7d86870ee4705090a5b46f3fbf5d0e48e2b3888e
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
8 $_q $_authors %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18$ENV{GIT_DIR} ||= '.git';
19$Git::SVN::default_repo_id = 'svn';
20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21$Git::SVN::Ra::_log_window_size = 100;
22$Git::SVN::Log::TZ = $ENV{TZ};
24$ENV{TZ} = 'UTC';
25$| = 1; # unbuffer STDOUT
26sub fatal (@) { print STDERR "@_\n"; exit 1 }
28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29require SVN::Ra;
30require SVN::Delta;
31if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33}
34push @Git::SVN::Ra::ISA, 'SVN::Ra';
35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37use Carp qw/croak/;
38use IO::File qw//;
39use File::Basename qw/dirname basename/;
40use File::Path qw/mkpath/;
41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
42use IPC::Open3;
43use Git;
44BEGIN {
46 # import functions from Git into our packages, en masse
47 no strict 'refs';
48 foreach (qw/command command_oneline command_noisy command_output_pipe
49 command_input_pipe command_close_pipe/) {
50 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
51 Git::SVN::Migration Git::SVN::Log Git::SVN
52 Git::SVN::Util),
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
56 }
57}
58my ($SVN);
60$sha1 = qr/[a-f\d]{40}/;
62$sha1_short = qr/[a-f\d]{4,40}/;
63my ($_stdin, $_help, $_edit,
64 $_message, $_file,
65 $_template, $_shared,
66 $_version, $_fetch_all, $_no_rebase,
67 $_merge, $_strategy, $_dry_run, $_local,
68 $_prefix, $_no_checkout, $_url, $_verbose);
69$Git::SVN::_follow_parent = 1;
70my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
71 'config-dir=s' => \$Git::SVN::Ra::config_dir,
72 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
73my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
74 'authors-file|A=s' => \$_authors,
75 'repack:i' => \$Git::SVN::_repack,
76 'noMetadata' => \$Git::SVN::_no_metadata,
77 'useSvmProps' => \$Git::SVN::_use_svm_props,
78 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
79 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
80 'no-checkout' => \$_no_checkout,
81 'quiet|q' => \$_q,
82 'repack-flags|repack-args|repack-opts=s' =>
83 \$Git::SVN::_repack_flags,
84 %remote_opts );
85my ($_trunk, $_tags, $_branches, $_stdlayout);
87my %icv;
88my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
89 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
90 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
91 'stdlayout|s' => \$_stdlayout,
92 'minimize-url|m' => \$Git::SVN::_minimize_url,
93 'no-metadata' => sub { $icv{noMetadata} = 1 },
94 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
95 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
96 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
97 %remote_opts );
98my %cmt_opts = ( 'edit|e' => \$_edit,
99 'rmdir' => \$SVN::Git::Editor::_rmdir,
100 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
101 'l=i' => \$SVN::Git::Editor::_rename_limit,
102 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
103);
104my %cmd = (
106 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
107 { 'revision|r=s' => \$_revision,
108 'fetch-all|all' => \$_fetch_all,
109 %fc_opts } ],
110 clone => [ \&cmd_clone, "Initialize and fetch revisions",
111 { 'revision|r=s' => \$_revision,
112 %fc_opts, %init_opts } ],
113 init => [ \&cmd_init, "Initialize a repo for tracking" .
114 " (requires URL argument)",
115 \%init_opts ],
116 'multi-init' => [ \&cmd_multi_init,
117 "Deprecated alias for ".
118 "'$0 init -T<trunk> -b<branches> -t<tags>'",
119 \%init_opts ],
120 dcommit => [ \&cmd_dcommit,
121 'Commit several diffs to merge with upstream',
122 { 'merge|m|M' => \$_merge,
123 'strategy|s=s' => \$_strategy,
124 'verbose|v' => \$_verbose,
125 'dry-run|n' => \$_dry_run,
126 'fetch-all|all' => \$_fetch_all,
127 'no-rebase' => \$_no_rebase,
128 %cmt_opts, %fc_opts } ],
129 'set-tree' => [ \&cmd_set_tree,
130 "Set an SVN repository to a git tree-ish",
131 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
132 'create-ignore' => [ \&cmd_create_ignore,
133 'Create a .gitignore per svn:ignore',
134 { 'revision|r=i' => \$_revision
135 } ],
136 'propget' => [ \&cmd_propget,
137 'Print the value of a property on a file or directory',
138 { 'revision|r=i' => \$_revision } ],
139 'proplist' => [ \&cmd_proplist,
140 'List all properties of a file or directory',
141 { 'revision|r=i' => \$_revision } ],
142 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
143 { 'revision|r=i' => \$_revision
144 } ],
145 'multi-fetch' => [ \&cmd_multi_fetch,
146 "Deprecated alias for $0 fetch --all",
147 { 'revision|r=s' => \$_revision, %fc_opts } ],
148 'migrate' => [ sub { },
149 # no-op, we automatically run this anyways,
150 'Migrate configuration/metadata/layout from
151 previous versions of git-svn',
152 { 'minimize' => \$Git::SVN::Migration::_minimize,
153 %remote_opts } ],
154 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
155 { 'limit=i' => \$Git::SVN::Log::limit,
156 'revision|r=s' => \$_revision,
157 'verbose|v' => \$Git::SVN::Log::verbose,
158 'incremental' => \$Git::SVN::Log::incremental,
159 'oneline' => \$Git::SVN::Log::oneline,
160 'show-commit' => \$Git::SVN::Log::show_commit,
161 'non-recursive' => \$Git::SVN::Log::non_recursive,
162 'authors-file|A=s' => \$_authors,
163 'color' => \$Git::SVN::Log::color,
164 'pager=s' => \$Git::SVN::Log::pager
165 } ],
166 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
167 {} ],
168 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
169 { 'merge|m|M' => \$_merge,
170 'verbose|v' => \$_verbose,
171 'strategy|s=s' => \$_strategy,
172 'local|l' => \$_local,
173 'fetch-all|all' => \$_fetch_all,
174 %fc_opts } ],
175 'commit-diff' => [ \&cmd_commit_diff,
176 'Commit a diff between two trees',
177 { 'message|m=s' => \$_message,
178 'file|F=s' => \$_file,
179 'revision|r=s' => \$_revision,
180 %cmt_opts } ],
181 'info' => [ \&cmd_info,
182 "Show info about the latest SVN revision
183 on the current branch",
184 { 'url' => \$_url, } ],
185);
186my $cmd;
188for (my $i = 0; $i < @ARGV; $i++) {
189 if (defined $cmd{$ARGV[$i]}) {
190 $cmd = $ARGV[$i];
191 splice @ARGV, $i, 1;
192 last;
193 }
194};
195my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
197read_repo_config(\%opts);
199Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
200my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
201 'minimize-connections' => \$Git::SVN::Migration::_minimize,
202 'id|i=s' => \$Git::SVN::default_ref_id,
203 'svn-remote|remote|R=s' => sub {
204 $Git::SVN::no_reuse_existing = 1;
205 $Git::SVN::default_repo_id = $_[1] });
206exit 1 if (!$rv && $cmd && $cmd ne 'log');
207usage(0) if $_help;
209version() if $_version;
210usage(1) unless defined $cmd;
211load_authors() if $_authors;
212# make sure we're always running
214unless ($cmd =~ /(?:clone|init|multi-init)$/) {
215 unless (-d $ENV{GIT_DIR}) {
216 if ($git_dir_user_set) {
217 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
218 "but it is not a directory\n";
219 }
220 my $git_dir = delete $ENV{GIT_DIR};
221 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
222 unless (length $cdup) {
223 die "Already at toplevel, but $git_dir ",
224 "not found '$cdup'\n";
225 }
226 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
227 unless (-d $git_dir) {
228 die "$git_dir still not found after going to ",
229 "'$cdup'\n";
230 }
231 $ENV{GIT_DIR} = $git_dir;
232 }
233}
234unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
235 Git::SVN::Migration::migration_check();
236}
237Git::SVN::init_vars();
238eval {
239 Git::SVN::verify_remotes_sanity();
240 $cmd{$cmd}->[0]->(@ARGV);
241};
242fatal $@ if $@;
243post_fetch_checkout();
244exit 0;
245####################### primary functions ######################
247sub usage {
248 my $exit = shift || 0;
249 my $fd = $exit ? \*STDERR : \*STDOUT;
250 print $fd <<"";
251git-svn - bidirectional operations between a single Subversion tree and git
252Usage: $0 <command> [options] [arguments]\n
253 print $fd "Available commands:\n" unless $cmd;
255 foreach (sort keys %cmd) {
257 next if $cmd && $cmd ne $_;
258 next if /^multi-/; # don't show deprecated commands
259 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
260 foreach (sort keys %{$cmd{$_}->[2]}) {
261 # mixed-case options are for .git/config only
262 next if /[A-Z]/ && /^[a-z]+$/i;
263 # prints out arguments as they should be passed:
264 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
265 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
266 "--$_" : "-$_" }
267 split /\|/,$_)," $x\n";
268 }
269 }
270 print $fd <<"";
271\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
272arbitrary identifier if you're tracking multiple SVN branches/repositories in
273one git repository and want to keep them separate. See git-svn(1) for more
274information.
275 exit $exit;
277}
278sub version {
280 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
281 exit 0;
282}
283sub do_git_init_db {
285 unless (-d $ENV{GIT_DIR}) {
286 my @init_db = ('init');
287 push @init_db, "--template=$_template" if defined $_template;
288 if (defined $_shared) {
289 if ($_shared =~ /[a-z]/) {
290 push @init_db, "--shared=$_shared";
291 } else {
292 push @init_db, "--shared";
293 }
294 }
295 command_noisy(@init_db);
296 }
297 my $set;
298 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
299 foreach my $i (keys %icv) {
300 die "'$set' and '$i' cannot both be set\n" if $set;
301 next unless defined $icv{$i};
302 command_noisy('config', "$pfx.$i", $icv{$i});
303 $set = $i;
304 }
305}
306sub init_subdir {
308 my $repo_path = shift or return;
309 mkpath([$repo_path]) unless -d $repo_path;
310 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
311 $ENV{GIT_DIR} = '.git';
312}
313sub cmd_clone {
315 my ($url, $path) = @_;
316 if (!defined $path &&
317 (defined $_trunk || defined $_branches || defined $_tags ||
318 defined $_stdlayout) &&
319 $url !~ m#^[a-z\+]+://#) {
320 $path = $url;
321 }
322 $path = basename($url) if !defined $path || !length $path;
323 cmd_init($url, $path);
324 Git::SVN::fetch_all($Git::SVN::default_repo_id);
325}
326sub cmd_init {
328 if (defined $_stdlayout) {
329 $_trunk = 'trunk' if (!defined $_trunk);
330 $_tags = 'tags' if (!defined $_tags);
331 $_branches = 'branches' if (!defined $_branches);
332 }
333 if (defined $_trunk || defined $_branches || defined $_tags) {
334 return cmd_multi_init(@_);
335 }
336 my $url = shift or die "SVN repository location required ",
337 "as a command-line argument\n";
338 init_subdir(@_);
339 do_git_init_db();
340 Git::SVN->init($url);
342}
343sub cmd_fetch {
345 if (grep /^\d+=./, @_) {
346 die "'<rev>=<commit>' fetch arguments are ",
347 "no longer supported.\n";
348 }
349 my ($remote) = @_;
350 if (@_ > 1) {
351 die "Usage: $0 fetch [--all] [svn-remote]\n";
352 }
353 $remote ||= $Git::SVN::default_repo_id;
354 if ($_fetch_all) {
355 cmd_multi_fetch();
356 } else {
357 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
358 }
359}
360sub cmd_set_tree {
362 my (@commits) = @_;
363 if ($_stdin || !@commits) {
364 print "Reading from stdin...\n";
365 @commits = ();
366 while (<STDIN>) {
367 if (/\b($sha1_short)\b/o) {
368 unshift @commits, $1;
369 }
370 }
371 }
372 my @revs;
373 foreach my $c (@commits) {
374 my @tmp = command('rev-parse',$c);
375 if (scalar @tmp == 1) {
376 push @revs, $tmp[0];
377 } elsif (scalar @tmp > 1) {
378 push @revs, reverse(command('rev-list',@tmp));
379 } else {
380 fatal "Failed to rev-parse $c";
381 }
382 }
383 my $gs = Git::SVN->new;
384 my ($r_last, $cmt_last) = $gs->last_rev_commit;
385 $gs->fetch;
386 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
387 fatal "There are new revisions that were fetched ",
388 "and need to be merged (or acknowledged) ",
389 "before committing.\nlast rev: $r_last\n",
390 " current: $gs->{last_rev}";
391 }
392 $gs->set_tree($_) foreach @revs;
393 print "Done committing ",scalar @revs," revisions to SVN\n";
394}
395sub cmd_dcommit {
397 my $head = shift;
398 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
399 'Cannot dcommit with a dirty index. Commit your changes first, '
400 . "or stash them with `git stash'.\n";
401 $head ||= 'HEAD';
402 my @refs;
403 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
404 print "Committing to $url ...\n";
405 unless ($gs) {
406 die "Unable to determine upstream SVN information from ",
407 "$head history\n";
408 }
409 my $last_rev;
410 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
411 if ($_no_rebase && scalar(@$linear_refs) > 1) {
412 warn "Attempting to commit more than one change while ",
413 "--no-rebase is enabled.\n",
414 "If these changes depend on each other, re-running ",
415 "without --no-rebase will be required."
416 }
417 while (1) {
418 my $d = shift @$linear_refs or last;
419 unless (defined $last_rev) {
420 (undef, $last_rev, undef) = cmt_metadata("$d~1");
421 unless (defined $last_rev) {
422 fatal "Unable to extract revision information ",
423 "from commit $d~1";
424 }
425 }
426 if ($_dry_run) {
427 print "diff-tree $d~1 $d\n";
428 } else {
429 my $cmt_rev;
430 my %ed_opts = ( r => $last_rev,
431 log => get_commit_entry($d)->{log},
432 ra => Git::SVN::Ra->new($gs->full_url),
433 config => SVN::Core::config_get_config(
434 $Git::SVN::Ra::config_dir
435 ),
436 tree_a => "$d~1",
437 tree_b => $d,
438 editor_cb => sub {
439 print "Committed r$_[0]\n";
440 $cmt_rev = $_[0];
441 },
442 svn_path => '');
443 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
444 print "No changes\n$d~1 == $d\n";
445 } elsif ($parents->{$d} && @{$parents->{$d}}) {
446 $gs->{inject_parents_dcommit}->{$cmt_rev} =
447 $parents->{$d};
448 }
449 $_fetch_all ? $gs->fetch_all : $gs->fetch;
450 next if $_no_rebase;
451 # we always want to rebase against the current HEAD,
453 # not any head that was passed to us
454 my @diff = command('diff-tree', $d,
455 $gs->refname, '--');
456 my @finish;
457 if (@diff) {
458 @finish = rebase_cmd();
459 print STDERR "W: $d and ", $gs->refname,
460 " differ, using @finish:\n",
461 join("\n", @diff), "\n";
462 } else {
463 print "No changes between current HEAD and ",
464 $gs->refname,
465 "\nResetting to the latest ",
466 $gs->refname, "\n";
467 @finish = qw/reset --mixed/;
468 }
469 command_noisy(@finish, $gs->refname);
470 if (@diff) {
471 @refs = ();
472 my ($url_, $rev_, $uuid_, $gs_) =
473 working_head_info($head, \@refs);
474 my ($linear_refs_, $parents_) =
475 linearize_history($gs_, \@refs);
476 if (scalar(@$linear_refs) !=
477 scalar(@$linear_refs_)) {
478 fatal "# of revisions changed ",
479 "\nbefore:\n",
480 join("\n", @$linear_refs),
481 "\n\nafter:\n",
482 join("\n", @$linear_refs_), "\n",
483 'If you are attempting to commit ',
484 "merges, try running:\n\t",
485 'git rebase --interactive',
486 '--preserve-merges ',
487 $gs->refname,
488 "\nBefore dcommitting";
489 }
490 if ($url_ ne $url) {
491 fatal "URL mismatch after rebase: ",
492 "$url_ != $url";
493 }
494 if ($uuid_ ne $uuid) {
495 fatal "uuid mismatch after rebase: ",
496 "$uuid_ != $uuid";
497 }
498 # remap parents
499 my (%p, @l, $i);
500 for ($i = 0; $i < scalar @$linear_refs; $i++) {
501 my $new = $linear_refs_->[$i] or next;
502 $p{$new} =
503 $parents->{$linear_refs->[$i]};
504 push @l, $new;
505 }
506 $parents = \%p;
507 $linear_refs = \@l;
508 }
509 $last_rev = $cmt_rev;
510 }
511 }
512}
513sub cmd_find_rev {
515 my $revision_or_hash = shift;
516 my $result;
517 if ($revision_or_hash =~ /^r\d+$/) {
518 my $head = shift;
519 $head ||= 'HEAD';
520 my @refs;
521 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
522 unless ($gs) {
523 die "Unable to determine upstream SVN information from ",
524 "$head history\n";
525 }
526 my $desired_revision = substr($revision_or_hash, 1);
527 $result = $gs->rev_db_get($desired_revision);
528 } else {
529 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
530 $result = $rev;
531 }
532 print "$result\n" if $result;
533}
534sub cmd_rebase {
536 command_noisy(qw/update-index --refresh/);
537 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
538 unless ($gs) {
539 die "Unable to determine upstream SVN information from ",
540 "working tree history\n";
541 }
542 if (command(qw/diff-index HEAD --/)) {
543 print STDERR "Cannot rebase with uncommited changes:\n";
544 command_noisy('status');
545 exit 1;
546 }
547 unless ($_local) {
548 $_fetch_all ? $gs->fetch_all : $gs->fetch;
549 }
550 command_noisy(rebase_cmd(), $gs->refname);
551}
552sub cmd_show_ignore {
554 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
555 $gs ||= Git::SVN->new;
556 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
557 $gs->prop_walk($gs->{path}, $r, sub {
558 my ($gs, $path, $props) = @_;
559 print STDOUT "\n# $path\n";
560 my $s = $props->{'svn:ignore'} or return;
561 $s =~ s/[\r\n]+/\n/g;
562 chomp $s;
563 $s =~ s#^#$path#gm;
564 print STDOUT "$s\n";
565 });
566}
567sub cmd_create_ignore {
569 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
570 $gs ||= Git::SVN->new;
571 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
572 $gs->prop_walk($gs->{path}, $r, sub {
573 my ($gs, $path, $props) = @_;
574 # $path is of the form /path/to/dir/
575 my $ignore = '.' . $path . '.gitignore';
576 my $s = $props->{'svn:ignore'} or return;
577 open(GITIGNORE, '>', $ignore)
578 or fatal("Failed to open `$ignore' for writing: $!");
579 $s =~ s/[\r\n]+/\n/g;
580 chomp $s;
581 # Prefix all patterns so that the ignore doesn't apply
582 # to sub-directories.
583 $s =~ s#^#/#gm;
584 print GITIGNORE "$s\n";
585 close(GITIGNORE)
586 or fatal("Failed to close `$ignore': $!");
587 command_noisy('add', $ignore);
588 });
589}
590sub canonicalize_path {
592 my ($path) = @_;
593 my $dot_slash_added = 0;
594 if (substr($path, 0, 1) ne "/") {
595 $path = "./" . $path;
596 $dot_slash_added = 1;
597 }
598 # File::Spec->canonpath doesn't collapse x/../y into y (for a
599 # good reason), so let's do this manually.
600 $path =~ s#/+#/#g;
601 $path =~ s#/\.(?:/|$)#/#g;
602 $path =~ s#/[^/]+/\.\.##g;
603 $path =~ s#/$##g;
604 $path =~ s#^\./## if $dot_slash_added;
605 return $path;
606}
607# get_svnprops(PATH)
609# ------------------
610# Helper for cmd_propget and cmd_proplist below.
611sub get_svnprops {
612 my $path = shift;
613 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
614 $gs ||= Git::SVN->new;
615 # prefix THE PATH by the sub-directory from which the user
617 # invoked us.
618 $path = $cmd_dir_prefix . $path;
619 fatal("No such file or directory: $path") unless -e $path;
620 my $is_dir = -d $path ? 1 : 0;
621 $path = $gs->{path} . '/' . $path;
622 # canonicalize the path (otherwise libsvn will abort or fail to
624 # find the file)
625 $path = canonicalize_path($path);
626 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
628 my $props;
629 if ($is_dir) {
630 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
631 }
632 else {
633 (undef, $props) = $gs->ra->get_file($path, $r, undef);
634 }
635 return $props;
636}
637# cmd_propget (PROP, PATH)
639# ------------------------
640# Print the SVN property PROP for PATH.
641sub cmd_propget {
642 my ($prop, $path) = @_;
643 $path = '.' if not defined $path;
644 usage(1) if not defined $prop;
645 my $props = get_svnprops($path);
646 if (not defined $props->{$prop}) {
647 fatal("`$path' does not have a `$prop' SVN property.");
648 }
649 print $props->{$prop} . "\n";
650}
651# cmd_proplist (PATH)
653# -------------------
654# Print the list of SVN properties for PATH.
655sub cmd_proplist {
656 my $path = shift;
657 $path = '.' if not defined $path;
658 my $props = get_svnprops($path);
659 print "Properties on '$path':\n";
660 foreach (sort keys %{$props}) {
661 print " $_\n";
662 }
663}
664sub cmd_multi_init {
666 my $url = shift;
667 unless (defined $_trunk || defined $_branches || defined $_tags) {
668 usage(1);
669 }
670 # there are currently some bugs that prevent multi-init/multi-fetch
672 # setups from working well without this.
673 $Git::SVN::_minimize_url = 1;
674 $_prefix = '' unless defined $_prefix;
676 if (defined $url) {
677 $url =~ s#/+$##;
678 init_subdir(@_);
679 }
680 do_git_init_db();
681 if (defined $_trunk) {
682 my $trunk_ref = $_prefix . 'trunk';
683 # try both old-style and new-style lookups:
684 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
685 unless ($gs_trunk) {
686 my ($trunk_url, $trunk_path) =
687 complete_svn_url($url, $_trunk);
688 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
689 undef, $trunk_ref);
690 }
691 }
692 return unless defined $_branches || defined $_tags;
693 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
694 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
695 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
696}
697sub cmd_multi_fetch {
699 my $remotes = Git::SVN::read_all_remotes();
700 foreach my $repo_id (sort keys %$remotes) {
701 if ($remotes->{$repo_id}->{url}) {
702 Git::SVN::fetch_all($repo_id, $remotes);
703 }
704 }
705}
706# this command is special because it requires no metadata
708sub cmd_commit_diff {
709 my ($ta, $tb, $url) = @_;
710 my $usage = "Usage: $0 commit-diff -r<revision> ".
711 "<tree-ish> <tree-ish> [<URL>]";
712 fatal($usage) if (!defined $ta || !defined $tb);
713 my $svn_path;
714 if (!defined $url) {
715 my $gs = eval { Git::SVN->new };
716 if (!$gs) {
717 fatal("Needed URL or usable git-svn --id in ",
718 "the command-line\n", $usage);
719 }
720 $url = $gs->{url};
721 $svn_path = $gs->{path};
722 }
723 unless (defined $_revision) {
724 fatal("-r|--revision is a required argument\n", $usage);
725 }
726 if (defined $_message && defined $_file) {
727 fatal("Both --message/-m and --file/-F specified ",
728 "for the commit message.\n",
729 "I have no idea what you mean");
730 }
731 if (defined $_file) {
732 $_message = file_to_s($_file);
733 } else {
734 $_message ||= get_commit_entry($tb)->{log};
735 }
736 my $ra ||= Git::SVN::Ra->new($url);
737 $svn_path ||= $ra->{svn_path};
738 my $r = $_revision;
739 if ($r eq 'HEAD') {
740 $r = $ra->get_latest_revnum;
741 } elsif ($r !~ /^\d+$/) {
742 die "revision argument: $r not understood by git-svn\n";
743 }
744 my %ed_opts = ( r => $r,
745 log => $_message,
746 ra => $ra,
747 tree_a => $ta,
748 tree_b => $tb,
749 editor_cb => sub { print "Committed r$_[0]\n" },
750 svn_path => $svn_path );
751 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
752 print "No changes\n$ta == $tb\n";
753 }
754}
755sub cmd_info {
757 my $path = canonicalize_path(shift or ".");
758 unless (scalar(@_) == 0) {
759 die "Too many arguments specified\n";
760 }
761 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
763 if (!$file_type && !$diff_status) {
765 print STDERR "$path: (Not a versioned resource)\n\n";
766 return;
767 }
768 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
770 unless ($gs) {
771 die "Unable to determine upstream SVN information from ",
772 "working tree history\n";
773 }
774 my $full_url = $url . ($path eq "." ? "" : "/$path");
775 if ($_url) {
777 print $full_url, "\n";
778 return;
779 }
780 my $result = "Path: $path\n";
782 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
783 $result .= "URL: " . $full_url . "\n";
784 my $repos_root = $gs->ra->{repos_root};
786 Git::SVN::remove_username($repos_root);
787 $result .= "Repository Root: $repos_root\n";
788 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
789 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
790 $result .= "Node Kind: " .
792 ($file_type eq "dir" ? "directory" : "file") . "\n";
793 my $schedule = $diff_status eq "A"
795 ? "add"
796 : ($diff_status eq "D" ? "delete" : "normal");
797 $result .= "Schedule: $schedule\n";
798 if ($diff_status eq "A") {
800 print $result, "\n";
801 return;
802 }
803 my ($lc_author, $lc_rev, $lc_date_utc);
805 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
806 my $log = command_output_pipe(@args);
807 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
808 while (<$log>) {
809 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
810 $lc_author = $1;
811 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
812 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
813 (undef, $lc_rev, undef) = ::extract_metadata($1);
814 }
815 }
816 close $log;
817 Git::SVN::Log::set_local_timezone();
819 $result .= "Last Changed Author: $lc_author\n";
821 $result .= "Last Changed Rev: $lc_rev\n";
822 $result .= "Last Changed Date: " .
823 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
824 if ($file_type ne "dir") {
826 my $text_last_updated_date =
827 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
828 $result .=
829 "Text Last Updated: " .
830 Git::SVN::Log::format_svn_date($text_last_updated_date) .
831 "\n";
832 my $checksum;
833 if ($diff_status eq "D") {
834 my ($fh, $ctx) =
835 command_output_pipe(qw(cat-file blob), "HEAD:$path");
836 if ($file_type eq "link") {
837 my $file_name = <$fh>;
838 $checksum = Git::SVN::Util::md5sum("link $file_name");
839 } else {
840 $checksum = Git::SVN::Util::md5sum($fh);
841 }
842 command_close_pipe($fh, $ctx);
843 } elsif ($file_type eq "link") {
844 my $file_name =
845 command(qw(cat-file blob), "HEAD:$path");
846 $checksum =
847 Git::SVN::Util::md5sum("link " . $file_name);
848 } else {
849 open FILE, "<", $path or die $!;
850 $checksum = Git::SVN::Util::md5sum(\*FILE);
851 close FILE or die $!;
852 }
853 $result .= "Checksum: " . $checksum . "\n";
854 }
855 print $result, "\n";
857}
858########################### utility functions #########################
860sub rebase_cmd {
862 my @cmd = qw/rebase/;
863 push @cmd, '-v' if $_verbose;
864 push @cmd, qw/--merge/ if $_merge;
865 push @cmd, "--strategy=$_strategy" if $_strategy;
866 @cmd;
867}
868sub post_fetch_checkout {
870 return if $_no_checkout;
871 my $gs = $Git::SVN::_head or return;
872 return if verify_ref('refs/heads/master^0');
873 my $valid_head = verify_ref('HEAD^0');
875 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
876 return if ($valid_head || !verify_ref('HEAD^0'));
877 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
879 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
880 return if -f $index;
881 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
883 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
884 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
885 print STDERR "Checked out HEAD:\n ",
886 $gs->full_url, " r", $gs->last_rev, "\n";
887}
888sub complete_svn_url {
890 my ($url, $path) = @_;
891 $path =~ s#/+$##;
892 if ($path !~ m#^[a-z\+]+://#) {
893 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
894 fatal("E: '$path' is not a complete URL ",
895 "and a separate URL is not specified");
896 }
897 return ($url, $path);
898 }
899 return ($path, '');
900}
901sub complete_url_ls_init {
903 my ($ra, $repo_path, $switch, $pfx) = @_;
904 unless ($repo_path) {
905 print STDERR "W: $switch not specified\n";
906 return;
907 }
908 $repo_path =~ s#/+$##;
909 if ($repo_path =~ m#^[a-z\+]+://#) {
910 $ra = Git::SVN::Ra->new($repo_path);
911 $repo_path = '';
912 } else {
913 $repo_path =~ s#^/+##;
914 unless ($ra) {
915 fatal("E: '$repo_path' is not a complete URL ",
916 "and a separate URL is not specified");
917 }
918 }
919 my $url = $ra->{url};
920 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
921 my $k = "svn-remote.$gs->{repo_id}.url";
922 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
923 if ($orig_url && ($orig_url ne $gs->{url})) {
924 die "$k already set: $orig_url\n",
925 "wanted to set to: $gs->{url}\n";
926 }
927 command_oneline('config', $k, $gs->{url}) unless $orig_url;
928 my $remote_path = "$ra->{svn_path}/$repo_path/*";
929 $remote_path =~ s#/+#/#g;
930 $remote_path =~ s#^/##g;
931 my ($n) = ($switch =~ /^--(\w+)/);
932 if (length $pfx && $pfx !~ m#/$#) {
933 die "--prefix='$pfx' must have a trailing slash '/'\n";
934 }
935 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
936 "$remote_path:refs/remotes/$pfx*");
937}
938sub verify_ref {
940 my ($ref) = @_;
941 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
942 { STDERR => 0 }); };
943}
944sub get_tree_from_treeish {
946 my ($treeish) = @_;
947 # $treeish can be a symbolic ref, too:
948 my $type = command_oneline(qw/cat-file -t/, $treeish);
949 my $expected;
950 while ($type eq 'tag') {
951 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
952 }
953 if ($type eq 'commit') {
954 $expected = (grep /^tree /, command(qw/cat-file commit/,
955 $treeish))[0];
956 ($expected) = ($expected =~ /^tree ($sha1)$/o);
957 die "Unable to get tree from $treeish\n" unless $expected;
958 } elsif ($type eq 'tree') {
959 $expected = $treeish;
960 } else {
961 die "$treeish is a $type, expected tree, tag or commit\n";
962 }
963 return $expected;
964}
965sub get_commit_entry {
967 my ($treeish) = shift;
968 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
969 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
970 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
971 open my $log_fh, '>', $commit_editmsg or croak $!;
972 my $type = command_oneline(qw/cat-file -t/, $treeish);
974 if ($type eq 'commit' || $type eq 'tag') {
975 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
976 $type, $treeish);
977 my $in_msg = 0;
978 while (<$msg_fh>) {
979 if (!$in_msg) {
980 $in_msg = 1 if (/^\s*$/);
981 } elsif (/^git-svn-id: /) {
982 # skip this for now, we regenerate the
983 # correct one on re-fetch anyways
984 # TODO: set *:merge properties or like...
985 } else {
986 print $log_fh $_ or croak $!;
987 }
988 }
989 command_close_pipe($msg_fh, $ctx);
990 }
991 close $log_fh or croak $!;
992 if ($_edit || ($type eq 'tree')) {
994 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
995 # TODO: strip out spaces, comments, like git-commit.sh
996 system($editor, $commit_editmsg);
997 }
998 rename $commit_editmsg, $commit_msg or croak $!;
999 open $log_fh, '<', $commit_msg or croak $!;
1000 { local $/; chomp($log_entry{log} = <$log_fh>); }
1001 close $log_fh or croak $!;
1002 unlink $commit_msg;
1003 \%log_entry;
1004}
1005sub s_to_file {
1007 my ($str, $file, $mode) = @_;
1008 open my $fd,'>',$file or croak $!;
1009 print $fd $str,"\n" or croak $!;
1010 close $fd or croak $!;
1011 chmod ($mode &~ umask, $file) if (defined $mode);
1012}
1013sub file_to_s {
1015 my $file = shift;
1016 open my $fd,'<',$file or croak "$!: file: $file\n";
1017 local $/;
1018 my $ret = <$fd>;
1019 close $fd or croak $!;
1020 $ret =~ s/\s*$//s;
1021 return $ret;
1022}
1023# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1025sub load_authors {
1026 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1027 my $log = $cmd eq 'log';
1028 while (<$authors>) {
1029 chomp;
1030 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1031 my ($user, $name, $email) = ($1, $2, $3);
1032 if ($log) {
1033 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1034 } else {
1035 $users{$user} = [$name, $email];
1036 }
1037 }
1038 close $authors or croak $!;
1039}
1040# convert GetOpt::Long specs for use by git-config
1042sub read_repo_config {
1043 return unless -d $ENV{GIT_DIR};
1044 my $opts = shift;
1045 my @config_only;
1046 foreach my $o (keys %$opts) {
1047 # if we have mixedCase and a long option-only, then
1048 # it's a config-only variable that we don't need for
1049 # the command-line.
1050 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1051 my $v = $opts->{$o};
1052 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1053 $key =~ s/-//g;
1054 my $arg = 'git-config';
1055 $arg .= ' --int' if ($o =~ /[:=]i$/);
1056 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1057 if (ref $v eq 'ARRAY') {
1058 chomp(my @tmp = `$arg --get-all svn.$key`);
1059 @$v = @tmp if @tmp;
1060 } else {
1061 chomp(my $tmp = `$arg --get svn.$key`);
1062 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1063 $$v = $tmp;
1064 }
1065 }
1066 }
1067 delete @$opts{@config_only} if @config_only;
1068}
1069sub extract_metadata {
1071 my $id = shift or return (undef, undef, undef);
1072 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1073 \s([a-f\d\-]+)$/x);
1074 if (!defined $rev || !$uuid || !$url) {
1075 # some of the original repositories I made had
1076 # identifiers like this:
1077 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1078 }
1079 return ($url, $rev, $uuid);
1080}
1081sub cmt_metadata {
1083 return extract_metadata((grep(/^git-svn-id: /,
1084 command(qw/cat-file commit/, shift)))[-1]);
1085}
1086sub working_head_info {
1088 my ($head, $refs) = @_;
1089 my @args = ('log', '--no-color', '--first-parent');
1090 my ($fh, $ctx) = command_output_pipe(@args, $head);
1091 my $hash;
1092 my %max;
1093 while (<$fh>) {
1094 if ( m{^commit ($::sha1)$} ) {
1095 unshift @$refs, $hash if $hash and $refs;
1096 $hash = $1;
1097 next;
1098 }
1099 next unless s{^\s*(git-svn-id:)}{$1};
1100 my ($url, $rev, $uuid) = extract_metadata($_);
1101 if (defined $url && defined $rev) {
1102 next if $max{$url} and $max{$url} < $rev;
1103 if (my $gs = Git::SVN->find_by_url($url)) {
1104 my $c = $gs->rev_db_get($rev);
1105 if ($c && $c eq $hash) {
1106 close $fh; # break the pipe
1107 return ($url, $rev, $uuid, $gs);
1108 } else {
1109 $max{$url} ||= $gs->rev_db_max;
1110 }
1111 }
1112 }
1113 }
1114 command_close_pipe($fh, $ctx);
1115 (undef, undef, undef, undef);
1116}
1117sub read_commit_parents {
1119 my ($parents, $c) = @_;
1120 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1121 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1122 @{$parents->{$c}} = split(/ /, $p);
1123}
1124sub linearize_history {
1126 my ($gs, $refs) = @_;
1127 my %parents;
1128 foreach my $c (@$refs) {
1129 read_commit_parents(\%parents, $c);
1130 }
1131 my @linear_refs;
1133 my %skip = ();
1134 my $last_svn_commit = $gs->last_commit;
1135 foreach my $c (reverse @$refs) {
1136 next if $c eq $last_svn_commit;
1137 last if $skip{$c};
1138 unshift @linear_refs, $c;
1140 $skip{$c} = 1;
1141 # we only want the first parent to diff against for linear
1143 # history, we save the rest to inject when we finalize the
1144 # svn commit
1145 my $fp_a = verify_ref("$c~1");
1146 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1147 if (!$fp_a || !$fp_b) {
1148 die "Commit $c\n",
1149 "has no parent commit, and therefore ",
1150 "nothing to diff against.\n",
1151 "You should be working from a repository ",
1152 "originally created by git-svn\n";
1153 }
1154 if ($fp_a ne $fp_b) {
1155 die "$c~1 = $fp_a, however parsing commit $c ",
1156 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1157 }
1158 foreach my $p (@{$parents{$c}}) {
1160 $skip{$p} = 1;
1161 }
1162 }
1163 (\@linear_refs, \%parents);
1164}
1165sub find_file_type_and_diff_status {
1167 my ($path) = @_;
1168 return ('dir', '') if $path eq '.';
1169 my $diff_output =
1171 command_oneline(qw(diff --cached --name-status --), $path) || "";
1172 my $diff_status = (split(' ', $diff_output))[0] || "";
1173 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1175 return (undef, undef) if !$diff_status && !$ls_tree;
1177 if ($diff_status eq "A") {
1179 return ("link", $diff_status) if -l $path;
1180 return ("dir", $diff_status) if -d $path;
1181 return ("file", $diff_status);
1182 }
1183 my $mode = (split(' ', $ls_tree))[0] || "";
1185 return ("link", $diff_status) if $mode eq "120000";
1187 return ("dir", $diff_status) if $mode eq "040000";
1188 return ("file", $diff_status);
1189}
1190package Git::SVN::Util;
1192use strict;
1193use warnings;
1194use Digest::MD5;
1195sub md5sum {
1197 my $arg = shift;
1198 my $ref = ref $arg;
1199 my $md5 = Digest::MD5->new();
1200 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1201 $md5->addfile($arg) or croak $!;
1202 } elsif ($ref eq 'SCALAR') {
1203 $md5->add($$arg) or croak $!;
1204 } elsif (!$ref) {
1205 $md5->add($arg) or croak $!;
1206 } else {
1207 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1208 }
1209 return $md5->hexdigest();
1210}
1211package Git::SVN;
1213use strict;
1214use warnings;
1215use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1216 $_repack $_repack_flags $_use_svm_props $_head
1217 $_use_svnsync_props $no_reuse_existing $_minimize_url/;
1218use Carp qw/croak/;
1219use File::Path qw/mkpath/;
1220use File::Copy qw/copy/;
1221use IPC::Open3;
1222my $_repack_nr;
1224# properties that we do not log:
1225my %SKIP_PROP;
1226BEGIN {
1227 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1228 svn:special svn:executable
1229 svn:entry:committed-rev
1230 svn:entry:last-author
1231 svn:entry:uuid
1232 svn:entry:committed-date/;
1233 # some options are read globally, but can be overridden locally
1235 # per [svn-remote "..."] section. Command-line options will *NOT*
1236 # override options set in an [svn-remote "..."] section
1237 no strict 'refs';
1238 for my $option (qw/follow_parent no_metadata use_svm_props
1239 use_svnsync_props/) {
1240 my $key = $option;
1241 $key =~ tr/_//d;
1242 my $prop = "-$option";
1243 *$option = sub {
1244 my ($self) = @_;
1245 return $self->{$prop} if exists $self->{$prop};
1246 my $k = "svn-remote.$self->{repo_id}.$key";
1247 eval { command_oneline(qw/config --get/, $k) };
1248 if ($@) {
1249 $self->{$prop} = ${"Git::SVN::_$option"};
1250 } else {
1251 my $v = command_oneline(qw/config --bool/,$k);
1252 $self->{$prop} = $v eq 'false' ? 0 : 1;
1253 }
1254 return $self->{$prop};
1255 }
1256 }
1257}
1258my %LOCKFILES;
1260END { unlink keys %LOCKFILES if %LOCKFILES }
1261sub resolve_local_globs {
1263 my ($url, $fetch, $glob_spec) = @_;
1264 return unless defined $glob_spec;
1265 my $ref = $glob_spec->{ref};
1266 my $path = $glob_spec->{path};
1267 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1268 next unless m#^refs/remotes/$ref->{regex}$#;
1269 my $p = $1;
1270 my $pathname = desanitize_refname($path->full_path($p));
1271 my $refname = desanitize_refname($ref->full_path($p));
1272 if (my $existing = $fetch->{$pathname}) {
1273 if ($existing ne $refname) {
1274 die "Refspec conflict:\n",
1275 "existing: refs/remotes/$existing\n",
1276 " globbed: refs/remotes/$refname\n";
1277 }
1278 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1279 $u =~ s!^\Q$url\E(/|$)!! or die
1280 "refs/remotes/$refname: '$url' not found in '$u'\n";
1281 if ($pathname ne $u) {
1282 warn "W: Refspec glob conflict ",
1283 "(ref: refs/remotes/$refname):\n",
1284 "expected path: $pathname\n",
1285 " real path: $u\n",
1286 "Continuing ahead with $u\n";
1287 next;
1288 }
1289 } else {
1290 $fetch->{$pathname} = $refname;
1291 }
1292 }
1293}
1294sub parse_revision_argument {
1296 my ($base, $head) = @_;
1297 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1298 return ($base, $head);
1299 }
1300 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1301 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1302 return ($head, $head) if ($::_revision eq 'HEAD');
1303 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1304 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1305 die "revision argument: $::_revision not understood by git-svn\n";
1306}
1307sub fetch_all {
1309 my ($repo_id, $remotes) = @_;
1310 if (ref $repo_id) {
1311 my $gs = $repo_id;
1312 $repo_id = undef;
1313 $repo_id = $gs->{repo_id};
1314 }
1315 $remotes ||= read_all_remotes();
1316 my $remote = $remotes->{$repo_id} or
1317 die "[svn-remote \"$repo_id\"] unknown\n";
1318 my $fetch = $remote->{fetch};
1319 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1320 my (@gs, @globs);
1321 my $ra = Git::SVN::Ra->new($url);
1322 my $uuid = $ra->get_uuid;
1323 my $head = $ra->get_latest_revnum;
1324 my $base = defined $fetch ? $head : 0;
1325 # read the max revs for wildcard expansion (branches/*, tags/*)
1327 foreach my $t (qw/branches tags/) {
1328 defined $remote->{$t} or next;
1329 push @globs, $remote->{$t};
1330 my $max_rev = eval { tmp_config(qw/--int --get/,
1331 "svn-remote.$repo_id.${t}-maxRev") };
1332 if (defined $max_rev && ($max_rev < $base)) {
1333 $base = $max_rev;
1334 } elsif (!defined $max_rev) {
1335 $base = 0;
1336 }
1337 }
1338 if ($fetch) {
1340 foreach my $p (sort keys %$fetch) {
1341 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1342 my $lr = $gs->rev_db_max;
1343 if (defined $lr) {
1344 $base = $lr if ($lr < $base);
1345 }
1346 push @gs, $gs;
1347 }
1348 }
1349 ($base, $head) = parse_revision_argument($base, $head);
1351 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1352}
1353sub read_all_remotes {
1355 my $r = {};
1356 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1357 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1358 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1359 $local_ref =~ s{^/}{};
1360 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1361 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1362 $r->{$1}->{url} = $2;
1363 } elsif (m!^(.+)\.(branches|tags)=
1364 (.*):refs/remotes/(.+)\s*$/!x) {
1365 my ($p, $g) = ($3, $4);
1366 my $rs = $r->{$1}->{$2} = {
1367 t => $2,
1368 remote => $1,
1369 path => Git::SVN::GlobSpec->new($p),
1370 ref => Git::SVN::GlobSpec->new($g) };
1371 if (length($rs->{ref}->{right}) != 0) {
1372 die "The '*' glob character must be the last ",
1373 "character of '$g'\n";
1374 }
1375 }
1376 }
1377 $r;
1378}
1379sub init_vars {
1381 if (defined $_repack) {
1382 $_repack = 1000 if ($_repack <= 0);
1383 $_repack_nr = $_repack;
1384 $_repack_flags ||= '-d';
1385 }
1386}
1387sub verify_remotes_sanity {
1389 return unless -d $ENV{GIT_DIR};
1390 my %seen;
1391 foreach (command(qw/config -l/)) {
1392 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1393 if ($seen{$1}) {
1394 die "Remote ref refs/remote/$1 is tracked by",
1395 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1396 "Please resolve this ambiguity in ",
1397 "your git configuration file before ",
1398 "continuing\n";
1399 }
1400 $seen{$1} = $_;
1401 }
1402 }
1403}
1404# we allow more chars than remotes2config.sh...
1406sub sanitize_remote_name {
1407 my ($name) = @_;
1408 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1409 $name;
1410}
1411sub find_existing_remote {
1413 my ($url, $remotes) = @_;
1414 return undef if $no_reuse_existing;
1415 my $existing;
1416 foreach my $repo_id (keys %$remotes) {
1417 my $u = $remotes->{$repo_id}->{url} or next;
1418 next if $u ne $url;
1419 $existing = $repo_id;
1420 last;
1421 }
1422 $existing;
1423}
1424sub init_remote_config {
1426 my ($self, $url, $no_write) = @_;
1427 $url =~ s!/+$!!; # strip trailing slash
1428 my $r = read_all_remotes();
1429 my $existing = find_existing_remote($url, $r);
1430 if ($existing) {
1431 unless ($no_write) {
1432 print STDERR "Using existing ",
1433 "[svn-remote \"$existing\"]\n";
1434 }
1435 $self->{repo_id} = $existing;
1436 } elsif ($_minimize_url) {
1437 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1438 $existing = find_existing_remote($min_url, $r);
1439 if ($existing) {
1440 unless ($no_write) {
1441 print STDERR "Using existing ",
1442 "[svn-remote \"$existing\"]\n";
1443 }
1444 $self->{repo_id} = $existing;
1445 }
1446 if ($min_url ne $url) {
1447 unless ($no_write) {
1448 print STDERR "Using higher level of URL: ",
1449 "$url => $min_url\n";
1450 }
1451 my $old_path = $self->{path};
1452 $self->{path} = $url;
1453 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1454 if (length $old_path) {
1455 $self->{path} .= "/$old_path";
1456 }
1457 $url = $min_url;
1458 }
1459 }
1460 my $orig_url;
1461 if (!$existing) {
1462 # verify that we aren't overwriting anything:
1463 $orig_url = eval {
1464 command_oneline('config', '--get',
1465 "svn-remote.$self->{repo_id}.url")
1466 };
1467 if ($orig_url && ($orig_url ne $url)) {
1468 die "svn-remote.$self->{repo_id}.url already set: ",
1469 "$orig_url\nwanted to set to: $url\n";
1470 }
1471 }
1472 my ($xrepo_id, $xpath) = find_ref($self->refname);
1473 if (defined $xpath) {
1474 die "svn-remote.$xrepo_id.fetch already set to track ",
1475 "$xpath:refs/remotes/", $self->refname, "\n";
1476 }
1477 unless ($no_write) {
1478 command_noisy('config',
1479 "svn-remote.$self->{repo_id}.url", $url);
1480 $self->{path} =~ s{^/}{};
1481 command_noisy('config', '--add',
1482 "svn-remote.$self->{repo_id}.fetch",
1483 "$self->{path}:".$self->refname);
1484 }
1485 $self->{url} = $url;
1486}
1487sub find_by_url { # repos_root and, path are optional
1489 my ($class, $full_url, $repos_root, $path) = @_;
1490 return undef unless defined $full_url;
1492 remove_username($full_url);
1493 remove_username($repos_root) if defined $repos_root;
1494 my $remotes = read_all_remotes();
1495 if (defined $full_url && defined $repos_root && !defined $path) {
1496 $path = $full_url;
1497 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1498 }
1499 foreach my $repo_id (keys %$remotes) {
1500 my $u = $remotes->{$repo_id}->{url} or next;
1501 remove_username($u);
1502 next if defined $repos_root && $repos_root ne $u;
1503 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1505 foreach (qw/branches tags/) {
1506 resolve_local_globs($u, $fetch,
1507 $remotes->{$repo_id}->{$_});
1508 }
1509 my $p = $path;
1510 unless (defined $p) {
1511 $p = $full_url;
1512 $p =~ s#^\Q$u\E(?:/|$)## or next;
1513 }
1514 foreach my $f (keys %$fetch) {
1515 next if $f ne $p;
1516 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1517 }
1518 }
1519 undef;
1520}
1521sub init {
1523 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1524 my $self = _new($class, $repo_id, $ref_id, $path);
1525 if (defined $url) {
1526 $self->init_remote_config($url, $no_write);
1527 }
1528 $self;
1529}
1530sub find_ref {
1532 my ($ref_id) = @_;
1533 foreach (command(qw/config -l/)) {
1534 next unless m!^svn-remote\.(.+)\.fetch=
1535 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1536 my ($repo_id, $path, $ref) = ($1, $2, $3);
1537 if ($ref eq $ref_id) {
1538 $path = '' if ($path =~ m#^\./?#);
1539 return ($repo_id, $path);
1540 }
1541 }
1542 (undef, undef, undef);
1543}
1544sub new {
1546 my ($class, $ref_id, $repo_id, $path) = @_;
1547 if (defined $ref_id && !defined $repo_id && !defined $path) {
1548 ($repo_id, $path) = find_ref($ref_id);
1549 if (!defined $repo_id) {
1550 die "Could not find a \"svn-remote.*.fetch\" key ",
1551 "in the repository configuration matching: ",
1552 "refs/remotes/$ref_id\n";
1553 }
1554 }
1555 my $self = _new($class, $repo_id, $ref_id, $path);
1556 if (!defined $self->{path} || !length $self->{path}) {
1557 my $fetch = command_oneline('config', '--get',
1558 "svn-remote.$repo_id.fetch",
1559 ":refs/remotes/$ref_id\$") or
1560 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1561 "\":refs/remotes/$ref_id\$\" in config\n";
1562 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1563 }
1564 $self->{url} = command_oneline('config', '--get',
1565 "svn-remote.$repo_id.url") or
1566 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1567 $self->rebuild;
1568 $self;
1569}
1570sub refname {
1572 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1573 # It cannot end with a slash /, we'll throw up on this because
1575 # SVN can't have directories with a slash in their name, either:
1576 if ($refname =~ m{/$}) {
1577 die "ref: '$refname' ends with a trailing slash, this is ",
1578 "not permitted by git nor Subversion\n";
1579 }
1580 # It cannot have ASCII control character space, tilde ~, caret ^,
1582 # colon :, question-mark ?, asterisk *, space, or open bracket [
1583 # anywhere.
1584 #
1585 # Additionally, % must be escaped because it is used for escaping
1586 # and we want our escaped refname to be reversible
1587 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1588 # no slash-separated component can begin with a dot .
1590 # /.* becomes /%2E*
1591 $refname =~ s{/\.}{/%2E}g;
1592 # It cannot have two consecutive dots .. anywhere
1594 # .. becomes %2E%2E
1595 $refname =~ s{\.\.}{%2E%2E}g;
1596 return $refname;
1598}
1599sub desanitize_refname {
1601 my ($refname) = @_;
1602 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1603 return $refname;
1604}
1605sub svm_uuid {
1607 my ($self) = @_;
1608 return $self->{svm}->{uuid} if $self->svm;
1609 $self->ra;
1610 unless ($self->{svm}) {
1611 die "SVM UUID not cached, and reading remotely failed\n";
1612 }
1613 $self->{svm}->{uuid};
1614}
1615sub svm {
1617 my ($self) = @_;
1618 return $self->{svm} if $self->{svm};
1619 my $svm;
1620 # see if we have it in our config, first:
1621 eval {
1622 my $section = "svn-remote.$self->{repo_id}";
1623 $svm = {
1624 source => tmp_config('--get', "$section.svm-source"),
1625 uuid => tmp_config('--get', "$section.svm-uuid"),
1626 replace => tmp_config('--get', "$section.svm-replace"),
1627 }
1628 };
1629 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1630 $self->{svm} = $svm;
1631 }
1632 $self->{svm};
1633}
1634sub _set_svm_vars {
1636 my ($self, $ra) = @_;
1637 return $ra if $self->svm;
1638 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1640 "(svm:source, svm:uuid) ",
1641 "from the following URLs:\n" );
1642 sub read_svm_props {
1643 my ($self, $ra, $path, $r) = @_;
1644 my $props = ($ra->get_dir($path, $r))[2];
1645 my $src = $props->{'svm:source'};
1646 my $uuid = $props->{'svm:uuid'};
1647 return undef if (!$src || !$uuid);
1648 chomp($src, $uuid);
1650 $uuid =~ m{^[0-9a-f\-]{30,}$}
1652 or die "doesn't look right - svm:uuid is '$uuid'\n";
1653 # the '!' is used to mark the repos_root!/relative/path
1655 $src =~ s{/?!/?}{/};
1656 $src =~ s{/+$}{}; # no trailing slashes please
1657 # username is of no interest
1658 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1659 my $replace = $ra->{url};
1661 $replace .= "/$path" if length $path;
1662 my $section = "svn-remote.$self->{repo_id}";
1664 tmp_config("$section.svm-source", $src);
1665 tmp_config("$section.svm-replace", $replace);
1666 tmp_config("$section.svm-uuid", $uuid);
1667 $self->{svm} = {
1668 source => $src,
1669 uuid => $uuid,
1670 replace => $replace
1671 };
1672 }
1673 my $r = $ra->get_latest_revnum;
1675 my $path = $self->{path};
1676 my %tried;
1677 while (length $path) {
1678 unless ($tried{"$self->{url}/$path"}) {
1679 return $ra if $self->read_svm_props($ra, $path, $r);
1680 $tried{"$self->{url}/$path"} = 1;
1681 }
1682 $path =~ s#/?[^/]+$##;
1683 }
1684 die "Path: '$path' should be ''\n" if $path ne '';
1685 return $ra if $self->read_svm_props($ra, $path, $r);
1686 $tried{"$self->{url}/$path"} = 1;
1687 if ($ra->{repos_root} eq $self->{url}) {
1689 die @err, (map { " $_\n" } keys %tried), "\n";
1690 }
1691 # nope, make sure we're connected to the repository root:
1693 my $ok;
1694 my @tried_b;
1695 $path = $ra->{svn_path};
1696 $ra = Git::SVN::Ra->new($ra->{repos_root});
1697 while (length $path) {
1698 unless ($tried{"$ra->{url}/$path"}) {
1699 $ok = $self->read_svm_props($ra, $path, $r);
1700 last if $ok;
1701 $tried{"$ra->{url}/$path"} = 1;
1702 }
1703 $path =~ s#/?[^/]+$##;
1704 }
1705 die "Path: '$path' should be ''\n" if $path ne '';
1706 $ok ||= $self->read_svm_props($ra, $path, $r);
1707 $tried{"$ra->{url}/$path"} = 1;
1708 if (!$ok) {
1709 die @err, (map { " $_\n" } keys %tried), "\n";
1710 }
1711 Git::SVN::Ra->new($self->{url});
1712}
1713sub svnsync {
1715 my ($self) = @_;
1716 return $self->{svnsync} if $self->{svnsync};
1717 if ($self->no_metadata) {
1719 die "Can't have both 'noMetadata' and ",
1720 "'useSvnsyncProps' options set!\n";
1721 }
1722 if ($self->rewrite_root) {
1723 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1724 "options set!\n";
1725 }
1726 my $svnsync;
1728 # see if we have it in our config, first:
1729 eval {
1730 my $section = "svn-remote.$self->{repo_id}";
1731 $svnsync = {
1732 url => tmp_config('--get', "$section.svnsync-url"),
1733 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1734 }
1735 };
1736 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1737 return $self->{svnsync} = $svnsync;
1738 }
1739 my $err = "useSvnsyncProps set, but failed to read " .
1741 "svnsync property: svn:sync-from-";
1742 my $rp = $self->ra->rev_proplist(0);
1743 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1745 $url =~ m{^[a-z\+]+://} or
1746 die "doesn't look right - svn:sync-from-url is '$url'\n";
1747 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1749 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1750 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1751 my $section = "svn-remote.$self->{repo_id}";
1753 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1754 tmp_config('--add', "$section.svnsync-url", $url);
1755 return $self->{svnsync} = { url => $url, uuid => $uuid };
1756}
1757# this allows us to memoize our SVN::Ra UUID locally and avoid a
1759# remote lookup (useful for 'git svn log').
1760sub ra_uuid {
1761 my ($self) = @_;
1762 unless ($self->{ra_uuid}) {
1763 my $key = "svn-remote.$self->{repo_id}.uuid";
1764 my $uuid = eval { tmp_config('--get', $key) };
1765 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1766 $self->{ra_uuid} = $uuid;
1767 } else {
1768 die "ra_uuid called without URL\n" unless $self->{url};
1769 $self->{ra_uuid} = $self->ra->get_uuid;
1770 tmp_config('--add', $key, $self->{ra_uuid});
1771 }
1772 }
1773 $self->{ra_uuid};
1774}
1775sub ra {
1777 my ($self) = shift;
1778 my $ra = Git::SVN::Ra->new($self->{url});
1779 if ($self->use_svm_props && !$self->{svm}) {
1780 if ($self->no_metadata) {
1781 die "Can't have both 'noMetadata' and ",
1782 "'useSvmProps' options set!\n";
1783 } elsif ($self->use_svnsync_props) {
1784 die "Can't have both 'useSvnsyncProps' and ",
1785 "'useSvmProps' options set!\n";
1786 }
1787 $ra = $self->_set_svm_vars($ra);
1788 $self->{-want_revprops} = 1;
1789 }
1790 $ra;
1791}
1792sub rel_path {
1794 my ($self) = @_;
1795 my $repos_root = $self->ra->{repos_root};
1796 return $self->{path} if ($self->{url} eq $repos_root);
1797 my $url = $self->{url} .
1798 (length $self->{path} ? "/$self->{path}" : $self->{path});
1799 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1800 $url;
1801}
1802# prop_walk(PATH, REV, SUB)
1804# -------------------------
1805# Recursively traverse PATH at revision REV and invoke SUB for each
1806# directory that contains a SVN property. SUB will be invoked as
1807# follows: &SUB(gs, path, props); where `gs' is this instance of
1808# Git::SVN, `path' the path to the directory where the properties
1809# `props' were found. The `path' will be relative to point of checkout,
1810# that is, if url://repo/trunk is the current Git branch, and that
1811# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1812# as `path' (note the trailing `/').
1813sub prop_walk {
1814 my ($self, $path, $rev, $sub) = @_;
1815 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1817 $path =~ s#^/*#/#g;
1818 my $p = $path;
1819 # Strip the irrelevant part of the path.
1820 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1821 # Ensure the path is terminated by a `/'.
1822 $p =~ s#/*$#/#;
1823 # The properties contain all the internal SVN stuff nobody
1825 # (usually) cares about.
1826 my $interesting_props = 0;
1827 foreach (keys %{$props}) {
1828 # If it doesn't start with `svn:', it must be a
1829 # user-defined property.
1830 ++$interesting_props and next if $_ !~ /^svn:/;
1831 # FIXME: Fragile, if SVN adds new public properties,
1832 # this needs to be updated.
1833 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1834 |eol-style|mime-type
1835 |externals|needs-lock)$/x;
1836 }
1837 &$sub($self, $p, $props) if $interesting_props;
1838 foreach (sort keys %$dirent) {
1840 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1841 $self->prop_walk($path . '/' . $_, $rev, $sub);
1842 }
1843}
1844sub last_rev { ($_[0]->last_rev_commit)[0] }
1846sub last_commit { ($_[0]->last_rev_commit)[1] }
1847# returns the newest SVN revision number and newest commit SHA1
1849sub last_rev_commit {
1850 my ($self) = @_;
1851 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1852 return ($self->{last_rev}, $self->{last_commit});
1853 }
1854 my $c = ::verify_ref($self->refname.'^0');
1855 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1856 my $rev = (::cmt_metadata($c))[1];
1857 if (defined $rev) {
1858 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1859 return ($rev, $c);
1860 }
1861 }
1862 my $db_path = $self->db_path;
1863 unless (-e $db_path) {
1864 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1865 return (undef, undef);
1866 }
1867 my $offset = -41; # from tail
1868 my $rl;
1869 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1870 sysseek($fh, $offset, 2); # don't care for errors
1871 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1872 chomp $rl;
1873 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1874 $offset -= 41;
1875 sysseek($fh, $offset, 2); # don't care for errors
1876 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1877 chomp $rl;
1878 }
1879 if ($c && $c ne $rl) {
1880 die "$db_path and ", $self->refname,
1881 " inconsistent!:\n$c != $rl\n";
1882 }
1883 my $rev = sysseek($fh, 0, 1) or croak $!;
1884 $rev = ($rev - 41) / 41;
1885 close $fh or croak $!;
1886 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1887 return ($rev, $c);
1888}
1889sub get_fetch_range {
1891 my ($self, $min, $max) = @_;
1892 $max ||= $self->ra->get_latest_revnum;
1893 $min ||= $self->rev_db_max;
1894 (++$min, $max);
1895}
1896sub tmp_config {
1898 my (@args) = @_;
1899 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1900 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1901 if (! -f $config && -f $old_def_config) {
1902 rename $old_def_config, $config or
1903 die "Failed rename $old_def_config => $config: $!\n";
1904 }
1905 my $old_config = $ENV{GIT_CONFIG};
1906 $ENV{GIT_CONFIG} = $config;
1907 $@ = undef;
1908 my @ret = eval {
1909 unless (-f $config) {
1910 mkfile($config);
1911 open my $fh, '>', $config or
1912 die "Can't open $config: $!\n";
1913 print $fh "; This file is used internally by ",
1914 "git-svn\n" or die
1915 "Couldn't write to $config: $!\n";
1916 print $fh "; You should not have to edit it\n" or
1917 die "Couldn't write to $config: $!\n";
1918 close $fh or die "Couldn't close $config: $!\n";
1919 }
1920 command('config', @args);
1921 };
1922 my $err = $@;
1923 if (defined $old_config) {
1924 $ENV{GIT_CONFIG} = $old_config;
1925 } else {
1926 delete $ENV{GIT_CONFIG};
1927 }
1928 die $err if $err;
1929 wantarray ? @ret : $ret[0];
1930}
1931sub tmp_index_do {
1933 my ($self, $sub) = @_;
1934 my $old_index = $ENV{GIT_INDEX_FILE};
1935 $ENV{GIT_INDEX_FILE} = $self->{index};
1936 $@ = undef;
1937 my @ret = eval {
1938 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1939 mkpath([$dir]) unless -d $dir;
1940 &$sub;
1941 };
1942 my $err = $@;
1943 if (defined $old_index) {
1944 $ENV{GIT_INDEX_FILE} = $old_index;
1945 } else {
1946 delete $ENV{GIT_INDEX_FILE};
1947 }
1948 die $err if $err;
1949 wantarray ? @ret : $ret[0];
1950}
1951sub assert_index_clean {
1953 my ($self, $treeish) = @_;
1954 $self->tmp_index_do(sub {
1956 command_noisy('read-tree', $treeish) unless -e $self->{index};
1957 my $x = command_oneline('write-tree');
1958 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1959 /^tree ($::sha1)/mo);
1960 return if $y eq $x;
1961 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1963 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1964 command_noisy('read-tree', $treeish);
1965 $x = command_oneline('write-tree');
1966 if ($y ne $x) {
1967 ::fatal "trees ($treeish) $y != $x\n",
1968 "Something is seriously wrong...";
1969 }
1970 });
1971}
1972sub get_commit_parents {
1974 my ($self, $log_entry) = @_;
1975 my (%seen, @ret, @tmp);
1976 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1977 if (my $ip = $self->{inject_parents}) {
1978 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1979 push @tmp, $commit;
1980 }
1981 }
1982 if (my $cur = ::verify_ref($self->refname.'^0')) {
1983 push @tmp, $cur;
1984 }
1985 if (my $ipd = $self->{inject_parents_dcommit}) {
1986 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
1987 push @tmp, @$commit;
1988 }
1989 }
1990 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1991 while (my $p = shift @tmp) {
1992 next if $seen{$p};
1993 $seen{$p} = 1;
1994 push @ret, $p;
1995 # MAXPARENT is defined to 16 in commit-tree.c:
1996 last if @ret >= 16;
1997 }
1998 if (@tmp) {
1999 die "r$log_entry->{revision}: No room for parents:\n\t",
2000 join("\n\t", @tmp), "\n";
2001 }
2002 @ret;
2003}
2004sub rewrite_root {
2006 my ($self) = @_;
2007 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2008 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2009 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2010 if ($rwr) {
2011 $rwr =~ s#/+$##;
2012 if ($rwr !~ m#^[a-z\+]+://#) {
2013 die "$rwr is not a valid URL (key: $k)\n";
2014 }
2015 }
2016 $self->{-rewrite_root} = $rwr;
2017}
2018sub metadata_url {
2020 my ($self) = @_;
2021 ($self->rewrite_root || $self->{url}) .
2022 (length $self->{path} ? '/' . $self->{path} : '');
2023}
2024sub full_url {
2026 my ($self) = @_;
2027 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2028}
2029sub do_git_commit {
2031 my ($self, $log_entry) = @_;
2032 my $lr = $self->last_rev;
2033 if (defined $lr && $lr >= $log_entry->{revision}) {
2034 die "Last fetched revision of ", $self->refname,
2035 " was r$lr, but we are about to fetch: ",
2036 "r$log_entry->{revision}!\n";
2037 }
2038 if (my $c = $self->rev_db_get($log_entry->{revision})) {
2039 croak "$log_entry->{revision} = $c already exists! ",
2040 "Why are we refetching it?\n";
2041 }
2042 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
2043 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
2044 $log_entry->{email};
2045 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2046 my $tree = $log_entry->{tree};
2048 if (!defined $tree) {
2049 $tree = $self->tmp_index_do(sub {
2050 command_oneline('write-tree') });
2051 }
2052 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2053 my @exec = ('git-commit-tree', $tree);
2055 foreach ($self->get_commit_parents($log_entry)) {
2056 push @exec, '-p', $_;
2057 }
2058 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2059 or croak $!;
2060 print $msg_fh $log_entry->{log} or croak $!;
2061 unless ($self->no_metadata) {
2062 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2063 or croak $!;
2064 }
2065 $msg_fh->flush == 0 or croak $!;
2066 close $msg_fh or croak $!;
2067 chomp(my $commit = do { local $/; <$out_fh> });
2068 close $out_fh or croak $!;
2069 waitpid $pid, 0;
2070 croak $? if $?;
2071 if ($commit !~ /^$::sha1$/o) {
2072 die "Failed to commit, invalid sha1: $commit\n";
2073 }
2074 $self->rev_db_set($log_entry->{revision}, $commit, 1);
2076 $self->{last_rev} = $log_entry->{revision};
2078 $self->{last_commit} = $commit;
2079 print "r$log_entry->{revision}";
2080 if (defined $log_entry->{svm_revision}) {
2081 print " (\@$log_entry->{svm_revision})";
2082 $self->rev_db_set($log_entry->{svm_revision}, $commit,
2083 0, $self->svm_uuid);
2084 }
2085 print " = $commit ($self->{ref_id})\n";
2086 if (defined $_repack && (--$_repack_nr == 0)) {
2087 $_repack_nr = $_repack;
2088 # repack doesn't use any arguments with spaces in them, does it?
2089 print "Running git repack $_repack_flags ...\n";
2090 command_noisy('repack', split(/\s+/, $_repack_flags));
2091 print "Done repacking\n";
2092 }
2093 return $commit;
2094}
2095sub match_paths {
2097 my ($self, $paths, $r) = @_;
2098 return 1 if $self->{path} eq '';
2099 if (my $path = $paths->{"/$self->{path}"}) {
2100 return ($path->{action} eq 'D') ? 0 : 1;
2101 }
2102 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2103 if (grep /$self->{path_regex}/, keys %$paths) {
2104 return 1;
2105 }
2106 my $c = '';
2107 foreach (split m#/#, $self->{path}) {
2108 $c .= "/$_";
2109 next unless ($paths->{$c} &&
2110 ($paths->{$c}->{action} =~ /^[AR]$/));
2111 if ($self->ra->check_path($self->{path}, $r) ==
2112 $SVN::Node::dir) {
2113 return 1;
2114 }
2115 }
2116 return 0;
2117}
2118sub find_parent_branch {
2120 my ($self, $paths, $rev) = @_;
2121 return undef unless $self->follow_parent;
2122 unless (defined $paths) {
2123 my $err_handler = $SVN::Error::handler;
2124 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2125 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2126 $paths =
2127 Git::SVN::Ra::dup_changed_paths($_[0]) });
2128 $SVN::Error::handler = $err_handler;
2129 }
2130 return undef unless defined $paths;
2131 # look for a parent from another branch:
2133 my @b_path_components = split m#/#, $self->rel_path;
2134 my @a_path_components;
2135 my $i;
2136 while (@b_path_components) {
2137 $i = $paths->{'/'.join('/', @b_path_components)};
2138 last if $i && defined $i->{copyfrom_path};
2139 unshift(@a_path_components, pop(@b_path_components));
2140 }
2141 return undef unless defined $i && defined $i->{copyfrom_path};
2142 my $branch_from = $i->{copyfrom_path};
2143 if (@a_path_components) {
2144 print STDERR "branch_from: $branch_from => ";
2145 $branch_from .= '/'.join('/', @a_path_components);
2146 print STDERR $branch_from, "\n";
2147 }
2148 my $r = $i->{copyfrom_rev};
2149 my $repos_root = $self->ra->{repos_root};
2150 my $url = $self->ra->{url};
2151 my $new_url = $repos_root . $branch_from;
2152 print STDERR "Found possible branch point: ",
2153 "$new_url => ", $self->full_url, ", $r\n";
2154 $branch_from =~ s#^/##;
2155 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2156 unless ($gs) {
2157 my $ref_id = $self->{ref_id};
2158 $ref_id =~ s/\@\d+$//;
2159 $ref_id .= "\@$r";
2160 # just grow a tail if we're not unique enough :x
2161 $ref_id .= '-' while find_ref($ref_id);
2162 print STDERR "Initializing parent: $ref_id\n";
2163 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
2164 }
2165 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2166 if (!defined $r0 || !defined $parent) {
2167 my ($base, $head) = parse_revision_argument(0, $r);
2168 if ($base <= $r) {
2169 $gs->fetch($base, $r);
2170 }
2171 ($r0, $parent) = $gs->last_rev_commit;
2172 }
2173 if (defined $r0 && defined $parent) {
2174 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2175 my $ed;
2176 if ($self->ra->can_do_switch) {
2177 $self->assert_index_clean($parent);
2178 print STDERR "Following parent with do_switch\n";
2179 # do_switch works with svn/trunk >= r22312, but that
2180 # is not included with SVN 1.4.3 (the latest version
2181 # at the moment), so we can't rely on it
2182 $self->{last_commit} = $parent;
2183 $ed = SVN::Git::Fetcher->new($self);
2184 $gs->ra->gs_do_switch($r0, $rev, $gs,
2185 $self->full_url, $ed)
2186 or die "SVN connection failed somewhere...\n";
2187 } elsif ($self->ra->trees_match($new_url, $r0,
2188 $self->full_url, $rev)) {
2189 print STDERR "Trees match:\n",
2190 " $new_url\@$r0\n",
2191 " ${\$self->full_url}\@$rev\n",
2192 "Following parent with no changes\n";
2193 $self->tmp_index_do(sub {
2194 command_noisy('read-tree', $parent);
2195 });
2196 $self->{last_commit} = $parent;
2197 } else {
2198 print STDERR "Following parent with do_update\n";
2199 $ed = SVN::Git::Fetcher->new($self);
2200 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2201 or die "SVN connection failed somewhere...\n";
2202 }
2203 print STDERR "Successfully followed parent\n";
2204 return $self->make_log_entry($rev, [$parent], $ed);
2205 }
2206 return undef;
2207}
2208sub do_fetch {
2210 my ($self, $paths, $rev) = @_;
2211 my $ed;
2212 my ($last_rev, @parents);
2213 if (my $lc = $self->last_commit) {
2214 # we can have a branch that was deleted, then re-added
2215 # under the same name but copied from another path, in
2216 # which case we'll have multiple parents (we don't
2217 # want to break the original ref, nor lose copypath info):
2218 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2219 push @{$log_entry->{parents}}, $lc;
2220 return $log_entry;
2221 }
2222 $ed = SVN::Git::Fetcher->new($self);
2223 $last_rev = $self->{last_rev};
2224 $ed->{c} = $lc;
2225 @parents = ($lc);
2226 } else {
2227 $last_rev = $rev;
2228 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2229 return $log_entry;
2230 }
2231 $ed = SVN::Git::Fetcher->new($self);
2232 }
2233 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2234 die "SVN connection failed somewhere...\n";
2235 }
2236 $self->make_log_entry($rev, \@parents, $ed);
2237}
2238sub get_untracked {
2240 my ($self, $ed) = @_;
2241 my @out;
2242 my $h = $ed->{empty};
2243 foreach (sort keys %$h) {
2244 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2245 push @out, " $act: " . uri_encode($_);
2246 warn "W: $act: $_\n";
2247 }
2248 foreach my $t (qw/dir_prop file_prop/) {
2249 $h = $ed->{$t} or next;
2250 foreach my $path (sort keys %$h) {
2251 my $ppath = $path eq '' ? '.' : $path;
2252 foreach my $prop (sort keys %{$h->{$path}}) {
2253 next if $SKIP_PROP{$prop};
2254 my $v = $h->{$path}->{$prop};
2255 my $t_ppath_prop = "$t: " .
2256 uri_encode($ppath) . ' ' .
2257 uri_encode($prop);
2258 if (defined $v) {
2259 push @out, " +$t_ppath_prop " .
2260 uri_encode($v);
2261 } else {
2262 push @out, " -$t_ppath_prop";
2263 }
2264 }
2265 }
2266 }
2267 foreach my $t (qw/absent_file absent_directory/) {
2268 $h = $ed->{$t} or next;
2269 foreach my $parent (sort keys %$h) {
2270 foreach my $path (sort @{$h->{$parent}}) {
2271 push @out, " $t: " .
2272 uri_encode("$parent/$path");
2273 warn "W: $t: $parent/$path ",
2274 "Insufficient permissions?\n";
2275 }
2276 }
2277 }
2278 \@out;
2279}
2280sub parse_svn_date {
2282 my $date = shift || return '+0000 1970-01-01 00:00:00';
2283 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2284 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2285 croak "Unable to parse date: $date\n";
2286 "+0000 $Y-$m-$d $H:$M:$S";
2287}
2288sub check_author {
2290 my ($author) = @_;
2291 if (!defined $author || length $author == 0) {
2292 $author = '(no author)';
2293 }
2294 if (defined $::_authors && ! defined $::users{$author}) {
2295 die "Author: $author not defined in $::_authors file\n";
2296 }
2297 $author;
2298}
2299sub make_log_entry {
2301 my ($self, $rev, $parents, $ed) = @_;
2302 my $untracked = $self->get_untracked($ed);
2303 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2305 print $un "r$rev\n" or croak $!;
2306 print $un $_, "\n" foreach @$untracked;
2307 my %log_entry = ( parents => $parents || [], revision => $rev,
2308 log => '');
2309 my $headrev;
2311 my $logged = delete $self->{logged_rev_props};
2312 if (!$logged || $self->{-want_revprops}) {
2313 my $rp = $self->ra->rev_proplist($rev);
2314 foreach (sort keys %$rp) {
2315 my $v = $rp->{$_};
2316 if (/^svn:(author|date|log)$/) {
2317 $log_entry{$1} = $v;
2318 } elsif ($_ eq 'svm:headrev') {
2319 $headrev = $v;
2320 } else {
2321 print $un " rev_prop: ", uri_encode($_), ' ',
2322 uri_encode($v), "\n";
2323 }
2324 }
2325 } else {
2326 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2327 }
2328 close $un or croak $!;
2329 $log_entry{date} = parse_svn_date($log_entry{date});
2331 $log_entry{log} .= "\n";
2332 my $author = $log_entry{author} = check_author($log_entry{author});
2333 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2334 : ($author, undef);
2335 if (defined $headrev && $self->use_svm_props) {
2336 if ($self->rewrite_root) {
2337 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2338 "options set!\n";
2339 }
2340 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2341 # we don't want "SVM: initializing mirror for junk" ...
2342 return undef if $r == 0;
2343 my $svm = $self->svm;
2344 if ($uuid ne $svm->{uuid}) {
2345 die "UUID mismatch on SVM path:\n",
2346 "expected: $svm->{uuid}\n",
2347 " got: $uuid\n";
2348 }
2349 my $full_url = $self->full_url;
2350 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2351 die "Failed to replace '$svm->{replace}' with ",
2352 "'$svm->{source}' in $full_url\n";
2353 # throw away username for storing in records
2354 remove_username($full_url);
2355 $log_entry{metadata} = "$full_url\@$r $uuid";
2356 $log_entry{svm_revision} = $r;
2357 $email ||= "$author\@$uuid"
2358 } elsif ($self->use_svnsync_props) {
2359 my $full_url = $self->svnsync->{url};
2360 $full_url .= "/$self->{path}" if length $self->{path};
2361 remove_username($full_url);
2362 my $uuid = $self->svnsync->{uuid};
2363 $log_entry{metadata} = "$full_url\@$rev $uuid";
2364 $email ||= "$author\@$uuid"
2365 } else {
2366 my $url = $self->metadata_url;
2367 remove_username($url);
2368 $log_entry{metadata} = "$url\@$rev " .
2369 $self->ra->get_uuid;
2370 $email ||= "$author\@" . $self->ra->get_uuid;
2371 }
2372 $log_entry{name} = $name;
2373 $log_entry{email} = $email;
2374 \%log_entry;
2375}
2376sub fetch {
2378 my ($self, $min_rev, $max_rev, @parents) = @_;
2379 my ($last_rev, $last_commit) = $self->last_rev_commit;
2380 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2381 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2382}
2383sub set_tree_cb {
2385 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2386 $self->{inject_parents} = { $rev => $tree };
2387 $self->fetch(undef, undef);
2388}
2389sub set_tree {
2391 my ($self, $tree) = (shift, shift);
2392 my $log_entry = ::get_commit_entry($tree);
2393 unless ($self->{last_rev}) {
2394 fatal("Must have an existing revision to commit");
2395 }
2396 my %ed_opts = ( r => $self->{last_rev},
2397 log => $log_entry->{log},
2398 ra => $self->ra,
2399 tree_a => $self->{last_commit},
2400 tree_b => $tree,
2401 editor_cb => sub {
2402 $self->set_tree_cb($log_entry, $tree, @_) },
2403 svn_path => $self->{path} );
2404 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2405 print "No changes\nr$self->{last_rev} = $tree\n";
2406 }
2407}
2408sub rebuild {
2410 my ($self) = @_;
2411 my $db_path = $self->db_path;
2412 return if (-e $db_path && ! -z $db_path);
2413 return unless ::verify_ref($self->refname.'^0');
2414 if (-f $self->{db_root}) {
2415 rename $self->{db_root}, $db_path or die
2416 "rename $self->{db_root} => $db_path failed: $!\n";
2417 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2418 symlink $base, $self->{db_root} or die
2419 "symlink $base => $self->{db_root} failed: $!\n";
2420 return;
2421 }
2422 print "Rebuilding $db_path ...\n";
2423 my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2424 my $latest;
2425 my $full_url = $self->full_url;
2426 remove_username($full_url);
2427 my $svn_uuid;
2428 my $c;
2429 while (<$log>) {
2430 if ( m{^commit ($::sha1)$} ) {
2431 $c = $1;
2432 next;
2433 }
2434 next unless s{^\s*(git-svn-id:)}{$1};
2435 my ($url, $rev, $uuid) = ::extract_metadata($_);
2436 remove_username($url);
2437 # ignore merges (from set-tree)
2439 next if (!defined $rev || !$uuid);
2440 # if we merged or otherwise started elsewhere, this is
2442 # how we break out of it
2443 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2444 ($full_url && $url && ($url ne $full_url))) {
2445 next;
2446 }
2447 $latest ||= $rev;
2448 $svn_uuid ||= $uuid;
2449 $self->rev_db_set($rev, $c);
2451 print "r$rev = $c\n";
2452 }
2453 command_close_pipe($log, $ctx);
2454 print "Done rebuilding $db_path\n";
2455}
2456# rev_db:
2458# Tie::File seems to be prone to offset errors if revisions get sparse,
2459# it's not that fast, either. Tie::File is also not in Perl 5.6. So
2460# one of my favorite modules is out :< Next up would be one of the DBM
2461# modules, but I'm not sure which is most portable... So I'll just
2462# go with something that's plain-text, but still capable of
2463# being randomly accessed. So here's my ultra-simple fixed-width
2464# database. All records are 40 characters + "\n", so it's easy to seek
2465# to a revision: (41 * rev) is the byte offset.
2466# A record of 40 0s denotes an empty revision.
2467# And yes, it's still pretty fast (faster than Tie::File).
2468# These files are disposable unless noMetadata or useSvmProps is set
2469sub _rev_db_set {
2471 my ($fh, $rev, $commit) = @_;
2472 my $offset = $rev * 41;
2473 # assume that append is the common case:
2474 seek $fh, 0, 2 or croak $!;
2475 my $pos = tell $fh;
2476 if ($pos < $offset) {
2477 for (1 .. (($offset - $pos) / 41)) {
2478 print $fh (('0' x 40),"\n") or croak $!;
2479 }
2480 }
2481 seek $fh, $offset, 0 or croak $!;
2482 print $fh $commit,"\n" or croak $!;
2483}
2484sub mkfile {
2486 my ($path) = @_;
2487 unless (-e $path) {
2488 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2489 mkpath([$dir]) unless -d $dir;
2490 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2491 close $fh or die "Couldn't close (create) $path: $!\n";
2492 }
2493}
2494sub rev_db_set {
2496 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2497 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2498 my $db = $self->db_path($uuid);
2499 my $db_lock = "$db.lock";
2500 my $sig;
2501 if ($update_ref) {
2502 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2503 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2504 }
2505 mkfile($db);
2506 $LOCKFILES{$db_lock} = 1;
2508 my $sync;
2509 # both of these options make our .rev_db file very, very important
2510 # and we can't afford to lose it because rebuild() won't work
2511 if ($self->use_svm_props || $self->no_metadata) {
2512 $sync = 1;
2513 copy($db, $db_lock) or die "rev_db_set(@_): ",
2514 "Failed to copy: ",
2515 "$db => $db_lock ($!)\n";
2516 } else {
2517 rename $db, $db_lock or die "rev_db_set(@_): ",
2518 "Failed to rename: ",
2519 "$db => $db_lock ($!)\n";
2520 }
2521 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2522 _rev_db_set($fh, $rev, $commit);
2523 if ($sync) {
2524 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2525 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2526 }
2527 close $fh or croak $!;
2528 if ($update_ref) {
2529 $_head = $self;
2530 command_noisy('update-ref', '-m', "r$rev",
2531 $self->refname, $commit);
2532 }
2533 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2534 "$db_lock => $db ($!)\n";
2535 delete $LOCKFILES{$db_lock};
2536 if ($update_ref) {
2537 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2538 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2539 kill $sig, $$ if defined $sig;
2540 }
2541}
2542sub rev_db_max {
2544 my ($self) = @_;
2545 $self->rebuild;
2546 my $db_path = $self->db_path;
2547 my @stat = stat $db_path or return 0;
2548 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2549 my $max = $stat[7] / 41;
2550 (($max > 0) ? $max - 1 : 0);
2551}
2552sub rev_db_get {
2554 my ($self, $rev, $uuid) = @_;
2555 my $ret;
2556 my $offset = $rev * 41;
2557 my $db_path = $self->db_path($uuid);
2558 return undef unless -e $db_path;
2559 open my $fh, '<', $db_path or croak $!;
2560 if (sysseek($fh, $offset, 0) == $offset) {
2561 my $read = sysread($fh, $ret, 40);
2562 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2563 }
2564 close $fh or croak $!;
2565 $ret;
2566}
2567# Finds the first svn revision that exists on (if $eq_ok is true) or
2569# before $rev for the current branch. It will not search any lower
2570# than $min_rev. Returns the git commit hash and svn revision number
2571# if found, else (undef, undef).
2572sub find_rev_before {
2573 my ($self, $rev, $eq_ok, $min_rev) = @_;
2574 --$rev unless $eq_ok;
2575 $min_rev ||= 1;
2576 while ($rev >= $min_rev) {
2577 if (my $c = $self->rev_db_get($rev)) {
2578 return ($rev, $c);
2579 }
2580 --$rev;
2581 }
2582 return (undef, undef);
2583}
2584# Finds the first svn revision that exists on (if $eq_ok is true) or
2586# after $rev for the current branch. It will not search any higher
2587# than $max_rev. Returns the git commit hash and svn revision number
2588# if found, else (undef, undef).
2589sub find_rev_after {
2590 my ($self, $rev, $eq_ok, $max_rev) = @_;
2591 ++$rev unless $eq_ok;
2592 $max_rev ||= $self->rev_db_max();
2593 while ($rev <= $max_rev) {
2594 if (my $c = $self->rev_db_get($rev)) {
2595 return ($rev, $c);
2596 }
2597 ++$rev;
2598 }
2599 return (undef, undef);
2600}
2601sub _new {
2603 my ($class, $repo_id, $ref_id, $path) = @_;
2604 unless (defined $repo_id && length $repo_id) {
2605 $repo_id = $Git::SVN::default_repo_id;
2606 }
2607 unless (defined $ref_id && length $ref_id) {
2608 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2609 }
2610 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2611 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2612 $_[3] = $path = '' unless (defined $path);
2613 mkpath(["$ENV{GIT_DIR}/svn"]);
2614 bless {
2615 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2616 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2617 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2618}
2619sub db_path {
2621 my ($self, $uuid) = @_;
2622 $uuid ||= $self->ra_uuid;
2623 "$self->{db_root}.$uuid";
2624}
2625sub uri_encode {
2627 my ($f) = @_;
2628 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2629 $f
2630}
2631sub remove_username {
2633 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2634}
2635package Git::SVN::Prompt;
2637use strict;
2638use warnings;
2639require SVN::Core;
2640use vars qw/$_no_auth_cache $_username/;
2641sub simple {
2643 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2644 $may_save = undef if $_no_auth_cache;
2645 $default_username = $_username if defined $_username;
2646 if (defined $default_username && length $default_username) {
2647 if (defined $realm && length $realm) {
2648 print STDERR "Authentication realm: $realm\n";
2649 STDERR->flush;
2650 }
2651 $cred->username($default_username);
2652 } else {
2653 username($cred, $realm, $may_save, $pool);
2654 }
2655 $cred->password(_read_password("Password for '" .
2656 $cred->username . "': ", $realm));
2657 $cred->may_save($may_save);
2658 $SVN::_Core::SVN_NO_ERROR;
2659}
2660sub ssl_server_trust {
2662 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2663 $may_save = undef if $_no_auth_cache;
2664 print STDERR "Error validating server certificate for '$realm':\n";
2665 {
2666 no warnings 'once';
2667 # All variables SVN::Auth::SSL::* are used only once,
2668 # so we're shutting up Perl warnings about this.
2669 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2670 print STDERR " - The certificate is not issued ",
2671 "by a trusted authority. Use the\n",
2672 " fingerprint to validate ",
2673 "the certificate manually!\n";
2674 }
2675 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2676 print STDERR " - The certificate hostname ",
2677 "does not match.\n";
2678 }
2679 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2680 print STDERR " - The certificate is not yet valid.\n";
2681 }
2682 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2683 print STDERR " - The certificate has expired.\n";
2684 }
2685 if ($failures & $SVN::Auth::SSL::OTHER) {
2686 print STDERR " - The certificate has ",
2687 "an unknown error.\n";
2688 }
2689 } # no warnings 'once'
2690 printf STDERR
2691 "Certificate information:\n".
2692 " - Hostname: %s\n".
2693 " - Valid: from %s until %s\n".
2694 " - Issuer: %s\n".
2695 " - Fingerprint: %s\n",
2696 map $cert_info->$_, qw(hostname valid_from valid_until
2697 issuer_dname fingerprint);
2698 my $choice;
2699prompt:
2700 print STDERR $may_save ?
2701 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2702 "(R)eject or accept (t)emporarily? ";
2703 STDERR->flush;
2704 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2705 if ($choice =~ /^t$/i) {
2706 $cred->may_save(undef);
2707 } elsif ($choice =~ /^r$/i) {
2708 return -1;
2709 } elsif ($may_save && $choice =~ /^p$/i) {
2710 $cred->may_save($may_save);
2711 } else {
2712 goto prompt;
2713 }
2714 $cred->accepted_failures($failures);
2715 $SVN::_Core::SVN_NO_ERROR;
2716}
2717sub ssl_client_cert {
2719 my ($cred, $realm, $may_save, $pool) = @_;
2720 $may_save = undef if $_no_auth_cache;
2721 print STDERR "Client certificate filename: ";
2722 STDERR->flush;
2723 chomp(my $filename = <STDIN>);
2724 $cred->cert_file($filename);
2725 $cred->may_save($may_save);
2726 $SVN::_Core::SVN_NO_ERROR;
2727}
2728sub ssl_client_cert_pw {
2730 my ($cred, $realm, $may_save, $pool) = @_;
2731 $may_save = undef if $_no_auth_cache;
2732 $cred->password(_read_password("Password: ", $realm));
2733 $cred->may_save($may_save);
2734 $SVN::_Core::SVN_NO_ERROR;
2735}
2736sub username {
2738 my ($cred, $realm, $may_save, $pool) = @_;
2739 $may_save = undef if $_no_auth_cache;
2740 if (defined $realm && length $realm) {
2741 print STDERR "Authentication realm: $realm\n";
2742 }
2743 my $username;
2744 if (defined $_username) {
2745 $username = $_username;
2746 } else {
2747 print STDERR "Username: ";
2748 STDERR->flush;
2749 chomp($username = <STDIN>);
2750 }
2751 $cred->username($username);
2752 $cred->may_save($may_save);
2753 $SVN::_Core::SVN_NO_ERROR;
2754}
2755sub _read_password {
2757 my ($prompt, $realm) = @_;
2758 print STDERR $prompt;
2759 STDERR->flush;
2760 require Term::ReadKey;
2761 Term::ReadKey::ReadMode('noecho');
2762 my $password = '';
2763 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2764 last if $key =~ /[\012\015]/; # \n\r
2765 $password .= $key;
2766 }
2767 Term::ReadKey::ReadMode('restore');
2768 print STDERR "\n";
2769 STDERR->flush;
2770 $password;
2771}
2772package SVN::Git::Fetcher;
2774use vars qw/@ISA/;
2775use strict;
2776use warnings;
2777use Carp qw/croak/;
2778use IO::File qw//;
2779# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2781sub new {
2782 my ($class, $git_svn) = @_;
2783 my $self = SVN::Delta::Editor->new;
2784 bless $self, $class;
2785 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2786 $self->{empty} = {};
2787 $self->{dir_prop} = {};
2788 $self->{file_prop} = {};
2789 $self->{absent_dir} = {};
2790 $self->{absent_file} = {};
2791 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2792 $self;
2793}
2794sub set_path_strip {
2796 my ($self, $path) = @_;
2797 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2798}
2799sub open_root {
2801 { path => '' };
2802}
2803sub open_directory {
2805 my ($self, $path, $pb, $rev) = @_;
2806 { path => $path };
2807}
2808sub git_path {
2810 my ($self, $path) = @_;
2811 if ($self->{path_strip}) {
2812 $path =~ s!$self->{path_strip}!! or
2813 die "Failed to strip path '$path' ($self->{path_strip})\n";
2814 }
2815 $path;
2816}
2817sub delete_entry {
2819 my ($self, $path, $rev, $pb) = @_;
2820 my $gpath = $self->git_path($path);
2822 return undef if ($gpath eq '');
2823 # remove entire directories.
2825 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2826 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2827 -r --name-only -z/,
2828 $self->{c}, '--', $gpath);
2829 local $/ = "\0";
2830 while (<$ls>) {
2831 chomp;
2832 $self->{gii}->remove($_);
2833 print "\tD\t$_\n" unless $::_q;
2834 }
2835 print "\tD\t$gpath/\n" unless $::_q;
2836 command_close_pipe($ls, $ctx);
2837 $self->{empty}->{$path} = 0
2838 } else {
2839 $self->{gii}->remove($gpath);
2840 print "\tD\t$gpath\n" unless $::_q;
2841 }
2842 undef;
2843}
2844sub open_file {
2846 my ($self, $path, $pb, $rev) = @_;
2847 my $gpath = $self->git_path($path);
2848 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2849 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2850 unless (defined $mode && defined $blob) {
2851 die "$path was not found in commit $self->{c} (r$rev)\n";
2852 }
2853 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2854 pool => SVN::Pool->new, action => 'M' };
2855}
2856sub add_file {
2858 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2859 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2860 delete $self->{empty}->{$dir};
2861 { path => $path, mode_a => 100644, mode_b => 100644,
2862 pool => SVN::Pool->new, action => 'A' };
2863}
2864sub add_directory {
2866 my ($self, $path, $cp_path, $cp_rev) = @_;
2867 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2868 delete $self->{empty}->{$dir};
2869 $self->{empty}->{$path} = 1;
2870 { path => $path };
2871}
2872sub change_dir_prop {
2874 my ($self, $db, $prop, $value) = @_;
2875 $self->{dir_prop}->{$db->{path}} ||= {};
2876 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2877 undef;
2878}
2879sub absent_directory {
2881 my ($self, $path, $pb) = @_;
2882 $self->{absent_dir}->{$pb->{path}} ||= [];
2883 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2884 undef;
2885}
2886sub absent_file {
2888 my ($self, $path, $pb) = @_;
2889 $self->{absent_file}->{$pb->{path}} ||= [];
2890 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2891 undef;
2892}
2893sub change_file_prop {
2895 my ($self, $fb, $prop, $value) = @_;
2896 if ($prop eq 'svn:executable') {
2897 if ($fb->{mode_b} != 120000) {
2898 $fb->{mode_b} = defined $value ? 100755 : 100644;
2899 }
2900 } elsif ($prop eq 'svn:special') {
2901 $fb->{mode_b} = defined $value ? 120000 : 100644;
2902 } else {
2903 $self->{file_prop}->{$fb->{path}} ||= {};
2904 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2905 }
2906 undef;
2907}
2908sub apply_textdelta {
2910 my ($self, $fb, $exp) = @_;
2911 my $fh = IO::File->new_tmpfile;
2912 $fh->autoflush(1);
2913 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2914 # (but $base does not,) so dup() it for reading in close_file
2915 open my $dup, '<&', $fh or croak $!;
2916 my $base = IO::File->new_tmpfile;
2917 $base->autoflush(1);
2918 if ($fb->{blob}) {
2919 defined (my $pid = fork) or croak $!;
2920 if (!$pid) {
2921 open STDOUT, '>&', $base or croak $!;
2922 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2923 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2924 }
2925 waitpid $pid, 0;
2926 croak $? if $?;
2927 if (defined $exp) {
2929 seek $base, 0, 0 or croak $!;
2930 my $got = Git::SVN::Util::md5sum($base);
2931 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2932 "expected: $exp\n",
2933 " got: $got\n" if ($got ne $exp);
2934 }
2935 }
2936 seek $base, 0, 0 or croak $!;
2937 $fb->{fh} = $dup;
2938 $fb->{base} = $base;
2939 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2940}
2941sub close_file {
2943 my ($self, $fb, $exp) = @_;
2944 my $hash;
2945 my $path = $self->git_path($fb->{path});
2946 if (my $fh = $fb->{fh}) {
2947 if (defined $exp) {
2948 seek($fh, 0, 0) or croak $!;
2949 my $got = Git::SVN::Util::md5sum($fh);
2950 if ($got ne $exp) {
2951 die "Checksum mismatch: $path\n",
2952 "expected: $exp\n got: $got\n";
2953 }
2954 }
2955 sysseek($fh, 0, 0) or croak $!;
2956 if ($fb->{mode_b} == 120000) {
2957 sysread($fh, my $buf, 5) == 5 or croak $!;
2958 $buf eq 'link ' or die "$path has mode 120000",
2959 "but is not a link\n";
2960 }
2961 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2962 if (!$pid) {
2963 open STDIN, '<&', $fh or croak $!;
2964 exec qw/git-hash-object -w --stdin/ or croak $!;
2965 }
2966 chomp($hash = do { local $/; <$out> });
2967 close $out or croak $!;
2968 close $fh or croak $!;
2969 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2970 close $fb->{base} or croak $!;
2971 } else {
2972 $hash = $fb->{blob} or die "no blob information\n";
2973 }
2974 $fb->{pool}->clear;
2975 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2976 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2977 undef;
2978}
2979sub abort_edit {
2981 my $self = shift;
2982 $self->{nr} = $self->{gii}->{nr};
2983 delete $self->{gii};
2984 $self->SUPER::abort_edit(@_);
2985}
2986sub close_edit {
2988 my $self = shift;
2989 $self->{git_commit_ok} = 1;
2990 $self->{nr} = $self->{gii}->{nr};
2991 delete $self->{gii};
2992 $self->SUPER::close_edit(@_);
2993}
2994package SVN::Git::Editor;
2996use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2997use strict;
2998use warnings;
2999use Carp qw/croak/;
3000use IO::File;
3001sub new {
3003 my ($class, $opts) = @_;
3004 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3005 die "$_ required!\n" unless (defined $opts->{$_});
3006 }
3007 my $pool = SVN::Pool->new;
3009 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3010 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3011 $opts->{r}, $mods);
3012 # $opts->{ra} functions should not be used after this:
3014 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3015 $opts->{editor_cb}, $pool);
3016 my $self = SVN::Delta::Editor->new(@ce, $pool);
3017 bless $self, $class;
3018 foreach (qw/svn_path r tree_a tree_b/) {
3019 $self->{$_} = $opts->{$_};
3020 }
3021 $self->{url} = $opts->{ra}->{url};
3022 $self->{mods} = $mods;
3023 $self->{types} = $types;
3024 $self->{pool} = $pool;
3025 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3026 $self->{rm} = { };
3027 $self->{path_prefix} = length $self->{svn_path} ?
3028 "$self->{svn_path}/" : '';
3029 return $self;
3030}
3031sub generate_diff {
3033 my ($tree_a, $tree_b) = @_;
3034 my @diff_tree = qw(diff-tree -z -r);
3035 if ($_cp_similarity) {
3036 push @diff_tree, "-C$_cp_similarity";
3037 } else {
3038 push @diff_tree, '-C';
3039 }
3040 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3041 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3042 push @diff_tree, $tree_a, $tree_b;
3043 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3044 local $/ = "\0";
3045 my $state = 'meta';
3046 my @mods;
3047 while (<$diff_fh>) {
3048 chomp $_; # this gets rid of the trailing "\0"
3049 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3050 $::sha1\s($::sha1)\s
3051 ([MTCRAD])\d*$/xo) {
3052 push @mods, { mode_a => $1, mode_b => $2,
3053 sha1_b => $3, chg => $4 };
3054 if ($4 =~ /^(?:C|R)$/) {
3055 $state = 'file_a';
3056 } else {
3057 $state = 'file_b';
3058 }
3059 } elsif ($state eq 'file_a') {
3060 my $x = $mods[$#mods] or croak "Empty array\n";
3061 if ($x->{chg} !~ /^(?:C|R)$/) {
3062 croak "Error parsing $_, $x->{chg}\n";
3063 }
3064 $x->{file_a} = $_;
3065 $state = 'file_b';
3066 } elsif ($state eq 'file_b') {
3067 my $x = $mods[$#mods] or croak "Empty array\n";
3068 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3069 croak "Error parsing $_, $x->{chg}\n";
3070 }
3071 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3072 croak "Error parsing $_, $x->{chg}\n";
3073 }
3074 $x->{file_b} = $_;
3075 $state = 'meta';
3076 } else {
3077 croak "Error parsing $_\n";
3078 }
3079 }
3080 command_close_pipe($diff_fh, $ctx);
3081 \@mods;
3082}
3083sub check_diff_paths {
3085 my ($ra, $pfx, $rev, $mods) = @_;
3086 my %types;
3087 $pfx .= '/' if length $pfx;
3088 sub type_diff_paths {
3090 my ($ra, $types, $path, $rev) = @_;
3091 my @p = split m#/+#, $path;
3092 my $c = shift @p;
3093 unless (defined $types->{$c}) {
3094 $types->{$c} = $ra->check_path($c, $rev);
3095 }
3096 while (@p) {
3097 $c .= '/' . shift @p;
3098 next if defined $types->{$c};
3099 $types->{$c} = $ra->check_path($c, $rev);
3100 }
3101 }
3102 foreach my $m (@$mods) {
3104 foreach my $f (qw/file_a file_b/) {
3105 next unless defined $m->{$f};
3106 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3107 if (length $pfx.$dir && ! defined $types{$dir}) {
3108 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3109 }
3110 }
3111 }
3112 \%types;
3113}
3114sub split_path {
3116 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3117}
3118sub repo_path {
3120 my ($self, $path) = @_;
3121 $self->{path_prefix}.(defined $path ? $path : '');
3122}
3123sub url_path {
3125 my ($self, $path) = @_;
3126 if ($self->{url} =~ m#^https?://#) {
3127 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3128 }
3129 $self->{url} . '/' . $self->repo_path($path);
3130}
3131sub rmdirs {
3133 my ($self) = @_;
3134 my $rm = $self->{rm};
3135 delete $rm->{''}; # we never delete the url we're tracking
3136 return unless %$rm;
3137 foreach (keys %$rm) {
3139 my @d = split m#/#, $_;
3140 my $c = shift @d;
3141 $rm->{$c} = 1;
3142 while (@d) {
3143 $c .= '/' . shift @d;
3144 $rm->{$c} = 1;
3145 }
3146 }
3147 delete $rm->{$self->{svn_path}};
3148 delete $rm->{''}; # we never delete the url we're tracking
3149 return unless %$rm;
3150 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3152 $self->{tree_b});
3153 local $/ = "\0";
3154 while (<$fh>) {
3155 chomp;
3156 my @dn = split m#/#, $_;
3157 while (pop @dn) {
3158 delete $rm->{join '/', @dn};
3159 }
3160 unless (%$rm) {
3161 close $fh;
3162 return;
3163 }
3164 }
3165 command_close_pipe($fh, $ctx);
3166 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3168 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3169 $self->close_directory($bat->{$d}, $p);
3170 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3171 print "\tD+\t$d/\n" unless $::_q;
3172 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3173 delete $bat->{$d};
3174 }
3175}
3176sub open_or_add_dir {
3178 my ($self, $full_path, $baton) = @_;
3179 my $t = $self->{types}->{$full_path};
3180 if (!defined $t) {
3181 die "$full_path not known in r$self->{r} or we have a bug!\n";
3182 }
3183 {
3184 no warnings 'once';
3185 # SVN::Node::none and SVN::Node::file are used only once,
3186 # so we're shutting up Perl's warnings about them.
3187 if ($t == $SVN::Node::none) {
3188 return $self->add_directory($full_path, $baton,
3189 undef, -1, $self->{pool});
3190 } elsif ($t == $SVN::Node::dir) {
3191 return $self->open_directory($full_path, $baton,
3192 $self->{r}, $self->{pool});
3193 } # no warnings 'once'
3194 print STDERR "$full_path already exists in repository at ",
3195 "r$self->{r} and it is not a directory (",
3196 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3197 } # no warnings 'once'
3198 exit 1;
3199}
3200sub ensure_path {
3202 my ($self, $path) = @_;
3203 my $bat = $self->{bat};
3204 my $repo_path = $self->repo_path($path);
3205 return $bat->{''} unless (length $repo_path);
3206 my @p = split m#/+#, $repo_path;
3207 my $c = shift @p;
3208 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3209 while (@p) {
3210 my $c0 = $c;
3211 $c .= '/' . shift @p;
3212 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3213 }
3214 return $bat->{$c};
3215}
3216sub A {
3218 my ($self, $m) = @_;
3219 my ($dir, $file) = split_path($m->{file_b});
3220 my $pbat = $self->ensure_path($dir);
3221 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3222 undef, -1);
3223 print "\tA\t$m->{file_b}\n" unless $::_q;
3224 $self->chg_file($fbat, $m);
3225 $self->close_file($fbat,undef,$self->{pool});
3226}
3227sub C {
3229 my ($self, $m) = @_;
3230 my ($dir, $file) = split_path($m->{file_b});
3231 my $pbat = $self->ensure_path($dir);
3232 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3233 $self->url_path($m->{file_a}), $self->{r});
3234 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3235 $self->chg_file($fbat, $m);
3236 $self->close_file($fbat,undef,$self->{pool});
3237}
3238sub delete_entry {
3240 my ($self, $path, $pbat) = @_;
3241 my $rpath = $self->repo_path($path);
3242 my ($dir, $file) = split_path($rpath);
3243 $self->{rm}->{$dir} = 1;
3244 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3245}
3246sub R {
3248 my ($self, $m) = @_;
3249 my ($dir, $file) = split_path($m->{file_b});
3250 my $pbat = $self->ensure_path($dir);
3251 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3252 $self->url_path($m->{file_a}), $self->{r});
3253 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3254 $self->chg_file($fbat, $m);
3255 $self->close_file($fbat,undef,$self->{pool});
3256 ($dir, $file) = split_path($m->{file_a});
3258 $pbat = $self->ensure_path($dir);
3259 $self->delete_entry($m->{file_a}, $pbat);
3260}
3261sub M {
3263 my ($self, $m) = @_;
3264 my ($dir, $file) = split_path($m->{file_b});
3265 my $pbat = $self->ensure_path($dir);
3266 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3267 $pbat,$self->{r},$self->{pool});
3268 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3269 $self->chg_file($fbat, $m);
3270 $self->close_file($fbat,undef,$self->{pool});
3271}
3272sub T { shift->M(@_) }
3274sub change_file_prop {
3276 my ($self, $fbat, $pname, $pval) = @_;
3277 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3278}
3279sub chg_file {
3281 my ($self, $fbat, $m) = @_;
3282 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3283 $self->change_file_prop($fbat,'svn:executable','*');
3284 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3285 $self->change_file_prop($fbat,'svn:executable',undef);
3286 }
3287 my $fh = IO::File->new_tmpfile or croak $!;
3288 if ($m->{mode_b} =~ /^120/) {
3289 print $fh 'link ' or croak $!;
3290 $self->change_file_prop($fbat,'svn:special','*');
3291 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3292 $self->change_file_prop($fbat,'svn:special',undef);
3293 }
3294 defined(my $pid = fork) or croak $!;
3295 if (!$pid) {
3296 open STDOUT, '>&', $fh or croak $!;
3297 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3298 }
3299 waitpid $pid, 0;
3300 croak $? if $?;
3301 $fh->flush == 0 or croak $!;
3302 seek $fh, 0, 0 or croak $!;
3303 my $exp = Git::SVN::Util::md5sum($fh);
3305 seek $fh, 0, 0 or croak $!;
3306 my $pool = SVN::Pool->new;
3308 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3309 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3310 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3311 $pool->clear;
3312 close $fh or croak $!;
3314}
3315sub D {
3317 my ($self, $m) = @_;
3318 my ($dir, $file) = split_path($m->{file_b});
3319 my $pbat = $self->ensure_path($dir);
3320 print "\tD\t$m->{file_b}\n" unless $::_q;
3321 $self->delete_entry($m->{file_b}, $pbat);
3322}
3323sub close_edit {
3325 my ($self) = @_;
3326 my ($p,$bat) = ($self->{pool}, $self->{bat});
3327 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3328 next if $_ eq '';
3329 $self->close_directory($bat->{$_}, $p);
3330 }
3331 $self->close_directory($bat->{''}, $p);
3332 $self->SUPER::close_edit($p);
3333 $p->clear;
3334}
3335sub abort_edit {
3337 my ($self) = @_;
3338 $self->SUPER::abort_edit($self->{pool});
3339}
3340sub DESTROY {
3342 my $self = shift;
3343 $self->SUPER::DESTROY(@_);
3344 $self->{pool}->clear;
3345}
3346# this drives the editor
3348sub apply_diff {
3349 my ($self) = @_;
3350 my $mods = $self->{mods};
3351 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3352 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3353 my $f = $m->{chg};
3354 if (defined $o{$f}) {
3355 $self->$f($m);
3356 } else {
3357 fatal("Invalid change type: $f");
3358 }
3359 }
3360 $self->rmdirs if $_rmdir;
3361 if (@$mods == 0) {
3362 $self->abort_edit;
3363 } else {
3364 $self->close_edit;
3365 }
3366 return scalar @$mods;
3367}
3368package Git::SVN::Ra;
3370use vars qw/@ISA $config_dir $_log_window_size/;
3371use strict;
3372use warnings;
3373my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3374BEGIN {
3376 # enforce temporary pool usage for some simple functions
3377 no strict 'refs';
3378 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3379 my $SUPER = "SUPER::$f";
3380 *$f = sub {
3381 my $self = shift;
3382 my $pool = SVN::Pool->new;
3383 my @ret = $self->$SUPER(@_,$pool);
3384 $pool->clear;
3385 wantarray ? @ret : $ret[0];
3386 };
3387 }
3388}
3389sub _auth_providers () {
3391 [
3392 SVN::Client::get_simple_provider(),
3393 SVN::Client::get_ssl_server_trust_file_provider(),
3394 SVN::Client::get_simple_prompt_provider(
3395 \&Git::SVN::Prompt::simple, 2),
3396 SVN::Client::get_ssl_client_cert_file_provider(),
3397 SVN::Client::get_ssl_client_cert_prompt_provider(
3398 \&Git::SVN::Prompt::ssl_client_cert, 2),
3399 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3400 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3401 SVN::Client::get_username_provider(),
3402 SVN::Client::get_ssl_server_trust_prompt_provider(
3403 \&Git::SVN::Prompt::ssl_server_trust),
3404 SVN::Client::get_username_prompt_provider(
3405 \&Git::SVN::Prompt::username, 2)
3406 ]
3407}
3408sub escape_uri_only {
3410 my ($uri) = @_;
3411 my @tmp;
3412 foreach (split m{/}, $uri) {
3413 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3414 push @tmp, $_;
3415 }
3416 join('/', @tmp);
3417}
3418sub escape_url {
3420 my ($url) = @_;
3421 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3422 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3423 $url = "$scheme://$domain$uri";
3424 }
3425 $url;
3426}
3427sub new {
3429 my ($class, $url) = @_;
3430 $url =~ s!/+$!!;
3431 return $RA if ($RA && $RA->{url} eq $url);
3432 SVN::_Core::svn_config_ensure($config_dir, undef);
3434 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3435 my $config = SVN::Core::config_get_config($config_dir);
3436 $RA = undef;
3437 my $dont_store_passwords = 1;
3438 my $conf_t = ${$config}{'config'};
3439 {
3440 no warnings 'once';
3441 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3442 # produces warnings that variables are used only once.
3443 # I had not found the better way to shut them up, so
3444 # the warnings of type 'once' are disabled in this block.
3445 if (SVN::_Core::svn_config_get_bool($conf_t,
3446 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3447 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3448 1) == 0) {
3449 SVN::_Core::svn_auth_set_parameter($baton,
3450 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3451 bless (\$dont_store_passwords, "_p_void"));
3452 }
3453 if (SVN::_Core::svn_config_get_bool($conf_t,
3454 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3455 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3456 1) == 0) {
3457 $Git::SVN::Prompt::_no_auth_cache = 1;
3458 }
3459 } # no warnings 'once'
3460 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3461 config => $config,
3462 pool => SVN::Pool->new,
3463 auth_provider_callbacks => $callbacks);
3464 $self->{url} = $url;
3465 $self->{svn_path} = $url;
3466 $self->{repos_root} = $self->get_repos_root;
3467 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3468 $self->{cache} = { check_path => { r => 0, data => {} },
3469 get_dir => { r => 0, data => {} } };
3470 $RA = bless $self, $class;
3471}
3472sub check_path {
3474 my ($self, $path, $r) = @_;
3475 my $cache = $self->{cache}->{check_path};
3476 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3477 return $cache->{data}->{$path};
3478 }
3479 my $pool = SVN::Pool->new;
3480 my $t = $self->SUPER::check_path($path, $r, $pool);
3481 $pool->clear;
3482 if ($r != $cache->{r}) {
3483 %{$cache->{data}} = ();
3484 $cache->{r} = $r;
3485 }
3486 $cache->{data}->{$path} = $t;
3487}
3488sub get_dir {
3490 my ($self, $dir, $r) = @_;
3491 my $cache = $self->{cache}->{get_dir};
3492 if ($r == $cache->{r}) {
3493 if (my $x = $cache->{data}->{$dir}) {
3494 return wantarray ? @$x : $x->[0];
3495 }
3496 }
3497 my $pool = SVN::Pool->new;
3498 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3499 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3500 $pool->clear;
3501 if ($r != $cache->{r}) {
3502 %{$cache->{data}} = ();
3503 $cache->{r} = $r;
3504 }
3505 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3506 wantarray ? (\%dirents, $r, $props) : \%dirents;
3507}
3508sub DESTROY {
3510 # do not call the real DESTROY since we store ourselves in $RA
3511}
3512sub get_log {
3514 my ($self, @args) = @_;
3515 my $pool = SVN::Pool->new;
3516 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3517 my $ret = $self->SUPER::get_log(@args, $pool);
3518 $pool->clear;
3519 $ret;
3520}
3521sub trees_match {
3523 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3524 my $ctx = SVN::Client->new(auth => _auth_providers);
3525 my $out = IO::File->new_tmpfile;
3526 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3528 # $ctx->diff(), so we'll create a default one
3529 my $pool = SVN::Pool->new_default_sub;
3530 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3532 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3533 $out->flush;
3534 my $ret = (($out->stat)[7] == 0);
3535 close $out or croak $!;
3536 $ret;
3538}
3539sub get_commit_editor {
3541 my ($self, $log, $cb, $pool) = @_;
3542 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3543 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3544}
3545sub gs_do_update {
3547 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3548 my $new = ($rev_a == $rev_b);
3549 my $path = $gs->{path};
3550 if ($new && -e $gs->{index}) {
3552 unlink $gs->{index} or die
3553 "Couldn't unlink index: $gs->{index}: $!\n";
3554 }
3555 my $pool = SVN::Pool->new;
3556 $editor->set_path_strip($path);
3557 my (@pc) = split m#/#, $path;
3558 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3559 1, $editor, $pool);
3560 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3561 # Since we can't rely on svn_ra_reparent being available, we'll
3563 # just have to do some magic with set_path to make it so
3564 # we only want a partial path.
3565 my $sp = '';
3566 my $final = join('/', @pc);
3567 while (@pc) {
3568 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3569 $sp .= '/' if length $sp;
3570 $sp .= shift @pc;
3571 }
3572 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3573 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3575 $reporter->finish_report($pool);
3577 $pool->clear;
3578 $editor->{git_commit_ok};
3579}
3580# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3582# svn_ra_reparent didn't work before 1.4)
3583sub gs_do_switch {
3584 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3585 my $path = $gs->{path};
3586 my $pool = SVN::Pool->new;
3587 my $full_url = $self->{url};
3589 my $old_url = $full_url;
3590 $full_url .= '/' . escape_uri_only($path) if length $path;
3591 my ($ra, $reparented);
3592 if ($old_url ne $full_url) {
3593 if ($old_url !~ m#^svn(\+ssh)?://#) {
3594 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3595 $pool);
3596 $self->{url} = $full_url;
3597 $reparented = 1;
3598 } else {
3599 $_[0] = undef;
3600 $self = undef;
3601 $RA = undef;
3602 $ra = Git::SVN::Ra->new($full_url);
3603 $ra_invalid = 1;
3604 }
3605 }
3606 $ra ||= $self;
3607 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3608 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3609 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3610 $reporter->finish_report($pool);
3611 if ($reparented) {
3613 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3614 $self->{url} = $old_url;
3615 }
3616 $pool->clear;
3618 $editor->{git_commit_ok};
3619}
3620sub longest_common_path {
3622 my ($gsv, $globs) = @_;
3623 my %common;
3624 my $common_max = scalar @$gsv;
3625 foreach my $gs (@$gsv) {
3627 my @tmp = split m#/#, $gs->{path};
3628 my $p = '';
3629 foreach (@tmp) {
3630 $p .= length($p) ? "/$_" : $_;
3631 $common{$p} ||= 0;
3632 $common{$p}++;
3633 }
3634 }
3635 $globs ||= [];
3636 $common_max += scalar @$globs;
3637 foreach my $glob (@$globs) {
3638 my @tmp = split m#/#, $glob->{path}->{left};
3639 my $p = '';
3640 foreach (@tmp) {
3641 $p .= length($p) ? "/$_" : $_;
3642 $common{$p} ||= 0;
3643 $common{$p}++;
3644 }
3645 }
3646 my $longest_path = '';
3648 foreach (sort {length $b <=> length $a} keys %common) {
3649 if ($common{$_} == $common_max) {
3650 $longest_path = $_;
3651 last;
3652 }
3653 }
3654 $longest_path;
3655}
3656sub gs_fetch_loop_common {
3658 my ($self, $base, $head, $gsv, $globs) = @_;
3659 return if ($base > $head);
3660 my $inc = $_log_window_size;
3661 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3662 my $longest_path = longest_common_path($gsv, $globs);
3663 my $ra_url = $self->{url};
3664 while (1) {
3665 my %revs;
3666 my $err;
3667 my $err_handler = $SVN::Error::handler;
3668 $SVN::Error::handler = sub {
3669 ($err) = @_;
3670 skip_unknown_revs($err);
3671 };
3672 sub _cb {
3673 my ($paths, $r, $author, $date, $log) = @_;
3674 [ dup_changed_paths($paths),
3675 { author => $author, date => $date, log => $log } ];
3676 }
3677 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3678 sub { $revs{$_[1]} = _cb(@_) });
3679 if ($err && $max >= $head) {
3680 print STDERR "Path '$longest_path' ",
3681 "was probably deleted:\n",
3682 $err->expanded_message,
3683 "\nWill attempt to follow ",
3684 "revisions r$min .. r$max ",
3685 "committed before the deletion\n";
3686 my $hi = $max;
3687 while (--$hi >= $min) {
3688 my $ok;
3689 $self->get_log([$longest_path], $min, $hi,
3690 0, 1, 1, sub {
3691 $ok ||= $_[1];
3692 $revs{$_[1]} = _cb(@_) });
3693 if ($ok) {
3694 print STDERR "r$min .. r$ok OK\n";
3695 last;
3696 }
3697 }
3698 }
3699 $SVN::Error::handler = $err_handler;
3700 my %exists = map { $_->{path} => $_ } @$gsv;
3702 foreach my $r (sort {$a <=> $b} keys %revs) {
3703 my ($paths, $logged) = @{$revs{$r}};
3704 foreach my $gs ($self->match_globs(\%exists, $paths,
3706 $globs, $r)) {
3707 if ($gs->rev_db_max >= $r) {
3708 next;
3709 }
3710 next unless $gs->match_paths($paths, $r);
3711 $gs->{logged_rev_props} = $logged;
3712 if (my $last_commit = $gs->last_commit) {
3713 $gs->assert_index_clean($last_commit);
3714 }
3715 my $log_entry = $gs->do_fetch($paths, $r);
3716 if ($log_entry) {
3717 $gs->do_git_commit($log_entry);
3718 }
3719 }
3720 foreach my $g (@$globs) {
3721 my $k = "svn-remote.$g->{remote}." .
3722 "$g->{t}-maxRev";
3723 Git::SVN::tmp_config($k, $r);
3724 }
3725 if ($ra_invalid) {
3726 $_[0] = undef;
3727 $self = undef;
3728 $RA = undef;
3729 $self = Git::SVN::Ra->new($ra_url);
3730 $ra_invalid = undef;
3731 }
3732 }
3733 # pre-fill the .rev_db since it'll eventually get filled in
3734 # with '0' x40 if something new gets committed
3735 foreach my $gs (@$gsv) {
3736 next if defined $gs->rev_db_get($max);
3737 $gs->rev_db_set($max, 0 x40);
3738 }
3739 foreach my $g (@$globs) {
3740 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3741 Git::SVN::tmp_config($k, $max);
3742 }
3743 last if $max >= $head;
3744 $min = $max + 1;
3745 $max += $inc;
3746 $max = $head if ($max > $head);
3747 }
3748}
3749sub match_globs {
3751 my ($self, $exists, $paths, $globs, $r) = @_;
3752 sub get_dir_check {
3754 my ($self, $exists, $g, $r) = @_;
3755 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3756 return unless scalar @x == 3;
3757 my $dirents = $x[0];
3758 foreach my $de (keys %$dirents) {
3759 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3760 my $p = $g->{path}->full_path($de);
3761 next if $exists->{$p};
3762 next if (length $g->{path}->{right} &&
3763 ($self->check_path($p, $r) !=
3764 $SVN::Node::dir));
3765 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3766 $g->{ref}->full_path($de), 1);
3767 }
3768 }
3769 foreach my $g (@$globs) {
3770 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3771 if ($path->{action} =~ /^[AR]$/) {
3772 get_dir_check($self, $exists, $g, $r);
3773 }
3774 }
3775 foreach (keys %$paths) {
3776 if (/$g->{path}->{left_regex}/ &&
3777 !/$g->{path}->{regex}/) {
3778 next if $paths->{$_}->{action} !~ /^[AR]$/;
3779 get_dir_check($self, $exists, $g, $r);
3780 }
3781 next unless /$g->{path}->{regex}/;
3782 my $p = $1;
3783 my $pathname = $g->{path}->full_path($p);
3784 next if $exists->{$pathname};
3785 next if ($self->check_path($pathname, $r) !=
3786 $SVN::Node::dir);
3787 $exists->{$pathname} = Git::SVN->init(
3788 $self->{url}, $pathname, undef,
3789 $g->{ref}->full_path($p), 1);
3790 }
3791 my $c = '';
3792 foreach (split m#/#, $g->{path}->{left}) {
3793 $c .= "/$_";
3794 next unless ($paths->{$c} &&
3795 ($paths->{$c}->{action} =~ /^[AR]$/));
3796 get_dir_check($self, $exists, $g, $r);
3797 }
3798 }
3799 values %$exists;
3800}
3801sub minimize_url {
3803 my ($self) = @_;
3804 return $self->{url} if ($self->{url} eq $self->{repos_root});
3805 my $url = $self->{repos_root};
3806 my @components = split(m!/!, $self->{svn_path});
3807 my $c = '';
3808 do {
3809 $url .= "/$c" if length $c;
3810 eval { (ref $self)->new($url)->get_latest_revnum };
3811 } while ($@ && ($c = shift @components));
3812 $url;
3813}
3814sub can_do_switch {
3816 my $self = shift;
3817 unless (defined $can_do_switch) {
3818 my $pool = SVN::Pool->new;
3819 my $rep = eval {
3820 $self->do_switch(1, '', 0, $self->{url},
3821 SVN::Delta::Editor->new, $pool);
3822 };
3823 if ($@) {
3824 $can_do_switch = 0;
3825 } else {
3826 $rep->abort_report($pool);
3827 $can_do_switch = 1;
3828 }
3829 $pool->clear;
3830 }
3831 $can_do_switch;
3832}
3833sub skip_unknown_revs {
3835 my ($err) = @_;
3836 my $errno = $err->apr_err();
3837 # Maybe the branch we're tracking didn't
3838 # exist when the repo started, so it's
3839 # not an error if it doesn't, just continue
3840 #
3841 # Wonderfully consistent library, eh?
3842 # 160013 - svn:// and file://
3843 # 175002 - http(s)://
3844 # 175007 - http(s):// (this repo required authorization, too...)
3845 # More codes may be discovered later...
3846 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3847 my $err_key = $err->expanded_message;
3848 # revision numbers change every time, filter them out
3849 $err_key =~ s/\d+/\0/g;
3850 $err_key = "$errno\0$err_key";
3851 unless ($ignored_err{$err_key}) {
3852 warn "W: Ignoring error from SVN, path probably ",
3853 "does not exist: ($errno): ",
3854 $err->expanded_message,"\n";
3855 $ignored_err{$err_key} = 1;
3856 }
3857 return;
3858 }
3859 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3860}
3861# svn_log_changed_path_t objects passed to get_log are likely to be
3863# overwritten even if only the refs are copied to an external variable,
3864# so we should dup the structures in their entirety. Using an externally
3865# passed pool (instead of our temporary and quickly cleared pool in
3866# Git::SVN::Ra) does not help matters at all...
3867sub dup_changed_paths {
3868 my ($paths) = @_;
3869 return undef unless $paths;
3870 my %ret;
3871 foreach my $p (keys %$paths) {
3872 my $i = $paths->{$p};
3873 my %s = map { $_ => $i->$_ }
3874 qw/copyfrom_path copyfrom_rev action/;
3875 $ret{$p} = \%s;
3876 }
3877 \%ret;
3878}
3879package Git::SVN::Log;
3881use strict;
3882use warnings;
3883use POSIX qw/strftime/;
3884use constant commit_log_separator => ('-' x 72) . "\n";
3885use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3886 %rusers $show_commit $incremental/;
3887my $l_fmt;
3888sub cmt_showable {
3890 my ($c) = @_;
3891 return 1 if defined $c->{r};
3892 # big commit message got truncated by the 16k pretty buffer in rev-list
3894 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3895 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3896 @{$c->{l}} = ();
3897 my @log = command(qw/cat-file commit/, $c->{c});
3898 # shift off the headers
3900 shift @log while ($log[0] ne '');
3901 shift @log;
3902 # TODO: make $c->{l} not have a trailing newline in the future
3904 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3905 (undef, $c->{r}, undef) = ::extract_metadata(
3907 (grep(/^git-svn-id: /, @log))[-1]);
3908 }
3909 return defined $c->{r};
3910}
3911sub log_use_color {
3913 return 1 if $color;
3914 my ($dc, $dcvar);
3915 $dcvar = 'color.diff';
3916 $dc = `git-config --get $dcvar`;
3917 if ($dc eq '') {
3918 # nothing at all; fallback to "diff.color"
3919 $dcvar = 'diff.color';
3920 $dc = `git-config --get $dcvar`;
3921 }
3922 chomp($dc);
3923 if ($dc eq 'auto') {
3924 my $pc;
3925 $pc = `git-config --get color.pager`;
3926 if ($pc eq '') {
3927 # does not have it -- fallback to pager.color
3928 $pc = `git-config --bool --get pager.color`;
3929 }
3930 else {
3931 $pc = `git-config --bool --get color.pager`;
3932 if ($?) {
3933 $pc = 'false';
3934 }
3935 }
3936 chomp($pc);
3937 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3938 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3939 }
3940 return 0;
3941 }
3942 return 0 if $dc eq 'never';
3943 return 1 if $dc eq 'always';
3944 chomp($dc = `git-config --bool --get $dcvar`);
3945 return ($dc eq 'true');
3946}
3947sub git_svn_log_cmd {
3949 my ($r_min, $r_max, @args) = @_;
3950 my $head = 'HEAD';
3951 my (@files, @log_opts);
3952 foreach my $x (@args) {
3953 if ($x eq '--' || @files) {
3954 push @files, $x;
3955 } else {
3956 if (::verify_ref("$x^0")) {
3957 $head = $x;
3958 } else {
3959 push @log_opts, $x;
3960 }
3961 }
3962 }
3963 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3965 $gs ||= Git::SVN->_new;
3966 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3967 $gs->refname);
3968 push @cmd, '-r' unless $non_recursive;
3969 push @cmd, qw/--raw --name-status/ if $verbose;
3970 push @cmd, '--color' if log_use_color();
3971 push @cmd, @log_opts;
3972 if (defined $r_max && $r_max == $r_min) {
3973 push @cmd, '--max-count=1';
3974 if (my $c = $gs->rev_db_get($r_max)) {
3975 push @cmd, $c;
3976 }
3977 } elsif (defined $r_max) {
3978 if ($r_max < $r_min) {
3979 ($r_min, $r_max) = ($r_max, $r_min);
3980 }
3981 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
3982 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
3983 # If there are no commits in the range, both $c_max and $c_min
3984 # will be undefined. If there is at least 1 commit in the
3985 # range, both will be defined.
3986 return () if !defined $c_min || !defined $c_max;
3987 if ($c_min eq $c_max) {
3988 push @cmd, '--max-count=1', $c_min;
3989 } else {
3990 push @cmd, '--boundary', "$c_min..$c_max";
3991 }
3992 }
3993 return (@cmd, @files);
3994}
3995# adapted from pager.c
3997sub config_pager {
3998 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3999 if (!defined $pager) {
4000 $pager = 'less';
4001 } elsif (length $pager == 0 || $pager eq 'cat') {
4002 $pager = undef;
4003 }
4004}
4005sub run_pager {
4007 return unless -t *STDOUT && defined $pager;
4008 pipe my $rfd, my $wfd or return;
4009 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4010 if (!$pid) {
4011 open STDOUT, '>&', $wfd or
4012 ::fatal "Can't redirect to stdout: $!";
4013 return;
4014 }
4015 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4016 $ENV{LESS} ||= 'FRSX';
4017 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4018}
4019sub format_svn_date {
4021 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4022}
4023sub parse_git_date {
4025 my ($t, $tz) = @_;
4026 # Date::Parse isn't in the standard Perl distro :(
4027 if ($tz =~ s/^\+//) {
4028 $t += tz_to_s_offset($tz);
4029 } elsif ($tz =~ s/^\-//) {
4030 $t -= tz_to_s_offset($tz);
4031 }
4032 return $t;
4033}
4034sub set_local_timezone {
4036 if (defined $TZ) {
4037 $ENV{TZ} = $TZ;
4038 } else {
4039 delete $ENV{TZ};
4040 }
4041}
4042sub tz_to_s_offset {
4044 my ($tz) = @_;
4045 $tz =~ s/(\d\d)$//;
4046 return ($1 * 60) + ($tz * 3600);
4047}
4048sub get_author_info {
4050 my ($dest, $author, $t, $tz) = @_;
4051 $author =~ s/(?:^\s*|\s*$)//g;
4052 $dest->{a_raw} = $author;
4053 my $au;
4054 if ($::_authors) {
4055 $au = $rusers{$author} || undef;
4056 }
4057 if (!$au) {
4058 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4059 }
4060 $dest->{t} = $t;
4061 $dest->{tz} = $tz;
4062 $dest->{a} = $au;
4063 $dest->{t_utc} = parse_git_date($t, $tz);
4064}
4065sub process_commit {
4067 my ($c, $r_min, $r_max, $defer) = @_;
4068 if (defined $r_min && defined $r_max) {
4069 if ($r_min == $c->{r} && $r_min == $r_max) {
4070 show_commit($c);
4071 return 0;
4072 }
4073 return 1 if $r_min == $r_max;
4074 if ($r_min < $r_max) {
4075 # we need to reverse the print order
4076 return 0 if (defined $limit && --$limit < 0);
4077 push @$defer, $c;
4078 return 1;
4079 }
4080 if ($r_min != $r_max) {
4081 return 1 if ($r_min < $c->{r});
4082 return 1 if ($r_max > $c->{r});
4083 }
4084 }
4085 return 0 if (defined $limit && --$limit < 0);
4086 show_commit($c);
4087 return 1;
4088}
4089sub show_commit {
4091 my $c = shift;
4092 if ($oneline) {
4093 my $x = "\n";
4094 if (my $l = $c->{l}) {
4095 while ($l->[0] =~ /^\s*$/) { shift @$l }
4096 $x = $l->[0];
4097 }
4098 $l_fmt ||= 'A' . length($c->{r});
4099 print 'r',pack($l_fmt, $c->{r}),' | ';
4100 print "$c->{c} | " if $show_commit;
4101 print $x;
4102 } else {
4103 show_commit_normal($c);
4104 }
4105}
4106sub show_commit_changed_paths {
4108 my ($c) = @_;
4109 return unless $c->{changed};
4110 print "Changed paths:\n", @{$c->{changed}};
4111}
4112sub show_commit_normal {
4114 my ($c) = @_;
4115 print commit_log_separator, "r$c->{r} | ";
4116 print "$c->{c} | " if $show_commit;
4117 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4118 my $nr_line = 0;
4119 if (my $l = $c->{l}) {
4121 while ($l->[$#$l] eq "\n" && $#$l > 0
4122 && $l->[($#$l - 1)] eq "\n") {
4123 pop @$l;
4124 }
4125 $nr_line = scalar @$l;
4126 if (!$nr_line) {
4127 print "1 line\n\n\n";
4128 } else {
4129 if ($nr_line == 1) {
4130 $nr_line = '1 line';
4131 } else {
4132 $nr_line .= ' lines';
4133 }
4134 print $nr_line, "\n";
4135 show_commit_changed_paths($c);
4136 print "\n";
4137 print $_ foreach @$l;
4138 }
4139 } else {
4140 print "1 line\n";
4141 show_commit_changed_paths($c);
4142 print "\n";
4143 }
4145 foreach my $x (qw/raw stat diff/) {
4146 if ($c->{$x}) {
4147 print "\n";
4148 print $_ foreach @{$c->{$x}}
4149 }
4150 }
4151}
4152sub cmd_show_log {
4154 my (@args) = @_;
4155 my ($r_min, $r_max);
4156 my $r_last = -1; # prevent dupes
4157 set_local_timezone();
4158 if (defined $::_revision) {
4159 if ($::_revision =~ /^(\d+):(\d+)$/) {
4160 ($r_min, $r_max) = ($1, $2);
4161 } elsif ($::_revision =~ /^\d+$/) {
4162 $r_min = $r_max = $::_revision;
4163 } else {
4164 ::fatal "-r$::_revision is not supported, use ",
4165 "standard 'git log' arguments instead";
4166 }
4167 }
4168 config_pager();
4170 @args = git_svn_log_cmd($r_min, $r_max, @args);
4171 if (!@args) {
4172 print commit_log_separator unless $incremental || $oneline;
4173 return;
4174 }
4175 my $log = command_output_pipe(@args);
4176 run_pager();
4177 my (@k, $c, $d, $stat);
4178 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4179 while (<$log>) {
4180 if (/^${esc_color}commit -?($::sha1_short)/o) {
4181 my $cmt = $1;
4182 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4183 $r_last = $c->{r};
4184 process_commit($c, $r_min, $r_max, \@k) or
4185 goto out;
4186 }
4187 $d = undef;
4188 $c = { c => $cmt };
4189 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4190 get_author_info($c, $1, $2, $3);
4191 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4192 # ignore
4193 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4194 push @{$c->{raw}}, $_;
4195 } elsif (/^${esc_color}[ACRMDT]\t/) {
4196 # we could add $SVN->{svn_path} here, but that requires
4197 # remote access at the moment (repo_path_split)...
4198 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4199 push @{$c->{changed}}, $_;
4200 } elsif (/^${esc_color}diff /o) {
4201 $d = 1;
4202 push @{$c->{diff}}, $_;
4203 } elsif ($d) {
4204 push @{$c->{diff}}, $_;
4205 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4206 $esc_color*[\+\-]*$esc_color$/x) {
4207 $stat = 1;
4208 push @{$c->{stat}}, $_;
4209 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4210 push @{$c->{stat}}, $_;
4211 $stat = undef;
4212 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4213 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4214 } elsif (s/^${esc_color} //o) {
4215 push @{$c->{l}}, $_;
4216 }
4217 }
4218 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4219 $r_last = $c->{r};
4220 process_commit($c, $r_min, $r_max, \@k);
4221 }
4222 if (@k) {
4223 ($r_min, $r_max) = ($r_max, $r_min);
4224 process_commit($_, $r_min, $r_max) foreach reverse @k;
4225 }
4226out:
4227 close $log;
4228 print commit_log_separator unless $incremental || $oneline;
4229}
4230package Git::SVN::Migration;
4232# these version numbers do NOT correspond to actual version numbers
4233# of git nor git-svn. They are just relative.
4234#
4235# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4236#
4237# v1 layout: .git/$id/info/url, refs/remotes/$id
4238#
4239# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4240#
4241# v3 layout: .git/svn/$id, refs/remotes/$id
4242# - info/url may remain for backwards compatibility
4243# - this is what we migrate up to this layout automatically,
4244# - this will be used by git svn init on single branches
4245# v3.1 layout (auto migrated):
4246# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4247# for backwards compatibility
4248#
4249# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4250# - this is only created for newly multi-init-ed
4251# repositories. Similar in spirit to the
4252# --use-separate-remotes option in git-clone (now default)
4253# - we do not automatically migrate to this (following
4254# the example set by core git)
4255use strict;
4256use warnings;
4257use Carp qw/croak/;
4258use File::Path qw/mkpath/;
4259use File::Basename qw/dirname basename/;
4260use vars qw/$_minimize/;
4261sub migrate_from_v0 {
4263 my $git_dir = $ENV{GIT_DIR};
4264 return undef unless -d $git_dir;
4265 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4266 my $migrated = 0;
4267 while (<$fh>) {
4268 chomp;
4269 my ($id, $orig_ref) = ($_, $_);
4270 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4271 next unless -f "$git_dir/$id/info/url";
4272 my $new_ref = "refs/remotes/$id";
4273 if (::verify_ref("$new_ref^0")) {
4274 print STDERR "W: $orig_ref is probably an old ",
4275 "branch used by an ancient version of ",
4276 "git-svn.\n",
4277 "However, $new_ref also exists.\n",
4278 "We will not be able ",
4279 "to use this branch until this ",
4280 "ambiguity is resolved.\n";
4281 next;
4282 }
4283 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4284 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4285 command_noisy('update-ref', $new_ref, $orig_ref);
4286 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4287 $migrated++;
4288 }
4289 command_close_pipe($fh, $ctx);
4290 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4291 $migrated;
4292}
4293sub migrate_from_v1 {
4295 my $git_dir = $ENV{GIT_DIR};
4296 my $migrated = 0;
4297 return $migrated unless -d $git_dir;
4298 my $svn_dir = "$git_dir/svn";
4299 # just in case somebody used 'svn' as their $id at some point...
4301 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4302 print STDERR "Migrating from a git-svn v1 layout...\n";
4304 mkpath([$svn_dir]);
4305 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4306 "$svn_dir\n\t(required for this version ",
4307 "($::VERSION) of git-svn) does not. exist\n";
4308 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4309 while (<$fh>) {
4310 my $x = $_;
4311 next unless $x =~ s#^refs/remotes/##;
4312 chomp $x;
4313 next unless -f "$git_dir/$x/info/url";
4314 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4315 next unless $u;
4316 my $dn = dirname("$git_dir/svn/$x");
4317 mkpath([$dn]) unless -d $dn;
4318 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4319 mkpath(["$git_dir/svn/svn"]);
4320 print STDERR " - $git_dir/$x/info => ",
4321 "$git_dir/svn/$x/info\n";
4322 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4323 croak "$!: $x";
4324 # don't worry too much about these, they probably
4325 # don't exist with repos this old (save for index,
4326 # and we can easily regenerate that)
4327 foreach my $f (qw/unhandled.log index .rev_db/) {
4328 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4329 }
4330 } else {
4331 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4332 rename "$git_dir/$x", "$git_dir/svn/$x" or
4333 croak "$!: $x";
4334 }
4335 $migrated++;
4336 }
4337 command_close_pipe($fh, $ctx);
4338 print STDERR "Done migrating from a git-svn v1 layout\n";
4339 $migrated;
4340}
4341sub read_old_urls {
4343 my ($l_map, $pfx, $path) = @_;
4344 my @dir;
4345 foreach (<$path/*>) {
4346 if (-r "$_/info/url") {
4347 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4348 my $ref_id = $pfx . basename $_;
4349 my $url = ::file_to_s("$_/info/url");
4350 $l_map->{$ref_id} = $url;
4351 } elsif (-d $_) {
4352 push @dir, $_;
4353 }
4354 }
4355 foreach (@dir) {
4356 my $x = $_;
4357 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4358 read_old_urls($l_map, $x, $_);
4359 }
4360}
4361sub migrate_from_v2 {
4363 my @cfg = command(qw/config -l/);
4364 return if grep /^svn-remote\..+\.url=/, @cfg;
4365 my %l_map;
4366 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4367 my $migrated = 0;
4368 foreach my $ref_id (sort keys %l_map) {
4370 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4371 if ($@) {
4372 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4373 }
4374 $migrated++;
4375 }
4376 $migrated;
4377}
4378sub minimize_connections {
4380 my $r = Git::SVN::read_all_remotes();
4381 my $new_urls = {};
4382 my $root_repos = {};
4383 foreach my $repo_id (keys %$r) {
4384 my $url = $r->{$repo_id}->{url} or next;
4385 my $fetch = $r->{$repo_id}->{fetch} or next;
4386 my $ra = Git::SVN::Ra->new($url);
4387 # skip existing cases where we already connect to the root
4389 if (($ra->{url} eq $ra->{repos_root}) ||
4390 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4391 $repo_id)) {
4392 $root_repos->{$ra->{url}} = $repo_id;
4393 next;
4394 }
4395 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4397 my $root_path = $ra->{url};
4398 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4399 foreach my $path (keys %$fetch) {
4400 my $ref_id = $fetch->{$path};
4401 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4402 # make sure we can read when connecting to
4404 # a higher level of a repository
4405 my ($last_rev, undef) = $gs->last_rev_commit;
4406 if (!defined $last_rev) {
4407 $last_rev = eval {
4408 $root_ra->get_latest_revnum;
4409 };
4410 next if $@;
4411 }
4412 my $new = $root_path;
4413 $new .= length $path ? "/$path" : '';
4414 eval {
4415 $root_ra->get_log([$new], $last_rev, $last_rev,
4416 0, 0, 1, sub { });
4417 };
4418 next if $@;
4419 $new_urls->{$ra->{repos_root}}->{$new} =
4420 { ref_id => $ref_id,
4421 old_repo_id => $repo_id,
4422 old_path => $path };
4423 }
4424 }
4425 my @emptied;
4427 foreach my $url (keys %$new_urls) {
4428 # see if we can re-use an existing [svn-remote "repo_id"]
4429 # instead of creating a(n ugly) new section:
4430 my $repo_id = $root_repos->{$url} ||
4431 Git::SVN::sanitize_remote_name($url);
4432 my $fetch = $new_urls->{$url};
4434 foreach my $path (keys %$fetch) {
4435 my $x = $fetch->{$path};
4436 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4437 my $pfx = "svn-remote.$x->{old_repo_id}";
4438 my $old_fetch = quotemeta("$x->{old_path}:".
4440 "refs/remotes/$x->{ref_id}");
4441 command_noisy(qw/config --unset/,
4442 "$pfx.fetch", '^'. $old_fetch . '$');
4443 delete $r->{$x->{old_repo_id}}->
4444 {fetch}->{$x->{old_path}};
4445 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4446 command_noisy(qw/config --unset/,
4447 "$pfx.url");
4448 push @emptied, $x->{old_repo_id}
4449 }
4450 }
4451 }
4452 if (@emptied) {
4453 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4454 "$ENV{GIT_DIR}/config";
4455 print STDERR <<EOF;
4456The following [svn-remote] sections in your config file ($file) are empty
4457and can be safely removed:
4458EOF
4459 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4460 }
4461}
4462sub migration_check {
4464 migrate_from_v0();
4465 migrate_from_v1();
4466 migrate_from_v2();
4467 minimize_connections() if $_minimize;
4468}
4469package Git::IndexInfo;
4471use strict;
4472use warnings;
4473use Git qw/command_input_pipe command_close_pipe/;
4474sub new {
4476 my ($class) = @_;
4477 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4478 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4479}
4480sub remove {
4482 my ($self, $path) = @_;
4483 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4484 return ++$self->{nr};
4485 }
4486 undef;
4487}
4488sub update {
4490 my ($self, $mode, $hash, $path) = @_;
4491 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4492 return ++$self->{nr};
4493 }
4494 undef;
4495}
4496sub DESTROY {
4498 my ($self) = @_;
4499 command_close_pipe($self->{gui}, $self->{ctx});
4500}
4501package Git::SVN::GlobSpec;
4503use strict;
4504use warnings;
4505sub new {
4507 my ($class, $glob) = @_;
4508 my $re = $glob;
4509 $re =~ s!/+$!!g; # no need for trailing slashes
4510 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4511 my ($left, $right) = ($1, $2);
4512 if ($nr > 1) {
4513 die "Only one '*' wildcard expansion ",
4514 "is supported (got $nr): '$glob'\n";
4515 } elsif ($nr == 0) {
4516 die "One '*' is needed for glob: '$glob'\n";
4517 }
4518 $re = quotemeta($left) . $re . quotemeta($right);
4519 if (length $left && !($left =~ s!/+$!!g)) {
4520 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4521 }
4522 if (length $right && !($right =~ s!^/+!!g)) {
4523 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4524 }
4525 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4526 bless { left => $left, right => $right, left_regex => $left_re,
4527 regex => qr/$re/, glob => $glob }, $class;
4528}
4529sub full_path {
4531 my ($self, $path) = @_;
4532 return (length $self->{left} ? "$self->{left}/" : '') .
4533 $path . (length $self->{right} ? "/$self->{right}" : '');
4534}
4535__END__
4537Data structures:
4539$remotes = { # returned by read_all_remotes()
4542 'svn' => {
4543 # svn-remote.svn.url=https://svn.musicpd.org
4544 url => 'https://svn.musicpd.org',
4545 # svn-remote.svn.fetch=mpd/trunk:trunk
4546 fetch => {
4547 'mpd/trunk' => 'trunk',
4548 },
4549 # svn-remote.svn.tags=mpd/tags/*:tags/*
4550 tags => {
4551 path => {
4552 left => 'mpd/tags',
4553 right => '',
4554 regex => qr!mpd/tags/([^/]+)$!,
4555 glob => 'tags/*',
4556 },
4557 ref => {
4558 left => 'tags',
4559 right => '',
4560 regex => qr!tags/([^/]+)$!,
4561 glob => 'tags/*',
4562 },
4563 }
4564 }
4565};
4566$log_entry hashref as returned by libsvn_log_entry()
4568{
4569 log => 'whitespace-formatted log entry
4570', # trailing newline is preserved
4571 revision => '8', # integer
4572 date => '2004-02-24T17:01:44.108345Z', # commit date
4573 author => 'committer name'
4574};
4575# this is generated by generate_diff();
4578@mods = array of diff-index line hashes, each element represents one line
4579 of diff-index output
4580diff-index line ($m hash)
4582{
4583 mode_a => first column of diff-index output, no leading ':',
4584 mode_b => second column of diff-index output,
4585 sha1_b => sha1sum of the final blob,
4586 chg => change type [MCRADT],
4587 file_a => original file name of a file (iff chg is 'C' or 'R')
4588 file_b => new/current file name of a file (any chg)
4589}
4590;
4591# retval of read_url_paths{,_all}();
4593$l_map = {
4594 # repository root url
4595 'https://svn.musicpd.org' => {
4596 # repository path # GIT_SVN_ID
4597 'mpd/trunk' => 'trunk',
4598 'mpd/tags/0.11.5' => 'tags/0.11.5',
4599 },
4600}
4601Notes:
4603 I don't trust the each() function on unless I created %hash myself
4604 because the internal iterator may not have started at base.