d111dc1442096a30431731b45aa0666604f07921
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@@';
11my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
13$ENV{GIT_DIR} ||= '.git';
14$Git::SVN::default_repo_id = 'svn';
15$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
16$Git::SVN::Ra::_log_window_size = 100;
17$Git::SVN::Log::TZ = $ENV{TZ};
19$ENV{TZ} = 'UTC';
20$| = 1; # unbuffer STDOUT
21sub fatal (@) { print STDERR @_; exit 1 }
23require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
24require SVN::Ra;
25require SVN::Delta;
26if ($SVN::Core::VERSION lt '1.1.0') {
27 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
28}
29push @Git::SVN::Ra::ISA, 'SVN::Ra';
30push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
31push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
32use Carp qw/croak/;
33use IO::File qw//;
34use File::Basename qw/dirname basename/;
35use File::Path qw/mkpath/;
36use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
37use IPC::Open3;
38use Git;
39BEGIN {
41 # import functions from Git into our packages, en masse
42 no strict 'refs';
43 foreach (qw/command command_oneline command_noisy command_output_pipe
44 command_input_pipe command_close_pipe/) {
45 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
46 Git::SVN::Migration Git::SVN::Log Git::SVN),
47 __PACKAGE__) {
48 *{"${package}::$_"} = \&{"Git::$_"};
49 }
50 }
51}
52my ($SVN);
54$sha1 = qr/[a-f\d]{40}/;
56$sha1_short = qr/[a-f\d]{4,40}/;
57my ($_stdin, $_help, $_edit,
58 $_message, $_file,
59 $_template, $_shared,
60 $_version, $_fetch_all, $_no_rebase,
61 $_merge, $_strategy, $_dry_run, $_local,
62 $_prefix, $_no_checkout, $_verbose);
63$Git::SVN::_follow_parent = 1;
64my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
65 'config-dir=s' => \$Git::SVN::Ra::config_dir,
66 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
67my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
68 'authors-file|A=s' => \$_authors,
69 'repack:i' => \$Git::SVN::_repack,
70 'noMetadata' => \$Git::SVN::_no_metadata,
71 'useSvmProps' => \$Git::SVN::_use_svm_props,
72 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
73 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
74 'no-checkout' => \$_no_checkout,
75 'quiet|q' => \$_q,
76 'repack-flags|repack-args|repack-opts=s' =>
77 \$Git::SVN::_repack_flags,
78 %remote_opts );
79my ($_trunk, $_tags, $_branches);
81my %icv;
82my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
83 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
84 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
85 'minimize-url|m' => \$Git::SVN::_minimize_url,
86 'no-metadata' => sub { $icv{noMetadata} = 1 },
87 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
88 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
89 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
90 %remote_opts );
91my %cmt_opts = ( 'edit|e' => \$_edit,
92 'rmdir' => \$SVN::Git::Editor::_rmdir,
93 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
94 'l=i' => \$SVN::Git::Editor::_rename_limit,
95 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
96);
97my %cmd = (
99 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
100 { 'revision|r=s' => \$_revision,
101 'fetch-all|all' => \$_fetch_all,
102 %fc_opts } ],
103 clone => [ \&cmd_clone, "Initialize and fetch revisions",
104 { 'revision|r=s' => \$_revision,
105 %fc_opts, %init_opts } ],
106 init => [ \&cmd_init, "Initialize a repo for tracking" .
107 " (requires URL argument)",
108 \%init_opts ],
109 'multi-init' => [ \&cmd_multi_init,
110 "Deprecated alias for ".
111 "'$0 init -T<trunk> -b<branches> -t<tags>'",
112 \%init_opts ],
113 dcommit => [ \&cmd_dcommit,
114 'Commit several diffs to merge with upstream',
115 { 'merge|m|M' => \$_merge,
116 'strategy|s=s' => \$_strategy,
117 'verbose|v' => \$_verbose,
118 'dry-run|n' => \$_dry_run,
119 'fetch-all|all' => \$_fetch_all,
120 'no-rebase' => \$_no_rebase,
121 %cmt_opts, %fc_opts } ],
122 'set-tree' => [ \&cmd_set_tree,
123 "Set an SVN repository to a git tree-ish",
124 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
125 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
126 { 'revision|r=i' => \$_revision } ],
127 'multi-fetch' => [ \&cmd_multi_fetch,
128 "Deprecated alias for $0 fetch --all",
129 { 'revision|r=s' => \$_revision, %fc_opts } ],
130 'migrate' => [ sub { },
131 # no-op, we automatically run this anyways,
132 'Migrate configuration/metadata/layout from
133 previous versions of git-svn',
134 { 'minimize' => \$Git::SVN::Migration::_minimize,
135 %remote_opts } ],
136 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
137 { 'limit=i' => \$Git::SVN::Log::limit,
138 'revision|r=s' => \$_revision,
139 'verbose|v' => \$Git::SVN::Log::verbose,
140 'incremental' => \$Git::SVN::Log::incremental,
141 'oneline' => \$Git::SVN::Log::oneline,
142 'show-commit' => \$Git::SVN::Log::show_commit,
143 'non-recursive' => \$Git::SVN::Log::non_recursive,
144 'authors-file|A=s' => \$_authors,
145 'color' => \$Git::SVN::Log::color,
146 'pager=s' => \$Git::SVN::Log::pager,
147 } ],
148 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
149 { } ],
150 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
151 { 'merge|m|M' => \$_merge,
152 'verbose|v' => \$_verbose,
153 'strategy|s=s' => \$_strategy,
154 'local|l' => \$_local,
155 'fetch-all|all' => \$_fetch_all,
156 %fc_opts } ],
157 'commit-diff' => [ \&cmd_commit_diff,
158 'Commit a diff between two trees',
159 { 'message|m=s' => \$_message,
160 'file|F=s' => \$_file,
161 'revision|r=s' => \$_revision,
162 %cmt_opts } ],
163);
164my $cmd;
166for (my $i = 0; $i < @ARGV; $i++) {
167 if (defined $cmd{$ARGV[$i]}) {
168 $cmd = $ARGV[$i];
169 splice @ARGV, $i, 1;
170 last;
171 }
172};
173my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
175read_repo_config(\%opts);
177Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
178my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
179 'minimize-connections' => \$Git::SVN::Migration::_minimize,
180 'id|i=s' => \$Git::SVN::default_ref_id,
181 'svn-remote|remote|R=s' => sub {
182 $Git::SVN::no_reuse_existing = 1;
183 $Git::SVN::default_repo_id = $_[1] });
184exit 1 if (!$rv && $cmd && $cmd ne 'log');
185usage(0) if $_help;
187version() if $_version;
188usage(1) unless defined $cmd;
189load_authors() if $_authors;
190# make sure we're always running
192unless ($cmd =~ /(?:clone|init|multi-init)$/) {
193 unless (-d $ENV{GIT_DIR}) {
194 if ($git_dir_user_set) {
195 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
196 "but it is not a directory\n";
197 }
198 my $git_dir = delete $ENV{GIT_DIR};
199 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
200 unless (length $cdup) {
201 die "Already at toplevel, but $git_dir ",
202 "not found '$cdup'\n";
203 }
204 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
205 unless (-d $git_dir) {
206 die "$git_dir still not found after going to ",
207 "'$cdup'\n";
208 }
209 $ENV{GIT_DIR} = $git_dir;
210 }
211}
212unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
213 Git::SVN::Migration::migration_check();
214}
215Git::SVN::init_vars();
216eval {
217 Git::SVN::verify_remotes_sanity();
218 $cmd{$cmd}->[0]->(@ARGV);
219};
220fatal $@ if $@;
221post_fetch_checkout();
222exit 0;
223####################### primary functions ######################
225sub usage {
226 my $exit = shift || 0;
227 my $fd = $exit ? \*STDERR : \*STDOUT;
228 print $fd <<"";
229git-svn - bidirectional operations between a single Subversion tree and git
230Usage: $0 <command> [options] [arguments]\n
231 print $fd "Available commands:\n" unless $cmd;
233 foreach (sort keys %cmd) {
235 next if $cmd && $cmd ne $_;
236 next if /^multi-/; # don't show deprecated commands
237 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
238 foreach (keys %{$cmd{$_}->[2]}) {
239 # mixed-case options are for .git/config only
240 next if /[A-Z]/ && /^[a-z]+$/i;
241 # prints out arguments as they should be passed:
242 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
243 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
244 "--$_" : "-$_" }
245 split /\|/,$_)," $x\n";
246 }
247 }
248 print $fd <<"";
249\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
250arbitrary identifier if you're tracking multiple SVN branches/repositories in
251one git repository and want to keep them separate. See git-svn(1) for more
252information.
253 exit $exit;
255}
256sub version {
258 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
259 exit 0;
260}
261sub do_git_init_db {
263 unless (-d $ENV{GIT_DIR}) {
264 my @init_db = ('init');
265 push @init_db, "--template=$_template" if defined $_template;
266 if (defined $_shared) {
267 if ($_shared =~ /[a-z]/) {
268 push @init_db, "--shared=$_shared";
269 } else {
270 push @init_db, "--shared";
271 }
272 }
273 command_noisy(@init_db);
274 }
275 my $set;
276 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
277 foreach my $i (keys %icv) {
278 die "'$set' and '$i' cannot both be set\n" if $set;
279 next unless defined $icv{$i};
280 command_noisy('config', "$pfx.$i", $icv{$i});
281 $set = $i;
282 }
283}
284sub init_subdir {
286 my $repo_path = shift or return;
287 mkpath([$repo_path]) unless -d $repo_path;
288 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
289 $ENV{GIT_DIR} = '.git';
290}
291sub cmd_clone {
293 my ($url, $path) = @_;
294 if (!defined $path &&
295 (defined $_trunk || defined $_branches || defined $_tags) &&
296 $url !~ m#^[a-z\+]+://#) {
297 $path = $url;
298 }
299 $path = basename($url) if !defined $path || !length $path;
300 cmd_init($url, $path);
301 Git::SVN::fetch_all($Git::SVN::default_repo_id);
302}
303sub cmd_init {
305 if (defined $_trunk || defined $_branches || defined $_tags) {
306 return cmd_multi_init(@_);
307 }
308 my $url = shift or die "SVN repository location required ",
309 "as a command-line argument\n";
310 init_subdir(@_);
311 do_git_init_db();
312 Git::SVN->init($url);
314}
315sub cmd_fetch {
317 if (grep /^\d+=./, @_) {
318 die "'<rev>=<commit>' fetch arguments are ",
319 "no longer supported.\n";
320 }
321 my ($remote) = @_;
322 if (@_ > 1) {
323 die "Usage: $0 fetch [--all] [svn-remote]\n";
324 }
325 $remote ||= $Git::SVN::default_repo_id;
326 if ($_fetch_all) {
327 cmd_multi_fetch();
328 } else {
329 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
330 }
331}
332sub cmd_set_tree {
334 my (@commits) = @_;
335 if ($_stdin || !@commits) {
336 print "Reading from stdin...\n";
337 @commits = ();
338 while (<STDIN>) {
339 if (/\b($sha1_short)\b/o) {
340 unshift @commits, $1;
341 }
342 }
343 }
344 my @revs;
345 foreach my $c (@commits) {
346 my @tmp = command('rev-parse',$c);
347 if (scalar @tmp == 1) {
348 push @revs, $tmp[0];
349 } elsif (scalar @tmp > 1) {
350 push @revs, reverse(command('rev-list',@tmp));
351 } else {
352 fatal "Failed to rev-parse $c\n";
353 }
354 }
355 my $gs = Git::SVN->new;
356 my ($r_last, $cmt_last) = $gs->last_rev_commit;
357 $gs->fetch;
358 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
359 fatal "There are new revisions that were fetched ",
360 "and need to be merged (or acknowledged) ",
361 "before committing.\nlast rev: $r_last\n",
362 " current: $gs->{last_rev}\n";
363 }
364 $gs->set_tree($_) foreach @revs;
365 print "Done committing ",scalar @revs," revisions to SVN\n";
366}
367sub cmd_dcommit {
369 my $head = shift;
370 $head ||= 'HEAD';
371 my @refs;
372 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
373 unless ($gs) {
374 die "Unable to determine upstream SVN information from ",
375 "$head history\n";
376 }
377 my $c = $refs[-1];
378 my $last_rev;
379 foreach my $d (@refs) {
380 if (!verify_ref("$d~1")) {
381 fatal "Commit $d\n",
382 "has no parent commit, and therefore ",
383 "nothing to diff against.\n",
384 "You should be working from a repository ",
385 "originally created by git-svn\n";
386 }
387 unless (defined $last_rev) {
388 (undef, $last_rev, undef) = cmt_metadata("$d~1");
389 unless (defined $last_rev) {
390 fatal "Unable to extract revision information ",
391 "from commit $d~1\n";
392 }
393 }
394 if ($_dry_run) {
395 print "diff-tree $d~1 $d\n";
396 } else {
397 my %ed_opts = ( r => $last_rev,
398 log => get_commit_entry($d)->{log},
399 ra => Git::SVN::Ra->new($gs->full_url),
400 tree_a => "$d~1",
401 tree_b => $d,
402 editor_cb => sub {
403 print "Committed r$_[0]\n";
404 $last_rev = $_[0]; },
405 svn_path => '');
406 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
407 print "No changes\n$d~1 == $d\n";
408 }
409 }
410 }
411 return if $_dry_run;
412 unless ($gs) {
413 warn "Could not determine fetch information for $url\n",
414 "Will not attempt to fetch and rebase commits.\n",
415 "This probably means you have useSvmProps and should\n",
416 "now resync your SVN::Mirror repository.\n";
417 return;
418 }
419 $_fetch_all ? $gs->fetch_all : $gs->fetch;
420 unless ($_no_rebase) {
421 # we always want to rebase against the current HEAD, not any
422 # head that was passed to us
423 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
424 my @finish;
425 if (@diff) {
426 @finish = rebase_cmd();
427 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
428 "using @finish:\n", "@diff";
429 } else {
430 print "No changes between current HEAD and ",
431 $gs->refname, "\nResetting to the latest ",
432 $gs->refname, "\n";
433 @finish = qw/reset --mixed/;
434 }
435 command_noisy(@finish, $gs->refname);
436 }
437}
438sub cmd_find_rev {
440 my $revision_or_hash = shift;
441 my $result;
442 if ($revision_or_hash =~ /^r\d+$/) {
443 my $head = shift;
444 $head ||= 'HEAD';
445 my @refs;
446 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
447 unless ($gs) {
448 die "Unable to determine upstream SVN information from ",
449 "$head history\n";
450 }
451 my $desired_revision = substr($revision_or_hash, 1);
452 $result = $gs->rev_db_get($desired_revision);
453 } else {
454 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
455 $result = $rev;
456 }
457 print "$result\n" if $result;
458}
459sub cmd_rebase {
461 command_noisy(qw/update-index --refresh/);
462 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
463 unless ($gs) {
464 die "Unable to determine upstream SVN information from ",
465 "working tree history\n";
466 }
467 if (command(qw/diff-index HEAD --/)) {
468 print STDERR "Cannot rebase with uncommited changes:\n";
469 command_noisy('status');
470 exit 1;
471 }
472 unless ($_local) {
473 $_fetch_all ? $gs->fetch_all : $gs->fetch;
474 }
475 command_noisy(rebase_cmd(), $gs->refname);
476}
477sub cmd_show_ignore {
479 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
480 $gs ||= Git::SVN->new;
481 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
482 $gs->traverse_ignore(\*STDOUT, $gs->{path}, $r);
483}
484sub cmd_multi_init {
486 my $url = shift;
487 unless (defined $_trunk || defined $_branches || defined $_tags) {
488 usage(1);
489 }
490 # there are currently some bugs that prevent multi-init/multi-fetch
492 # setups from working well without this.
493 $Git::SVN::_minimize_url = 1;
494 $_prefix = '' unless defined $_prefix;
496 if (defined $url) {
497 $url =~ s#/+$##;
498 init_subdir(@_);
499 }
500 do_git_init_db();
501 if (defined $_trunk) {
502 my $trunk_ref = $_prefix . 'trunk';
503 # try both old-style and new-style lookups:
504 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
505 unless ($gs_trunk) {
506 my ($trunk_url, $trunk_path) =
507 complete_svn_url($url, $_trunk);
508 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
509 undef, $trunk_ref);
510 }
511 }
512 return unless defined $_branches || defined $_tags;
513 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
514 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
515 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
516}
517sub cmd_multi_fetch {
519 my $remotes = Git::SVN::read_all_remotes();
520 foreach my $repo_id (sort keys %$remotes) {
521 if ($remotes->{$repo_id}->{url}) {
522 Git::SVN::fetch_all($repo_id, $remotes);
523 }
524 }
525}
526# this command is special because it requires no metadata
528sub cmd_commit_diff {
529 my ($ta, $tb, $url) = @_;
530 my $usage = "Usage: $0 commit-diff -r<revision> ".
531 "<tree-ish> <tree-ish> [<URL>]\n";
532 fatal($usage) if (!defined $ta || !defined $tb);
533 my $svn_path;
534 if (!defined $url) {
535 my $gs = eval { Git::SVN->new };
536 if (!$gs) {
537 fatal("Needed URL or usable git-svn --id in ",
538 "the command-line\n", $usage);
539 }
540 $url = $gs->{url};
541 $svn_path = $gs->{path};
542 }
543 unless (defined $_revision) {
544 fatal("-r|--revision is a required argument\n", $usage);
545 }
546 if (defined $_message && defined $_file) {
547 fatal("Both --message/-m and --file/-F specified ",
548 "for the commit message.\n",
549 "I have no idea what you mean\n");
550 }
551 if (defined $_file) {
552 $_message = file_to_s($_file);
553 } else {
554 $_message ||= get_commit_entry($tb)->{log};
555 }
556 my $ra ||= Git::SVN::Ra->new($url);
557 $svn_path ||= $ra->{svn_path};
558 my $r = $_revision;
559 if ($r eq 'HEAD') {
560 $r = $ra->get_latest_revnum;
561 } elsif ($r !~ /^\d+$/) {
562 die "revision argument: $r not understood by git-svn\n";
563 }
564 my %ed_opts = ( r => $r,
565 log => $_message,
566 ra => $ra,
567 tree_a => $ta,
568 tree_b => $tb,
569 editor_cb => sub { print "Committed r$_[0]\n" },
570 svn_path => $svn_path );
571 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
572 print "No changes\n$ta == $tb\n";
573 }
574}
575########################### utility functions #########################
577sub rebase_cmd {
579 my @cmd = qw/rebase/;
580 push @cmd, '-v' if $_verbose;
581 push @cmd, qw/--merge/ if $_merge;
582 push @cmd, "--strategy=$_strategy" if $_strategy;
583 @cmd;
584}
585sub post_fetch_checkout {
587 return if $_no_checkout;
588 my $gs = $Git::SVN::_head or return;
589 return if verify_ref('refs/heads/master^0');
590 my $valid_head = verify_ref('HEAD^0');
592 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
593 return if ($valid_head || !verify_ref('HEAD^0'));
594 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
596 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
597 return if -f $index;
598 chomp(my $bare = `git config --bool --get core.bare`);
600 return if $bare eq 'true';
601 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
602 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
603 print STDERR "Checked out HEAD:\n ",
604 $gs->full_url, " r", $gs->last_rev, "\n";
605}
606sub complete_svn_url {
608 my ($url, $path) = @_;
609 $path =~ s#/+$##;
610 if ($path !~ m#^[a-z\+]+://#) {
611 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
612 fatal("E: '$path' is not a complete URL ",
613 "and a separate URL is not specified\n");
614 }
615 return ($url, $path);
616 }
617 return ($path, '');
618}
619sub complete_url_ls_init {
621 my ($ra, $repo_path, $switch, $pfx) = @_;
622 unless ($repo_path) {
623 print STDERR "W: $switch not specified\n";
624 return;
625 }
626 $repo_path =~ s#/+$##;
627 if ($repo_path =~ m#^[a-z\+]+://#) {
628 $ra = Git::SVN::Ra->new($repo_path);
629 $repo_path = '';
630 } else {
631 $repo_path =~ s#^/+##;
632 unless ($ra) {
633 fatal("E: '$repo_path' is not a complete URL ",
634 "and a separate URL is not specified\n");
635 }
636 }
637 my $url = $ra->{url};
638 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
639 my $k = "svn-remote.$gs->{repo_id}.url";
640 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
641 if ($orig_url && ($orig_url ne $gs->{url})) {
642 die "$k already set: $orig_url\n",
643 "wanted to set to: $gs->{url}\n";
644 }
645 command_oneline('config', $k, $gs->{url}) unless $orig_url;
646 my $remote_path = "$ra->{svn_path}/$repo_path/*";
647 $remote_path =~ s#/+#/#g;
648 $remote_path =~ s#^/##g;
649 my ($n) = ($switch =~ /^--(\w+)/);
650 if (length $pfx && $pfx !~ m#/$#) {
651 die "--prefix='$pfx' must have a trailing slash '/'\n";
652 }
653 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
654 "$remote_path:refs/remotes/$pfx*");
655}
656sub verify_ref {
658 my ($ref) = @_;
659 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
660 { STDERR => 0 }); };
661}
662sub get_tree_from_treeish {
664 my ($treeish) = @_;
665 # $treeish can be a symbolic ref, too:
666 my $type = command_oneline(qw/cat-file -t/, $treeish);
667 my $expected;
668 while ($type eq 'tag') {
669 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
670 }
671 if ($type eq 'commit') {
672 $expected = (grep /^tree /, command(qw/cat-file commit/,
673 $treeish))[0];
674 ($expected) = ($expected =~ /^tree ($sha1)$/o);
675 die "Unable to get tree from $treeish\n" unless $expected;
676 } elsif ($type eq 'tree') {
677 $expected = $treeish;
678 } else {
679 die "$treeish is a $type, expected tree, tag or commit\n";
680 }
681 return $expected;
682}
683sub get_commit_entry {
685 my ($treeish) = shift;
686 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
687 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
688 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
689 open my $log_fh, '>', $commit_editmsg or croak $!;
690 my $type = command_oneline(qw/cat-file -t/, $treeish);
692 if ($type eq 'commit' || $type eq 'tag') {
693 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
694 $type, $treeish);
695 my $in_msg = 0;
696 while (<$msg_fh>) {
697 if (!$in_msg) {
698 $in_msg = 1 if (/^\s*$/);
699 } elsif (/^git-svn-id: /) {
700 # skip this for now, we regenerate the
701 # correct one on re-fetch anyways
702 # TODO: set *:merge properties or like...
703 } else {
704 print $log_fh $_ or croak $!;
705 }
706 }
707 command_close_pipe($msg_fh, $ctx);
708 }
709 close $log_fh or croak $!;
710 if ($_edit || ($type eq 'tree')) {
712 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
713 # TODO: strip out spaces, comments, like git-commit.sh
714 system($editor, $commit_editmsg);
715 }
716 rename $commit_editmsg, $commit_msg or croak $!;
717 open $log_fh, '<', $commit_msg or croak $!;
718 { local $/; chomp($log_entry{log} = <$log_fh>); }
719 close $log_fh or croak $!;
720 unlink $commit_msg;
721 \%log_entry;
722}
723sub s_to_file {
725 my ($str, $file, $mode) = @_;
726 open my $fd,'>',$file or croak $!;
727 print $fd $str,"\n" or croak $!;
728 close $fd or croak $!;
729 chmod ($mode &~ umask, $file) if (defined $mode);
730}
731sub file_to_s {
733 my $file = shift;
734 open my $fd,'<',$file or croak "$!: file: $file\n";
735 local $/;
736 my $ret = <$fd>;
737 close $fd or croak $!;
738 $ret =~ s/\s*$//s;
739 return $ret;
740}
741# '<svn username> = real-name <email address>' mapping based on git-svnimport:
743sub load_authors {
744 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
745 my $log = $cmd eq 'log';
746 while (<$authors>) {
747 chomp;
748 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
749 my ($user, $name, $email) = ($1, $2, $3);
750 if ($log) {
751 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
752 } else {
753 $users{$user} = [$name, $email];
754 }
755 }
756 close $authors or croak $!;
757}
758# convert GetOpt::Long specs for use by git-config
760sub read_repo_config {
761 return unless -d $ENV{GIT_DIR};
762 my $opts = shift;
763 my @config_only;
764 foreach my $o (keys %$opts) {
765 # if we have mixedCase and a long option-only, then
766 # it's a config-only variable that we don't need for
767 # the command-line.
768 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
769 my $v = $opts->{$o};
770 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
771 $key =~ s/-//g;
772 my $arg = 'git-config';
773 $arg .= ' --int' if ($o =~ /[:=]i$/);
774 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
775 if (ref $v eq 'ARRAY') {
776 chomp(my @tmp = `$arg --get-all svn.$key`);
777 @$v = @tmp if @tmp;
778 } else {
779 chomp(my $tmp = `$arg --get svn.$key`);
780 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
781 $$v = $tmp;
782 }
783 }
784 }
785 delete @$opts{@config_only} if @config_only;
786}
787sub extract_metadata {
789 my $id = shift or return (undef, undef, undef);
790 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
791 \s([a-f\d\-]+)$/x);
792 if (!defined $rev || !$uuid || !$url) {
793 # some of the original repositories I made had
794 # identifiers like this:
795 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
796 }
797 return ($url, $rev, $uuid);
798}
799sub cmt_metadata {
801 return extract_metadata((grep(/^git-svn-id: /,
802 command(qw/cat-file commit/, shift)))[-1]);
803}
804sub working_head_info {
806 my ($head, $refs) = @_;
807 my ($fh, $ctx) = command_output_pipe('log', $head);
808 my $hash;
809 while (<$fh>) {
810 if ( m{^commit ($::sha1)$} ) {
811 unshift @$refs, $hash if $hash and $refs;
812 $hash = $1;
813 next;
814 }
815 next unless s{^\s*(git-svn-id:)}{$1};
816 my ($url, $rev, $uuid) = extract_metadata($_);
817 if (defined $url && defined $rev) {
818 if (my $gs = Git::SVN->find_by_url($url)) {
819 my $c = $gs->rev_db_get($rev);
820 if ($c && $c eq $hash) {
821 close $fh; # break the pipe
822 return ($url, $rev, $uuid, $gs);
823 }
824 }
825 }
826 }
827 command_close_pipe($fh, $ctx);
828 (undef, undef, undef, undef);
829}
830package Git::SVN;
832use strict;
833use warnings;
834use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
835 $_repack $_repack_flags $_use_svm_props $_head
836 $_use_svnsync_props $no_reuse_existing $_minimize_url/;
837use Carp qw/croak/;
838use File::Path qw/mkpath/;
839use File::Copy qw/copy/;
840use IPC::Open3;
841my $_repack_nr;
843# properties that we do not log:
844my %SKIP_PROP;
845BEGIN {
846 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
847 svn:special svn:executable
848 svn:entry:committed-rev
849 svn:entry:last-author
850 svn:entry:uuid
851 svn:entry:committed-date/;
852 # some options are read globally, but can be overridden locally
854 # per [svn-remote "..."] section. Command-line options will *NOT*
855 # override options set in an [svn-remote "..."] section
856 no strict 'refs';
857 for my $option (qw/follow_parent no_metadata use_svm_props
858 use_svnsync_props/) {
859 my $key = $option;
860 $key =~ tr/_//d;
861 my $prop = "-$option";
862 *$option = sub {
863 my ($self) = @_;
864 return $self->{$prop} if exists $self->{$prop};
865 my $k = "svn-remote.$self->{repo_id}.$key";
866 eval { command_oneline(qw/config --get/, $k) };
867 if ($@) {
868 $self->{$prop} = ${"Git::SVN::_$option"};
869 } else {
870 my $v = command_oneline(qw/config --bool/,$k);
871 $self->{$prop} = $v eq 'false' ? 0 : 1;
872 }
873 return $self->{$prop};
874 }
875 }
876}
877my %LOCKFILES;
879END { unlink keys %LOCKFILES if %LOCKFILES }
880sub resolve_local_globs {
882 my ($url, $fetch, $glob_spec) = @_;
883 return unless defined $glob_spec;
884 my $ref = $glob_spec->{ref};
885 my $path = $glob_spec->{path};
886 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
887 next unless m#^refs/remotes/$ref->{regex}$#;
888 my $p = $1;
889 my $pathname = $path->full_path($p);
890 my $refname = $ref->full_path($p);
891 if (my $existing = $fetch->{$pathname}) {
892 if ($existing ne $refname) {
893 die "Refspec conflict:\n",
894 "existing: refs/remotes/$existing\n",
895 " globbed: refs/remotes/$refname\n";
896 }
897 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
898 $u =~ s!^\Q$url\E(/|$)!! or die
899 "refs/remotes/$refname: '$url' not found in '$u'\n";
900 if ($pathname ne $u) {
901 warn "W: Refspec glob conflict ",
902 "(ref: refs/remotes/$refname):\n",
903 "expected path: $pathname\n",
904 " real path: $u\n",
905 "Continuing ahead with $u\n";
906 next;
907 }
908 } else {
909 $fetch->{$pathname} = $refname;
910 }
911 }
912}
913sub parse_revision_argument {
915 my ($base, $head) = @_;
916 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
917 return ($base, $head);
918 }
919 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
920 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
921 return ($head, $head) if ($::_revision eq 'HEAD');
922 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
923 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
924 die "revision argument: $::_revision not understood by git-svn\n";
925}
926sub fetch_all {
928 my ($repo_id, $remotes) = @_;
929 if (ref $repo_id) {
930 my $gs = $repo_id;
931 $repo_id = undef;
932 $repo_id = $gs->{repo_id};
933 }
934 $remotes ||= read_all_remotes();
935 my $remote = $remotes->{$repo_id} or
936 die "[svn-remote \"$repo_id\"] unknown\n";
937 my $fetch = $remote->{fetch};
938 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
939 my (@gs, @globs);
940 my $ra = Git::SVN::Ra->new($url);
941 my $uuid = $ra->get_uuid;
942 my $head = $ra->get_latest_revnum;
943 my $base = defined $fetch ? $head : 0;
944 # read the max revs for wildcard expansion (branches/*, tags/*)
946 foreach my $t (qw/branches tags/) {
947 defined $remote->{$t} or next;
948 push @globs, $remote->{$t};
949 my $max_rev = eval { tmp_config(qw/--int --get/,
950 "svn-remote.$repo_id.${t}-maxRev") };
951 if (defined $max_rev && ($max_rev < $base)) {
952 $base = $max_rev;
953 } elsif (!defined $max_rev) {
954 $base = 0;
955 }
956 }
957 if ($fetch) {
959 foreach my $p (sort keys %$fetch) {
960 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
961 my $lr = $gs->rev_db_max;
962 if (defined $lr) {
963 $base = $lr if ($lr < $base);
964 }
965 push @gs, $gs;
966 }
967 }
968 ($base, $head) = parse_revision_argument($base, $head);
970 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
971}
972sub read_all_remotes {
974 my $r = {};
975 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
976 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
977 $r->{$1}->{fetch}->{$2} = $3;
978 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
979 $r->{$1}->{url} = $2;
980 } elsif (m!^(.+)\.(branches|tags)=
981 (.*):refs/remotes/(.+)\s*$/!x) {
982 my ($p, $g) = ($3, $4);
983 my $rs = $r->{$1}->{$2} = {
984 t => $2,
985 remote => $1,
986 path => Git::SVN::GlobSpec->new($p),
987 ref => Git::SVN::GlobSpec->new($g) };
988 if (length($rs->{ref}->{right}) != 0) {
989 die "The '*' glob character must be the last ",
990 "character of '$g'\n";
991 }
992 }
993 }
994 $r;
995}
996sub init_vars {
998 if (defined $_repack) {
999 $_repack = 1000 if ($_repack <= 0);
1000 $_repack_nr = $_repack;
1001 $_repack_flags ||= '-d';
1002 }
1003}
1004sub verify_remotes_sanity {
1006 return unless -d $ENV{GIT_DIR};
1007 my %seen;
1008 foreach (command(qw/config -l/)) {
1009 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1010 if ($seen{$1}) {
1011 die "Remote ref refs/remote/$1 is tracked by",
1012 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1013 "Please resolve this ambiguity in ",
1014 "your git configuration file before ",
1015 "continuing\n";
1016 }
1017 $seen{$1} = $_;
1018 }
1019 }
1020}
1021# we allow more chars than remotes2config.sh...
1023sub sanitize_remote_name {
1024 my ($name) = @_;
1025 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1026 $name;
1027}
1028sub find_existing_remote {
1030 my ($url, $remotes) = @_;
1031 return undef if $no_reuse_existing;
1032 my $existing;
1033 foreach my $repo_id (keys %$remotes) {
1034 my $u = $remotes->{$repo_id}->{url} or next;
1035 next if $u ne $url;
1036 $existing = $repo_id;
1037 last;
1038 }
1039 $existing;
1040}
1041sub init_remote_config {
1043 my ($self, $url, $no_write) = @_;
1044 $url =~ s!/+$!!; # strip trailing slash
1045 my $r = read_all_remotes();
1046 my $existing = find_existing_remote($url, $r);
1047 if ($existing) {
1048 unless ($no_write) {
1049 print STDERR "Using existing ",
1050 "[svn-remote \"$existing\"]\n";
1051 }
1052 $self->{repo_id} = $existing;
1053 } elsif ($_minimize_url) {
1054 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1055 $existing = find_existing_remote($min_url, $r);
1056 if ($existing) {
1057 unless ($no_write) {
1058 print STDERR "Using existing ",
1059 "[svn-remote \"$existing\"]\n";
1060 }
1061 $self->{repo_id} = $existing;
1062 }
1063 if ($min_url ne $url) {
1064 unless ($no_write) {
1065 print STDERR "Using higher level of URL: ",
1066 "$url => $min_url\n";
1067 }
1068 my $old_path = $self->{path};
1069 $self->{path} = $url;
1070 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1071 if (length $old_path) {
1072 $self->{path} .= "/$old_path";
1073 }
1074 $url = $min_url;
1075 }
1076 }
1077 my $orig_url;
1078 if (!$existing) {
1079 # verify that we aren't overwriting anything:
1080 $orig_url = eval {
1081 command_oneline('config', '--get',
1082 "svn-remote.$self->{repo_id}.url")
1083 };
1084 if ($orig_url && ($orig_url ne $url)) {
1085 die "svn-remote.$self->{repo_id}.url already set: ",
1086 "$orig_url\nwanted to set to: $url\n";
1087 }
1088 }
1089 my ($xrepo_id, $xpath) = find_ref($self->refname);
1090 if (defined $xpath) {
1091 die "svn-remote.$xrepo_id.fetch already set to track ",
1092 "$xpath:refs/remotes/", $self->refname, "\n";
1093 }
1094 unless ($no_write) {
1095 command_noisy('config',
1096 "svn-remote.$self->{repo_id}.url", $url);
1097 command_noisy('config', '--add',
1098 "svn-remote.$self->{repo_id}.fetch",
1099 "$self->{path}:".$self->refname);
1100 }
1101 $self->{url} = $url;
1102}
1103sub find_by_url { # repos_root and, path are optional
1105 my ($class, $full_url, $repos_root, $path) = @_;
1106 return undef unless defined $full_url;
1108 remove_username($full_url);
1109 remove_username($repos_root) if defined $repos_root;
1110 my $remotes = read_all_remotes();
1111 if (defined $full_url && defined $repos_root && !defined $path) {
1112 $path = $full_url;
1113 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1114 }
1115 foreach my $repo_id (keys %$remotes) {
1116 my $u = $remotes->{$repo_id}->{url} or next;
1117 remove_username($u);
1118 next if defined $repos_root && $repos_root ne $u;
1119 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1121 foreach (qw/branches tags/) {
1122 resolve_local_globs($u, $fetch,
1123 $remotes->{$repo_id}->{$_});
1124 }
1125 my $p = $path;
1126 unless (defined $p) {
1127 $p = $full_url;
1128 $p =~ s#^\Q$u\E(?:/|$)## or next;
1129 }
1130 foreach my $f (keys %$fetch) {
1131 next if $f ne $p;
1132 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1133 }
1134 }
1135 undef;
1136}
1137sub init {
1139 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1140 my $self = _new($class, $repo_id, $ref_id, $path);
1141 if (defined $url) {
1142 $self->init_remote_config($url, $no_write);
1143 }
1144 $self;
1145}
1146sub find_ref {
1148 my ($ref_id) = @_;
1149 foreach (command(qw/config -l/)) {
1150 next unless m!^svn-remote\.(.+)\.fetch=
1151 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1152 my ($repo_id, $path, $ref) = ($1, $2, $3);
1153 if ($ref eq $ref_id) {
1154 $path = '' if ($path =~ m#^\./?#);
1155 return ($repo_id, $path);
1156 }
1157 }
1158 (undef, undef, undef);
1159}
1160sub new {
1162 my ($class, $ref_id, $repo_id, $path) = @_;
1163 if (defined $ref_id && !defined $repo_id && !defined $path) {
1164 ($repo_id, $path) = find_ref($ref_id);
1165 if (!defined $repo_id) {
1166 die "Could not find a \"svn-remote.*.fetch\" key ",
1167 "in the repository configuration matching: ",
1168 "refs/remotes/$ref_id\n";
1169 }
1170 }
1171 my $self = _new($class, $repo_id, $ref_id, $path);
1172 if (!defined $self->{path} || !length $self->{path}) {
1173 my $fetch = command_oneline('config', '--get',
1174 "svn-remote.$repo_id.fetch",
1175 ":refs/remotes/$ref_id\$") or
1176 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1177 "\":refs/remotes/$ref_id\$\" in config\n";
1178 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1179 }
1180 $self->{url} = command_oneline('config', '--get',
1181 "svn-remote.$repo_id.url") or
1182 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1183 $self->rebuild;
1184 $self;
1185}
1186sub refname { "refs/remotes/$_[0]->{ref_id}" }
1188sub svm_uuid {
1190 my ($self) = @_;
1191 return $self->{svm}->{uuid} if $self->svm;
1192 $self->ra;
1193 unless ($self->{svm}) {
1194 die "SVM UUID not cached, and reading remotely failed\n";
1195 }
1196 $self->{svm}->{uuid};
1197}
1198sub svm {
1200 my ($self) = @_;
1201 return $self->{svm} if $self->{svm};
1202 my $svm;
1203 # see if we have it in our config, first:
1204 eval {
1205 my $section = "svn-remote.$self->{repo_id}";
1206 $svm = {
1207 source => tmp_config('--get', "$section.svm-source"),
1208 uuid => tmp_config('--get', "$section.svm-uuid"),
1209 replace => tmp_config('--get', "$section.svm-replace"),
1210 }
1211 };
1212 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1213 $self->{svm} = $svm;
1214 }
1215 $self->{svm};
1216}
1217sub _set_svm_vars {
1219 my ($self, $ra) = @_;
1220 return $ra if $self->svm;
1221 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1223 "(svm:source, svm:uuid) ",
1224 "from the following URLs:\n" );
1225 sub read_svm_props {
1226 my ($self, $ra, $path, $r) = @_;
1227 my $props = ($ra->get_dir($path, $r))[2];
1228 my $src = $props->{'svm:source'};
1229 my $uuid = $props->{'svm:uuid'};
1230 return undef if (!$src || !$uuid);
1231 chomp($src, $uuid);
1233 $uuid =~ m{^[0-9a-f\-]{30,}$}
1235 or die "doesn't look right - svm:uuid is '$uuid'\n";
1236 # the '!' is used to mark the repos_root!/relative/path
1238 $src =~ s{/?!/?}{/};
1239 $src =~ s{/+$}{}; # no trailing slashes please
1240 # username is of no interest
1241 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1242 my $replace = $ra->{url};
1244 $replace .= "/$path" if length $path;
1245 my $section = "svn-remote.$self->{repo_id}";
1247 tmp_config("$section.svm-source", $src);
1248 tmp_config("$section.svm-replace", $replace);
1249 tmp_config("$section.svm-uuid", $uuid);
1250 $self->{svm} = {
1251 source => $src,
1252 uuid => $uuid,
1253 replace => $replace
1254 };
1255 }
1256 my $r = $ra->get_latest_revnum;
1258 my $path = $self->{path};
1259 my %tried;
1260 while (length $path) {
1261 unless ($tried{"$self->{url}/$path"}) {
1262 return $ra if $self->read_svm_props($ra, $path, $r);
1263 $tried{"$self->{url}/$path"} = 1;
1264 }
1265 $path =~ s#/?[^/]+$##;
1266 }
1267 die "Path: '$path' should be ''\n" if $path ne '';
1268 return $ra if $self->read_svm_props($ra, $path, $r);
1269 $tried{"$self->{url}/$path"} = 1;
1270 if ($ra->{repos_root} eq $self->{url}) {
1272 die @err, (map { " $_\n" } keys %tried), "\n";
1273 }
1274 # nope, make sure we're connected to the repository root:
1276 my $ok;
1277 my @tried_b;
1278 $path = $ra->{svn_path};
1279 $ra = Git::SVN::Ra->new($ra->{repos_root});
1280 while (length $path) {
1281 unless ($tried{"$ra->{url}/$path"}) {
1282 $ok = $self->read_svm_props($ra, $path, $r);
1283 last if $ok;
1284 $tried{"$ra->{url}/$path"} = 1;
1285 }
1286 $path =~ s#/?[^/]+$##;
1287 }
1288 die "Path: '$path' should be ''\n" if $path ne '';
1289 $ok ||= $self->read_svm_props($ra, $path, $r);
1290 $tried{"$ra->{url}/$path"} = 1;
1291 if (!$ok) {
1292 die @err, (map { " $_\n" } keys %tried), "\n";
1293 }
1294 Git::SVN::Ra->new($self->{url});
1295}
1296sub svnsync {
1298 my ($self) = @_;
1299 return $self->{svnsync} if $self->{svnsync};
1300 if ($self->no_metadata) {
1302 die "Can't have both 'noMetadata' and ",
1303 "'useSvnsyncProps' options set!\n";
1304 }
1305 if ($self->rewrite_root) {
1306 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1307 "options set!\n";
1308 }
1309 my $svnsync;
1311 # see if we have it in our config, first:
1312 eval {
1313 my $section = "svn-remote.$self->{repo_id}";
1314 $svnsync = {
1315 url => tmp_config('--get', "$section.svnsync-url"),
1316 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1317 }
1318 };
1319 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1320 return $self->{svnsync} = $svnsync;
1321 }
1322 my $err = "useSvnsyncProps set, but failed to read " .
1324 "svnsync property: svn:sync-from-";
1325 my $rp = $self->ra->rev_proplist(0);
1326 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1328 $url =~ m{^[a-z\+]+://} or
1329 die "doesn't look right - svn:sync-from-url is '$url'\n";
1330 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1332 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1333 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1334 my $section = "svn-remote.$self->{repo_id}";
1336 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1337 tmp_config('--add', "$section.svnsync-url", $url);
1338 return $self->{svnsync} = { url => $url, uuid => $uuid };
1339}
1340# this allows us to memoize our SVN::Ra UUID locally and avoid a
1342# remote lookup (useful for 'git svn log').
1343sub ra_uuid {
1344 my ($self) = @_;
1345 unless ($self->{ra_uuid}) {
1346 my $key = "svn-remote.$self->{repo_id}.uuid";
1347 my $uuid = eval { tmp_config('--get', $key) };
1348 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1349 $self->{ra_uuid} = $uuid;
1350 } else {
1351 die "ra_uuid called without URL\n" unless $self->{url};
1352 $self->{ra_uuid} = $self->ra->get_uuid;
1353 tmp_config('--add', $key, $self->{ra_uuid});
1354 }
1355 }
1356 $self->{ra_uuid};
1357}
1358sub ra {
1360 my ($self) = shift;
1361 my $ra = Git::SVN::Ra->new($self->{url});
1362 if ($self->use_svm_props && !$self->{svm}) {
1363 if ($self->no_metadata) {
1364 die "Can't have both 'noMetadata' and ",
1365 "'useSvmProps' options set!\n";
1366 } elsif ($self->use_svnsync_props) {
1367 die "Can't have both 'useSvnsyncProps' and ",
1368 "'useSvmProps' options set!\n";
1369 }
1370 $ra = $self->_set_svm_vars($ra);
1371 $self->{-want_revprops} = 1;
1372 }
1373 $ra;
1374}
1375sub rel_path {
1377 my ($self) = @_;
1378 my $repos_root = $self->ra->{repos_root};
1379 return $self->{path} if ($self->{url} eq $repos_root);
1380 my $url = $self->{url} .
1381 (length $self->{path} ? "/$self->{path}" : $self->{path});
1382 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1383 $url;
1384}
1385sub traverse_ignore {
1387 my ($self, $fh, $path, $r) = @_;
1388 $path =~ s#^/+##g;
1389 my $ra = $self->ra;
1390 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1391 my $p = $path;
1392 $p =~ s#^\Q$self->{path}\E(/|$)##;
1393 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1394 if (my $s = $props->{'svn:ignore'}) {
1395 $s =~ s/[\r\n]+/\n/g;
1396 chomp $s;
1397 if (length $p == 0) {
1398 $s =~ s#\n#\n/$p#g;
1399 print $fh "/$s\n";
1400 } else {
1401 $s =~ s#\n#\n/$p/#g;
1402 print $fh "/$p/$s\n";
1403 }
1404 }
1405 foreach (sort keys %$dirent) {
1406 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1407 $self->traverse_ignore($fh, "$path/$_", $r);
1408 }
1409}
1410sub last_rev { ($_[0]->last_rev_commit)[0] }
1412sub last_commit { ($_[0]->last_rev_commit)[1] }
1413# returns the newest SVN revision number and newest commit SHA1
1415sub last_rev_commit {
1416 my ($self) = @_;
1417 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1418 return ($self->{last_rev}, $self->{last_commit});
1419 }
1420 my $c = ::verify_ref($self->refname.'^0');
1421 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1422 my $rev = (::cmt_metadata($c))[1];
1423 if (defined $rev) {
1424 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1425 return ($rev, $c);
1426 }
1427 }
1428 my $db_path = $self->db_path;
1429 unless (-e $db_path) {
1430 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1431 return (undef, undef);
1432 }
1433 my $offset = -41; # from tail
1434 my $rl;
1435 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1436 sysseek($fh, $offset, 2); # don't care for errors
1437 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1438 chomp $rl;
1439 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1440 $offset -= 41;
1441 sysseek($fh, $offset, 2); # don't care for errors
1442 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1443 chomp $rl;
1444 }
1445 if ($c && $c ne $rl) {
1446 die "$db_path and ", $self->refname,
1447 " inconsistent!:\n$c != $rl\n";
1448 }
1449 my $rev = sysseek($fh, 0, 1) or croak $!;
1450 $rev = ($rev - 41) / 41;
1451 close $fh or croak $!;
1452 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1453 return ($rev, $c);
1454}
1455sub get_fetch_range {
1457 my ($self, $min, $max) = @_;
1458 $max ||= $self->ra->get_latest_revnum;
1459 $min ||= $self->rev_db_max;
1460 (++$min, $max);
1461}
1462sub tmp_config {
1464 my (@args) = @_;
1465 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1466 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1467 if (! -f $config && -f $old_def_config) {
1468 rename $old_def_config, $config or
1469 die "Failed rename $old_def_config => $config: $!\n";
1470 }
1471 my $old_config = $ENV{GIT_CONFIG};
1472 $ENV{GIT_CONFIG} = $config;
1473 $@ = undef;
1474 my @ret = eval {
1475 unless (-f $config) {
1476 mkfile($config);
1477 open my $fh, '>', $config or
1478 die "Can't open $config: $!\n";
1479 print $fh "; This file is used internally by ",
1480 "git-svn\n" or die
1481 "Couldn't write to $config: $!\n";
1482 print $fh "; You should not have to edit it\n" or
1483 die "Couldn't write to $config: $!\n";
1484 close $fh or die "Couldn't close $config: $!\n";
1485 }
1486 command('config', @args);
1487 };
1488 my $err = $@;
1489 if (defined $old_config) {
1490 $ENV{GIT_CONFIG} = $old_config;
1491 } else {
1492 delete $ENV{GIT_CONFIG};
1493 }
1494 die $err if $err;
1495 wantarray ? @ret : $ret[0];
1496}
1497sub tmp_index_do {
1499 my ($self, $sub) = @_;
1500 my $old_index = $ENV{GIT_INDEX_FILE};
1501 $ENV{GIT_INDEX_FILE} = $self->{index};
1502 $@ = undef;
1503 my @ret = eval {
1504 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1505 mkpath([$dir]) unless -d $dir;
1506 &$sub;
1507 };
1508 my $err = $@;
1509 if (defined $old_index) {
1510 $ENV{GIT_INDEX_FILE} = $old_index;
1511 } else {
1512 delete $ENV{GIT_INDEX_FILE};
1513 }
1514 die $err if $err;
1515 wantarray ? @ret : $ret[0];
1516}
1517sub assert_index_clean {
1519 my ($self, $treeish) = @_;
1520 $self->tmp_index_do(sub {
1522 command_noisy('read-tree', $treeish) unless -e $self->{index};
1523 my $x = command_oneline('write-tree');
1524 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1525 /^tree ($::sha1)/mo);
1526 return if $y eq $x;
1527 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1529 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1530 command_noisy('read-tree', $treeish);
1531 $x = command_oneline('write-tree');
1532 if ($y ne $x) {
1533 ::fatal "trees ($treeish) $y != $x\n",
1534 "Something is seriously wrong...\n";
1535 }
1536 });
1537}
1538sub get_commit_parents {
1540 my ($self, $log_entry) = @_;
1541 my (%seen, @ret, @tmp);
1542 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1543 if (my $ip = $self->{inject_parents}) {
1544 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1545 push @tmp, $commit;
1546 }
1547 }
1548 if (my $cur = ::verify_ref($self->refname.'^0')) {
1549 push @tmp, $cur;
1550 }
1551 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1552 while (my $p = shift @tmp) {
1553 next if $seen{$p};
1554 $seen{$p} = 1;
1555 push @ret, $p;
1556 # MAXPARENT is defined to 16 in commit-tree.c:
1557 last if @ret >= 16;
1558 }
1559 if (@tmp) {
1560 die "r$log_entry->{revision}: No room for parents:\n\t",
1561 join("\n\t", @tmp), "\n";
1562 }
1563 @ret;
1564}
1565sub rewrite_root {
1567 my ($self) = @_;
1568 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1569 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1570 my $rwr = eval { command_oneline(qw/config --get/, $k) };
1571 if ($rwr) {
1572 $rwr =~ s#/+$##;
1573 if ($rwr !~ m#^[a-z\+]+://#) {
1574 die "$rwr is not a valid URL (key: $k)\n";
1575 }
1576 }
1577 $self->{-rewrite_root} = $rwr;
1578}
1579sub metadata_url {
1581 my ($self) = @_;
1582 ($self->rewrite_root || $self->{url}) .
1583 (length $self->{path} ? '/' . $self->{path} : '');
1584}
1585sub full_url {
1587 my ($self) = @_;
1588 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1589}
1590sub do_git_commit {
1592 my ($self, $log_entry) = @_;
1593 my $lr = $self->last_rev;
1594 if (defined $lr && $lr >= $log_entry->{revision}) {
1595 die "Last fetched revision of ", $self->refname,
1596 " was r$lr, but we are about to fetch: ",
1597 "r$log_entry->{revision}!\n";
1598 }
1599 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1600 croak "$log_entry->{revision} = $c already exists! ",
1601 "Why are we refetching it?\n";
1602 }
1603 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1604 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1605 $log_entry->{email};
1606 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1607 my $tree = $log_entry->{tree};
1609 if (!defined $tree) {
1610 $tree = $self->tmp_index_do(sub {
1611 command_oneline('write-tree') });
1612 }
1613 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1614 my @exec = ('git-commit-tree', $tree);
1616 foreach ($self->get_commit_parents($log_entry)) {
1617 push @exec, '-p', $_;
1618 }
1619 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1620 or croak $!;
1621 print $msg_fh $log_entry->{log} or croak $!;
1622 unless ($self->no_metadata) {
1623 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1624 or croak $!;
1625 }
1626 $msg_fh->flush == 0 or croak $!;
1627 close $msg_fh or croak $!;
1628 chomp(my $commit = do { local $/; <$out_fh> });
1629 close $out_fh or croak $!;
1630 waitpid $pid, 0;
1631 croak $? if $?;
1632 if ($commit !~ /^$::sha1$/o) {
1633 die "Failed to commit, invalid sha1: $commit\n";
1634 }
1635 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1637 $self->{last_rev} = $log_entry->{revision};
1639 $self->{last_commit} = $commit;
1640 print "r$log_entry->{revision}";
1641 if (defined $log_entry->{svm_revision}) {
1642 print " (\@$log_entry->{svm_revision})";
1643 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1644 0, $self->svm_uuid);
1645 }
1646 print " = $commit ($self->{ref_id})\n";
1647 if (defined $_repack && (--$_repack_nr == 0)) {
1648 $_repack_nr = $_repack;
1649 # repack doesn't use any arguments with spaces in them, does it?
1650 print "Running git repack $_repack_flags ...\n";
1651 command_noisy('repack', split(/\s+/, $_repack_flags));
1652 print "Done repacking\n";
1653 }
1654 return $commit;
1655}
1656sub match_paths {
1658 my ($self, $paths, $r) = @_;
1659 return 1 if $self->{path} eq '';
1660 if (my $path = $paths->{"/$self->{path}"}) {
1661 return ($path->{action} eq 'D') ? 0 : 1;
1662 }
1663 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1664 if (grep /$self->{path_regex}/, keys %$paths) {
1665 return 1;
1666 }
1667 my $c = '';
1668 foreach (split m#/#, $self->{path}) {
1669 $c .= "/$_";
1670 next unless ($paths->{$c} &&
1671 ($paths->{$c}->{action} =~ /^[AR]$/));
1672 if ($self->ra->check_path($self->{path}, $r) ==
1673 $SVN::Node::dir) {
1674 return 1;
1675 }
1676 }
1677 return 0;
1678}
1679sub find_parent_branch {
1681 my ($self, $paths, $rev) = @_;
1682 return undef unless $self->follow_parent;
1683 unless (defined $paths) {
1684 my $err_handler = $SVN::Error::handler;
1685 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1686 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1687 $paths =
1688 Git::SVN::Ra::dup_changed_paths($_[0]) });
1689 $SVN::Error::handler = $err_handler;
1690 }
1691 return undef unless defined $paths;
1692 # look for a parent from another branch:
1694 my @b_path_components = split m#/#, $self->rel_path;
1695 my @a_path_components;
1696 my $i;
1697 while (@b_path_components) {
1698 $i = $paths->{'/'.join('/', @b_path_components)};
1699 last if $i && defined $i->{copyfrom_path};
1700 unshift(@a_path_components, pop(@b_path_components));
1701 }
1702 return undef unless defined $i && defined $i->{copyfrom_path};
1703 my $branch_from = $i->{copyfrom_path};
1704 if (@a_path_components) {
1705 print STDERR "branch_from: $branch_from => ";
1706 $branch_from .= '/'.join('/', @a_path_components);
1707 print STDERR $branch_from, "\n";
1708 }
1709 my $r = $i->{copyfrom_rev};
1710 my $repos_root = $self->ra->{repos_root};
1711 my $url = $self->ra->{url};
1712 my $new_url = $repos_root . $branch_from;
1713 print STDERR "Found possible branch point: ",
1714 "$new_url => ", $self->full_url, ", $r\n";
1715 $branch_from =~ s#^/##;
1716 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1717 unless ($gs) {
1718 my $ref_id = $self->{ref_id};
1719 $ref_id =~ s/\@\d+$//;
1720 $ref_id .= "\@$r";
1721 # just grow a tail if we're not unique enough :x
1722 $ref_id .= '-' while find_ref($ref_id);
1723 print STDERR "Initializing parent: $ref_id\n";
1724 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1725 }
1726 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1727 if (!defined $r0 || !defined $parent) {
1728 my ($base, $head) = parse_revision_argument(0, $r);
1729 if ($base <= $r) {
1730 $gs->fetch($base, $r);
1731 }
1732 ($r0, $parent) = $gs->last_rev_commit;
1733 }
1734 if (defined $r0 && defined $parent) {
1735 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1736 my $ed;
1737 if ($self->ra->can_do_switch) {
1738 $self->assert_index_clean($parent);
1739 print STDERR "Following parent with do_switch\n";
1740 # do_switch works with svn/trunk >= r22312, but that
1741 # is not included with SVN 1.4.3 (the latest version
1742 # at the moment), so we can't rely on it
1743 $self->{last_commit} = $parent;
1744 $ed = SVN::Git::Fetcher->new($self);
1745 $gs->ra->gs_do_switch($r0, $rev, $gs,
1746 $self->full_url, $ed)
1747 or die "SVN connection failed somewhere...\n";
1748 } else {
1749 print STDERR "Following parent with do_update\n";
1750 $ed = SVN::Git::Fetcher->new($self);
1751 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1752 or die "SVN connection failed somewhere...\n";
1753 }
1754 print STDERR "Successfully followed parent\n";
1755 return $self->make_log_entry($rev, [$parent], $ed);
1756 }
1757 return undef;
1758}
1759sub do_fetch {
1761 my ($self, $paths, $rev) = @_;
1762 my $ed;
1763 my ($last_rev, @parents);
1764 if (my $lc = $self->last_commit) {
1765 # we can have a branch that was deleted, then re-added
1766 # under the same name but copied from another path, in
1767 # which case we'll have multiple parents (we don't
1768 # want to break the original ref, nor lose copypath info):
1769 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1770 push @{$log_entry->{parents}}, $lc;
1771 return $log_entry;
1772 }
1773 $ed = SVN::Git::Fetcher->new($self);
1774 $last_rev = $self->{last_rev};
1775 $ed->{c} = $lc;
1776 @parents = ($lc);
1777 } else {
1778 $last_rev = $rev;
1779 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1780 return $log_entry;
1781 }
1782 $ed = SVN::Git::Fetcher->new($self);
1783 }
1784 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1785 die "SVN connection failed somewhere...\n";
1786 }
1787 $self->make_log_entry($rev, \@parents, $ed);
1788}
1789sub get_untracked {
1791 my ($self, $ed) = @_;
1792 my @out;
1793 my $h = $ed->{empty};
1794 foreach (sort keys %$h) {
1795 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1796 push @out, " $act: " . uri_encode($_);
1797 warn "W: $act: $_\n";
1798 }
1799 foreach my $t (qw/dir_prop file_prop/) {
1800 $h = $ed->{$t} or next;
1801 foreach my $path (sort keys %$h) {
1802 my $ppath = $path eq '' ? '.' : $path;
1803 foreach my $prop (sort keys %{$h->{$path}}) {
1804 next if $SKIP_PROP{$prop};
1805 my $v = $h->{$path}->{$prop};
1806 my $t_ppath_prop = "$t: " .
1807 uri_encode($ppath) . ' ' .
1808 uri_encode($prop);
1809 if (defined $v) {
1810 push @out, " +$t_ppath_prop " .
1811 uri_encode($v);
1812 } else {
1813 push @out, " -$t_ppath_prop";
1814 }
1815 }
1816 }
1817 }
1818 foreach my $t (qw/absent_file absent_directory/) {
1819 $h = $ed->{$t} or next;
1820 foreach my $parent (sort keys %$h) {
1821 foreach my $path (sort @{$h->{$parent}}) {
1822 push @out, " $t: " .
1823 uri_encode("$parent/$path");
1824 warn "W: $t: $parent/$path ",
1825 "Insufficient permissions?\n";
1826 }
1827 }
1828 }
1829 \@out;
1830}
1831sub parse_svn_date {
1833 my $date = shift || return '+0000 1970-01-01 00:00:00';
1834 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1835 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1836 croak "Unable to parse date: $date\n";
1837 "+0000 $Y-$m-$d $H:$M:$S";
1838}
1839sub check_author {
1841 my ($author) = @_;
1842 if (!defined $author || length $author == 0) {
1843 $author = '(no author)';
1844 }
1845 if (defined $::_authors && ! defined $::users{$author}) {
1846 die "Author: $author not defined in $::_authors file\n";
1847 }
1848 $author;
1849}
1850sub make_log_entry {
1852 my ($self, $rev, $parents, $ed) = @_;
1853 my $untracked = $self->get_untracked($ed);
1854 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1856 print $un "r$rev\n" or croak $!;
1857 print $un $_, "\n" foreach @$untracked;
1858 my %log_entry = ( parents => $parents || [], revision => $rev,
1859 log => '');
1860 my $headrev;
1862 my $logged = delete $self->{logged_rev_props};
1863 if (!$logged || $self->{-want_revprops}) {
1864 my $rp = $self->ra->rev_proplist($rev);
1865 foreach (sort keys %$rp) {
1866 my $v = $rp->{$_};
1867 if (/^svn:(author|date|log)$/) {
1868 $log_entry{$1} = $v;
1869 } elsif ($_ eq 'svm:headrev') {
1870 $headrev = $v;
1871 } else {
1872 print $un " rev_prop: ", uri_encode($_), ' ',
1873 uri_encode($v), "\n";
1874 }
1875 }
1876 } else {
1877 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1878 }
1879 close $un or croak $!;
1880 $log_entry{date} = parse_svn_date($log_entry{date});
1882 $log_entry{log} .= "\n";
1883 my $author = $log_entry{author} = check_author($log_entry{author});
1884 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1885 : ($author, undef);
1886 if (defined $headrev && $self->use_svm_props) {
1887 if ($self->rewrite_root) {
1888 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1889 "options set!\n";
1890 }
1891 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1892 # we don't want "SVM: initializing mirror for junk" ...
1893 return undef if $r == 0;
1894 my $svm = $self->svm;
1895 if ($uuid ne $svm->{uuid}) {
1896 die "UUID mismatch on SVM path:\n",
1897 "expected: $svm->{uuid}\n",
1898 " got: $uuid\n";
1899 }
1900 my $full_url = $self->full_url;
1901 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1902 die "Failed to replace '$svm->{replace}' with ",
1903 "'$svm->{source}' in $full_url\n";
1904 # throw away username for storing in records
1905 remove_username($full_url);
1906 $log_entry{metadata} = "$full_url\@$r $uuid";
1907 $log_entry{svm_revision} = $r;
1908 $email ||= "$author\@$uuid"
1909 } elsif ($self->use_svnsync_props) {
1910 my $full_url = $self->svnsync->{url};
1911 $full_url .= "/$self->{path}" if length $self->{path};
1912 remove_username($full_url);
1913 my $uuid = $self->svnsync->{uuid};
1914 $log_entry{metadata} = "$full_url\@$rev $uuid";
1915 $email ||= "$author\@$uuid"
1916 } else {
1917 my $url = $self->metadata_url;
1918 remove_username($url);
1919 $log_entry{metadata} = "$url\@$rev " .
1920 $self->ra->get_uuid;
1921 $email ||= "$author\@" . $self->ra->get_uuid;
1922 }
1923 $log_entry{name} = $name;
1924 $log_entry{email} = $email;
1925 \%log_entry;
1926}
1927sub fetch {
1929 my ($self, $min_rev, $max_rev, @parents) = @_;
1930 my ($last_rev, $last_commit) = $self->last_rev_commit;
1931 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1932 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1933}
1934sub set_tree_cb {
1936 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1937 $self->{inject_parents} = { $rev => $tree };
1938 $self->fetch(undef, undef);
1939}
1940sub set_tree {
1942 my ($self, $tree) = (shift, shift);
1943 my $log_entry = ::get_commit_entry($tree);
1944 unless ($self->{last_rev}) {
1945 fatal("Must have an existing revision to commit\n");
1946 }
1947 my %ed_opts = ( r => $self->{last_rev},
1948 log => $log_entry->{log},
1949 ra => $self->ra,
1950 tree_a => $self->{last_commit},
1951 tree_b => $tree,
1952 editor_cb => sub {
1953 $self->set_tree_cb($log_entry, $tree, @_) },
1954 svn_path => $self->{path} );
1955 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1956 print "No changes\nr$self->{last_rev} = $tree\n";
1957 }
1958}
1959sub rebuild {
1961 my ($self) = @_;
1962 my $db_path = $self->db_path;
1963 return if (-e $db_path && ! -z $db_path);
1964 return unless ::verify_ref($self->refname.'^0');
1965 if (-f $self->{db_root}) {
1966 rename $self->{db_root}, $db_path or die
1967 "rename $self->{db_root} => $db_path failed: $!\n";
1968 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1969 symlink $base, $self->{db_root} or die
1970 "symlink $base => $self->{db_root} failed: $!\n";
1971 return;
1972 }
1973 print "Rebuilding $db_path ...\n";
1974 my ($log, $ctx) = command_output_pipe("log", $self->refname);
1975 my $latest;
1976 my $full_url = $self->full_url;
1977 remove_username($full_url);
1978 my $svn_uuid;
1979 my $c;
1980 while (<$log>) {
1981 if ( m{^commit ($::sha1)$} ) {
1982 $c = $1;
1983 next;
1984 }
1985 next unless s{^\s*(git-svn-id:)}{$1};
1986 my ($url, $rev, $uuid) = ::extract_metadata($_);
1987 remove_username($url);
1988 # ignore merges (from set-tree)
1990 next if (!defined $rev || !$uuid);
1991 # if we merged or otherwise started elsewhere, this is
1993 # how we break out of it
1994 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1995 ($full_url && $url && ($url ne $full_url))) {
1996 next;
1997 }
1998 $latest ||= $rev;
1999 $svn_uuid ||= $uuid;
2000 $self->rev_db_set($rev, $c);
2002 print "r$rev = $c\n";
2003 }
2004 command_close_pipe($log, $ctx);
2005 print "Done rebuilding $db_path\n";
2006}
2007# rev_db:
2009# Tie::File seems to be prone to offset errors if revisions get sparse,
2010# it's not that fast, either. Tie::File is also not in Perl 5.6. So
2011# one of my favorite modules is out :< Next up would be one of the DBM
2012# modules, but I'm not sure which is most portable... So I'll just
2013# go with something that's plain-text, but still capable of
2014# being randomly accessed. So here's my ultra-simple fixed-width
2015# database. All records are 40 characters + "\n", so it's easy to seek
2016# to a revision: (41 * rev) is the byte offset.
2017# A record of 40 0s denotes an empty revision.
2018# And yes, it's still pretty fast (faster than Tie::File).
2019# These files are disposable unless noMetadata or useSvmProps is set
2020sub _rev_db_set {
2022 my ($fh, $rev, $commit) = @_;
2023 my $offset = $rev * 41;
2024 # assume that append is the common case:
2025 seek $fh, 0, 2 or croak $!;
2026 my $pos = tell $fh;
2027 if ($pos < $offset) {
2028 for (1 .. (($offset - $pos) / 41)) {
2029 print $fh (('0' x 40),"\n") or croak $!;
2030 }
2031 }
2032 seek $fh, $offset, 0 or croak $!;
2033 print $fh $commit,"\n" or croak $!;
2034}
2035sub mkfile {
2037 my ($path) = @_;
2038 unless (-e $path) {
2039 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2040 mkpath([$dir]) unless -d $dir;
2041 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2042 close $fh or die "Couldn't close (create) $path: $!\n";
2043 }
2044}
2045sub rev_db_set {
2047 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2048 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2049 my $db = $self->db_path($uuid);
2050 my $db_lock = "$db.lock";
2051 my $sig;
2052 if ($update_ref) {
2053 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2054 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2055 }
2056 mkfile($db);
2057 $LOCKFILES{$db_lock} = 1;
2059 my $sync;
2060 # both of these options make our .rev_db file very, very important
2061 # and we can't afford to lose it because rebuild() won't work
2062 if ($self->use_svm_props || $self->no_metadata) {
2063 $sync = 1;
2064 copy($db, $db_lock) or die "rev_db_set(@_): ",
2065 "Failed to copy: ",
2066 "$db => $db_lock ($!)\n";
2067 } else {
2068 rename $db, $db_lock or die "rev_db_set(@_): ",
2069 "Failed to rename: ",
2070 "$db => $db_lock ($!)\n";
2071 }
2072 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2073 _rev_db_set($fh, $rev, $commit);
2074 if ($sync) {
2075 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2076 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2077 }
2078 close $fh or croak $!;
2079 if ($update_ref) {
2080 $_head = $self;
2081 command_noisy('update-ref', '-m', "r$rev",
2082 $self->refname, $commit);
2083 }
2084 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2085 "$db_lock => $db ($!)\n";
2086 delete $LOCKFILES{$db_lock};
2087 if ($update_ref) {
2088 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2089 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2090 kill $sig, $$ if defined $sig;
2091 }
2092}
2093sub rev_db_max {
2095 my ($self) = @_;
2096 $self->rebuild;
2097 my $db_path = $self->db_path;
2098 my @stat = stat $db_path or return 0;
2099 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2100 my $max = $stat[7] / 41;
2101 (($max > 0) ? $max - 1 : 0);
2102}
2103sub rev_db_get {
2105 my ($self, $rev, $uuid) = @_;
2106 my $ret;
2107 my $offset = $rev * 41;
2108 my $db_path = $self->db_path($uuid);
2109 return undef unless -e $db_path;
2110 open my $fh, '<', $db_path or croak $!;
2111 if (sysseek($fh, $offset, 0) == $offset) {
2112 my $read = sysread($fh, $ret, 40);
2113 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2114 }
2115 close $fh or croak $!;
2116 $ret;
2117}
2118sub find_rev_before {
2120 my ($self, $rev, $eq_ok) = @_;
2121 --$rev unless $eq_ok;
2122 while ($rev > 0) {
2123 if (my $c = $self->rev_db_get($rev)) {
2124 return ($rev, $c);
2125 }
2126 --$rev;
2127 }
2128 return (undef, undef);
2129}
2130sub _new {
2132 my ($class, $repo_id, $ref_id, $path) = @_;
2133 unless (defined $repo_id && length $repo_id) {
2134 $repo_id = $Git::SVN::default_repo_id;
2135 }
2136 unless (defined $ref_id && length $ref_id) {
2137 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2138 }
2139 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2140 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2141 $_[3] = $path = '' unless (defined $path);
2142 mkpath(["$ENV{GIT_DIR}/svn"]);
2143 bless {
2144 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2145 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2146 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2147}
2148sub db_path {
2150 my ($self, $uuid) = @_;
2151 $uuid ||= $self->ra_uuid;
2152 "$self->{db_root}.$uuid";
2153}
2154sub uri_encode {
2156 my ($f) = @_;
2157 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2158 $f
2159}
2160sub remove_username {
2162 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2163}
2164package Git::SVN::Prompt;
2166use strict;
2167use warnings;
2168require SVN::Core;
2169use vars qw/$_no_auth_cache $_username/;
2170sub simple {
2172 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2173 $may_save = undef if $_no_auth_cache;
2174 $default_username = $_username if defined $_username;
2175 if (defined $default_username && length $default_username) {
2176 if (defined $realm && length $realm) {
2177 print STDERR "Authentication realm: $realm\n";
2178 STDERR->flush;
2179 }
2180 $cred->username($default_username);
2181 } else {
2182 username($cred, $realm, $may_save, $pool);
2183 }
2184 $cred->password(_read_password("Password for '" .
2185 $cred->username . "': ", $realm));
2186 $cred->may_save($may_save);
2187 $SVN::_Core::SVN_NO_ERROR;
2188}
2189sub ssl_server_trust {
2191 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2192 $may_save = undef if $_no_auth_cache;
2193 print STDERR "Error validating server certificate for '$realm':\n";
2194 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2195 print STDERR " - The certificate is not issued by a trusted ",
2196 "authority. Use the\n",
2197 " fingerprint to validate the certificate manually!\n";
2198 }
2199 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2200 print STDERR " - The certificate hostname does not match.\n";
2201 }
2202 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2203 print STDERR " - The certificate is not yet valid.\n";
2204 }
2205 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2206 print STDERR " - The certificate has expired.\n";
2207 }
2208 if ($failures & $SVN::Auth::SSL::OTHER) {
2209 print STDERR " - The certificate has an unknown error.\n";
2210 }
2211 printf STDERR
2212 "Certificate information:\n".
2213 " - Hostname: %s\n".
2214 " - Valid: from %s until %s\n".
2215 " - Issuer: %s\n".
2216 " - Fingerprint: %s\n",
2217 map $cert_info->$_, qw(hostname valid_from valid_until
2218 issuer_dname fingerprint);
2219 my $choice;
2220prompt:
2221 print STDERR $may_save ?
2222 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2223 "(R)eject or accept (t)emporarily? ";
2224 STDERR->flush;
2225 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2226 if ($choice =~ /^t$/i) {
2227 $cred->may_save(undef);
2228 } elsif ($choice =~ /^r$/i) {
2229 return -1;
2230 } elsif ($may_save && $choice =~ /^p$/i) {
2231 $cred->may_save($may_save);
2232 } else {
2233 goto prompt;
2234 }
2235 $cred->accepted_failures($failures);
2236 $SVN::_Core::SVN_NO_ERROR;
2237}
2238sub ssl_client_cert {
2240 my ($cred, $realm, $may_save, $pool) = @_;
2241 $may_save = undef if $_no_auth_cache;
2242 print STDERR "Client certificate filename: ";
2243 STDERR->flush;
2244 chomp(my $filename = <STDIN>);
2245 $cred->cert_file($filename);
2246 $cred->may_save($may_save);
2247 $SVN::_Core::SVN_NO_ERROR;
2248}
2249sub ssl_client_cert_pw {
2251 my ($cred, $realm, $may_save, $pool) = @_;
2252 $may_save = undef if $_no_auth_cache;
2253 $cred->password(_read_password("Password: ", $realm));
2254 $cred->may_save($may_save);
2255 $SVN::_Core::SVN_NO_ERROR;
2256}
2257sub username {
2259 my ($cred, $realm, $may_save, $pool) = @_;
2260 $may_save = undef if $_no_auth_cache;
2261 if (defined $realm && length $realm) {
2262 print STDERR "Authentication realm: $realm\n";
2263 }
2264 my $username;
2265 if (defined $_username) {
2266 $username = $_username;
2267 } else {
2268 print STDERR "Username: ";
2269 STDERR->flush;
2270 chomp($username = <STDIN>);
2271 }
2272 $cred->username($username);
2273 $cred->may_save($may_save);
2274 $SVN::_Core::SVN_NO_ERROR;
2275}
2276sub _read_password {
2278 my ($prompt, $realm) = @_;
2279 print STDERR $prompt;
2280 STDERR->flush;
2281 require Term::ReadKey;
2282 Term::ReadKey::ReadMode('noecho');
2283 my $password = '';
2284 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2285 last if $key =~ /[\012\015]/; # \n\r
2286 $password .= $key;
2287 }
2288 Term::ReadKey::ReadMode('restore');
2289 print STDERR "\n";
2290 STDERR->flush;
2291 $password;
2292}
2293package main;
2295{
2297 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2298 $SVN::Node::dir.$SVN::Node::unknown.
2299 $SVN::Node::none.$SVN::Node::file.
2300 $SVN::Node::dir.$SVN::Node::unknown.
2301 $SVN::Auth::SSL::CNMISMATCH.
2302 $SVN::Auth::SSL::NOTYETVALID.
2303 $SVN::Auth::SSL::EXPIRED.
2304 $SVN::Auth::SSL::UNKNOWNCA.
2305 $SVN::Auth::SSL::OTHER;
2306}
2307package SVN::Git::Fetcher;
2309use vars qw/@ISA/;
2310use strict;
2311use warnings;
2312use Carp qw/croak/;
2313use IO::File qw//;
2314use Digest::MD5;
2315# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2317sub new {
2318 my ($class, $git_svn) = @_;
2319 my $self = SVN::Delta::Editor->new;
2320 bless $self, $class;
2321 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2322 $self->{empty} = {};
2323 $self->{dir_prop} = {};
2324 $self->{file_prop} = {};
2325 $self->{absent_dir} = {};
2326 $self->{absent_file} = {};
2327 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2328 $self;
2329}
2330sub set_path_strip {
2332 my ($self, $path) = @_;
2333 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2334}
2335sub open_root {
2337 { path => '' };
2338}
2339sub open_directory {
2341 my ($self, $path, $pb, $rev) = @_;
2342 { path => $path };
2343}
2344sub git_path {
2346 my ($self, $path) = @_;
2347 if ($self->{path_strip}) {
2348 $path =~ s!$self->{path_strip}!! or
2349 die "Failed to strip path '$path' ($self->{path_strip})\n";
2350 }
2351 $path;
2352}
2353sub delete_entry {
2355 my ($self, $path, $rev, $pb) = @_;
2356 my $gpath = $self->git_path($path);
2358 return undef if ($gpath eq '');
2359 # remove entire directories.
2361 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2362 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2363 -r --name-only -z/,
2364 $self->{c}, '--', $gpath);
2365 local $/ = "\0";
2366 while (<$ls>) {
2367 chomp;
2368 $self->{gii}->remove($_);
2369 print "\tD\t$_\n" unless $::_q;
2370 }
2371 print "\tD\t$gpath/\n" unless $::_q;
2372 command_close_pipe($ls, $ctx);
2373 $self->{empty}->{$path} = 0
2374 } else {
2375 $self->{gii}->remove($gpath);
2376 print "\tD\t$gpath\n" unless $::_q;
2377 }
2378 undef;
2379}
2380sub open_file {
2382 my ($self, $path, $pb, $rev) = @_;
2383 my $gpath = $self->git_path($path);
2384 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2385 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2386 unless (defined $mode && defined $blob) {
2387 die "$path was not found in commit $self->{c} (r$rev)\n";
2388 }
2389 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2390 pool => SVN::Pool->new, action => 'M' };
2391}
2392sub add_file {
2394 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2395 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2396 delete $self->{empty}->{$dir};
2397 { path => $path, mode_a => 100644, mode_b => 100644,
2398 pool => SVN::Pool->new, action => 'A' };
2399}
2400sub add_directory {
2402 my ($self, $path, $cp_path, $cp_rev) = @_;
2403 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2404 delete $self->{empty}->{$dir};
2405 $self->{empty}->{$path} = 1;
2406 { path => $path };
2407}
2408sub change_dir_prop {
2410 my ($self, $db, $prop, $value) = @_;
2411 $self->{dir_prop}->{$db->{path}} ||= {};
2412 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2413 undef;
2414}
2415sub absent_directory {
2417 my ($self, $path, $pb) = @_;
2418 $self->{absent_dir}->{$pb->{path}} ||= [];
2419 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2420 undef;
2421}
2422sub absent_file {
2424 my ($self, $path, $pb) = @_;
2425 $self->{absent_file}->{$pb->{path}} ||= [];
2426 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2427 undef;
2428}
2429sub change_file_prop {
2431 my ($self, $fb, $prop, $value) = @_;
2432 if ($prop eq 'svn:executable') {
2433 if ($fb->{mode_b} != 120000) {
2434 $fb->{mode_b} = defined $value ? 100755 : 100644;
2435 }
2436 } elsif ($prop eq 'svn:special') {
2437 $fb->{mode_b} = defined $value ? 120000 : 100644;
2438 } else {
2439 $self->{file_prop}->{$fb->{path}} ||= {};
2440 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2441 }
2442 undef;
2443}
2444sub apply_textdelta {
2446 my ($self, $fb, $exp) = @_;
2447 my $fh = IO::File->new_tmpfile;
2448 $fh->autoflush(1);
2449 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2450 # (but $base does not,) so dup() it for reading in close_file
2451 open my $dup, '<&', $fh or croak $!;
2452 my $base = IO::File->new_tmpfile;
2453 $base->autoflush(1);
2454 if ($fb->{blob}) {
2455 defined (my $pid = fork) or croak $!;
2456 if (!$pid) {
2457 open STDOUT, '>&', $base or croak $!;
2458 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2459 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2460 }
2461 waitpid $pid, 0;
2462 croak $? if $?;
2463 if (defined $exp) {
2465 seek $base, 0, 0 or croak $!;
2466 my $md5 = Digest::MD5->new;
2467 $md5->addfile($base);
2468 my $got = $md5->hexdigest;
2469 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2470 "expected: $exp\n",
2471 " got: $got\n" if ($got ne $exp);
2472 }
2473 }
2474 seek $base, 0, 0 or croak $!;
2475 $fb->{fh} = $dup;
2476 $fb->{base} = $base;
2477 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2478}
2479sub close_file {
2481 my ($self, $fb, $exp) = @_;
2482 my $hash;
2483 my $path = $self->git_path($fb->{path});
2484 if (my $fh = $fb->{fh}) {
2485 if (defined $exp) {
2486 seek($fh, 0, 0) or croak $!;
2487 my $md5 = Digest::MD5->new;
2488 $md5->addfile($fh);
2489 my $got = $md5->hexdigest;
2490 if ($got ne $exp) {
2491 die "Checksum mismatch: $path\n",
2492 "expected: $exp\n got: $got\n";
2493 }
2494 }
2495 sysseek($fh, 0, 0) or croak $!;
2496 if ($fb->{mode_b} == 120000) {
2497 sysread($fh, my $buf, 5) == 5 or croak $!;
2498 $buf eq 'link ' or die "$path has mode 120000",
2499 "but is not a link\n";
2500 }
2501 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2502 if (!$pid) {
2503 open STDIN, '<&', $fh or croak $!;
2504 exec qw/git-hash-object -w --stdin/ or croak $!;
2505 }
2506 chomp($hash = do { local $/; <$out> });
2507 close $out or croak $!;
2508 close $fh or croak $!;
2509 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2510 close $fb->{base} or croak $!;
2511 } else {
2512 $hash = $fb->{blob} or die "no blob information\n";
2513 }
2514 $fb->{pool}->clear;
2515 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2516 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2517 undef;
2518}
2519sub abort_edit {
2521 my $self = shift;
2522 $self->{nr} = $self->{gii}->{nr};
2523 delete $self->{gii};
2524 $self->SUPER::abort_edit(@_);
2525}
2526sub close_edit {
2528 my $self = shift;
2529 $self->{git_commit_ok} = 1;
2530 $self->{nr} = $self->{gii}->{nr};
2531 delete $self->{gii};
2532 $self->SUPER::close_edit(@_);
2533}
2534package SVN::Git::Editor;
2536use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2537use strict;
2538use warnings;
2539use Carp qw/croak/;
2540use IO::File;
2541use Digest::MD5;
2542sub new {
2544 my ($class, $opts) = @_;
2545 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2546 die "$_ required!\n" unless (defined $opts->{$_});
2547 }
2548 my $pool = SVN::Pool->new;
2550 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2551 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2552 $opts->{r}, $mods);
2553 # $opts->{ra} functions should not be used after this:
2555 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2556 $opts->{editor_cb}, $pool);
2557 my $self = SVN::Delta::Editor->new(@ce, $pool);
2558 bless $self, $class;
2559 foreach (qw/svn_path r tree_a tree_b/) {
2560 $self->{$_} = $opts->{$_};
2561 }
2562 $self->{url} = $opts->{ra}->{url};
2563 $self->{mods} = $mods;
2564 $self->{types} = $types;
2565 $self->{pool} = $pool;
2566 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2567 $self->{rm} = { };
2568 $self->{path_prefix} = length $self->{svn_path} ?
2569 "$self->{svn_path}/" : '';
2570 return $self;
2571}
2572sub generate_diff {
2574 my ($tree_a, $tree_b) = @_;
2575 my @diff_tree = qw(diff-tree -z -r);
2576 if ($_cp_similarity) {
2577 push @diff_tree, "-C$_cp_similarity";
2578 } else {
2579 push @diff_tree, '-C';
2580 }
2581 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2582 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2583 push @diff_tree, $tree_a, $tree_b;
2584 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2585 local $/ = "\0";
2586 my $state = 'meta';
2587 my @mods;
2588 while (<$diff_fh>) {
2589 chomp $_; # this gets rid of the trailing "\0"
2590 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2591 $::sha1\s($::sha1)\s
2592 ([MTCRAD])\d*$/xo) {
2593 push @mods, { mode_a => $1, mode_b => $2,
2594 sha1_b => $3, chg => $4 };
2595 if ($4 =~ /^(?:C|R)$/) {
2596 $state = 'file_a';
2597 } else {
2598 $state = 'file_b';
2599 }
2600 } elsif ($state eq 'file_a') {
2601 my $x = $mods[$#mods] or croak "Empty array\n";
2602 if ($x->{chg} !~ /^(?:C|R)$/) {
2603 croak "Error parsing $_, $x->{chg}\n";
2604 }
2605 $x->{file_a} = $_;
2606 $state = 'file_b';
2607 } elsif ($state eq 'file_b') {
2608 my $x = $mods[$#mods] or croak "Empty array\n";
2609 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2610 croak "Error parsing $_, $x->{chg}\n";
2611 }
2612 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2613 croak "Error parsing $_, $x->{chg}\n";
2614 }
2615 $x->{file_b} = $_;
2616 $state = 'meta';
2617 } else {
2618 croak "Error parsing $_\n";
2619 }
2620 }
2621 command_close_pipe($diff_fh, $ctx);
2622 \@mods;
2623}
2624sub check_diff_paths {
2626 my ($ra, $pfx, $rev, $mods) = @_;
2627 my %types;
2628 $pfx .= '/' if length $pfx;
2629 sub type_diff_paths {
2631 my ($ra, $types, $path, $rev) = @_;
2632 my @p = split m#/+#, $path;
2633 my $c = shift @p;
2634 unless (defined $types->{$c}) {
2635 $types->{$c} = $ra->check_path($c, $rev);
2636 }
2637 while (@p) {
2638 $c .= '/' . shift @p;
2639 next if defined $types->{$c};
2640 $types->{$c} = $ra->check_path($c, $rev);
2641 }
2642 }
2643 foreach my $m (@$mods) {
2645 foreach my $f (qw/file_a file_b/) {
2646 next unless defined $m->{$f};
2647 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2648 if (length $pfx.$dir && ! defined $types{$dir}) {
2649 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2650 }
2651 }
2652 }
2653 \%types;
2654}
2655sub split_path {
2657 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2658}
2659sub repo_path {
2661 my ($self, $path) = @_;
2662 $self->{path_prefix}.(defined $path ? $path : '');
2663}
2664sub url_path {
2666 my ($self, $path) = @_;
2667 $self->{url} . '/' . $self->repo_path($path);
2668}
2669sub rmdirs {
2671 my ($self) = @_;
2672 my $rm = $self->{rm};
2673 delete $rm->{''}; # we never delete the url we're tracking
2674 return unless %$rm;
2675 foreach (keys %$rm) {
2677 my @d = split m#/#, $_;
2678 my $c = shift @d;
2679 $rm->{$c} = 1;
2680 while (@d) {
2681 $c .= '/' . shift @d;
2682 $rm->{$c} = 1;
2683 }
2684 }
2685 delete $rm->{$self->{svn_path}};
2686 delete $rm->{''}; # we never delete the url we're tracking
2687 return unless %$rm;
2688 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2690 $self->{tree_b});
2691 local $/ = "\0";
2692 while (<$fh>) {
2693 chomp;
2694 my @dn = split m#/#, $_;
2695 while (pop @dn) {
2696 delete $rm->{join '/', @dn};
2697 }
2698 unless (%$rm) {
2699 close $fh;
2700 return;
2701 }
2702 }
2703 command_close_pipe($fh, $ctx);
2704 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2706 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2707 $self->close_directory($bat->{$d}, $p);
2708 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2709 print "\tD+\t$d/\n" unless $::_q;
2710 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2711 delete $bat->{$d};
2712 }
2713}
2714sub open_or_add_dir {
2716 my ($self, $full_path, $baton) = @_;
2717 my $t = $self->{types}->{$full_path};
2718 if (!defined $t) {
2719 die "$full_path not known in r$self->{r} or we have a bug!\n";
2720 }
2721 if ($t == $SVN::Node::none) {
2722 return $self->add_directory($full_path, $baton,
2723 undef, -1, $self->{pool});
2724 } elsif ($t == $SVN::Node::dir) {
2725 return $self->open_directory($full_path, $baton,
2726 $self->{r}, $self->{pool});
2727 }
2728 print STDERR "$full_path already exists in repository at ",
2729 "r$self->{r} and it is not a directory (",
2730 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2731 exit 1;
2732}
2733sub ensure_path {
2735 my ($self, $path) = @_;
2736 my $bat = $self->{bat};
2737 my $repo_path = $self->repo_path($path);
2738 return $bat->{''} unless (length $repo_path);
2739 my @p = split m#/+#, $repo_path;
2740 my $c = shift @p;
2741 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2742 while (@p) {
2743 my $c0 = $c;
2744 $c .= '/' . shift @p;
2745 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2746 }
2747 return $bat->{$c};
2748}
2749sub A {
2751 my ($self, $m) = @_;
2752 my ($dir, $file) = split_path($m->{file_b});
2753 my $pbat = $self->ensure_path($dir);
2754 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2755 undef, -1);
2756 print "\tA\t$m->{file_b}\n" unless $::_q;
2757 $self->chg_file($fbat, $m);
2758 $self->close_file($fbat,undef,$self->{pool});
2759}
2760sub C {
2762 my ($self, $m) = @_;
2763 my ($dir, $file) = split_path($m->{file_b});
2764 my $pbat = $self->ensure_path($dir);
2765 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2766 $self->url_path($m->{file_a}), $self->{r});
2767 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2768 $self->chg_file($fbat, $m);
2769 $self->close_file($fbat,undef,$self->{pool});
2770}
2771sub delete_entry {
2773 my ($self, $path, $pbat) = @_;
2774 my $rpath = $self->repo_path($path);
2775 my ($dir, $file) = split_path($rpath);
2776 $self->{rm}->{$dir} = 1;
2777 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2778}
2779sub R {
2781 my ($self, $m) = @_;
2782 my ($dir, $file) = split_path($m->{file_b});
2783 my $pbat = $self->ensure_path($dir);
2784 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2785 $self->url_path($m->{file_a}), $self->{r});
2786 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2787 $self->chg_file($fbat, $m);
2788 $self->close_file($fbat,undef,$self->{pool});
2789 ($dir, $file) = split_path($m->{file_a});
2791 $pbat = $self->ensure_path($dir);
2792 $self->delete_entry($m->{file_a}, $pbat);
2793}
2794sub M {
2796 my ($self, $m) = @_;
2797 my ($dir, $file) = split_path($m->{file_b});
2798 my $pbat = $self->ensure_path($dir);
2799 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2800 $pbat,$self->{r},$self->{pool});
2801 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2802 $self->chg_file($fbat, $m);
2803 $self->close_file($fbat,undef,$self->{pool});
2804}
2805sub T { shift->M(@_) }
2807sub change_file_prop {
2809 my ($self, $fbat, $pname, $pval) = @_;
2810 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2811}
2812sub chg_file {
2814 my ($self, $fbat, $m) = @_;
2815 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2816 $self->change_file_prop($fbat,'svn:executable','*');
2817 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2818 $self->change_file_prop($fbat,'svn:executable',undef);
2819 }
2820 my $fh = IO::File->new_tmpfile or croak $!;
2821 if ($m->{mode_b} =~ /^120/) {
2822 print $fh 'link ' or croak $!;
2823 $self->change_file_prop($fbat,'svn:special','*');
2824 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2825 $self->change_file_prop($fbat,'svn:special',undef);
2826 }
2827 defined(my $pid = fork) or croak $!;
2828 if (!$pid) {
2829 open STDOUT, '>&', $fh or croak $!;
2830 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2831 }
2832 waitpid $pid, 0;
2833 croak $? if $?;
2834 $fh->flush == 0 or croak $!;
2835 seek $fh, 0, 0 or croak $!;
2836 my $md5 = Digest::MD5->new;
2838 $md5->addfile($fh) or croak $!;
2839 seek $fh, 0, 0 or croak $!;
2840 my $exp = $md5->hexdigest;
2842 my $pool = SVN::Pool->new;
2843 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2844 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2845 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2846 $pool->clear;
2847 close $fh or croak $!;
2849}
2850sub D {
2852 my ($self, $m) = @_;
2853 my ($dir, $file) = split_path($m->{file_b});
2854 my $pbat = $self->ensure_path($dir);
2855 print "\tD\t$m->{file_b}\n" unless $::_q;
2856 $self->delete_entry($m->{file_b}, $pbat);
2857}
2858sub close_edit {
2860 my ($self) = @_;
2861 my ($p,$bat) = ($self->{pool}, $self->{bat});
2862 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2863 next if $_ eq '';
2864 $self->close_directory($bat->{$_}, $p);
2865 }
2866 $self->close_directory($bat->{''}, $p);
2867 $self->SUPER::close_edit($p);
2868 $p->clear;
2869}
2870sub abort_edit {
2872 my ($self) = @_;
2873 $self->SUPER::abort_edit($self->{pool});
2874}
2875sub DESTROY {
2877 my $self = shift;
2878 $self->SUPER::DESTROY(@_);
2879 $self->{pool}->clear;
2880}
2881# this drives the editor
2883sub apply_diff {
2884 my ($self) = @_;
2885 my $mods = $self->{mods};
2886 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2887 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2888 my $f = $m->{chg};
2889 if (defined $o{$f}) {
2890 $self->$f($m);
2891 } else {
2892 fatal("Invalid change type: $f\n");
2893 }
2894 }
2895 $self->rmdirs if $_rmdir;
2896 if (@$mods == 0) {
2897 $self->abort_edit;
2898 } else {
2899 $self->close_edit;
2900 }
2901 return scalar @$mods;
2902}
2903package Git::SVN::Ra;
2905use vars qw/@ISA $config_dir $_log_window_size/;
2906use strict;
2907use warnings;
2908my ($can_do_switch, %ignored_err, $RA);
2909BEGIN {
2911 # enforce temporary pool usage for some simple functions
2912 no strict 'refs';
2913 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
2914 my $SUPER = "SUPER::$f";
2915 *$f = sub {
2916 my $self = shift;
2917 my $pool = SVN::Pool->new;
2918 my @ret = $self->$SUPER(@_,$pool);
2919 $pool->clear;
2920 wantarray ? @ret : $ret[0];
2921 };
2922 }
2923}
2924sub new {
2926 my ($class, $url) = @_;
2927 $url =~ s!/+$!!;
2928 return $RA if ($RA && $RA->{url} eq $url);
2929 SVN::_Core::svn_config_ensure($config_dir, undef);
2931 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2932 SVN::Client::get_simple_provider(),
2933 SVN::Client::get_ssl_server_trust_file_provider(),
2934 SVN::Client::get_simple_prompt_provider(
2935 \&Git::SVN::Prompt::simple, 2),
2936 SVN::Client::get_ssl_client_cert_prompt_provider(
2937 \&Git::SVN::Prompt::ssl_client_cert, 2),
2938 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2939 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2940 SVN::Client::get_username_provider(),
2941 SVN::Client::get_ssl_server_trust_prompt_provider(
2942 \&Git::SVN::Prompt::ssl_server_trust),
2943 SVN::Client::get_username_prompt_provider(
2944 \&Git::SVN::Prompt::username, 2),
2945 ]);
2946 my $config = SVN::Core::config_get_config($config_dir);
2947 my $self = SVN::Ra->new(url => $url, auth => $baton,
2948 config => $config,
2949 pool => SVN::Pool->new,
2950 auth_provider_callbacks => $callbacks);
2951 $self->{svn_path} = $url;
2952 $self->{repos_root} = $self->get_repos_root;
2953 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2954 $self->{cache} = { check_path => { r => 0, data => {} },
2955 get_dir => { r => 0, data => {} } };
2956 $RA = bless $self, $class;
2957}
2958sub check_path {
2960 my ($self, $path, $r) = @_;
2961 my $cache = $self->{cache}->{check_path};
2962 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
2963 return $cache->{data}->{$path};
2964 }
2965 my $pool = SVN::Pool->new;
2966 my $t = $self->SUPER::check_path($path, $r, $pool);
2967 $pool->clear;
2968 if ($r != $cache->{r}) {
2969 %{$cache->{data}} = ();
2970 $cache->{r} = $r;
2971 }
2972 $cache->{data}->{$path} = $t;
2973}
2974sub get_dir {
2976 my ($self, $dir, $r) = @_;
2977 my $cache = $self->{cache}->{get_dir};
2978 if ($r == $cache->{r}) {
2979 if (my $x = $cache->{data}->{$dir}) {
2980 return wantarray ? @$x : $x->[0];
2981 }
2982 }
2983 my $pool = SVN::Pool->new;
2984 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
2985 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
2986 $pool->clear;
2987 if ($r != $cache->{r}) {
2988 %{$cache->{data}} = ();
2989 $cache->{r} = $r;
2990 }
2991 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
2992 wantarray ? (\%dirents, $r, $props) : \%dirents;
2993}
2994sub DESTROY {
2996 # do not call the real DESTROY since we store ourselves in $RA
2997}
2998sub get_log {
3000 my ($self, @args) = @_;
3001 my $pool = SVN::Pool->new;
3002 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3003 my $ret = $self->SUPER::get_log(@args, $pool);
3004 $pool->clear;
3005 $ret;
3006}
3007sub get_commit_editor {
3009 my ($self, $log, $cb, $pool) = @_;
3010 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3011 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3012}
3013sub gs_do_update {
3015 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3016 my $new = ($rev_a == $rev_b);
3017 my $path = $gs->{path};
3018 if ($new && -e $gs->{index}) {
3020 unlink $gs->{index} or die
3021 "Couldn't unlink index: $gs->{index}: $!\n";
3022 }
3023 my $pool = SVN::Pool->new;
3024 $editor->set_path_strip($path);
3025 my (@pc) = split m#/#, $path;
3026 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3027 1, $editor, $pool);
3028 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3029 # Since we can't rely on svn_ra_reparent being available, we'll
3031 # just have to do some magic with set_path to make it so
3032 # we only want a partial path.
3033 my $sp = '';
3034 my $final = join('/', @pc);
3035 while (@pc) {
3036 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3037 $sp .= '/' if length $sp;
3038 $sp .= shift @pc;
3039 }
3040 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3041 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3043 $reporter->finish_report($pool);
3045 $pool->clear;
3046 $editor->{git_commit_ok};
3047}
3048# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3050# svn_ra_reparent didn't work before 1.4)
3051sub gs_do_switch {
3052 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3053 my $path = $gs->{path};
3054 my $pool = SVN::Pool->new;
3055 my $full_url = $self->{url};
3057 my $old_url = $full_url;
3058 $full_url .= "/$path" if length $path;
3059 my ($ra, $reparented);
3060 if ($old_url ne $full_url) {
3061 if ($old_url !~ m#^svn(\+ssh)?://#) {
3062 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3063 $pool);
3064 $self->{url} = $full_url;
3065 $reparented = 1;
3066 } else {
3067 $ra = Git::SVN::Ra->new($full_url);
3068 }
3069 }
3070 $ra ||= $self;
3071 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3072 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3073 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3074 $reporter->finish_report($pool);
3075 if ($reparented) {
3077 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3078 $self->{url} = $old_url;
3079 }
3080 $pool->clear;
3082 $editor->{git_commit_ok};
3083}
3084sub longest_common_path {
3086 my ($gsv, $globs) = @_;
3087 my %common;
3088 my $common_max = scalar @$gsv;
3089 foreach my $gs (@$gsv) {
3091 my @tmp = split m#/#, $gs->{path};
3092 my $p = '';
3093 foreach (@tmp) {
3094 $p .= length($p) ? "/$_" : $_;
3095 $common{$p} ||= 0;
3096 $common{$p}++;
3097 }
3098 }
3099 $globs ||= [];
3100 $common_max += scalar @$globs;
3101 foreach my $glob (@$globs) {
3102 my @tmp = split m#/#, $glob->{path}->{left};
3103 my $p = '';
3104 foreach (@tmp) {
3105 $p .= length($p) ? "/$_" : $_;
3106 $common{$p} ||= 0;
3107 $common{$p}++;
3108 }
3109 }
3110 my $longest_path = '';
3112 foreach (sort {length $b <=> length $a} keys %common) {
3113 if ($common{$_} == $common_max) {
3114 $longest_path = $_;
3115 last;
3116 }
3117 }
3118 $longest_path;
3119}
3120sub gs_fetch_loop_common {
3122 my ($self, $base, $head, $gsv, $globs) = @_;
3123 return if ($base > $head);
3124 my $inc = $_log_window_size;
3125 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3126 my $longest_path = longest_common_path($gsv, $globs);
3127 while (1) {
3128 my %revs;
3129 my $err;
3130 my $err_handler = $SVN::Error::handler;
3131 $SVN::Error::handler = sub {
3132 ($err) = @_;
3133 skip_unknown_revs($err);
3134 };
3135 sub _cb {
3136 my ($paths, $r, $author, $date, $log) = @_;
3137 [ dup_changed_paths($paths),
3138 { author => $author, date => $date, log => $log } ];
3139 }
3140 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3141 sub { $revs{$_[1]} = _cb(@_) });
3142 if ($err && $max >= $head) {
3143 print STDERR "Path '$longest_path' ",
3144 "was probably deleted:\n",
3145 $err->expanded_message,
3146 "\nWill attempt to follow ",
3147 "revisions r$min .. r$max ",
3148 "committed before the deletion\n";
3149 my $hi = $max;
3150 while (--$hi >= $min) {
3151 my $ok;
3152 $self->get_log([$longest_path], $min, $hi,
3153 0, 1, 1, sub {
3154 $ok ||= $_[1];
3155 $revs{$_[1]} = _cb(@_) });
3156 if ($ok) {
3157 print STDERR "r$min .. r$ok OK\n";
3158 last;
3159 }
3160 }
3161 }
3162 $SVN::Error::handler = $err_handler;
3163 my %exists = map { $_->{path} => $_ } @$gsv;
3165 foreach my $r (sort {$a <=> $b} keys %revs) {
3166 my ($paths, $logged) = @{$revs{$r}};
3167 foreach my $gs ($self->match_globs(\%exists, $paths,
3169 $globs, $r)) {
3170 if ($gs->rev_db_max >= $r) {
3171 next;
3172 }
3173 next unless $gs->match_paths($paths, $r);
3174 $gs->{logged_rev_props} = $logged;
3175 if (my $last_commit = $gs->last_commit) {
3176 $gs->assert_index_clean($last_commit);
3177 }
3178 my $log_entry = $gs->do_fetch($paths, $r);
3179 if ($log_entry) {
3180 $gs->do_git_commit($log_entry);
3181 }
3182 }
3183 foreach my $g (@$globs) {
3184 my $k = "svn-remote.$g->{remote}." .
3185 "$g->{t}-maxRev";
3186 Git::SVN::tmp_config($k, $r);
3187 }
3188 }
3189 # pre-fill the .rev_db since it'll eventually get filled in
3190 # with '0' x40 if something new gets committed
3191 foreach my $gs (@$gsv) {
3192 next if defined $gs->rev_db_get($max);
3193 $gs->rev_db_set($max, 0 x40);
3194 }
3195 foreach my $g (@$globs) {
3196 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3197 Git::SVN::tmp_config($k, $max);
3198 }
3199 last if $max >= $head;
3200 $min = $max + 1;
3201 $max += $inc;
3202 $max = $head if ($max > $head);
3203 }
3204}
3205sub match_globs {
3207 my ($self, $exists, $paths, $globs, $r) = @_;
3208 sub get_dir_check {
3210 my ($self, $exists, $g, $r) = @_;
3211 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3212 return unless scalar @x == 3;
3213 my $dirents = $x[0];
3214 foreach my $de (keys %$dirents) {
3215 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3216 my $p = $g->{path}->full_path($de);
3217 next if $exists->{$p};
3218 next if (length $g->{path}->{right} &&
3219 ($self->check_path($p, $r) !=
3220 $SVN::Node::dir));
3221 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3222 $g->{ref}->full_path($de), 1);
3223 }
3224 }
3225 foreach my $g (@$globs) {
3226 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3227 if ($path->{action} =~ /^[AR]$/) {
3228 get_dir_check($self, $exists, $g, $r);
3229 }
3230 }
3231 foreach (keys %$paths) {
3232 if (/$g->{path}->{left_regex}/ &&
3233 !/$g->{path}->{regex}/) {
3234 next if $paths->{$_}->{action} !~ /^[AR]$/;
3235 get_dir_check($self, $exists, $g, $r);
3236 }
3237 next unless /$g->{path}->{regex}/;
3238 my $p = $1;
3239 my $pathname = $g->{path}->full_path($p);
3240 next if $exists->{$pathname};
3241 next if ($self->check_path($pathname, $r) !=
3242 $SVN::Node::dir);
3243 $exists->{$pathname} = Git::SVN->init(
3244 $self->{url}, $pathname, undef,
3245 $g->{ref}->full_path($p), 1);
3246 }
3247 my $c = '';
3248 foreach (split m#/#, $g->{path}->{left}) {
3249 $c .= "/$_";
3250 next unless ($paths->{$c} &&
3251 ($paths->{$c}->{action} =~ /^[AR]$/));
3252 get_dir_check($self, $exists, $g, $r);
3253 }
3254 }
3255 values %$exists;
3256}
3257sub minimize_url {
3259 my ($self) = @_;
3260 return $self->{url} if ($self->{url} eq $self->{repos_root});
3261 my $url = $self->{repos_root};
3262 my @components = split(m!/!, $self->{svn_path});
3263 my $c = '';
3264 do {
3265 $url .= "/$c" if length $c;
3266 eval { (ref $self)->new($url)->get_latest_revnum };
3267 } while ($@ && ($c = shift @components));
3268 $url;
3269}
3270sub can_do_switch {
3272 my $self = shift;
3273 unless (defined $can_do_switch) {
3274 my $pool = SVN::Pool->new;
3275 my $rep = eval {
3276 $self->do_switch(1, '', 0, $self->{url},
3277 SVN::Delta::Editor->new, $pool);
3278 };
3279 if ($@) {
3280 $can_do_switch = 0;
3281 } else {
3282 $rep->abort_report($pool);
3283 $can_do_switch = 1;
3284 }
3285 $pool->clear;
3286 }
3287 $can_do_switch;
3288}
3289sub skip_unknown_revs {
3291 my ($err) = @_;
3292 my $errno = $err->apr_err();
3293 # Maybe the branch we're tracking didn't
3294 # exist when the repo started, so it's
3295 # not an error if it doesn't, just continue
3296 #
3297 # Wonderfully consistent library, eh?
3298 # 160013 - svn:// and file://
3299 # 175002 - http(s)://
3300 # 175007 - http(s):// (this repo required authorization, too...)
3301 # More codes may be discovered later...
3302 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3303 my $err_key = $err->expanded_message;
3304 # revision numbers change every time, filter them out
3305 $err_key =~ s/\d+/\0/g;
3306 $err_key = "$errno\0$err_key";
3307 unless ($ignored_err{$err_key}) {
3308 warn "W: Ignoring error from SVN, path probably ",
3309 "does not exist: ($errno): ",
3310 $err->expanded_message,"\n";
3311 $ignored_err{$err_key} = 1;
3312 }
3313 return;
3314 }
3315 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3316}
3317# svn_log_changed_path_t objects passed to get_log are likely to be
3319# overwritten even if only the refs are copied to an external variable,
3320# so we should dup the structures in their entirety. Using an externally
3321# passed pool (instead of our temporary and quickly cleared pool in
3322# Git::SVN::Ra) does not help matters at all...
3323sub dup_changed_paths {
3324 my ($paths) = @_;
3325 return undef unless $paths;
3326 my %ret;
3327 foreach my $p (keys %$paths) {
3328 my $i = $paths->{$p};
3329 my %s = map { $_ => $i->$_ }
3330 qw/copyfrom_path copyfrom_rev action/;
3331 $ret{$p} = \%s;
3332 }
3333 \%ret;
3334}
3335package Git::SVN::Log;
3337use strict;
3338use warnings;
3339use POSIX qw/strftime/;
3340use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3341 %rusers $show_commit $incremental/;
3342my $l_fmt;
3343sub cmt_showable {
3345 my ($c) = @_;
3346 return 1 if defined $c->{r};
3347 # big commit message got truncated by the 16k pretty buffer in rev-list
3349 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3350 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3351 @{$c->{l}} = ();
3352 my @log = command(qw/cat-file commit/, $c->{c});
3353 # shift off the headers
3355 shift @log while ($log[0] ne '');
3356 shift @log;
3357 # TODO: make $c->{l} not have a trailing newline in the future
3359 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3360 (undef, $c->{r}, undef) = ::extract_metadata(
3362 (grep(/^git-svn-id: /, @log))[-1]);
3363 }
3364 return defined $c->{r};
3365}
3366sub log_use_color {
3368 return 1 if $color;
3369 my ($dc, $dcvar);
3370 $dcvar = 'color.diff';
3371 $dc = `git-config --get $dcvar`;
3372 if ($dc eq '') {
3373 # nothing at all; fallback to "diff.color"
3374 $dcvar = 'diff.color';
3375 $dc = `git-config --get $dcvar`;
3376 }
3377 chomp($dc);
3378 if ($dc eq 'auto') {
3379 my $pc;
3380 $pc = `git-config --get color.pager`;
3381 if ($pc eq '') {
3382 # does not have it -- fallback to pager.color
3383 $pc = `git-config --bool --get pager.color`;
3384 }
3385 else {
3386 $pc = `git-config --bool --get color.pager`;
3387 if ($?) {
3388 $pc = 'false';
3389 }
3390 }
3391 chomp($pc);
3392 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3393 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3394 }
3395 return 0;
3396 }
3397 return 0 if $dc eq 'never';
3398 return 1 if $dc eq 'always';
3399 chomp($dc = `git-config --bool --get $dcvar`);
3400 return ($dc eq 'true');
3401}
3402sub git_svn_log_cmd {
3404 my ($r_min, $r_max, @args) = @_;
3405 my $head = 'HEAD';
3406 foreach my $x (@args) {
3407 last if $x eq '--';
3408 next unless ::verify_ref("$x^0");
3409 $head = $x;
3410 last;
3411 }
3412 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3414 $gs ||= Git::SVN->_new;
3415 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3416 $gs->refname);
3417 push @cmd, '-r' unless $non_recursive;
3418 push @cmd, qw/--raw --name-status/ if $verbose;
3419 push @cmd, '--color' if log_use_color();
3420 return @cmd unless defined $r_max;
3421 if ($r_max == $r_min) {
3422 push @cmd, '--max-count=1';
3423 if (my $c = $gs->rev_db_get($r_max)) {
3424 push @cmd, $c;
3425 }
3426 } else {
3427 my ($c_min, $c_max);
3428 $c_max = $gs->rev_db_get($r_max);
3429 $c_min = $gs->rev_db_get($r_min);
3430 if (defined $c_min && defined $c_max) {
3431 if ($r_max > $r_max) {
3432 push @cmd, "$c_min..$c_max";
3433 } else {
3434 push @cmd, "$c_max..$c_min";
3435 }
3436 } elsif ($r_max > $r_min) {
3437 push @cmd, $c_max;
3438 } else {
3439 push @cmd, $c_min;
3440 }
3441 }
3442 return @cmd;
3443}
3444# adapted from pager.c
3446sub config_pager {
3447 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3448 if (!defined $pager) {
3449 $pager = 'less';
3450 } elsif (length $pager == 0 || $pager eq 'cat') {
3451 $pager = undef;
3452 }
3453}
3454sub run_pager {
3456 return unless -t *STDOUT;
3457 pipe my $rfd, my $wfd or return;
3458 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3459 if (!$pid) {
3460 open STDOUT, '>&', $wfd or
3461 ::fatal "Can't redirect to stdout: $!\n";
3462 return;
3463 }
3464 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3465 $ENV{LESS} ||= 'FRSX';
3466 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3467}
3468sub tz_to_s_offset {
3470 my ($tz) = @_;
3471 $tz =~ s/(\d\d)$//;
3472 return ($1 * 60) + ($tz * 3600);
3473}
3474sub get_author_info {
3476 my ($dest, $author, $t, $tz) = @_;
3477 $author =~ s/(?:^\s*|\s*$)//g;
3478 $dest->{a_raw} = $author;
3479 my $au;
3480 if ($::_authors) {
3481 $au = $rusers{$author} || undef;
3482 }
3483 if (!$au) {
3484 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3485 }
3486 $dest->{t} = $t;
3487 $dest->{tz} = $tz;
3488 $dest->{a} = $au;
3489 # Date::Parse isn't in the standard Perl distro :(
3490 if ($tz =~ s/^\+//) {
3491 $t += tz_to_s_offset($tz);
3492 } elsif ($tz =~ s/^\-//) {
3493 $t -= tz_to_s_offset($tz);
3494 }
3495 $dest->{t_utc} = $t;
3496}
3497sub process_commit {
3499 my ($c, $r_min, $r_max, $defer) = @_;
3500 if (defined $r_min && defined $r_max) {
3501 if ($r_min == $c->{r} && $r_min == $r_max) {
3502 show_commit($c);
3503 return 0;
3504 }
3505 return 1 if $r_min == $r_max;
3506 if ($r_min < $r_max) {
3507 # we need to reverse the print order
3508 return 0 if (defined $limit && --$limit < 0);
3509 push @$defer, $c;
3510 return 1;
3511 }
3512 if ($r_min != $r_max) {
3513 return 1 if ($r_min < $c->{r});
3514 return 1 if ($r_max > $c->{r});
3515 }
3516 }
3517 return 0 if (defined $limit && --$limit < 0);
3518 show_commit($c);
3519 return 1;
3520}
3521sub show_commit {
3523 my $c = shift;
3524 if ($oneline) {
3525 my $x = "\n";
3526 if (my $l = $c->{l}) {
3527 while ($l->[0] =~ /^\s*$/) { shift @$l }
3528 $x = $l->[0];
3529 }
3530 $l_fmt ||= 'A' . length($c->{r});
3531 print 'r',pack($l_fmt, $c->{r}),' | ';
3532 print "$c->{c} | " if $show_commit;
3533 print $x;
3534 } else {
3535 show_commit_normal($c);
3536 }
3537}
3538sub show_commit_changed_paths {
3540 my ($c) = @_;
3541 return unless $c->{changed};
3542 print "Changed paths:\n", @{$c->{changed}};
3543}
3544sub show_commit_normal {
3546 my ($c) = @_;
3547 print '-' x72, "\nr$c->{r} | ";
3548 print "$c->{c} | " if $show_commit;
3549 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3550 localtime($c->{t_utc})), ' | ';
3551 my $nr_line = 0;
3552 if (my $l = $c->{l}) {
3554 while ($l->[$#$l] eq "\n" && $#$l > 0
3555 && $l->[($#$l - 1)] eq "\n") {
3556 pop @$l;
3557 }
3558 $nr_line = scalar @$l;
3559 if (!$nr_line) {
3560 print "1 line\n\n\n";
3561 } else {
3562 if ($nr_line == 1) {
3563 $nr_line = '1 line';
3564 } else {
3565 $nr_line .= ' lines';
3566 }
3567 print $nr_line, "\n";
3568 show_commit_changed_paths($c);
3569 print "\n";
3570 print $_ foreach @$l;
3571 }
3572 } else {
3573 print "1 line\n";
3574 show_commit_changed_paths($c);
3575 print "\n";
3576 }
3578 foreach my $x (qw/raw stat diff/) {
3579 if ($c->{$x}) {
3580 print "\n";
3581 print $_ foreach @{$c->{$x}}
3582 }
3583 }
3584}
3585sub cmd_show_log {
3587 my (@args) = @_;
3588 my ($r_min, $r_max);
3589 my $r_last = -1; # prevent dupes
3590 if (defined $TZ) {
3591 $ENV{TZ} = $TZ;
3592 } else {
3593 delete $ENV{TZ};
3594 }
3595 if (defined $::_revision) {
3596 if ($::_revision =~ /^(\d+):(\d+)$/) {
3597 ($r_min, $r_max) = ($1, $2);
3598 } elsif ($::_revision =~ /^\d+$/) {
3599 $r_min = $r_max = $::_revision;
3600 } else {
3601 ::fatal "-r$::_revision is not supported, use ",
3602 "standard \'git log\' arguments instead\n";
3603 }
3604 }
3605 config_pager();
3607 @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3608 my $log = command_output_pipe(@args);
3609 run_pager();
3610 my (@k, $c, $d, $stat);
3611 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3612 while (<$log>) {
3613 if (/^${esc_color}commit ($::sha1_short)/o) {
3614 my $cmt = $1;
3615 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3616 $r_last = $c->{r};
3617 process_commit($c, $r_min, $r_max, \@k) or
3618 goto out;
3619 }
3620 $d = undef;
3621 $c = { c => $cmt };
3622 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3623 get_author_info($c, $1, $2, $3);
3624 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3625 # ignore
3626 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3627 push @{$c->{raw}}, $_;
3628 } elsif (/^${esc_color}[ACRMDT]\t/) {
3629 # we could add $SVN->{svn_path} here, but that requires
3630 # remote access at the moment (repo_path_split)...
3631 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3632 push @{$c->{changed}}, $_;
3633 } elsif (/^${esc_color}diff /o) {
3634 $d = 1;
3635 push @{$c->{diff}}, $_;
3636 } elsif ($d) {
3637 push @{$c->{diff}}, $_;
3638 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3639 $esc_color*[\+\-]*$esc_color$/x) {
3640 $stat = 1;
3641 push @{$c->{stat}}, $_;
3642 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3643 push @{$c->{stat}}, $_;
3644 $stat = undef;
3645 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3646 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3647 } elsif (s/^${esc_color} //o) {
3648 push @{$c->{l}}, $_;
3649 }
3650 }
3651 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3652 $r_last = $c->{r};
3653 process_commit($c, $r_min, $r_max, \@k);
3654 }
3655 if (@k) {
3656 my $swap = $r_max;
3657 $r_max = $r_min;
3658 $r_min = $swap;
3659 process_commit($_, $r_min, $r_max) foreach reverse @k;
3660 }
3661out:
3662 close $log;
3663 print '-' x72,"\n" unless $incremental || $oneline;
3664}
3665package Git::SVN::Migration;
3667# these version numbers do NOT correspond to actual version numbers
3668# of git nor git-svn. They are just relative.
3669#
3670# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3671#
3672# v1 layout: .git/$id/info/url, refs/remotes/$id
3673#
3674# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3675#
3676# v3 layout: .git/svn/$id, refs/remotes/$id
3677# - info/url may remain for backwards compatibility
3678# - this is what we migrate up to this layout automatically,
3679# - this will be used by git svn init on single branches
3680# v3.1 layout (auto migrated):
3681# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3682# for backwards compatibility
3683#
3684# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3685# - this is only created for newly multi-init-ed
3686# repositories. Similar in spirit to the
3687# --use-separate-remotes option in git-clone (now default)
3688# - we do not automatically migrate to this (following
3689# the example set by core git)
3690use strict;
3691use warnings;
3692use Carp qw/croak/;
3693use File::Path qw/mkpath/;
3694use File::Basename qw/dirname basename/;
3695use vars qw/$_minimize/;
3696sub migrate_from_v0 {
3698 my $git_dir = $ENV{GIT_DIR};
3699 return undef unless -d $git_dir;
3700 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3701 my $migrated = 0;
3702 while (<$fh>) {
3703 chomp;
3704 my ($id, $orig_ref) = ($_, $_);
3705 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3706 next unless -f "$git_dir/$id/info/url";
3707 my $new_ref = "refs/remotes/$id";
3708 if (::verify_ref("$new_ref^0")) {
3709 print STDERR "W: $orig_ref is probably an old ",
3710 "branch used by an ancient version of ",
3711 "git-svn.\n",
3712 "However, $new_ref also exists.\n",
3713 "We will not be able ",
3714 "to use this branch until this ",
3715 "ambiguity is resolved.\n";
3716 next;
3717 }
3718 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3719 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3720 command_noisy('update-ref', $new_ref, $orig_ref);
3721 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3722 $migrated++;
3723 }
3724 command_close_pipe($fh, $ctx);
3725 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3726 $migrated;
3727}
3728sub migrate_from_v1 {
3730 my $git_dir = $ENV{GIT_DIR};
3731 my $migrated = 0;
3732 return $migrated unless -d $git_dir;
3733 my $svn_dir = "$git_dir/svn";
3734 # just in case somebody used 'svn' as their $id at some point...
3736 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3737 print STDERR "Migrating from a git-svn v1 layout...\n";
3739 mkpath([$svn_dir]);
3740 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3741 "$svn_dir\n\t(required for this version ",
3742 "($::VERSION) of git-svn) does not. exist\n";
3743 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3744 while (<$fh>) {
3745 my $x = $_;
3746 next unless $x =~ s#^refs/remotes/##;
3747 chomp $x;
3748 next unless -f "$git_dir/$x/info/url";
3749 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3750 next unless $u;
3751 my $dn = dirname("$git_dir/svn/$x");
3752 mkpath([$dn]) unless -d $dn;
3753 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3754 mkpath(["$git_dir/svn/svn"]);
3755 print STDERR " - $git_dir/$x/info => ",
3756 "$git_dir/svn/$x/info\n";
3757 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3758 croak "$!: $x";
3759 # don't worry too much about these, they probably
3760 # don't exist with repos this old (save for index,
3761 # and we can easily regenerate that)
3762 foreach my $f (qw/unhandled.log index .rev_db/) {
3763 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3764 }
3765 } else {
3766 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3767 rename "$git_dir/$x", "$git_dir/svn/$x" or
3768 croak "$!: $x";
3769 }
3770 $migrated++;
3771 }
3772 command_close_pipe($fh, $ctx);
3773 print STDERR "Done migrating from a git-svn v1 layout\n";
3774 $migrated;
3775}
3776sub read_old_urls {
3778 my ($l_map, $pfx, $path) = @_;
3779 my @dir;
3780 foreach (<$path/*>) {
3781 if (-r "$_/info/url") {
3782 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3783 my $ref_id = $pfx . basename $_;
3784 my $url = ::file_to_s("$_/info/url");
3785 $l_map->{$ref_id} = $url;
3786 } elsif (-d $_) {
3787 push @dir, $_;
3788 }
3789 }
3790 foreach (@dir) {
3791 my $x = $_;
3792 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3793 read_old_urls($l_map, $x, $_);
3794 }
3795}
3796sub migrate_from_v2 {
3798 my @cfg = command(qw/config -l/);
3799 return if grep /^svn-remote\..+\.url=/, @cfg;
3800 my %l_map;
3801 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3802 my $migrated = 0;
3803 foreach my $ref_id (sort keys %l_map) {
3805 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3806 if ($@) {
3807 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3808 }
3809 $migrated++;
3810 }
3811 $migrated;
3812}
3813sub minimize_connections {
3815 my $r = Git::SVN::read_all_remotes();
3816 my $new_urls = {};
3817 my $root_repos = {};
3818 foreach my $repo_id (keys %$r) {
3819 my $url = $r->{$repo_id}->{url} or next;
3820 my $fetch = $r->{$repo_id}->{fetch} or next;
3821 my $ra = Git::SVN::Ra->new($url);
3822 # skip existing cases where we already connect to the root
3824 if (($ra->{url} eq $ra->{repos_root}) ||
3825 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3826 $repo_id)) {
3827 $root_repos->{$ra->{url}} = $repo_id;
3828 next;
3829 }
3830 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3832 my $root_path = $ra->{url};
3833 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3834 foreach my $path (keys %$fetch) {
3835 my $ref_id = $fetch->{$path};
3836 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3837 # make sure we can read when connecting to
3839 # a higher level of a repository
3840 my ($last_rev, undef) = $gs->last_rev_commit;
3841 if (!defined $last_rev) {
3842 $last_rev = eval {
3843 $root_ra->get_latest_revnum;
3844 };
3845 next if $@;
3846 }
3847 my $new = $root_path;
3848 $new .= length $path ? "/$path" : '';
3849 eval {
3850 $root_ra->get_log([$new], $last_rev, $last_rev,
3851 0, 0, 1, sub { });
3852 };
3853 next if $@;
3854 $new_urls->{$ra->{repos_root}}->{$new} =
3855 { ref_id => $ref_id,
3856 old_repo_id => $repo_id,
3857 old_path => $path };
3858 }
3859 }
3860 my @emptied;
3862 foreach my $url (keys %$new_urls) {
3863 # see if we can re-use an existing [svn-remote "repo_id"]
3864 # instead of creating a(n ugly) new section:
3865 my $repo_id = $root_repos->{$url} ||
3866 Git::SVN::sanitize_remote_name($url);
3867 my $fetch = $new_urls->{$url};
3869 foreach my $path (keys %$fetch) {
3870 my $x = $fetch->{$path};
3871 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3872 my $pfx = "svn-remote.$x->{old_repo_id}";
3873 my $old_fetch = quotemeta("$x->{old_path}:".
3875 "refs/remotes/$x->{ref_id}");
3876 command_noisy(qw/config --unset/,
3877 "$pfx.fetch", '^'. $old_fetch . '$');
3878 delete $r->{$x->{old_repo_id}}->
3879 {fetch}->{$x->{old_path}};
3880 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3881 command_noisy(qw/config --unset/,
3882 "$pfx.url");
3883 push @emptied, $x->{old_repo_id}
3884 }
3885 }
3886 }
3887 if (@emptied) {
3888 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3889 "$ENV{GIT_DIR}/config";
3890 print STDERR <<EOF;
3891The following [svn-remote] sections in your config file ($file) are empty
3892and can be safely removed:
3893EOF
3894 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3895 }
3896}
3897sub migration_check {
3899 migrate_from_v0();
3900 migrate_from_v1();
3901 migrate_from_v2();
3902 minimize_connections() if $_minimize;
3903}
3904package Git::IndexInfo;
3906use strict;
3907use warnings;
3908use Git qw/command_input_pipe command_close_pipe/;
3909sub new {
3911 my ($class) = @_;
3912 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3913 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3914}
3915sub remove {
3917 my ($self, $path) = @_;
3918 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3919 return ++$self->{nr};
3920 }
3921 undef;
3922}
3923sub update {
3925 my ($self, $mode, $hash, $path) = @_;
3926 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3927 return ++$self->{nr};
3928 }
3929 undef;
3930}
3931sub DESTROY {
3933 my ($self) = @_;
3934 command_close_pipe($self->{gui}, $self->{ctx});
3935}
3936package Git::SVN::GlobSpec;
3938use strict;
3939use warnings;
3940sub new {
3942 my ($class, $glob) = @_;
3943 my $re = $glob;
3944 $re =~ s!/+$!!g; # no need for trailing slashes
3945 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3946 my ($left, $right) = ($1, $2);
3947 if ($nr > 1) {
3948 die "Only one '*' wildcard expansion ",
3949 "is supported (got $nr): '$glob'\n";
3950 } elsif ($nr == 0) {
3951 die "One '*' is needed for glob: '$glob'\n";
3952 }
3953 $re = quotemeta($left) . $re . quotemeta($right);
3954 if (length $left && !($left =~ s!/+$!!g)) {
3955 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3956 }
3957 if (length $right && !($right =~ s!^/+!!g)) {
3958 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3959 }
3960 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3961 bless { left => $left, right => $right, left_regex => $left_re,
3962 regex => qr/$re/, glob => $glob }, $class;
3963}
3964sub full_path {
3966 my ($self, $path) = @_;
3967 return (length $self->{left} ? "$self->{left}/" : '') .
3968 $path . (length $self->{right} ? "/$self->{right}" : '');
3969}
3970__END__
3972Data structures:
3974$remotes = { # returned by read_all_remotes()
3977 'svn' => {
3978 # svn-remote.svn.url=https://svn.musicpd.org
3979 url => 'https://svn.musicpd.org',
3980 # svn-remote.svn.fetch=mpd/trunk:trunk
3981 fetch => {
3982 'mpd/trunk' => 'trunk',
3983 },
3984 # svn-remote.svn.tags=mpd/tags/*:tags/*
3985 tags => {
3986 path => {
3987 left => 'mpd/tags',
3988 right => '',
3989 regex => qr!mpd/tags/([^/]+)$!,
3990 glob => 'tags/*',
3991 },
3992 ref => {
3993 left => 'tags',
3994 right => '',
3995 regex => qr!tags/([^/]+)$!,
3996 glob => 'tags/*',
3997 },
3998 }
3999 }
4000};
4001$log_entry hashref as returned by libsvn_log_entry()
4003{
4004 log => 'whitespace-formatted log entry
4005', # trailing newline is preserved
4006 revision => '8', # integer
4007 date => '2004-02-24T17:01:44.108345Z', # commit date
4008 author => 'committer name'
4009};
4010# this is generated by generate_diff();
4013@mods = array of diff-index line hashes, each element represents one line
4014 of diff-index output
4015diff-index line ($m hash)
4017{
4018 mode_a => first column of diff-index output, no leading ':',
4019 mode_b => second column of diff-index output,
4020 sha1_b => sha1sum of the final blob,
4021 chg => change type [MCRADT],
4022 file_a => original file name of a file (iff chg is 'C' or 'R')
4023 file_b => new/current file name of a file (any chg)
4024}
4025;
4026# retval of read_url_paths{,_all}();
4028$l_map = {
4029 # repository root url
4030 'https://svn.musicpd.org' => {
4031 # repository path # GIT_SVN_ID
4032 'mpd/trunk' => 'trunk',
4033 'mpd/tags/0.11.5' => 'tags/0.11.5',
4034 },
4035}
4036Notes:
4038 I don't trust the each() function on unless I created %hash myself
4039 because the internal iterator may not have started at base.