//) {
5555 $t += tz_to_s_offset($tz);
5556 } elsif ($tz =~ s/^\-//) {
5557 $t -= tz_to_s_offset($tz);
5558 }
5559 return $t;
5560}
5561sub set_local_timezone {
5563 if (defined $TZ) {
5564 $ENV{TZ} = $TZ;
5565 } else {
5566 delete $ENV{TZ};
5567 }
5568}
5569sub tz_to_s_offset {
5571 my ($tz) = @_;
5572 $tz =~ s/(\d\d)$//;
5573 return ($1 * 60) + ($tz * 3600);
5574}
5575sub get_author_info {
5577 my ($dest, $author, $t, $tz) = @_;
5578 $author =~ s/(?:^\s*|\s*$)//g;
5579 $dest->{a_raw} = $author;
5580 my $au;
5581 if ($::_authors) {
5582 $au = $rusers{$author} || undef;
5583 }
5584 if (!$au) {
5585 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5586 }
5587 $dest->{t} = $t;
5588 $dest->{tz} = $tz;
5589 $dest->{a} = $au;
5590 $dest->{t_utc} = parse_git_date($t, $tz);
5591}
5592sub process_commit {
5594 my ($c, $r_min, $r_max, $defer) = @_;
5595 if (defined $r_min && defined $r_max) {
5596 if ($r_min == $c->{r} && $r_min == $r_max) {
5597 show_commit($c);
5598 return 0;
5599 }
5600 return 1 if $r_min == $r_max;
5601 if ($r_min < $r_max) {
5602 # we need to reverse the print order
5603 return 0 if (defined $limit && --$limit < 0);
5604 push @$defer, $c;
5605 return 1;
5606 }
5607 if ($r_min != $r_max) {
5608 return 1 if ($r_min < $c->{r});
5609 return 1 if ($r_max > $c->{r});
5610 }
5611 }
5612 return 0 if (defined $limit && --$limit < 0);
5613 show_commit($c);
5614 return 1;
5615}
5616sub show_commit {
5618 my $c = shift;
5619 if ($oneline) {
5620 my $x = "\n";
5621 if (my $l = $c->{l}) {
5622 while ($l->[0] =~ /^\s*$/) { shift @$l }
5623 $x = $l->[0];
5624 }
5625 $l_fmt ||= 'A' . length($c->{r});
5626 print 'r',pack($l_fmt, $c->{r}),' | ';
5627 print "$c->{c} | " if $show_commit;
5628 print $x;
5629 } else {
5630 show_commit_normal($c);
5631 }
5632}
5633sub show_commit_changed_paths {
5635 my ($c) = @_;
5636 return unless $c->{changed};
5637 print "Changed paths:\n", @{$c->{changed}};
5638}
5639sub show_commit_normal {
5641 my ($c) = @_;
5642 print commit_log_separator, "r$c->{r} | ";
5643 print "$c->{c} | " if $show_commit;
5644 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5645 my $nr_line = 0;
5646 if (my $l = $c->{l}) {
5648 while ($l->[$#$l] eq "\n" && $#$l > 0
5649 && $l->[($#$l - 1)] eq "\n") {
5650 pop @$l;
5651 }
5652 $nr_line = scalar @$l;
5653 if (!$nr_line) {
5654 print "1 line\n\n\n";
5655 } else {
5656 if ($nr_line == 1) {
5657 $nr_line = '1 line';
5658 } else {
5659 $nr_line .= ' lines';
5660 }
5661 print $nr_line, "\n";
5662 show_commit_changed_paths($c);
5663 print "\n";
5664 print $_ foreach @$l;
5665 }
5666 } else {
5667 print "1 line\n";
5668 show_commit_changed_paths($c);
5669 print "\n";
5670 }
5672 foreach my $x (qw/raw stat diff/) {
5673 if ($c->{$x}) {
5674 print "\n";
5675 print $_ foreach @{$c->{$x}}
5676 }
5677 }
5678}
5679sub cmd_show_log {
5681 my (@args) = @_;
5682 my ($r_min, $r_max);
5683 my $r_last = -1; # prevent dupes
5684 set_local_timezone();
5685 if (defined $::_revision) {
5686 if ($::_revision =~ /^(\d+):(\d+)$/) {
5687 ($r_min, $r_max) = ($1, $2);
5688 } elsif ($::_revision =~ /^\d+$/) {
5689 $r_min = $r_max = $::_revision;
5690 } else {
5691 ::fatal "-r$::_revision is not supported, use ",
5692 "standard 'git log' arguments instead";
5693 }
5694 }
5695 config_pager();
5697 @args = git_svn_log_cmd($r_min, $r_max, @args);
5698 if (!@args) {
5699 print commit_log_separator unless $incremental || $oneline;
5700 return;
5701 }
5702 my $log = command_output_pipe(@args);
5703 run_pager();
5704 my (@k, $c, $d, $stat);
5705 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5706 while (<$log>) {
5707 if (/^${esc_color}commit -?($::sha1_short)/o) {
5708 my $cmt = $1;
5709 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5710 $r_last = $c->{r};
5711 process_commit($c, $r_min, $r_max, \@k) or
5712 goto out;
5713 }
5714 $d = undef;
5715 $c = { c => $cmt };
5716 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5717 get_author_info($c, $1, $2, $3);
5718 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5719 # ignore
5720 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5721 push @{$c->{raw}}, $_;
5722 } elsif (/^${esc_color}[ACRMDT]\t/) {
5723 # we could add $SVN->{svn_path} here, but that requires
5724 # remote access at the moment (repo_path_split)...
5725 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5726 push @{$c->{changed}}, $_;
5727 } elsif (/^${esc_color}diff /o) {
5728 $d = 1;
5729 push @{$c->{diff}}, $_;
5730 } elsif ($d) {
5731 push @{$c->{diff}}, $_;
5732 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5733 $esc_color*[\+\-]*$esc_color$/x) {
5734 $stat = 1;
5735 push @{$c->{stat}}, $_;
5736 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5737 push @{$c->{stat}}, $_;
5738 $stat = undef;
5739 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5740 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5741 } elsif (s/^${esc_color} //o) {
5742 push @{$c->{l}}, $_;
5743 }
5744 }
5745 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5746 $r_last = $c->{r};
5747 process_commit($c, $r_min, $r_max, \@k);
5748 }
5749 if (@k) {
5750 ($r_min, $r_max) = ($r_max, $r_min);
5751 process_commit($_, $r_min, $r_max) foreach reverse @k;
5752 }
5753out:
5754 close $log;
5755 print commit_log_separator unless $incremental || $oneline;
5756}
5757sub cmd_blame {
5759 my $path = pop;
5760 config_pager();
5762 run_pager();
5763 my ($fh, $ctx, $rev);
5765 if ($_git_format) {
5767 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5768 while (my $line = <$fh>) {
5769 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5770 # Uncommitted edits show up as a rev ID of
5771 # all zeros, which we can't look up with
5772 # cmt_metadata
5773 if ($1 !~ /^0+$/) {
5774 (undef, $rev, undef) =
5775 ::cmt_metadata($1);
5776 $rev = '0' if (!$rev);
5777 } else {
5778 $rev = '0';
5779 }
5780 $rev = sprintf('%-10s', $rev);
5781 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5782 }
5783 print $line;
5784 }
5785 } else {
5786 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5787 '--', $path);
5788 my ($sha1);
5789 my %authors;
5790 my @buffer;
5791 my %dsha; #distinct sha keys
5792 while (my $line = <$fh>) {
5794 push @buffer, $line;
5795 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5796 $dsha{$1} = 1;
5797 }
5798 }
5799 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5801 foreach my $line (@buffer) {
5803 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5804 $rev = $s2r->{$1};
5805 $rev = '0' if (!$rev)
5806 }
5807 elsif ($line =~ /^author (.*)/) {
5808 $authors{$rev} = $1;
5809 $authors{$rev} =~ s/\s/_/g;
5810 }
5811 elsif ($line =~ /^\t(.*)$/) {
5812 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5813 }
5814 }
5815 }
5816 command_close_pipe($fh, $ctx);
5817}
5818package Git::SVN::Migration;
5820# these version numbers do NOT correspond to actual version numbers
5821# of git nor git-svn. They are just relative.
5822#
5823# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5824#
5825# v1 layout: .git/$id/info/url, refs/remotes/$id
5826#
5827# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5828#
5829# v3 layout: .git/svn/$id, refs/remotes/$id
5830# - info/url may remain for backwards compatibility
5831# - this is what we migrate up to this layout automatically,
5832# - this will be used by git svn init on single branches
5833# v3.1 layout (auto migrated):
5834# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5835# for backwards compatibility
5836#
5837# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5838# - this is only created for newly multi-init-ed
5839# repositories. Similar in spirit to the
5840# --use-separate-remotes option in git-clone (now default)
5841# - we do not automatically migrate to this (following
5842# the example set by core git)
5843#
5844# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5845# - newer, more-efficient format that uses 24-bytes per record
5846# with no filler space.
5847# - use xxd -c24 < .rev_map.$UUID to view and debug
5848# - This is a one-way migration, repositories updated to the
5849# new format will not be able to use old git-svn without
5850# rebuilding the .rev_db. Rebuilding the rev_db is not
5851# possible if noMetadata or useSvmProps are set; but should
5852# be no problem for users that use the (sensible) defaults.
5853use strict;
5854use warnings;
5855use Carp qw/croak/;
5856use File::Path qw/mkpath/;
5857use File::Basename qw/dirname basename/;
5858use vars qw/$_minimize/;
5859sub migrate_from_v0 {
5861 my $git_dir = $ENV{GIT_DIR};
5862 return undef unless -d $git_dir;
5863 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5864 my $migrated = 0;
5865 while (<$fh>) {
5866 chomp;
5867 my ($id, $orig_ref) = ($_, $_);
5868 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5869 next unless -f "$git_dir/$id/info/url";
5870 my $new_ref = "refs/remotes/$id";
5871 if (::verify_ref("$new_ref^0")) {
5872 print STDERR "W: $orig_ref is probably an old ",
5873 "branch used by an ancient version of ",
5874 "git-svn.\n",
5875 "However, $new_ref also exists.\n",
5876 "We will not be able ",
5877 "to use this branch until this ",
5878 "ambiguity is resolved.\n";
5879 next;
5880 }
5881 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5882 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5883 command_noisy('update-ref', $new_ref, $orig_ref);
5884 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5885 $migrated++;
5886 }
5887 command_close_pipe($fh, $ctx);
5888 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5889 $migrated;
5890}
5891sub migrate_from_v1 {
5893 my $git_dir = $ENV{GIT_DIR};
5894 my $migrated = 0;
5895 return $migrated unless -d $git_dir;
5896 my $svn_dir = "$git_dir/svn";
5897 # just in case somebody used 'svn' as their $id at some point...
5899 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5900 print STDERR "Migrating from a git-svn v1 layout...\n";
5902 mkpath([$svn_dir]);
5903 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5904 "$svn_dir\n\t(required for this version ",
5905 "($::VERSION) of git-svn) does not exist.\n";
5906 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5907 while (<$fh>) {
5908 my $x = $_;
5909 next unless $x =~ s#^refs/remotes/##;
5910 chomp $x;
5911 next unless -f "$git_dir/$x/info/url";
5912 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5913 next unless $u;
5914 my $dn = dirname("$git_dir/svn/$x");
5915 mkpath([$dn]) unless -d $dn;
5916 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5917 mkpath(["$git_dir/svn/svn"]);
5918 print STDERR " - $git_dir/$x/info => ",
5919 "$git_dir/svn/$x/info\n";
5920 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5921 croak "$!: $x";
5922 # don't worry too much about these, they probably
5923 # don't exist with repos this old (save for index,
5924 # and we can easily regenerate that)
5925 foreach my $f (qw/unhandled.log index .rev_db/) {
5926 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5927 }
5928 } else {
5929 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5930 rename "$git_dir/$x", "$git_dir/svn/$x" or
5931 croak "$!: $x";
5932 }
5933 $migrated++;
5934 }
5935 command_close_pipe($fh, $ctx);
5936 print STDERR "Done migrating from a git-svn v1 layout\n";
5937 $migrated;
5938}
5939sub read_old_urls {
5941 my ($l_map, $pfx, $path) = @_;
5942 my @dir;
5943 foreach (<$path/*>) {
5944 if (-r "$_/info/url") {
5945 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5946 my $ref_id = $pfx . basename $_;
5947 my $url = ::file_to_s("$_/info/url");
5948 $l_map->{$ref_id} = $url;
5949 } elsif (-d $_) {
5950 push @dir, $_;
5951 }
5952 }
5953 foreach (@dir) {
5954 my $x = $_;
5955 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5956 read_old_urls($l_map, $x, $_);
5957 }
5958}
5959sub migrate_from_v2 {
5961 my @cfg = command(qw/config -l/);
5962 return if grep /^svn-remote\..+\.url=/, @cfg;
5963 my %l_map;
5964 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5965 my $migrated = 0;
5966 foreach my $ref_id (sort keys %l_map) {
5968 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5969 if ($@) {
5970 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5971 }
5972 $migrated++;
5973 }
5974 $migrated;
5975}
5976sub minimize_connections {
5978 my $r = Git::SVN::read_all_remotes();
5979 my $new_urls = {};
5980 my $root_repos = {};
5981 foreach my $repo_id (keys %$r) {
5982 my $url = $r->{$repo_id}->{url} or next;
5983 my $fetch = $r->{$repo_id}->{fetch} or next;
5984 my $ra = Git::SVN::Ra->new($url);
5985 # skip existing cases where we already connect to the root
5987 if (($ra->{url} eq $ra->{repos_root}) ||
5988 ($ra->{repos_root} eq $repo_id)) {
5989 $root_repos->{$ra->{url}} = $repo_id;
5990 next;
5991 }
5992 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5994 my $root_path = $ra->{url};
5995 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5996 foreach my $path (keys %$fetch) {
5997 my $ref_id = $fetch->{$path};
5998 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5999 # make sure we can read when connecting to
6001 # a higher level of a repository
6002 my ($last_rev, undef) = $gs->last_rev_commit;
6003 if (!defined $last_rev) {
6004 $last_rev = eval {
6005 $root_ra->get_latest_revnum;
6006 };
6007 next if $@;
6008 }
6009 my $new = $root_path;
6010 $new .= length $path ? "/$path" : '';
6011 eval {
6012 $root_ra->get_log([$new], $last_rev, $last_rev,
6013 0, 0, 1, sub { });
6014 };
6015 next if $@;
6016 $new_urls->{$ra->{repos_root}}->{$new} =
6017 { ref_id => $ref_id,
6018 old_repo_id => $repo_id,
6019 old_path => $path };
6020 }
6021 }
6022 my @emptied;
6024 foreach my $url (keys %$new_urls) {
6025 # see if we can re-use an existing [svn-remote "repo_id"]
6026 # instead of creating a(n ugly) new section:
6027 my $repo_id = $root_repos->{$url} || $url;
6028 my $fetch = $new_urls->{$url};
6030 foreach my $path (keys %$fetch) {
6031 my $x = $fetch->{$path};
6032 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6033 my $pfx = "svn-remote.$x->{old_repo_id}";
6034 my $old_fetch = quotemeta("$x->{old_path}:".
6036 "$x->{ref_id}");
6037 command_noisy(qw/config --unset/,
6038 "$pfx.fetch", '^'. $old_fetch . '$');
6039 delete $r->{$x->{old_repo_id}}->
6040 {fetch}->{$x->{old_path}};
6041 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6042 command_noisy(qw/config --unset/,
6043 "$pfx.url");
6044 push @emptied, $x->{old_repo_id}
6045 }
6046 }
6047 }
6048 if (@emptied) {
6049 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6050 print STDERR <<EOF;
6051The following [svn-remote] sections in your config file ($file) are empty
6052and can be safely removed:
6053EOF
6054 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6055 }
6056}
6057sub migration_check {
6059 migrate_from_v0();
6060 migrate_from_v1();
6061 migrate_from_v2();
6062 minimize_connections() if $_minimize;
6063}
6064package Git::IndexInfo;
6066use strict;
6067use warnings;
6068use Git qw/command_input_pipe command_close_pipe/;
6069sub new {
6071 my ($class) = @_;
6072 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6073 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6074}
6075sub remove {
6077 my ($self, $path) = @_;
6078 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6079 return ++$self->{nr};
6080 }
6081 undef;
6082}
6083sub update {
6085 my ($self, $mode, $hash, $path) = @_;
6086 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6087 return ++$self->{nr};
6088 }
6089 undef;
6090}
6091sub DESTROY {
6093 my ($self) = @_;
6094 command_close_pipe($self->{gui}, $self->{ctx});
6095}
6096package Git::SVN::GlobSpec;
6098use strict;
6099use warnings;
6100sub new {
6102 my ($class, $glob, $pattern_ok) = @_;
6103 my $re = $glob;
6104 $re =~ s!/+$!!g; # no need for trailing slashes
6105 my (@left, @right, @patterns);
6106 my $state = "left";
6107 my $die_msg = "Only one set of wildcard directories " .
6108 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6109 for my $part (split(m|/|, $glob)) {
6110 if ($part =~ /\*/ && $part ne "*") {
6111 die "Invalid pattern in '$glob': $part\n";
6112 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6113 $part !~ /^\{[^{}]+\}/) {
6114 die "Invalid pattern in '$glob': $part\n";
6115 }
6116 if ($part eq "*") {
6117 die $die_msg if $state eq "right";
6118 $state = "pattern";
6119 push(@patterns, "[^/]*");
6120 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6121 die $die_msg if $state eq "right";
6122 $state = "pattern";
6123 my $p = quotemeta($1);
6124 $p =~ s/\\,/|/g;
6125 push(@patterns, "(?:$p)");
6126 } else {
6127 if ($state eq "left") {
6128 push(@left, $part);
6129 } else {
6130 push(@right, $part);
6131 $state = "right";
6132 }
6133 }
6134 }
6135 my $depth = @patterns;
6136 if ($depth == 0) {
6137 die "One '*' is needed in glob: '$glob'\n";
6138 }
6139 my $left = join('/', @left);
6140 my $right = join('/', @right);
6141 $re = join('/', @patterns);
6142 $re = join('\/',
6143 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6144 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6145 bless { left => $left, right => $right, left_regex => $left_re,
6146 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6147}
6148sub full_path {
6150 my ($self, $path) = @_;
6151 return (length $self->{left} ? "$self->{left}/" : '') .
6152 $path . (length $self->{right} ? "/$self->{right}" : '');
6153}
6154__END__
6156Data structures:
6158$remotes = { # returned by read_all_remotes()
6161 'svn' => {
6162 # svn-remote.svn.url=https://svn.musicpd.org
6163 url => 'https://svn.musicpd.org',
6164 # svn-remote.svn.fetch=mpd/trunk:trunk
6165 fetch => {
6166 'mpd/trunk' => 'trunk',
6167 },
6168 # svn-remote.svn.tags=mpd/tags/*:tags/*
6169 tags => {
6170 path => {
6171 left => 'mpd/tags',
6172 right => '',
6173 regex => qr!mpd/tags/([^/]+)$!,
6174 glob => 'tags/*',
6175 },
6176 ref => {
6177 left => 'tags',
6178 right => '',
6179 regex => qr!tags/([^/]+)$!,
6180 glob => 'tags/*',
6181 },
6182 }
6183 }
6184};
6185$log_entry hashref as returned by libsvn_log_entry()
6187{
6188 log => 'whitespace-formatted log entry
6189', # trailing newline is preserved
6190 revision => '8', # integer
6191 date => '2004-02-24T17:01:44.108345Z', # commit date
6192 author => 'committer name'
6193};
6194# this is generated by generate_diff();
6197@mods = array of diff-index line hashes, each element represents one line
6198 of diff-index output
6199diff-index line ($m hash)
6201{
6202 mode_a => first column of diff-index output, no leading ':',
6203 mode_b => second column of diff-index output,
6204 sha1_b => sha1sum of the final blob,
6205 chg => change type [MCRADT],
6206 file_a => original file name of a file (iff chg is 'C' or 'R')
6207 file_b => new/current file name of a file (any chg)
6208}
6209;
6210# retval of read_url_paths{,_all}();
6212$l_map = {
6213 # repository root url
6214 'https://svn.musicpd.org' => {
6215 # repository path # GIT_SVN_ID
6216 'mpd/trunk' => 'trunk',
6217 'mpd/tags/0.11.5' => 'tags/0.11.5',
6218 },
6219}
6220Notes:
6222 I don't trust the each() function on unless I created %hash myself
6223 because the internal iterator may not have started at base.