?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: don't consider SVN URL usernames significant when comparing
[git/mjg.git] / git-svn.perl
blobea5afb7f8089732b0f2a1c2cb2eacf9632a245cd
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 my $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
18 $Git::SVN::Log::TZ = $ENV{TZ};
19 $ENV{TZ} = 'UTC';
20 $| = 1; # unbuffer STDOUT
21
22 sub fatal (@) { print STDERR @_; exit 1 }
23 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
24 require SVN::Ra;
25 require SVN::Delta;
26 if ($SVN::Core::VERSION lt '1.1.0') {
27 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
28 }
29 push @Git::SVN::Ra::ISA, 'SVN::Ra';
30 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
31 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
32 use Carp qw/croak/;
33 use IO::File qw//;
34 use File::Basename qw/dirname basename/;
35 use File::Path qw/mkpath/;
36 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
37 use IPC::Open3;
38 use Git;
39
40 BEGIN {
41 my $s;
42 foreach (qw/command command_oneline command_noisy command_output_pipe
43 command_input_pipe command_close_pipe/) {
44 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
45 "*Git::SVN::Migration::$_ = ".
46 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
47 }
48 eval $s;
49 }
50
51 my ($SVN);
52
53 $sha1 = qr/[a-f\d]{40}/;
54 $sha1_short = qr/[a-f\d]{4,40}/;
55 my ($_stdin, $_help, $_edit,
56 $_message, $_file,
57 $_template, $_shared,
58 $_version, $_fetch_all,
59 $_merge, $_strategy, $_dry_run,
60 $_prefix, $_no_checkout, $_verbose);
61 $Git::SVN::_follow_parent = 1;
62 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
63 'config-dir=s' => \$Git::SVN::Ra::config_dir,
64 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
65 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
66 'authors-file|A=s' => \$_authors,
67 'repack:i' => \$Git::SVN::_repack,
68 'noMetadata' => \$Git::SVN::_no_metadata,
69 'useSvmProps' => \$Git::SVN::_use_svm_props,
70 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
71 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
72 'no-checkout' => \$_no_checkout,
73 'quiet|q' => \$_q,
74 'repack-flags|repack-args|repack-opts=s' =>
75 \$Git::SVN::_repack_flags,
76 %remote_opts );
77
78 my ($_trunk, $_tags, $_branches);
79 my %icv;
80 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
81 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
82 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
83 'no-metadata' => sub { $icv{noMetadata} = 1 },
84 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
85 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
86 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
87 %remote_opts );
88 my %cmt_opts = ( 'edit|e' => \$_edit,
89 'rmdir' => \$SVN::Git::Editor::_rmdir,
90 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
91 'l=i' => \$SVN::Git::Editor::_rename_limit,
92 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
93 );
94
95 my %cmd = (
96 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
97 { 'revision|r=s' => \$_revision,
98 'fetch-all|all' => \$_fetch_all,
99 %fc_opts } ],
100 clone => [ \&cmd_clone, "Initialize and fetch revisions",
101 { 'revision|r=s' => \$_revision,
102 %fc_opts, %init_opts } ],
103 init => [ \&cmd_init, "Initialize a repo for tracking" .
104 " (requires URL argument)",
105 \%init_opts ],
106 'multi-init' => [ \&cmd_multi_init,
107 "Deprecated alias for ".
108 "'$0 init -T<trunk> -b<branches> -t<tags>'",
109 \%init_opts ],
110 dcommit => [ \&cmd_dcommit,
111 'Commit several diffs to merge with upstream',
112 { 'merge|m|M' => \$_merge,
113 'strategy|s=s' => \$_strategy,
114 'verbose|v' => \$_verbose,
115 'dry-run|n' => \$_dry_run,
116 'fetch-all|all' => \$_fetch_all,
117 %cmt_opts, %fc_opts } ],
118 'set-tree' => [ \&cmd_set_tree,
119 "Set an SVN repository to a git tree-ish",
120 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
121 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
122 { 'revision|r=i' => \$_revision } ],
123 'multi-fetch' => [ \&cmd_multi_fetch,
124 "Deprecated alias for $0 fetch --all",
125 { 'revision|r=s' => \$_revision, %fc_opts } ],
126 'migrate' => [ sub { },
127 # no-op, we automatically run this anyways,
128 'Migrate configuration/metadata/layout from
129 previous versions of git-svn',
130 { 'minimize' => \$Git::SVN::Migration::_minimize,
131 %remote_opts } ],
132 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
133 { 'limit=i' => \$Git::SVN::Log::limit,
134 'revision|r=s' => \$_revision,
135 'verbose|v' => \$Git::SVN::Log::verbose,
136 'incremental' => \$Git::SVN::Log::incremental,
137 'oneline' => \$Git::SVN::Log::oneline,
138 'show-commit' => \$Git::SVN::Log::show_commit,
139 'non-recursive' => \$Git::SVN::Log::non_recursive,
140 'authors-file|A=s' => \$_authors,
141 'color' => \$Git::SVN::Log::color,
142 'pager=s' => \$Git::SVN::Log::pager,
143 } ],
144 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
145 { 'merge|m|M' => \$_merge,
146 'verbose|v' => \$_verbose,
147 'strategy|s=s' => \$_strategy,
148 'fetch-all|all' => \$_fetch_all,
149 %fc_opts } ],
150 'commit-diff' => [ \&cmd_commit_diff,
151 'Commit a diff between two trees',
152 { 'message|m=s' => \$_message,
153 'file|F=s' => \$_file,
154 'revision|r=s' => \$_revision,
155 %cmt_opts } ],
156 );
157
158 my $cmd;
159 for (my $i = 0; $i < @ARGV; $i++) {
160 if (defined $cmd{$ARGV[$i]}) {
161 $cmd = $ARGV[$i];
162 splice @ARGV, $i, 1;
163 last;
164 }
165 };
166
167 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
168
169 read_repo_config(\%opts);
170 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
171 'minimize-connections' => \$Git::SVN::Migration::_minimize,
172 'id|i=s' => \$Git::SVN::default_ref_id,
173 'svn-remote|remote|R=s' => sub {
174 $Git::SVN::no_reuse_existing = 1;
175 $Git::SVN::default_repo_id = $_[1] });
176 exit 1 if (!$rv && $cmd ne 'log');
177
178 usage(0) if $_help;
179 version() if $_version;
180 usage(1) unless defined $cmd;
181 load_authors() if $_authors;
182
183 # make sure we're always running
184 unless ($cmd =~ /(?:clone|init|multi-init)$/) {
185 unless (-d $ENV{GIT_DIR}) {
186 if ($git_dir_user_set) {
187 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
188 "but it is not a directory\n";
189 }
190 my $git_dir = delete $ENV{GIT_DIR};
191 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
192 unless (length $cdup) {
193 die "Already at toplevel, but $git_dir ",
194 "not found '$cdup'\n";
195 }
196 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
197 unless (-d $git_dir) {
198 die "$git_dir still not found after going to ",
199 "'$cdup'\n";
200 }
201 $ENV{GIT_DIR} = $git_dir;
202 }
203 }
204 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
205 Git::SVN::Migration::migration_check();
206 }
207 Git::SVN::init_vars();
208 eval {
209 Git::SVN::verify_remotes_sanity();
210 $cmd{$cmd}->[0]->(@ARGV);
211 };
212 fatal $@ if $@;
213 post_fetch_checkout();
214 exit 0;
215
216 ####################### primary functions ######################
217 sub usage {
218 my $exit = shift || 0;
219 my $fd = $exit ? \*STDERR : \*STDOUT;
220 print $fd <<"";
221 git-svn - bidirectional operations between a single Subversion tree and git
222 Usage: $0 <command> [options] [arguments]\n
223
224 print $fd "Available commands:\n" unless $cmd;
225
226 foreach (sort keys %cmd) {
227 next if $cmd && $cmd ne $_;
228 next if /^multi-/; # don't show deprecated commands
229 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
230 foreach (keys %{$cmd{$_}->[2]}) {
231 # prints out arguments as they should be passed:
232 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
233 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
234 "--$_" : "-$_" }
235 split /\|/,$_)," $x\n";
236 }
237 }
238 print $fd <<"";
239 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
240 arbitrary identifier if you're tracking multiple SVN branches/repositories in
241 one git repository and want to keep them separate. See git-svn(1) for more
242 information.
243
244 exit $exit;
245 }
246
247 sub version {
248 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
249 exit 0;
250 }
251
252 sub do_git_init_db {
253 unless (-d $ENV{GIT_DIR}) {
254 my @init_db = ('init');
255 push @init_db, "--template=$_template" if defined $_template;
256 if (defined $_shared) {
257 if ($_shared =~ /[a-z]/) {
258 push @init_db, "--shared=$_shared";
259 } else {
260 push @init_db, "--shared";
261 }
262 }
263 command_noisy(@init_db);
264 }
265 my $set;
266 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
267 foreach my $i (keys %icv) {
268 die "'$set' and '$i' cannot both be set\n" if $set;
269 next unless defined $icv{$i};
270 command_noisy('config', "$pfx.$i", $icv{$i});
271 $set = $i;
272 }
273 }
274
275 sub init_subdir {
276 my $repo_path = shift or return;
277 mkpath([$repo_path]) unless -d $repo_path;
278 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
279 $ENV{GIT_DIR} = $repo_path . "/.git";
280 }
281
282 sub cmd_clone {
283 my ($url, $path) = @_;
284 if (!defined $path &&
285 (defined $_trunk || defined $_branches || defined $_tags) &&
286 $url !~ m#^[a-z\+]+://#) {
287 $path = $url;
288 }
289 warn "--path: $path\n" if defined $path;
290 $path = basename($url) if !defined $path || !length $path;
291 warn "++path: $path\n" if defined $path;
292 mkpath([$path]);
293 chdir $path or die "Couldn't chdir to $path\n";
294 cmd_init(@_);
295 Git::SVN::fetch_all($Git::SVN::default_repo_id);
296 }
297
298 sub cmd_init {
299 if (defined $_trunk || defined $_branches || defined $_tags) {
300 return cmd_multi_init(@_);
301 }
302 my $url = shift or die "SVN repository location required ",
303 "as a command-line argument\n";
304 init_subdir(@_);
305 do_git_init_db();
306
307 Git::SVN->init($url);
308 }
309
310 sub cmd_fetch {
311 if (grep /^\d+=./, @_) {
312 die "'<rev>=<commit>' fetch arguments are ",
313 "no longer supported.\n";
314 }
315 my ($remote) = @_;
316 if (@_ > 1) {
317 die "Usage: $0 fetch [--all] [svn-remote]\n";
318 }
319 $remote ||= $Git::SVN::default_repo_id;
320 if ($_fetch_all) {
321 cmd_multi_fetch();
322 } else {
323 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
324 }
325 }
326
327 sub cmd_set_tree {
328 my (@commits) = @_;
329 if ($_stdin || !@commits) {
330 print "Reading from stdin...\n";
331 @commits = ();
332 while (<STDIN>) {
333 if (/\b($sha1_short)\b/o) {
334 unshift @commits, $1;
335 }
336 }
337 }
338 my @revs;
339 foreach my $c (@commits) {
340 my @tmp = command('rev-parse',$c);
341 if (scalar @tmp == 1) {
342 push @revs, $tmp[0];
343 } elsif (scalar @tmp > 1) {
344 push @revs, reverse(command('rev-list',@tmp));
345 } else {
346 fatal "Failed to rev-parse $c\n";
347 }
348 }
349 my $gs = Git::SVN->new;
350 my ($r_last, $cmt_last) = $gs->last_rev_commit;
351 $gs->fetch;
352 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
353 fatal "There are new revisions that were fetched ",
354 "and need to be merged (or acknowledged) ",
355 "before committing.\nlast rev: $r_last\n",
356 " current: $gs->{last_rev}\n";
357 }
358 $gs->set_tree($_) foreach @revs;
359 print "Done committing ",scalar @revs," revisions to SVN\n";
360 }
361
362 sub cmd_dcommit {
363 my $head = shift;
364 $head ||= 'HEAD';
365 my @refs;
366 my ($url, $rev, $uuid) = working_head_info($head, \@refs);
367 my $c = $refs[-1];
368 unless (defined $url && defined $rev && defined $uuid) {
369 die "Unable to determine upstream SVN information from ",
370 "$head history\n";
371 }
372 my $gs = Git::SVN->find_by_url($url);
373 my $last_rev;
374 foreach my $d (@refs) {
375 if (!verify_ref("$d~1")) {
376 fatal "Commit $d\n",
377 "has no parent commit, and therefore ",
378 "nothing to diff against.\n",
379 "You should be working from a repository ",
380 "originally created by git-svn\n";
381 }
382 unless (defined $last_rev) {
383 (undef, $last_rev, undef) = cmt_metadata("$d~1");
384 unless (defined $last_rev) {
385 fatal "Unable to extract revision information ",
386 "from commit $d~1\n";
387 }
388 }
389 if ($_dry_run) {
390 print "diff-tree $d~1 $d\n";
391 } else {
392 my %ed_opts = ( r => $last_rev,
393 log => get_commit_entry($d)->{log},
394 ra => Git::SVN::Ra->new($url),
395 tree_a => "$d~1",
396 tree_b => $d,
397 editor_cb => sub {
398 print "Committed r$_[0]\n";
399 $last_rev = $_[0]; },
400 svn_path => '');
401 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
402 print "No changes\n$d~1 == $d\n";
403 }
404 }
405 }
406 return if $_dry_run;
407 unless ($gs) {
408 warn "Could not determine fetch information for $url\n",
409 "Will not attempt to fetch and rebase commits.\n",
410 "This probably means you have useSvmProps and should\n",
411 "now resync your SVN::Mirror repository.\n";
412 return;
413 }
414 $_fetch_all ? $gs->fetch_all : $gs->fetch;
415 # we always want to rebase against the current HEAD, not any
416 # head that was passed to us
417 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
418 my @finish;
419 if (@diff) {
420 @finish = rebase_cmd();
421 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
422 "using @finish:\n", "@diff";
423 } else {
424 print "No changes between current HEAD and ",
425 $gs->refname, "\nResetting to the latest ",
426 $gs->refname, "\n";
427 @finish = qw/reset --mixed/;
428 }
429 command_noisy(@finish, $gs->refname);
430 }
431
432 sub cmd_rebase {
433 command_noisy(qw/update-index --refresh/);
434 my $url = (working_head_info('HEAD'))[0];
435 if (!defined $url) {
436 die "Unable to determine upstream SVN information from ",
437 "working tree history\n";
438 }
439
440 my $gs = Git::SVN->find_by_url($url);
441 if (command(qw/diff-index HEAD --/)) {
442 print STDERR "Cannot rebase with uncommited changes:\n";
443 command_noisy('status');
444 exit 1;
445 }
446 $_fetch_all ? $gs->fetch_all : $gs->fetch;
447 command_noisy(rebase_cmd(), $gs->refname);
448 }
449
450 sub cmd_show_ignore {
451 my $url = (::working_head_info('HEAD'))[0];
452 my $gs = Git::SVN->find_by_url($url) || Git::SVN->new;
453 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
454 $gs->traverse_ignore(\*STDOUT, '', $r);
455 }
456
457 sub cmd_multi_init {
458 my $url = shift;
459 unless (defined $_trunk || defined $_branches || defined $_tags) {
460 usage(1);
461 }
462 do_git_init_db();
463 $_prefix = '' unless defined $_prefix;
464 if (defined $url) {
465 $url =~ s#/+$##;
466 init_subdir(@_);
467 }
468 if (defined $_trunk) {
469 my $trunk_ref = $_prefix . 'trunk';
470 # try both old-style and new-style lookups:
471 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
472 unless ($gs_trunk) {
473 my ($trunk_url, $trunk_path) =
474 complete_svn_url($url, $_trunk);
475 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
476 undef, $trunk_ref);
477 }
478 }
479 return unless defined $_branches || defined $_tags;
480 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
481 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
482 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
483 }
484
485 sub cmd_multi_fetch {
486 my $remotes = Git::SVN::read_all_remotes();
487 foreach my $repo_id (sort keys %$remotes) {
488 if ($remotes->{$repo_id}->{url}) {
489 Git::SVN::fetch_all($repo_id, $remotes);
490 }
491 }
492 }
493
494 # this command is special because it requires no metadata
495 sub cmd_commit_diff {
496 my ($ta, $tb, $url) = @_;
497 my $usage = "Usage: $0 commit-diff -r<revision> ".
498 "<tree-ish> <tree-ish> [<URL>]\n";
499 fatal($usage) if (!defined $ta || !defined $tb);
500 my $svn_path;
501 if (!defined $url) {
502 my $gs = eval { Git::SVN->new };
503 if (!$gs) {
504 fatal("Needed URL or usable git-svn --id in ",
505 "the command-line\n", $usage);
506 }
507 $url = $gs->{url};
508 $svn_path = $gs->{path};
509 }
510 unless (defined $_revision) {
511 fatal("-r|--revision is a required argument\n", $usage);
512 }
513 if (defined $_message && defined $_file) {
514 fatal("Both --message/-m and --file/-F specified ",
515 "for the commit message.\n",
516 "I have no idea what you mean\n");
517 }
518 if (defined $_file) {
519 $_message = file_to_s($_file);
520 } else {
521 $_message ||= get_commit_entry($tb)->{log};
522 }
523 my $ra ||= Git::SVN::Ra->new($url);
524 $svn_path ||= $ra->{svn_path};
525 my $r = $_revision;
526 if ($r eq 'HEAD') {
527 $r = $ra->get_latest_revnum;
528 } elsif ($r !~ /^\d+$/) {
529 die "revision argument: $r not understood by git-svn\n";
530 }
531 my %ed_opts = ( r => $r,
532 log => $_message,
533 ra => $ra,
534 tree_a => $ta,
535 tree_b => $tb,
536 editor_cb => sub { print "Committed r$_[0]\n" },
537 svn_path => $svn_path );
538 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
539 print "No changes\n$ta == $tb\n";
540 }
541 }
542
543 ########################### utility functions #########################
544
545 sub rebase_cmd {
546 my @cmd = qw/rebase/;
547 push @cmd, '-v' if $_verbose;
548 push @cmd, qw/--merge/ if $_merge;
549 push @cmd, "--strategy=$_strategy" if $_strategy;
550 @cmd;
551 }
552
553 sub post_fetch_checkout {
554 return if $_no_checkout;
555 my $gs = $Git::SVN::_head or return;
556 return if verify_ref('refs/heads/master^0');
557
558 my $valid_head = verify_ref('HEAD^0');
559 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
560 return if ($valid_head || !verify_ref('HEAD^0'));
561
562 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
563 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
564 return if -f $index;
565
566 chomp(my $bare = `git config --bool --get core.bare`);
567 return if $bare eq 'true';
568 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
569 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
570 print STDERR "Checked out HEAD:\n ",
571 $gs->full_url, " r", $gs->last_rev, "\n";
572 }
573
574 sub complete_svn_url {
575 my ($url, $path) = @_;
576 $path =~ s#/+$##;
577 if ($path !~ m#^[a-z\+]+://#) {
578 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
579 fatal("E: '$path' is not a complete URL ",
580 "and a separate URL is not specified\n");
581 }
582 return ($url, $path);
583 }
584 return ($path, '');
585 }
586
587 sub complete_url_ls_init {
588 my ($ra, $repo_path, $switch, $pfx) = @_;
589 unless ($repo_path) {
590 print STDERR "W: $switch not specified\n";
591 return;
592 }
593 $repo_path =~ s#/+$##;
594 if ($repo_path =~ m#^[a-z\+]+://#) {
595 $ra = Git::SVN::Ra->new($repo_path);
596 $repo_path = '';
597 } else {
598 $repo_path =~ s#^/+##;
599 unless ($ra) {
600 fatal("E: '$repo_path' is not a complete URL ",
601 "and a separate URL is not specified\n");
602 }
603 }
604 my $url = $ra->{url};
605 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
606 my $k = "svn-remote.$gs->{repo_id}.url";
607 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
608 if ($orig_url && ($orig_url ne $gs->{url})) {
609 die "$k already set: $orig_url\n",
610 "wanted to set to: $gs->{url}\n";
611 }
612 command_oneline('config', $k, $gs->{url}) unless $orig_url;
613 my $remote_path = "$ra->{svn_path}/$repo_path/*";
614 $remote_path =~ s#/+#/#g;
615 $remote_path =~ s#^/##g;
616 my ($n) = ($switch =~ /^--(\w+)/);
617 if (length $pfx && $pfx !~ m#/$#) {
618 die "--prefix='$pfx' must have a trailing slash '/'\n";
619 }
620 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
621 "$remote_path:refs/remotes/$pfx*");
622 }
623
624 sub verify_ref {
625 my ($ref) = @_;
626 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
627 { STDERR => 0 }); };
628 }
629
630 sub get_tree_from_treeish {
631 my ($treeish) = @_;
632 # $treeish can be a symbolic ref, too:
633 my $type = command_oneline(qw/cat-file -t/, $treeish);
634 my $expected;
635 while ($type eq 'tag') {
636 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
637 }
638 if ($type eq 'commit') {
639 $expected = (grep /^tree /, command(qw/cat-file commit/,
640 $treeish))[0];
641 ($expected) = ($expected =~ /^tree ($sha1)$/o);
642 die "Unable to get tree from $treeish\n" unless $expected;
643 } elsif ($type eq 'tree') {
644 $expected = $treeish;
645 } else {
646 die "$treeish is a $type, expected tree, tag or commit\n";
647 }
648 return $expected;
649 }
650
651 sub get_commit_entry {
652 my ($treeish) = shift;
653 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
654 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
655 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
656 open my $log_fh, '>', $commit_editmsg or croak $!;
657
658 my $type = command_oneline(qw/cat-file -t/, $treeish);
659 if ($type eq 'commit' || $type eq 'tag') {
660 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
661 $type, $treeish);
662 my $in_msg = 0;
663 while (<$msg_fh>) {
664 if (!$in_msg) {
665 $in_msg = 1 if (/^\s*$/);
666 } elsif (/^git-svn-id: /) {
667 # skip this for now, we regenerate the
668 # correct one on re-fetch anyways
669 # TODO: set *:merge properties or like...
670 } else {
671 print $log_fh $_ or croak $!;
672 }
673 }
674 command_close_pipe($msg_fh, $ctx);
675 }
676 close $log_fh or croak $!;
677
678 if ($_edit || ($type eq 'tree')) {
679 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
680 # TODO: strip out spaces, comments, like git-commit.sh
681 system($editor, $commit_editmsg);
682 }
683 rename $commit_editmsg, $commit_msg or croak $!;
684 open $log_fh, '<', $commit_msg or croak $!;
685 { local $/; chomp($log_entry{log} = <$log_fh>); }
686 close $log_fh or croak $!;
687 unlink $commit_msg;
688 \%log_entry;
689 }
690
691 sub s_to_file {
692 my ($str, $file, $mode) = @_;
693 open my $fd,'>',$file or croak $!;
694 print $fd $str,"\n" or croak $!;
695 close $fd or croak $!;
696 chmod ($mode &~ umask, $file) if (defined $mode);
697 }
698
699 sub file_to_s {
700 my $file = shift;
701 open my $fd,'<',$file or croak "$!: file: $file\n";
702 local $/;
703 my $ret = <$fd>;
704 close $fd or croak $!;
705 $ret =~ s/\s*$//s;
706 return $ret;
707 }
708
709 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
710 sub load_authors {
711 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
712 my $log = $cmd eq 'log';
713 while (<$authors>) {
714 chomp;
715 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
716 my ($user, $name, $email) = ($1, $2, $3);
717 if ($log) {
718 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
719 } else {
720 $users{$user} = [$name, $email];
721 }
722 }
723 close $authors or croak $!;
724 }
725
726 # convert GetOpt::Long specs for use by git-config
727 sub read_repo_config {
728 return unless -d $ENV{GIT_DIR};
729 my $opts = shift;
730 my @config_only;
731 foreach my $o (keys %$opts) {
732 # if we have mixedCase and a long option-only, then
733 # it's a config-only variable that we don't need for
734 # the command-line.
735 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
736 my $v = $opts->{$o};
737 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
738 $key =~ s/-//g;
739 my $arg = 'git-config';
740 $arg .= ' --int' if ($o =~ /[:=]i$/);
741 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
742 if (ref $v eq 'ARRAY') {
743 chomp(my @tmp = `$arg --get-all svn.$key`);
744 @$v = @tmp if @tmp;
745 } else {
746 chomp(my $tmp = `$arg --get svn.$key`);
747 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
748 $$v = $tmp;
749 }
750 }
751 }
752 delete @$opts{@config_only} if @config_only;
753 }
754
755 sub extract_metadata {
756 my $id = shift or return (undef, undef, undef);
757 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
758 \s([a-f\d\-]+)$/x);
759 if (!defined $rev || !$uuid || !$url) {
760 # some of the original repositories I made had
761 # identifiers like this:
762 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
763 }
764 return ($url, $rev, $uuid);
765 }
766
767 sub cmt_metadata {
768 return extract_metadata((grep(/^git-svn-id: /,
769 command(qw/cat-file commit/, shift)))[-1]);
770 }
771
772 sub working_head_info {
773 my ($head, $refs) = @_;
774 my ($url, $rev, $uuid);
775 my ($fh, $ctx) = command_output_pipe('rev-list', $head);
776 while (<$fh>) {
777 chomp;
778 ($url, $rev, $uuid) = cmt_metadata($_);
779 last if (defined $url && defined $rev && defined $uuid);
780 unshift @$refs, $_ if $refs;
781 }
782 close $fh; # break the pipe
783 ($url, $rev, $uuid);
784 }
785
786 package Git::SVN;
787 use strict;
788 use warnings;
789 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
790 $_repack $_repack_flags $_use_svm_props $_head
791 $_use_svnsync_props $no_reuse_existing/;
792 use Carp qw/croak/;
793 use File::Path qw/mkpath/;
794 use File::Copy qw/copy/;
795 use IPC::Open3;
796
797 my $_repack_nr;
798 # properties that we do not log:
799 my %SKIP_PROP;
800 BEGIN {
801 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
802 svn:special svn:executable
803 svn:entry:committed-rev
804 svn:entry:last-author
805 svn:entry:uuid
806 svn:entry:committed-date/;
807
808 # some options are read globally, but can be overridden locally
809 # per [svn-remote "..."] section. Command-line options will *NOT*
810 # override options set in an [svn-remote "..."] section
811 my $e;
812 foreach (qw/follow_parent no_metadata use_svm_props
813 use_svnsync_props/) {
814 my $key = $_;
815 $key =~ tr/_//d;
816 $e .= "sub $_ {
817 my (\$self) = \@_;
818 return \$self->{-$_} if exists \$self->{-$_};
819 my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
820 eval { command_oneline(qw/config --get/, \$k) };
821 if (\$@) {
822 \$self->{-$_} = \$Git::SVN::_$_;
823 } else {
824 my \$v = command_oneline(qw/config --bool/,\$k);
825 \$self->{-$_} = \$v eq 'false' ? 0 : 1;
826 }
827 return \$self->{-$_} }\n";
828 }
829 $e .= "1;\n";
830 eval $e or die $@;
831 }
832
833 my %LOCKFILES;
834 END { unlink keys %LOCKFILES if %LOCKFILES }
835
836 sub resolve_local_globs {
837 my ($url, $fetch, $glob_spec) = @_;
838 return unless defined $glob_spec;
839 my $ref = $glob_spec->{ref};
840 my $path = $glob_spec->{path};
841 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
842 next unless m#^refs/remotes/$ref->{regex}$#;
843 my $p = $1;
844 my $pathname = $path->full_path($p);
845 my $refname = $ref->full_path($p);
846 if (my $existing = $fetch->{$pathname}) {
847 if ($existing ne $refname) {
848 die "Refspec conflict:\n",
849 "existing: refs/remotes/$existing\n",
850 " globbed: refs/remotes/$refname\n";
851 }
852 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
853 $u =~ s!^\Q$url\E(/|$)!! or die
854 "refs/remotes/$refname: '$url' not found in '$u'\n";
855 if ($pathname ne $u) {
856 warn "W: Refspec glob conflict ",
857 "(ref: refs/remotes/$refname):\n",
858 "expected path: $pathname\n",
859 " real path: $u\n",
860 "Continuing ahead with $u\n";
861 next;
862 }
863 } else {
864 $fetch->{$pathname} = $refname;
865 }
866 }
867 }
868
869 sub parse_revision_argument {
870 my ($base, $head) = @_;
871 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
872 return ($base, $head);
873 }
874 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
875 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
876 return ($head, $head) if ($::_revision eq 'HEAD');
877 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
878 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
879 die "revision argument: $::_revision not understood by git-svn\n";
880 }
881
882 sub fetch_all {
883 my ($repo_id, $remotes) = @_;
884 if (ref $repo_id) {
885 my $gs = $repo_id;
886 $repo_id = undef;
887 $repo_id = $gs->{repo_id};
888 }
889 $remotes ||= read_all_remotes();
890 my $remote = $remotes->{$repo_id} or
891 die "[svn-remote \"$repo_id\"] unknown\n";
892 my $fetch = $remote->{fetch};
893 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
894 my (@gs, @globs);
895 my $ra = Git::SVN::Ra->new($url);
896 my $uuid = $ra->get_uuid;
897 my $head = $ra->get_latest_revnum;
898 my $base = defined $fetch ? $head : 0;
899
900 # read the max revs for wildcard expansion (branches/*, tags/*)
901 foreach my $t (qw/branches tags/) {
902 defined $remote->{$t} or next;
903 push @globs, $remote->{$t};
904 my $max_rev = eval { tmp_config(qw/--int --get/,
905 "svn-remote.$repo_id.${t}-maxRev") };
906 if (defined $max_rev && ($max_rev < $base)) {
907 $base = $max_rev;
908 } elsif (!defined $max_rev) {
909 $base = 0;
910 }
911 }
912
913 if ($fetch) {
914 foreach my $p (sort keys %$fetch) {
915 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
916 my $lr = $gs->rev_db_max;
917 if (defined $lr) {
918 $base = $lr if ($lr < $base);
919 }
920 push @gs, $gs;
921 }
922 }
923
924 ($base, $head) = parse_revision_argument($base, $head);
925 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
926 }
927
928 sub read_all_remotes {
929 my $r = {};
930 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
931 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
932 $r->{$1}->{fetch}->{$2} = $3;
933 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
934 $r->{$1}->{url} = $2;
935 } elsif (m!^(.+)\.(branches|tags)=
936 (.*):refs/remotes/(.+)\s*$/!x) {
937 my ($p, $g) = ($3, $4);
938 my $rs = $r->{$1}->{$2} = {
939 t => $2,
940 remote => $1,
941 path => Git::SVN::GlobSpec->new($p),
942 ref => Git::SVN::GlobSpec->new($g) };
943 if (length($rs->{ref}->{right}) != 0) {
944 die "The '*' glob character must be the last ",
945 "character of '$g'\n";
946 }
947 }
948 }
949 $r;
950 }
951
952 sub init_vars {
953 if (defined $_repack) {
954 $_repack = 1000 if ($_repack <= 0);
955 $_repack_nr = $_repack;
956 $_repack_flags ||= '-d';
957 }
958 }
959
960 sub verify_remotes_sanity {
961 return unless -d $ENV{GIT_DIR};
962 my %seen;
963 foreach (command(qw/config -l/)) {
964 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
965 if ($seen{$1}) {
966 die "Remote ref refs/remote/$1 is tracked by",
967 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
968 "Please resolve this ambiguity in ",
969 "your git configuration file before ",
970 "continuing\n";
971 }
972 $seen{$1} = $_;
973 }
974 }
975 }
976
977 # we allow more chars than remotes2config.sh...
978 sub sanitize_remote_name {
979 my ($name) = @_;
980 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
981 $name;
982 }
983
984 sub find_existing_remote {
985 my ($url, $remotes) = @_;
986 return undef if $no_reuse_existing;
987 my $existing;
988 foreach my $repo_id (keys %$remotes) {
989 my $u = $remotes->{$repo_id}->{url} or next;
990 next if $u ne $url;
991 $existing = $repo_id;
992 last;
993 }
994 $existing;
995 }
996
997 sub init_remote_config {
998 my ($self, $url, $no_write) = @_;
999 $url =~ s!/+$!!; # strip trailing slash
1000 my $r = read_all_remotes();
1001 my $existing = find_existing_remote($url, $r);
1002 if ($existing) {
1003 unless ($no_write) {
1004 print STDERR "Using existing ",
1005 "[svn-remote \"$existing\"]\n";
1006 }
1007 $self->{repo_id} = $existing;
1008 } else {
1009 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1010 $existing = find_existing_remote($min_url, $r);
1011 if ($existing) {
1012 unless ($no_write) {
1013 print STDERR "Using existing ",
1014 "[svn-remote \"$existing\"]\n";
1015 }
1016 $self->{repo_id} = $existing;
1017 }
1018 if ($min_url ne $url) {
1019 unless ($no_write) {
1020 print STDERR "Using higher level of URL: ",
1021 "$url => $min_url\n";
1022 }
1023 my $old_path = $self->{path};
1024 $self->{path} = $url;
1025 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1026 if (length $old_path) {
1027 $self->{path} .= "/$old_path";
1028 }
1029 $url = $min_url;
1030 }
1031 }
1032 my $orig_url;
1033 if (!$existing) {
1034 # verify that we aren't overwriting anything:
1035 $orig_url = eval {
1036 command_oneline('config', '--get',
1037 "svn-remote.$self->{repo_id}.url")
1038 };
1039 if ($orig_url && ($orig_url ne $url)) {
1040 die "svn-remote.$self->{repo_id}.url already set: ",
1041 "$orig_url\nwanted to set to: $url\n";
1042 }
1043 }
1044 my ($xrepo_id, $xpath) = find_ref($self->refname);
1045 if (defined $xpath) {
1046 die "svn-remote.$xrepo_id.fetch already set to track ",
1047 "$xpath:refs/remotes/", $self->refname, "\n";
1048 }
1049 unless ($no_write) {
1050 command_noisy('config',
1051 "svn-remote.$self->{repo_id}.url", $url);
1052 command_noisy('config', '--add',
1053 "svn-remote.$self->{repo_id}.fetch",
1054 "$self->{path}:".$self->refname);
1055 }
1056 $self->{url} = $url;
1057 }
1058
1059 sub find_by_url { # repos_root and, path are optional
1060 my ($class, $full_url, $repos_root, $path) = @_;
1061 return undef unless defined $full_url;
1062 my $remotes = read_all_remotes();
1063 if (defined $full_url && defined $repos_root && !defined $path) {
1064 $path = $full_url;
1065 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1066 }
1067 foreach my $repo_id (keys %$remotes) {
1068 my $u = $remotes->{$repo_id}->{url} or next;
1069 next if defined $repos_root && $repos_root ne $u;
1070
1071 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1072 foreach (qw/branches tags/) {
1073 resolve_local_globs($u, $fetch,
1074 $remotes->{$repo_id}->{$_});
1075 }
1076 my $p = $path;
1077 unless (defined $p) {
1078 $p = $full_url;
1079 $p =~ s#^\Q$u\E(?:/|$)## or next;
1080 }
1081 foreach my $f (keys %$fetch) {
1082 next if $f ne $p;
1083 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1084 }
1085 }
1086 undef;
1087 }
1088
1089 sub init {
1090 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1091 my $self = _new($class, $repo_id, $ref_id, $path);
1092 if (defined $url) {
1093 $self->init_remote_config($url, $no_write);
1094 }
1095 $self;
1096 }
1097
1098 sub find_ref {
1099 my ($ref_id) = @_;
1100 foreach (command(qw/config -l/)) {
1101 next unless m!^svn-remote\.(.+)\.fetch=
1102 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1103 my ($repo_id, $path, $ref) = ($1, $2, $3);
1104 if ($ref eq $ref_id) {
1105 $path = '' if ($path =~ m#^\./?#);
1106 return ($repo_id, $path);
1107 }
1108 }
1109 (undef, undef, undef);
1110 }
1111
1112 sub new {
1113 my ($class, $ref_id, $repo_id, $path) = @_;
1114 if (defined $ref_id && !defined $repo_id && !defined $path) {
1115 ($repo_id, $path) = find_ref($ref_id);
1116 if (!defined $repo_id) {
1117 die "Could not find a \"svn-remote.*.fetch\" key ",
1118 "in the repository configuration matching: ",
1119 "refs/remotes/$ref_id\n";
1120 }
1121 }
1122 my $self = _new($class, $repo_id, $ref_id, $path);
1123 if (!defined $self->{path} || !length $self->{path}) {
1124 my $fetch = command_oneline('config', '--get',
1125 "svn-remote.$repo_id.fetch",
1126 ":refs/remotes/$ref_id\$") or
1127 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1128 "\":refs/remotes/$ref_id\$\" in config\n";
1129 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1130 }
1131 $self->{url} = command_oneline('config', '--get',
1132 "svn-remote.$repo_id.url") or
1133 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1134 $self->rebuild;
1135 $self;
1136 }
1137
1138 sub refname { "refs/remotes/$_[0]->{ref_id}" }
1139
1140 sub svm_uuid {
1141 my ($self) = @_;
1142 return $self->{svm}->{uuid} if $self->svm;
1143 $self->ra;
1144 unless ($self->{svm}) {
1145 die "SVM UUID not cached, and reading remotely failed\n";
1146 }
1147 $self->{svm}->{uuid};
1148 }
1149
1150 sub svm {
1151 my ($self) = @_;
1152 return $self->{svm} if $self->{svm};
1153 my $svm;
1154 # see if we have it in our config, first:
1155 eval {
1156 my $section = "svn-remote.$self->{repo_id}";
1157 $svm = {
1158 source => tmp_config('--get', "$section.svm-source"),
1159 uuid => tmp_config('--get', "$section.svm-uuid"),
1160 replace => tmp_config('--get', "$section.svm-replace"),
1161 }
1162 };
1163 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1164 $self->{svm} = $svm;
1165 }
1166 $self->{svm};
1167 }
1168
1169 sub _set_svm_vars {
1170 my ($self, $ra) = @_;
1171 return $ra if $self->svm;
1172
1173 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1174 "(svm:source, svm:uuid) ",
1175 "from the following URLs:\n" );
1176 sub read_svm_props {
1177 my ($self, $ra, $path, $r) = @_;
1178 my $props = ($ra->get_dir($path, $r))[2];
1179 my $src = $props->{'svm:source'};
1180 my $uuid = $props->{'svm:uuid'};
1181 return undef if (!$src || !$uuid);
1182
1183 chomp($src, $uuid);
1184
1185 $uuid =~ m{^[0-9a-f\-]{30,}$}
1186 or die "doesn't look right - svm:uuid is '$uuid'\n";
1187
1188 # the '!' is used to mark the repos_root!/relative/path
1189 $src =~ s{/?!/?}{/};
1190 $src =~ s{/+$}{}; # no trailing slashes please
1191 # username is of no interest
1192 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1193
1194 my $replace = $ra->{url};
1195 $replace .= "/$path" if length $path;
1196
1197 my $section = "svn-remote.$self->{repo_id}";
1198 tmp_config("$section.svm-source", $src);
1199 tmp_config("$section.svm-replace", $replace);
1200 tmp_config("$section.svm-uuid", $uuid);
1201 $self->{svm} = {
1202 source => $src,
1203 uuid => $uuid,
1204 replace => $replace
1205 };
1206 }
1207
1208 my $r = $ra->get_latest_revnum;
1209 my $path = $self->{path};
1210 my %tried;
1211 while (length $path) {
1212 unless ($tried{"$self->{url}/$path"}) {
1213 return $ra if $self->read_svm_props($ra, $path, $r);
1214 $tried{"$self->{url}/$path"} = 1;
1215 }
1216 $path =~ s#/?[^/]+$##;
1217 }
1218 die "Path: '$path' should be ''\n" if $path ne '';
1219 return $ra if $self->read_svm_props($ra, $path, $r);
1220 $tried{"$self->{url}/$path"} = 1;
1221
1222 if ($ra->{repos_root} eq $self->{url}) {
1223 die @err, (map { " $_\n" } keys %tried), "\n";
1224 }
1225
1226 # nope, make sure we're connected to the repository root:
1227 my $ok;
1228 my @tried_b;
1229 $path = $ra->{svn_path};
1230 $ra = Git::SVN::Ra->new($ra->{repos_root});
1231 while (length $path) {
1232 unless ($tried{"$ra->{url}/$path"}) {
1233 $ok = $self->read_svm_props($ra, $path, $r);
1234 last if $ok;
1235 $tried{"$ra->{url}/$path"} = 1;
1236 }
1237 $path =~ s#/?[^/]+$##;
1238 }
1239 die "Path: '$path' should be ''\n" if $path ne '';
1240 $ok ||= $self->read_svm_props($ra, $path, $r);
1241 $tried{"$ra->{url}/$path"} = 1;
1242 if (!$ok) {
1243 die @err, (map { " $_\n" } keys %tried), "\n";
1244 }
1245 Git::SVN::Ra->new($self->{url});
1246 }
1247
1248 sub svnsync {
1249 my ($self) = @_;
1250 return $self->{svnsync} if $self->{svnsync};
1251
1252 if ($self->no_metadata) {
1253 die "Can't have both 'noMetadata' and ",
1254 "'useSvnsyncProps' options set!\n";
1255 }
1256 if ($self->rewrite_root) {
1257 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1258 "options set!\n";
1259 }
1260
1261 my $svnsync;
1262 # see if we have it in our config, first:
1263 eval {
1264 my $section = "svn-remote.$self->{repo_id}";
1265 $svnsync = {
1266 url => tmp_config('--get', "$section.svnsync-url"),
1267 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1268 }
1269 };
1270 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1271 return $self->{svnsync} = $svnsync;
1272 }
1273
1274 my $err = "useSvnsyncProps set, but failed to read " .
1275 "svnsync property: svn:sync-from-";
1276 my $rp = $self->ra->rev_proplist(0);
1277
1278 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1279 $url =~ m{^[a-z\+]+://} or
1280 die "doesn't look right - svn:sync-from-url is '$url'\n";
1281
1282 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1283 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1284 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1285
1286 my $section = "svn-remote.$self->{repo_id}";
1287 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1288 tmp_config('--add', "$section.svnsync-url", $url);
1289 return $self->{svnsync} = { url => $url, uuid => $uuid };
1290 }
1291
1292 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1293 # remote lookup (useful for 'git svn log').
1294 sub ra_uuid {
1295 my ($self) = @_;
1296 unless ($self->{ra_uuid}) {
1297 my $key = "svn-remote.$self->{repo_id}.uuid";
1298 my $uuid = eval { tmp_config('--get', $key) };
1299 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1300 $self->{ra_uuid} = $uuid;
1301 } else {
1302 die "ra_uuid called without URL\n" unless $self->{url};
1303 $self->{ra_uuid} = $self->ra->get_uuid;
1304 tmp_config('--add', $key, $self->{ra_uuid});
1305 }
1306 }
1307 $self->{ra_uuid};
1308 }
1309
1310 sub ra {
1311 my ($self) = shift;
1312 my $ra = Git::SVN::Ra->new($self->{url});
1313 if ($self->use_svm_props && !$self->{svm}) {
1314 if ($self->no_metadata) {
1315 die "Can't have both 'noMetadata' and ",
1316 "'useSvmProps' options set!\n";
1317 } elsif ($self->use_svnsync_props) {
1318 die "Can't have both 'useSvnsyncProps' and ",
1319 "'useSvmProps' options set!\n";
1320 }
1321 $ra = $self->_set_svm_vars($ra);
1322 $self->{-want_revprops} = 1;
1323 }
1324 $ra;
1325 }
1326
1327 sub rel_path {
1328 my ($self) = @_;
1329 my $repos_root = $self->ra->{repos_root};
1330 return $self->{path} if ($self->{url} eq $repos_root);
1331 die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1332 $self->ra->{url}, " path: $self->{path}, URL: $self->{url}\n";
1333 }
1334
1335 sub traverse_ignore {
1336 my ($self, $fh, $path, $r) = @_;
1337 $path =~ s#^/+##g;
1338 my $ra = $self->ra;
1339 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1340 my $p = $path;
1341 $p =~ s#^\Q$ra->{svn_path}\E/##;
1342 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1343 if (my $s = $props->{'svn:ignore'}) {
1344 $s =~ s/[\r\n]+/\n/g;
1345 chomp $s;
1346 if (length $p == 0) {
1347 $s =~ s#\n#\n/$p#g;
1348 print $fh "/$s\n";
1349 } else {
1350 $s =~ s#\n#\n/$p/#g;
1351 print $fh "/$p/$s\n";
1352 }
1353 }
1354 foreach (sort keys %$dirent) {
1355 next if $dirent->{$_}->kind != $SVN::Node::dir;
1356 $self->traverse_ignore($fh, "$path/$_", $r);
1357 }
1358 }
1359
1360 sub last_rev { ($_[0]->last_rev_commit)[0] }
1361 sub last_commit { ($_[0]->last_rev_commit)[1] }
1362
1363 # returns the newest SVN revision number and newest commit SHA1
1364 sub last_rev_commit {
1365 my ($self) = @_;
1366 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1367 return ($self->{last_rev}, $self->{last_commit});
1368 }
1369 my $c = ::verify_ref($self->refname.'^0');
1370 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1371 my $rev = (::cmt_metadata($c))[1];
1372 if (defined $rev) {
1373 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1374 return ($rev, $c);
1375 }
1376 }
1377 my $db_path = $self->db_path;
1378 unless (-e $db_path) {
1379 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1380 return (undef, undef);
1381 }
1382 my $offset = -41; # from tail
1383 my $rl;
1384 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1385 sysseek($fh, $offset, 2); # don't care for errors
1386 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1387 chomp $rl;
1388 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1389 $offset -= 41;
1390 sysseek($fh, $offset, 2); # don't care for errors
1391 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1392 chomp $rl;
1393 }
1394 if ($c && $c ne $rl) {
1395 die "$db_path and ", $self->refname,
1396 " inconsistent!:\n$c != $rl\n";
1397 }
1398 my $rev = sysseek($fh, 0, 1) or croak $!;
1399 $rev = ($rev - 41) / 41;
1400 close $fh or croak $!;
1401 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1402 return ($rev, $c);
1403 }
1404
1405 sub get_fetch_range {
1406 my ($self, $min, $max) = @_;
1407 $max ||= $self->ra->get_latest_revnum;
1408 $min ||= $self->rev_db_max;
1409 (++$min, $max);
1410 }
1411
1412 sub tmp_config {
1413 my (@args) = @_;
1414 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1415 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1416 if (-e $old_def_config && ! -e $config) {
1417 rename $old_def_config, $config or
1418 die "Failed rename $old_def_config => $config: $!\n";
1419 }
1420 my $old_config = $ENV{GIT_CONFIG};
1421 $ENV{GIT_CONFIG} = $config;
1422 $@ = undef;
1423 my @ret = eval {
1424 unless (-f $config) {
1425 mkfile($config);
1426 open my $fh, '>', $config or
1427 die "Can't open $config: $!\n";
1428 print $fh "; This file is used internally by ",
1429 "git-svn\n" or die
1430 "Couldn't write to $config: $!\n";
1431 print $fh "; You should not have to edit it\n" or
1432 die "Couldn't write to $config: $!\n";
1433 close $fh or die "Couldn't close $config: $!\n";
1434 }
1435 command('config', @args);
1436 };
1437 my $err = $@;
1438 if (defined $old_config) {
1439 $ENV{GIT_CONFIG} = $old_config;
1440 } else {
1441 delete $ENV{GIT_CONFIG};
1442 }
1443 die $err if $err;
1444 wantarray ? @ret : $ret[0];
1445 }
1446
1447 sub tmp_index_do {
1448 my ($self, $sub) = @_;
1449 my $old_index = $ENV{GIT_INDEX_FILE};
1450 $ENV{GIT_INDEX_FILE} = $self->{index};
1451 $@ = undef;
1452 my @ret = eval {
1453 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1454 mkpath([$dir]) unless -d $dir;
1455 &$sub;
1456 };
1457 my $err = $@;
1458 if (defined $old_index) {
1459 $ENV{GIT_INDEX_FILE} = $old_index;
1460 } else {
1461 delete $ENV{GIT_INDEX_FILE};
1462 }
1463 die $err if $err;
1464 wantarray ? @ret : $ret[0];
1465 }
1466
1467 sub assert_index_clean {
1468 my ($self, $treeish) = @_;
1469
1470 $self->tmp_index_do(sub {
1471 command_noisy('read-tree', $treeish) unless -e $self->{index};
1472 my $x = command_oneline('write-tree');
1473 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1474 /^tree ($::sha1)/mo);
1475 return if $y eq $x;
1476
1477 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1478 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1479 command_noisy('read-tree', $treeish);
1480 $x = command_oneline('write-tree');
1481 if ($y ne $x) {
1482 ::fatal "trees ($treeish) $y != $x\n",
1483 "Something is seriously wrong...\n";
1484 }
1485 });
1486 }
1487
1488 sub get_commit_parents {
1489 my ($self, $log_entry) = @_;
1490 my (%seen, @ret, @tmp);
1491 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1492 if (my $ip = $self->{inject_parents}) {
1493 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1494 push @tmp, $commit;
1495 }
1496 }
1497 if (my $cur = ::verify_ref($self->refname.'^0')) {
1498 push @tmp, $cur;
1499 }
1500 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1501 while (my $p = shift @tmp) {
1502 next if $seen{$p};
1503 $seen{$p} = 1;
1504 push @ret, $p;
1505 # MAXPARENT is defined to 16 in commit-tree.c:
1506 last if @ret >= 16;
1507 }
1508 if (@tmp) {
1509 die "r$log_entry->{revision}: No room for parents:\n\t",
1510 join("\n\t", @tmp), "\n";
1511 }
1512 @ret;
1513 }
1514
1515 sub rewrite_root {
1516 my ($self) = @_;
1517 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1518 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1519 my $rwr = eval { command_oneline(qw/config --get/, $k) };
1520 if ($rwr) {
1521 $rwr =~ s#/+$##;
1522 if ($rwr !~ m#^[a-z\+]+://#) {
1523 die "$rwr is not a valid URL (key: $k)\n";
1524 }
1525 }
1526 $self->{-rewrite_root} = $rwr;
1527 }
1528
1529 sub metadata_url {
1530 my ($self) = @_;
1531 ($self->rewrite_root || $self->{url}) .
1532 (length $self->{path} ? '/' . $self->{path} : '');
1533 }
1534
1535 sub full_url {
1536 my ($self) = @_;
1537 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1538 }
1539
1540 sub do_git_commit {
1541 my ($self, $log_entry) = @_;
1542 my $lr = $self->last_rev;
1543 if (defined $lr && $lr >= $log_entry->{revision}) {
1544 die "Last fetched revision of ", $self->refname,
1545 " was r$lr, but we are about to fetch: ",
1546 "r$log_entry->{revision}!\n";
1547 }
1548 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1549 croak "$log_entry->{revision} = $c already exists! ",
1550 "Why are we refetching it?\n";
1551 }
1552 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1553 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1554 $log_entry->{email};
1555 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1556
1557 my $tree = $log_entry->{tree};
1558 if (!defined $tree) {
1559 $tree = $self->tmp_index_do(sub {
1560 command_oneline('write-tree') });
1561 }
1562 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1563
1564 my @exec = ('git-commit-tree', $tree);
1565 foreach ($self->get_commit_parents($log_entry)) {
1566 push @exec, '-p', $_;
1567 }
1568 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1569 or croak $!;
1570 print $msg_fh $log_entry->{log} or croak $!;
1571 unless ($self->no_metadata) {
1572 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1573 or croak $!;
1574 }
1575 $msg_fh->flush == 0 or croak $!;
1576 close $msg_fh or croak $!;
1577 chomp(my $commit = do { local $/; <$out_fh> });
1578 close $out_fh or croak $!;
1579 waitpid $pid, 0;
1580 croak $? if $?;
1581 if ($commit !~ /^$::sha1$/o) {
1582 die "Failed to commit, invalid sha1: $commit\n";
1583 }
1584
1585 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1586
1587 $self->{last_rev} = $log_entry->{revision};
1588 $self->{last_commit} = $commit;
1589 print "r$log_entry->{revision}";
1590 if (defined $log_entry->{svm_revision}) {
1591 print " (\@$log_entry->{svm_revision})";
1592 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1593 0, $self->svm_uuid);
1594 }
1595 print " = $commit ($self->{ref_id})\n";
1596 if (defined $_repack && (--$_repack_nr == 0)) {
1597 $_repack_nr = $_repack;
1598 # repack doesn't use any arguments with spaces in them, does it?
1599 print "Running git repack $_repack_flags ...\n";
1600 command_noisy('repack', split(/\s+/, $_repack_flags));
1601 print "Done repacking\n";
1602 }
1603 return $commit;
1604 }
1605
1606 sub match_paths {
1607 my ($self, $paths, $r) = @_;
1608 return 1 if $self->{path} eq '';
1609 if (my $path = $paths->{"/$self->{path}"}) {
1610 return ($path->{action} eq 'D') ? 0 : 1;
1611 }
1612 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1613 if (grep /$self->{path_regex}/, keys %$paths) {
1614 return 1;
1615 }
1616 my $c = '';
1617 foreach (split m#/#, $self->{path}) {
1618 $c .= "/$_";
1619 next unless ($paths->{$c} &&
1620 ($paths->{$c}->{action} =~ /^[AR]$/));
1621 if ($self->ra->check_path($self->{path}, $r) ==
1622 $SVN::Node::dir) {
1623 return 1;
1624 }
1625 }
1626 return 0;
1627 }
1628
1629 sub find_parent_branch {
1630 my ($self, $paths, $rev) = @_;
1631 return undef unless $self->follow_parent;
1632 unless (defined $paths) {
1633 my $err_handler = $SVN::Error::handler;
1634 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1635 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1636 $paths =
1637 Git::SVN::Ra::dup_changed_paths($_[0]) });
1638 $SVN::Error::handler = $err_handler;
1639 }
1640 return undef unless defined $paths;
1641
1642 # look for a parent from another branch:
1643 my @b_path_components = split m#/#, $self->rel_path;
1644 my @a_path_components;
1645 my $i;
1646 while (@b_path_components) {
1647 $i = $paths->{'/'.join('/', @b_path_components)};
1648 last if $i && defined $i->{copyfrom_path};
1649 unshift(@a_path_components, pop(@b_path_components));
1650 }
1651 return undef unless defined $i && defined $i->{copyfrom_path};
1652 my $branch_from = $i->{copyfrom_path};
1653 if (@a_path_components) {
1654 print STDERR "branch_from: $branch_from => ";
1655 $branch_from .= '/'.join('/', @a_path_components);
1656 print STDERR $branch_from, "\n";
1657 }
1658 my $r = $i->{copyfrom_rev};
1659 my $repos_root = $self->ra->{repos_root};
1660 my $url = $self->ra->{url};
1661 my $new_url = $repos_root . $branch_from;
1662 print STDERR "Found possible branch point: ",
1663 "$new_url => ", $self->full_url, ", $r\n";
1664 $branch_from =~ s#^/##;
1665 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1666 unless ($gs) {
1667 my $ref_id = $self->{ref_id};
1668 $ref_id =~ s/\@\d+$//;
1669 $ref_id .= "\@$r";
1670 # just grow a tail if we're not unique enough :x
1671 $ref_id .= '-' while find_ref($ref_id);
1672 print STDERR "Initializing parent: $ref_id\n";
1673 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1674 }
1675 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1676 if (!defined $r0 || !defined $parent) {
1677 $gs->fetch(0, $r);
1678 ($r0, $parent) = $gs->last_rev_commit;
1679 }
1680 if (defined $r0 && defined $parent) {
1681 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1682 $self->assert_index_clean($parent);
1683 my $ed;
1684 if ($self->ra->can_do_switch) {
1685 print STDERR "Following parent with do_switch\n";
1686 # do_switch works with svn/trunk >= r22312, but that
1687 # is not included with SVN 1.4.3 (the latest version
1688 # at the moment), so we can't rely on it
1689 $self->{last_commit} = $parent;
1690 $ed = SVN::Git::Fetcher->new($self);
1691 $gs->ra->gs_do_switch($r0, $rev, $gs,
1692 $self->full_url, $ed)
1693 or die "SVN connection failed somewhere...\n";
1694 } else {
1695 print STDERR "Following parent with do_update\n";
1696 $ed = SVN::Git::Fetcher->new($self);
1697 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1698 or die "SVN connection failed somewhere...\n";
1699 }
1700 print STDERR "Successfully followed parent\n";
1701 return $self->make_log_entry($rev, [$parent], $ed);
1702 }
1703 return undef;
1704 }
1705
1706 sub do_fetch {
1707 my ($self, $paths, $rev) = @_;
1708 my $ed;
1709 my ($last_rev, @parents);
1710 if (my $lc = $self->last_commit) {
1711 # we can have a branch that was deleted, then re-added
1712 # under the same name but copied from another path, in
1713 # which case we'll have multiple parents (we don't
1714 # want to break the original ref, nor lose copypath info):
1715 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1716 push @{$log_entry->{parents}}, $lc;
1717 return $log_entry;
1718 }
1719 $ed = SVN::Git::Fetcher->new($self);
1720 $last_rev = $self->{last_rev};
1721 $ed->{c} = $lc;
1722 @parents = ($lc);
1723 } else {
1724 $last_rev = $rev;
1725 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1726 return $log_entry;
1727 }
1728 $ed = SVN::Git::Fetcher->new($self);
1729 }
1730 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1731 die "SVN connection failed somewhere...\n";
1732 }
1733 $self->make_log_entry($rev, \@parents, $ed);
1734 }
1735
1736 sub get_untracked {
1737 my ($self, $ed) = @_;
1738 my @out;
1739 my $h = $ed->{empty};
1740 foreach (sort keys %$h) {
1741 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1742 push @out, " $act: " . uri_encode($_);
1743 warn "W: $act: $_\n";
1744 }
1745 foreach my $t (qw/dir_prop file_prop/) {
1746 $h = $ed->{$t} or next;
1747 foreach my $path (sort keys %$h) {
1748 my $ppath = $path eq '' ? '.' : $path;
1749 foreach my $prop (sort keys %{$h->{$path}}) {
1750 next if $SKIP_PROP{$prop};
1751 my $v = $h->{$path}->{$prop};
1752 my $t_ppath_prop = "$t: " .
1753 uri_encode($ppath) . ' ' .
1754 uri_encode($prop);
1755 if (defined $v) {
1756 push @out, " +$t_ppath_prop " .
1757 uri_encode($v);
1758 } else {
1759 push @out, " -$t_ppath_prop";
1760 }
1761 }
1762 }
1763 }
1764 foreach my $t (qw/absent_file absent_directory/) {
1765 $h = $ed->{$t} or next;
1766 foreach my $parent (sort keys %$h) {
1767 foreach my $path (sort @{$h->{$parent}}) {
1768 push @out, " $t: " .
1769 uri_encode("$parent/$path");
1770 warn "W: $t: $parent/$path ",
1771 "Insufficient permissions?\n";
1772 }
1773 }
1774 }
1775 \@out;
1776 }
1777
1778 sub parse_svn_date {
1779 my $date = shift || return '+0000 1970-01-01 00:00:00';
1780 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1781 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1782 croak "Unable to parse date: $date\n";
1783 "+0000 $Y-$m-$d $H:$M:$S";
1784 }
1785
1786 sub check_author {
1787 my ($author) = @_;
1788 if (!defined $author || length $author == 0) {
1789 $author = '(no author)';
1790 }
1791 if (defined $::_authors && ! defined $::users{$author}) {
1792 die "Author: $author not defined in $::_authors file\n";
1793 }
1794 $author;
1795 }
1796
1797 sub make_log_entry {
1798 my ($self, $rev, $parents, $ed) = @_;
1799 my $untracked = $self->get_untracked($ed);
1800
1801 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1802 print $un "r$rev\n" or croak $!;
1803 print $un $_, "\n" foreach @$untracked;
1804 my %log_entry = ( parents => $parents || [], revision => $rev,
1805 log => '');
1806
1807 my $headrev;
1808 my $logged = delete $self->{logged_rev_props};
1809 if (!$logged || $self->{-want_revprops}) {
1810 my $rp = $self->ra->rev_proplist($rev);
1811 foreach (sort keys %$rp) {
1812 my $v = $rp->{$_};
1813 if (/^svn:(author|date|log)$/) {
1814 $log_entry{$1} = $v;
1815 } elsif ($_ eq 'svm:headrev') {
1816 $headrev = $v;
1817 } else {
1818 print $un " rev_prop: ", uri_encode($_), ' ',
1819 uri_encode($v), "\n";
1820 }
1821 }
1822 } else {
1823 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1824 }
1825 close $un or croak $!;
1826
1827 $log_entry{date} = parse_svn_date($log_entry{date});
1828 $log_entry{log} .= "\n";
1829 my $author = $log_entry{author} = check_author($log_entry{author});
1830 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1831 : ($author, undef);
1832 if (defined $headrev && $self->use_svm_props) {
1833 if ($self->rewrite_root) {
1834 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1835 "options set!\n";
1836 }
1837 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1838 # we don't want "SVM: initializing mirror for junk" ...
1839 return undef if $r == 0;
1840 my $svm = $self->svm;
1841 if ($uuid ne $svm->{uuid}) {
1842 die "UUID mismatch on SVM path:\n",
1843 "expected: $svm->{uuid}\n",
1844 " got: $uuid\n";
1845 }
1846 my $full_url = $self->full_url;
1847 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1848 die "Failed to replace '$svm->{replace}' with ",
1849 "'$svm->{source}' in $full_url\n";
1850 # throw away username for storing in records
1851 remove_username($full_url);
1852 $log_entry{metadata} = "$full_url\@$r $uuid";
1853 $log_entry{svm_revision} = $r;
1854 $email ||= "$author\@$uuid"
1855 } elsif ($self->use_svnsync_props) {
1856 my $full_url = $self->svnsync->{url};
1857 $full_url .= "/$self->{path}" if length $self->{path};
1858 my $uuid = $self->svnsync->{uuid};
1859 $log_entry{metadata} = "$full_url\@$rev $uuid";
1860 $email ||= "$author\@$uuid"
1861 } else {
1862 $log_entry{metadata} = $self->metadata_url. "\@$rev " .
1863 $self->ra->get_uuid;
1864 $email ||= "$author\@" . $self->ra->get_uuid;
1865 }
1866 $log_entry{name} = $name;
1867 $log_entry{email} = $email;
1868 \%log_entry;
1869 }
1870
1871 sub fetch {
1872 my ($self, $min_rev, $max_rev, @parents) = @_;
1873 my ($last_rev, $last_commit) = $self->last_rev_commit;
1874 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1875 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1876 }
1877
1878 sub set_tree_cb {
1879 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1880 $self->{inject_parents} = { $rev => $tree };
1881 $self->fetch(undef, undef);
1882 }
1883
1884 sub set_tree {
1885 my ($self, $tree) = (shift, shift);
1886 my $log_entry = ::get_commit_entry($tree);
1887 unless ($self->{last_rev}) {
1888 fatal("Must have an existing revision to commit\n");
1889 }
1890 my %ed_opts = ( r => $self->{last_rev},
1891 log => $log_entry->{log},
1892 ra => $self->ra,
1893 tree_a => $self->{last_commit},
1894 tree_b => $tree,
1895 editor_cb => sub {
1896 $self->set_tree_cb($log_entry, $tree, @_) },
1897 svn_path => $self->{path} );
1898 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1899 print "No changes\nr$self->{last_rev} = $tree\n";
1900 }
1901 }
1902
1903 sub rebuild {
1904 my ($self) = @_;
1905 my $db_path = $self->db_path;
1906 return if (-e $db_path && ! -z $db_path);
1907 return unless ::verify_ref($self->refname.'^0');
1908 if (-f $self->{db_root}) {
1909 rename $self->{db_root}, $db_path or die
1910 "rename $self->{db_root} => $db_path failed: $!\n";
1911 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1912 symlink $base, $self->{db_root} or die
1913 "symlink $base => $self->{db_root} failed: $!\n";
1914 return;
1915 }
1916 print "Rebuilding $db_path ...\n";
1917 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1918 my $latest;
1919 my $full_url = $self->full_url;
1920 remove_username($full_url);
1921 my $svn_uuid;
1922 while (<$rev_list>) {
1923 chomp;
1924 my $c = $_;
1925 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1926 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1927 remove_username($url);
1928
1929 # ignore merges (from set-tree)
1930 next if (!defined $rev || !$uuid);
1931
1932 # if we merged or otherwise started elsewhere, this is
1933 # how we break out of it
1934 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1935 ($full_url && $url && ($url ne $full_url))) {
1936 next;
1937 }
1938 $latest ||= $rev;
1939 $svn_uuid ||= $uuid;
1940
1941 $self->rev_db_set($rev, $c);
1942 print "r$rev = $c\n";
1943 }
1944 command_close_pipe($rev_list, $ctx);
1945 print "Done rebuilding $db_path\n";
1946 }
1947
1948 # rev_db:
1949 # Tie::File seems to be prone to offset errors if revisions get sparse,
1950 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1951 # one of my favorite modules is out :< Next up would be one of the DBM
1952 # modules, but I'm not sure which is most portable... So I'll just
1953 # go with something that's plain-text, but still capable of
1954 # being randomly accessed. So here's my ultra-simple fixed-width
1955 # database. All records are 40 characters + "\n", so it's easy to seek
1956 # to a revision: (41 * rev) is the byte offset.
1957 # A record of 40 0s denotes an empty revision.
1958 # And yes, it's still pretty fast (faster than Tie::File).
1959 # These files are disposable unless noMetadata or useSvmProps is set
1960
1961 sub _rev_db_set {
1962 my ($fh, $rev, $commit) = @_;
1963 my $offset = $rev * 41;
1964 # assume that append is the common case:
1965 seek $fh, 0, 2 or croak $!;
1966 my $pos = tell $fh;
1967 if ($pos < $offset) {
1968 for (1 .. (($offset - $pos) / 41)) {
1969 print $fh (('0' x 40),"\n") or croak $!;
1970 }
1971 }
1972 seek $fh, $offset, 0 or croak $!;
1973 print $fh $commit,"\n" or croak $!;
1974 }
1975
1976 sub mkfile {
1977 my ($path) = @_;
1978 unless (-e $path) {
1979 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1980 mkpath([$dir]) unless -d $dir;
1981 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1982 close $fh or die "Couldn't close (create) $path: $!\n";
1983 }
1984 }
1985
1986 sub rev_db_set {
1987 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1988 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1989 my $db = $self->db_path($uuid);
1990 my $db_lock = "$db.lock";
1991 my $sig;
1992 if ($update_ref) {
1993 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1994 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1995 }
1996 mkfile($db);
1997
1998 $LOCKFILES{$db_lock} = 1;
1999 my $sync;
2000 # both of these options make our .rev_db file very, very important
2001 # and we can't afford to lose it because rebuild() won't work
2002 if ($self->use_svm_props || $self->no_metadata) {
2003 $sync = 1;
2004 copy($db, $db_lock) or die "rev_db_set(@_): ",
2005 "Failed to copy: ",
2006 "$db => $db_lock ($!)\n";
2007 } else {
2008 rename $db, $db_lock or die "rev_db_set(@_): ",
2009 "Failed to rename: ",
2010 "$db => $db_lock ($!)\n";
2011 }
2012 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2013 _rev_db_set($fh, $rev, $commit);
2014 if ($sync) {
2015 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2016 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2017 }
2018 close $fh or croak $!;
2019 if ($update_ref) {
2020 $_head = $self;
2021 command_noisy('update-ref', '-m', "r$rev",
2022 $self->refname, $commit);
2023 }
2024 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2025 "$db_lock => $db ($!)\n";
2026 delete $LOCKFILES{$db_lock};
2027 if ($update_ref) {
2028 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2029 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2030 kill $sig, $$ if defined $sig;
2031 }
2032 }
2033
2034 sub rev_db_max {
2035 my ($self) = @_;
2036 $self->rebuild;
2037 my $db_path = $self->db_path;
2038 my @stat = stat $db_path or return 0;
2039 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2040 my $max = $stat[7] / 41;
2041 (($max > 0) ? $max - 1 : 0);
2042 }
2043
2044 sub rev_db_get {
2045 my ($self, $rev, $uuid) = @_;
2046 my $ret;
2047 my $offset = $rev * 41;
2048 my $db_path = $self->db_path($uuid);
2049 return undef unless -e $db_path;
2050 open my $fh, '<', $db_path or croak $!;
2051 if (sysseek($fh, $offset, 0) == $offset) {
2052 my $read = sysread($fh, $ret, 40);
2053 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2054 }
2055 close $fh or croak $!;
2056 $ret;
2057 }
2058
2059 sub find_rev_before {
2060 my ($self, $rev, $eq_ok) = @_;
2061 --$rev unless $eq_ok;
2062 while ($rev > 0) {
2063 if (my $c = $self->rev_db_get($rev)) {
2064 return ($rev, $c);
2065 }
2066 --$rev;
2067 }
2068 return (undef, undef);
2069 }
2070
2071 sub _new {
2072 my ($class, $repo_id, $ref_id, $path) = @_;
2073 unless (defined $repo_id && length $repo_id) {
2074 $repo_id = $Git::SVN::default_repo_id;
2075 }
2076 unless (defined $ref_id && length $ref_id) {
2077 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2078 }
2079 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2080 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2081 $_[3] = $path = '' unless (defined $path);
2082 mkpath(["$ENV{GIT_DIR}/svn"]);
2083 bless {
2084 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2085 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2086 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2087 }
2088
2089 sub db_path {
2090 my ($self, $uuid) = @_;
2091 $uuid ||= $self->ra_uuid;
2092 "$self->{db_root}.$uuid";
2093 }
2094
2095 sub uri_encode {
2096 my ($f) = @_;
2097 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2098 $f
2099 }
2100
2101 sub remove_username {
2102 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2103 }
2104
2105 package Git::SVN::Prompt;
2106 use strict;
2107 use warnings;
2108 require SVN::Core;
2109 use vars qw/$_no_auth_cache $_username/;
2110
2111 sub simple {
2112 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2113 $may_save = undef if $_no_auth_cache;
2114 $default_username = $_username if defined $_username;
2115 if (defined $default_username && length $default_username) {
2116 if (defined $realm && length $realm) {
2117 print STDERR "Authentication realm: $realm\n";
2118 STDERR->flush;
2119 }
2120 $cred->username($default_username);
2121 } else {
2122 username($cred, $realm, $may_save, $pool);
2123 }
2124 $cred->password(_read_password("Password for '" .
2125 $cred->username . "': ", $realm));
2126 $cred->may_save($may_save);
2127 $SVN::_Core::SVN_NO_ERROR;
2128 }
2129
2130 sub ssl_server_trust {
2131 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2132 $may_save = undef if $_no_auth_cache;
2133 print STDERR "Error validating server certificate for '$realm':\n";
2134 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2135 print STDERR " - The certificate is not issued by a trusted ",
2136 "authority. Use the\n",
2137 " fingerprint to validate the certificate manually!\n";
2138 }
2139 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2140 print STDERR " - The certificate hostname does not match.\n";
2141 }
2142 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2143 print STDERR " - The certificate is not yet valid.\n";
2144 }
2145 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2146 print STDERR " - The certificate has expired.\n";
2147 }
2148 if ($failures & $SVN::Auth::SSL::OTHER) {
2149 print STDERR " - The certificate has an unknown error.\n";
2150 }
2151 printf STDERR
2152 "Certificate information:\n".
2153 " - Hostname: %s\n".
2154 " - Valid: from %s until %s\n".
2155 " - Issuer: %s\n".
2156 " - Fingerprint: %s\n",
2157 map $cert_info->$_, qw(hostname valid_from valid_until
2158 issuer_dname fingerprint);
2159 my $choice;
2160 prompt:
2161 print STDERR $may_save ?
2162 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2163 "(R)eject or accept (t)emporarily? ";
2164 STDERR->flush;
2165 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2166 if ($choice =~ /^t$/i) {
2167 $cred->may_save(undef);
2168 } elsif ($choice =~ /^r$/i) {
2169 return -1;
2170 } elsif ($may_save && $choice =~ /^p$/i) {
2171 $cred->may_save($may_save);
2172 } else {
2173 goto prompt;
2174 }
2175 $cred->accepted_failures($failures);
2176 $SVN::_Core::SVN_NO_ERROR;
2177 }
2178
2179 sub ssl_client_cert {
2180 my ($cred, $realm, $may_save, $pool) = @_;
2181 $may_save = undef if $_no_auth_cache;
2182 print STDERR "Client certificate filename: ";
2183 STDERR->flush;
2184 chomp(my $filename = <STDIN>);
2185 $cred->cert_file($filename);
2186 $cred->may_save($may_save);
2187 $SVN::_Core::SVN_NO_ERROR;
2188 }
2189
2190 sub ssl_client_cert_pw {
2191 my ($cred, $realm, $may_save, $pool) = @_;
2192 $may_save = undef if $_no_auth_cache;
2193 $cred->password(_read_password("Password: ", $realm));
2194 $cred->may_save($may_save);
2195 $SVN::_Core::SVN_NO_ERROR;
2196 }
2197
2198 sub username {
2199 my ($cred, $realm, $may_save, $pool) = @_;
2200 $may_save = undef if $_no_auth_cache;
2201 if (defined $realm && length $realm) {
2202 print STDERR "Authentication realm: $realm\n";
2203 }
2204 my $username;
2205 if (defined $_username) {
2206 $username = $_username;
2207 } else {
2208 print STDERR "Username: ";
2209 STDERR->flush;
2210 chomp($username = <STDIN>);
2211 }
2212 $cred->username($username);
2213 $cred->may_save($may_save);
2214 $SVN::_Core::SVN_NO_ERROR;
2215 }
2216
2217 sub _read_password {
2218 my ($prompt, $realm) = @_;
2219 print STDERR $prompt;
2220 STDERR->flush;
2221 require Term::ReadKey;
2222 Term::ReadKey::ReadMode('noecho');
2223 my $password = '';
2224 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2225 last if $key =~ /[\012\015]/; # \n\r
2226 $password .= $key;
2227 }
2228 Term::ReadKey::ReadMode('restore');
2229 print STDERR "\n";
2230 STDERR->flush;
2231 $password;
2232 }
2233
2234 package main;
2235
2236 {
2237 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2238 $SVN::Node::dir.$SVN::Node::unknown.
2239 $SVN::Node::none.$SVN::Node::file.
2240 $SVN::Node::dir.$SVN::Node::unknown.
2241 $SVN::Auth::SSL::CNMISMATCH.
2242 $SVN::Auth::SSL::NOTYETVALID.
2243 $SVN::Auth::SSL::EXPIRED.
2244 $SVN::Auth::SSL::UNKNOWNCA.
2245 $SVN::Auth::SSL::OTHER;
2246 }
2247
2248 package SVN::Git::Fetcher;
2249 use vars qw/@ISA/;
2250 use strict;
2251 use warnings;
2252 use Carp qw/croak/;
2253 use IO::File qw//;
2254 use Digest::MD5;
2255
2256 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2257 sub new {
2258 my ($class, $git_svn) = @_;
2259 my $self = SVN::Delta::Editor->new;
2260 bless $self, $class;
2261 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2262 $self->{empty} = {};
2263 $self->{dir_prop} = {};
2264 $self->{file_prop} = {};
2265 $self->{absent_dir} = {};
2266 $self->{absent_file} = {};
2267 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2268 $self;
2269 }
2270
2271 sub set_path_strip {
2272 my ($self, $path) = @_;
2273 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2274 }
2275
2276 sub open_root {
2277 { path => '' };
2278 }
2279
2280 sub open_directory {
2281 my ($self, $path, $pb, $rev) = @_;
2282 { path => $path };
2283 }
2284
2285 sub git_path {
2286 my ($self, $path) = @_;
2287 if ($self->{path_strip}) {
2288 $path =~ s!$self->{path_strip}!! or
2289 die "Failed to strip path '$path' ($self->{path_strip})\n";
2290 }
2291 $path;
2292 }
2293
2294 sub delete_entry {
2295 my ($self, $path, $rev, $pb) = @_;
2296
2297 my $gpath = $self->git_path($path);
2298 return undef if ($gpath eq '');
2299
2300 # remove entire directories.
2301 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2302 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2303 -r --name-only -z/,
2304 $self->{c}, '--', $gpath);
2305 local $/ = "\0";
2306 while (<$ls>) {
2307 chomp;
2308 $self->{gii}->remove($_);
2309 print "\tD\t$_\n" unless $::_q;
2310 }
2311 print "\tD\t$gpath/\n" unless $::_q;
2312 command_close_pipe($ls, $ctx);
2313 $self->{empty}->{$path} = 0
2314 } else {
2315 $self->{gii}->remove($gpath);
2316 print "\tD\t$gpath\n" unless $::_q;
2317 }
2318 undef;
2319 }
2320
2321 sub open_file {
2322 my ($self, $path, $pb, $rev) = @_;
2323 my $gpath = $self->git_path($path);
2324 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2325 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2326 unless (defined $mode && defined $blob) {
2327 die "$path was not found in commit $self->{c} (r$rev)\n";
2328 }
2329 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2330 pool => SVN::Pool->new, action => 'M' };
2331 }
2332
2333 sub add_file {
2334 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2335 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2336 delete $self->{empty}->{$dir};
2337 { path => $path, mode_a => 100644, mode_b => 100644,
2338 pool => SVN::Pool->new, action => 'A' };
2339 }
2340
2341 sub add_directory {
2342 my ($self, $path, $cp_path, $cp_rev) = @_;
2343 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2344 delete $self->{empty}->{$dir};
2345 $self->{empty}->{$path} = 1;
2346 { path => $path };
2347 }
2348
2349 sub change_dir_prop {
2350 my ($self, $db, $prop, $value) = @_;
2351 $self->{dir_prop}->{$db->{path}} ||= {};
2352 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2353 undef;
2354 }
2355
2356 sub absent_directory {
2357 my ($self, $path, $pb) = @_;
2358 $self->{absent_dir}->{$pb->{path}} ||= [];
2359 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2360 undef;
2361 }
2362
2363 sub absent_file {
2364 my ($self, $path, $pb) = @_;
2365 $self->{absent_file}->{$pb->{path}} ||= [];
2366 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2367 undef;
2368 }
2369
2370 sub change_file_prop {
2371 my ($self, $fb, $prop, $value) = @_;
2372 if ($prop eq 'svn:executable') {
2373 if ($fb->{mode_b} != 120000) {
2374 $fb->{mode_b} = defined $value ? 100755 : 100644;
2375 }
2376 } elsif ($prop eq 'svn:special') {
2377 $fb->{mode_b} = defined $value ? 120000 : 100644;
2378 } else {
2379 $self->{file_prop}->{$fb->{path}} ||= {};
2380 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2381 }
2382 undef;
2383 }
2384
2385 sub apply_textdelta {
2386 my ($self, $fb, $exp) = @_;
2387 my $fh = IO::File->new_tmpfile;
2388 $fh->autoflush(1);
2389 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2390 # (but $base does not,) so dup() it for reading in close_file
2391 open my $dup, '<&', $fh or croak $!;
2392 my $base = IO::File->new_tmpfile;
2393 $base->autoflush(1);
2394 if ($fb->{blob}) {
2395 defined (my $pid = fork) or croak $!;
2396 if (!$pid) {
2397 open STDOUT, '>&', $base or croak $!;
2398 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2399 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2400 }
2401 waitpid $pid, 0;
2402 croak $? if $?;
2403
2404 if (defined $exp) {
2405 seek $base, 0, 0 or croak $!;
2406 my $md5 = Digest::MD5->new;
2407 $md5->addfile($base);
2408 my $got = $md5->hexdigest;
2409 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2410 "expected: $exp\n",
2411 " got: $got\n" if ($got ne $exp);
2412 }
2413 }
2414 seek $base, 0, 0 or croak $!;
2415 $fb->{fh} = $dup;
2416 $fb->{base} = $base;
2417 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2418 }
2419
2420 sub close_file {
2421 my ($self, $fb, $exp) = @_;
2422 my $hash;
2423 my $path = $self->git_path($fb->{path});
2424 if (my $fh = $fb->{fh}) {
2425 seek($fh, 0, 0) or croak $!;
2426 my $md5 = Digest::MD5->new;
2427 $md5->addfile($fh);
2428 my $got = $md5->hexdigest;
2429 die "Checksum mismatch: $path\n",
2430 "expected: $exp\n got: $got\n" if ($got ne $exp);
2431 seek($fh, 0, 0) or croak $!;
2432 if ($fb->{mode_b} == 120000) {
2433 read($fh, my $buf, 5) == 5 or croak $!;
2434 $buf eq 'link ' or die "$path has mode 120000",
2435 "but is not a link\n";
2436 }
2437 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2438 if (!$pid) {
2439 open STDIN, '<&', $fh or croak $!;
2440 exec qw/git-hash-object -w --stdin/ or croak $!;
2441 }
2442 chomp($hash = do { local $/; <$out> });
2443 close $out or croak $!;
2444 close $fh or croak $!;
2445 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2446 close $fb->{base} or croak $!;
2447 } else {
2448 $hash = $fb->{blob} or die "no blob information\n";
2449 }
2450 $fb->{pool}->clear;
2451 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2452 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2453 undef;
2454 }
2455
2456 sub abort_edit {
2457 my $self = shift;
2458 $self->{nr} = $self->{gii}->{nr};
2459 delete $self->{gii};
2460 $self->SUPER::abort_edit(@_);
2461 }
2462
2463 sub close_edit {
2464 my $self = shift;
2465 $self->{git_commit_ok} = 1;
2466 $self->{nr} = $self->{gii}->{nr};
2467 delete $self->{gii};
2468 $self->SUPER::close_edit(@_);
2469 }
2470
2471 package SVN::Git::Editor;
2472 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2473 use strict;
2474 use warnings;
2475 use Carp qw/croak/;
2476 use IO::File;
2477 use Digest::MD5;
2478
2479 sub new {
2480 my ($class, $opts) = @_;
2481 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2482 die "$_ required!\n" unless (defined $opts->{$_});
2483 }
2484
2485 my $pool = SVN::Pool->new;
2486 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2487 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2488 $opts->{r}, $mods);
2489
2490 # $opts->{ra} functions should not be used after this:
2491 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2492 $opts->{editor_cb}, $pool);
2493 my $self = SVN::Delta::Editor->new(@ce, $pool);
2494 bless $self, $class;
2495 foreach (qw/svn_path r tree_a tree_b/) {
2496 $self->{$_} = $opts->{$_};
2497 }
2498 $self->{url} = $opts->{ra}->{url};
2499 $self->{mods} = $mods;
2500 $self->{types} = $types;
2501 $self->{pool} = $pool;
2502 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2503 $self->{rm} = { };
2504 $self->{path_prefix} = length $self->{svn_path} ?
2505 "$self->{svn_path}/" : '';
2506 return $self;
2507 }
2508
2509 sub generate_diff {
2510 my ($tree_a, $tree_b) = @_;
2511 my @diff_tree = qw(diff-tree -z -r);
2512 if ($_cp_similarity) {
2513 push @diff_tree, "-C$_cp_similarity";
2514 } else {
2515 push @diff_tree, '-C';
2516 }
2517 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2518 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2519 push @diff_tree, $tree_a, $tree_b;
2520 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2521 local $/ = "\0";
2522 my $state = 'meta';
2523 my @mods;
2524 while (<$diff_fh>) {
2525 chomp $_; # this gets rid of the trailing "\0"
2526 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2527 $::sha1\s($::sha1)\s
2528 ([MTCRAD])\d*$/xo) {
2529 push @mods, { mode_a => $1, mode_b => $2,
2530 sha1_b => $3, chg => $4 };
2531 if ($4 =~ /^(?:C|R)$/) {
2532 $state = 'file_a';
2533 } else {
2534 $state = 'file_b';
2535 }
2536 } elsif ($state eq 'file_a') {
2537 my $x = $mods[$#mods] or croak "Empty array\n";
2538 if ($x->{chg} !~ /^(?:C|R)$/) {
2539 croak "Error parsing $_, $x->{chg}\n";
2540 }
2541 $x->{file_a} = $_;
2542 $state = 'file_b';
2543 } elsif ($state eq 'file_b') {
2544 my $x = $mods[$#mods] or croak "Empty array\n";
2545 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2546 croak "Error parsing $_, $x->{chg}\n";
2547 }
2548 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2549 croak "Error parsing $_, $x->{chg}\n";
2550 }
2551 $x->{file_b} = $_;
2552 $state = 'meta';
2553 } else {
2554 croak "Error parsing $_\n";
2555 }
2556 }
2557 command_close_pipe($diff_fh, $ctx);
2558 \@mods;
2559 }
2560
2561 sub check_diff_paths {
2562 my ($ra, $pfx, $rev, $mods) = @_;
2563 my %types;
2564 $pfx .= '/' if length $pfx;
2565
2566 sub type_diff_paths {
2567 my ($ra, $types, $path, $rev) = @_;
2568 my @p = split m#/+#, $path;
2569 my $c = shift @p;
2570 unless (defined $types->{$c}) {
2571 $types->{$c} = $ra->check_path($c, $rev);
2572 }
2573 while (@p) {
2574 $c .= '/' . shift @p;
2575 next if defined $types->{$c};
2576 $types->{$c} = $ra->check_path($c, $rev);
2577 }
2578 }
2579
2580 foreach my $m (@$mods) {
2581 foreach my $f (qw/file_a file_b/) {
2582 next unless defined $m->{$f};
2583 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2584 if (length $pfx.$dir && ! defined $types{$dir}) {
2585 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2586 }
2587 }
2588 }
2589 \%types;
2590 }
2591
2592 sub split_path {
2593 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2594 }
2595
2596 sub repo_path {
2597 my ($self, $path) = @_;
2598 $self->{path_prefix}.(defined $path ? $path : '');
2599 }
2600
2601 sub url_path {
2602 my ($self, $path) = @_;
2603 $self->{url} . '/' . $self->repo_path($path);
2604 }
2605
2606 sub rmdirs {
2607 my ($self) = @_;
2608 my $rm = $self->{rm};
2609 delete $rm->{''}; # we never delete the url we're tracking
2610 return unless %$rm;
2611
2612 foreach (keys %$rm) {
2613 my @d = split m#/#, $_;
2614 my $c = shift @d;
2615 $rm->{$c} = 1;
2616 while (@d) {
2617 $c .= '/' . shift @d;
2618 $rm->{$c} = 1;
2619 }
2620 }
2621 delete $rm->{$self->{svn_path}};
2622 delete $rm->{''}; # we never delete the url we're tracking
2623 return unless %$rm;
2624
2625 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2626 $self->{tree_b});
2627 local $/ = "\0";
2628 while (<$fh>) {
2629 chomp;
2630 my @dn = split m#/#, $_;
2631 while (pop @dn) {
2632 delete $rm->{join '/', @dn};
2633 }
2634 unless (%$rm) {
2635 close $fh;
2636 return;
2637 }
2638 }
2639 command_close_pipe($fh, $ctx);
2640
2641 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2642 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2643 $self->close_directory($bat->{$d}, $p);
2644 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2645 print "\tD+\t$d/\n" unless $::_q;
2646 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2647 delete $bat->{$d};
2648 }
2649 }
2650
2651 sub open_or_add_dir {
2652 my ($self, $full_path, $baton) = @_;
2653 my $t = $self->{types}->{$full_path};
2654 if (!defined $t) {
2655 die "$full_path not known in r$self->{r} or we have a bug!\n";
2656 }
2657 if ($t == $SVN::Node::none) {
2658 return $self->add_directory($full_path, $baton,
2659 undef, -1, $self->{pool});
2660 } elsif ($t == $SVN::Node::dir) {
2661 return $self->open_directory($full_path, $baton,
2662 $self->{r}, $self->{pool});
2663 }
2664 print STDERR "$full_path already exists in repository at ",
2665 "r$self->{r} and it is not a directory (",
2666 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2667 exit 1;
2668 }
2669
2670 sub ensure_path {
2671 my ($self, $path) = @_;
2672 my $bat = $self->{bat};
2673 my $repo_path = $self->repo_path($path);
2674 return $bat->{''} unless (length $repo_path);
2675 my @p = split m#/+#, $repo_path;
2676 my $c = shift @p;
2677 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2678 while (@p) {
2679 my $c0 = $c;
2680 $c .= '/' . shift @p;
2681 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2682 }
2683 return $bat->{$c};
2684 }
2685
2686 sub A {
2687 my ($self, $m) = @_;
2688 my ($dir, $file) = split_path($m->{file_b});
2689 my $pbat = $self->ensure_path($dir);
2690 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2691 undef, -1);
2692 print "\tA\t$m->{file_b}\n" unless $::_q;
2693 $self->chg_file($fbat, $m);
2694 $self->close_file($fbat,undef,$self->{pool});
2695 }
2696
2697 sub C {
2698 my ($self, $m) = @_;
2699 my ($dir, $file) = split_path($m->{file_b});
2700 my $pbat = $self->ensure_path($dir);
2701 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2702 $self->url_path($m->{file_a}), $self->{r});
2703 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2704 $self->chg_file($fbat, $m);
2705 $self->close_file($fbat,undef,$self->{pool});
2706 }
2707
2708 sub delete_entry {
2709 my ($self, $path, $pbat) = @_;
2710 my $rpath = $self->repo_path($path);
2711 my ($dir, $file) = split_path($rpath);
2712 $self->{rm}->{$dir} = 1;
2713 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2714 }
2715
2716 sub R {
2717 my ($self, $m) = @_;
2718 my ($dir, $file) = split_path($m->{file_b});
2719 my $pbat = $self->ensure_path($dir);
2720 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2721 $self->url_path($m->{file_a}), $self->{r});
2722 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2723 $self->chg_file($fbat, $m);
2724 $self->close_file($fbat,undef,$self->{pool});
2725
2726 ($dir, $file) = split_path($m->{file_a});
2727 $pbat = $self->ensure_path($dir);
2728 $self->delete_entry($m->{file_a}, $pbat);
2729 }
2730
2731 sub M {
2732 my ($self, $m) = @_;
2733 my ($dir, $file) = split_path($m->{file_b});
2734 my $pbat = $self->ensure_path($dir);
2735 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2736 $pbat,$self->{r},$self->{pool});
2737 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2738 $self->chg_file($fbat, $m);
2739 $self->close_file($fbat,undef,$self->{pool});
2740 }
2741
2742 sub T { shift->M(@_) }
2743
2744 sub change_file_prop {
2745 my ($self, $fbat, $pname, $pval) = @_;
2746 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2747 }
2748
2749 sub chg_file {
2750 my ($self, $fbat, $m) = @_;
2751 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2752 $self->change_file_prop($fbat,'svn:executable','*');
2753 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2754 $self->change_file_prop($fbat,'svn:executable',undef);
2755 }
2756 my $fh = IO::File->new_tmpfile or croak $!;
2757 if ($m->{mode_b} =~ /^120/) {
2758 print $fh 'link ' or croak $!;
2759 $self->change_file_prop($fbat,'svn:special','*');
2760 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2761 $self->change_file_prop($fbat,'svn:special',undef);
2762 }
2763 defined(my $pid = fork) or croak $!;
2764 if (!$pid) {
2765 open STDOUT, '>&', $fh or croak $!;
2766 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2767 }
2768 waitpid $pid, 0;
2769 croak $? if $?;
2770 $fh->flush == 0 or croak $!;
2771 seek $fh, 0, 0 or croak $!;
2772
2773 my $md5 = Digest::MD5->new;
2774 $md5->addfile($fh) or croak $!;
2775 seek $fh, 0, 0 or croak $!;
2776
2777 my $exp = $md5->hexdigest;
2778 my $pool = SVN::Pool->new;
2779 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2780 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2781 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2782 $pool->clear;
2783
2784 close $fh or croak $!;
2785 }
2786
2787 sub D {
2788 my ($self, $m) = @_;
2789 my ($dir, $file) = split_path($m->{file_b});
2790 my $pbat = $self->ensure_path($dir);
2791 print "\tD\t$m->{file_b}\n" unless $::_q;
2792 $self->delete_entry($m->{file_b}, $pbat);
2793 }
2794
2795 sub close_edit {
2796 my ($self) = @_;
2797 my ($p,$bat) = ($self->{pool}, $self->{bat});
2798 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2799 $self->close_directory($bat->{$_}, $p);
2800 }
2801 $self->SUPER::close_edit($p);
2802 $p->clear;
2803 }
2804
2805 sub abort_edit {
2806 my ($self) = @_;
2807 $self->SUPER::abort_edit($self->{pool});
2808 }
2809
2810 sub DESTROY {
2811 my $self = shift;
2812 $self->SUPER::DESTROY(@_);
2813 $self->{pool}->clear;
2814 }
2815
2816 # this drives the editor
2817 sub apply_diff {
2818 my ($self) = @_;
2819 my $mods = $self->{mods};
2820 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2821 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2822 my $f = $m->{chg};
2823 if (defined $o{$f}) {
2824 $self->$f($m);
2825 } else {
2826 fatal("Invalid change type: $f\n");
2827 }
2828 }
2829 $self->rmdirs if $_rmdir;
2830 if (@$mods == 0) {
2831 $self->abort_edit;
2832 } else {
2833 $self->close_edit;
2834 }
2835 return scalar @$mods;
2836 }
2837
2838 package Git::SVN::Ra;
2839 use vars qw/@ISA $config_dir $_log_window_size/;
2840 use strict;
2841 use warnings;
2842 my ($can_do_switch);
2843 my $RA;
2844
2845 BEGIN {
2846 # enforce temporary pool usage for some simple functions
2847 my $e;
2848 foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2849 $e .= "sub $_ {
2850 my \$self = shift;
2851 my \$pool = SVN::Pool->new;
2852 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2853 \$pool->clear;
2854 wantarray ? \@ret : \$ret[0]; }\n";
2855 }
2856
2857 # get_dir needs $pool held in cache for dirents to work,
2858 # check_path is cacheable and rev_proplist is close enough
2859 # for our purposes.
2860 foreach (qw/check_path get_dir rev_proplist/) {
2861 $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2862 my \$self = shift;
2863 my \$r = pop;
2864 my \$k = join(\"\\0\", \@_);
2865 if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2866 return wantarray ? \@\$x : \$x->[0];
2867 }
2868 my \$pool = SVN::Pool->new;
2869 my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2870 if (\$r != \$${_}_rev) {
2871 \%${_}_cache = ( pool => [] );
2872 \$${_}_rev = \$r;
2873 }
2874 \$${_}_cache{\$r}->{\$k} = \\\@ret;
2875 push \@{\$${_}_cache{pool}}, \$pool;
2876 wantarray ? \@ret : \$ret[0]; }\n";
2877 }
2878 $e .= "\n1;";
2879 eval $e or die $@;
2880 }
2881
2882 sub new {
2883 my ($class, $url) = @_;
2884 $url =~ s!/+$!!;
2885 return $RA if ($RA && $RA->{url} eq $url);
2886
2887 SVN::_Core::svn_config_ensure($config_dir, undef);
2888 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2889 SVN::Client::get_simple_provider(),
2890 SVN::Client::get_ssl_server_trust_file_provider(),
2891 SVN::Client::get_simple_prompt_provider(
2892 \&Git::SVN::Prompt::simple, 2),
2893 SVN::Client::get_ssl_client_cert_prompt_provider(
2894 \&Git::SVN::Prompt::ssl_client_cert, 2),
2895 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2896 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2897 SVN::Client::get_username_provider(),
2898 SVN::Client::get_ssl_server_trust_prompt_provider(
2899 \&Git::SVN::Prompt::ssl_server_trust),
2900 SVN::Client::get_username_prompt_provider(
2901 \&Git::SVN::Prompt::username, 2),
2902 ]);
2903 my $config = SVN::Core::config_get_config($config_dir);
2904 my $self = SVN::Ra->new(url => $url, auth => $baton,
2905 config => $config,
2906 pool => SVN::Pool->new,
2907 auth_provider_callbacks => $callbacks);
2908 $self->{svn_path} = $url;
2909 $self->{repos_root} = $self->get_repos_root;
2910 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2911 $RA = bless $self, $class;
2912 }
2913
2914 sub DESTROY {
2915 # do not call the real DESTROY since we store ourselves in $RA
2916 }
2917
2918 sub get_log {
2919 my ($self, @args) = @_;
2920 my $pool = SVN::Pool->new;
2921 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2922 my $ret = $self->SUPER::get_log(@args, $pool);
2923 $pool->clear;
2924 $ret;
2925 }
2926
2927 sub get_commit_editor {
2928 my ($self, $log, $cb, $pool) = @_;
2929 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2930 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2931 }
2932
2933 sub gs_do_update {
2934 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2935 my $new = ($rev_a == $rev_b);
2936 my $path = $gs->{path};
2937
2938 my $pool = SVN::Pool->new;
2939 $editor->set_path_strip($path);
2940 my (@pc) = split m#/#, $path;
2941 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2942 1, $editor, $pool);
2943 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2944
2945 # Since we can't rely on svn_ra_reparent being available, we'll
2946 # just have to do some magic with set_path to make it so
2947 # we only want a partial path.
2948 my $sp = '';
2949 my $final = join('/', @pc);
2950 while (@pc) {
2951 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2952 $sp .= '/' if length $sp;
2953 $sp .= shift @pc;
2954 }
2955 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2956
2957 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2958
2959 $reporter->finish_report($pool);
2960 $pool->clear;
2961 $editor->{git_commit_ok};
2962 }
2963
2964 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2965 # svn_ra_reparent didn't work before 1.4)
2966 sub gs_do_switch {
2967 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2968 my $path = $gs->{path};
2969 my $pool = SVN::Pool->new;
2970
2971 my $full_url = $self->{url};
2972 my $old_url = $full_url;
2973 $full_url .= "/$path" if length $path;
2974 my ($ra, $reparented);
2975 if ($old_url ne $full_url) {
2976 if ($old_url !~ m#^svn(\+ssh)?://#) {
2977 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2978 $pool);
2979 $self->{url} = $full_url;
2980 $reparented = 1;
2981 } else {
2982 $ra = Git::SVN::Ra->new($full_url);
2983 }
2984 }
2985 $ra ||= $self;
2986 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2987 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2988 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2989 $reporter->finish_report($pool);
2990
2991 if ($reparented) {
2992 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2993 $self->{url} = $old_url;
2994 }
2995
2996 $pool->clear;
2997 $editor->{git_commit_ok};
2998 }
2999
3000 sub gs_fetch_loop_common {
3001 my ($self, $base, $head, $gsv, $globs) = @_;
3002 return if ($base > $head);
3003 my $inc = $_log_window_size;
3004 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3005 my %common;
3006 my $common_max = scalar @$gsv;
3007
3008 foreach my $gs (@$gsv) {
3009 my @tmp = split m#/#, $gs->{path};
3010 my $p = '';
3011 foreach (@tmp) {
3012 $p .= length($p) ? "/$_" : $_;
3013 $common{$p} ||= 0;
3014 $common{$p}++;
3015 }
3016 }
3017 $globs ||= [];
3018 $common_max += scalar @$globs;
3019 foreach my $glob (@$globs) {
3020 my @tmp = split m#/#, $glob->{path}->{left};
3021 my $p = '';
3022 foreach (@tmp) {
3023 $p .= length($p) ? "/$_" : $_;
3024 $common{$p} ||= 0;
3025 $common{$p}++;
3026 }
3027 }
3028
3029 my $longest_path = '';
3030 foreach (sort {length $b <=> length $a} keys %common) {
3031 if ($common{$_} == $common_max) {
3032 $longest_path = $_;
3033 last;
3034 }
3035 }
3036 while (1) {
3037 my %revs;
3038 my $err;
3039 my $err_handler = $SVN::Error::handler;
3040 $SVN::Error::handler = sub {
3041 ($err) = @_;
3042 skip_unknown_revs($err);
3043 };
3044 sub _cb {
3045 my ($paths, $r, $author, $date, $log) = @_;
3046 [ dup_changed_paths($paths),
3047 { author => $author, date => $date, log => $log } ];
3048 }
3049 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3050 sub { $revs{$_[1]} = _cb(@_) });
3051 if ($err && $max >= $head) {
3052 print STDERR "Path '$longest_path' ",
3053 "was probably deleted:\n",
3054 $err->expanded_message,
3055 "\nWill attempt to follow ",
3056 "revisions r$min .. r$max ",
3057 "committed before the deletion\n";
3058 my $hi = $max;
3059 while (--$hi >= $min) {
3060 my $ok;
3061 $self->get_log([$longest_path], $min, $hi,
3062 0, 1, 1, sub {
3063 $ok ||= $_[1];
3064 $revs{$_[1]} = _cb(@_) });
3065 if ($ok) {
3066 print STDERR "r$min .. r$ok OK\n";
3067 last;
3068 }
3069 }
3070 }
3071 $SVN::Error::handler = $err_handler;
3072
3073 my %exists = map { $_->{path} => $_ } @$gsv;
3074 foreach my $r (sort {$a <=> $b} keys %revs) {
3075 my ($paths, $logged) = @{$revs{$r}};
3076
3077 foreach my $gs ($self->match_globs(\%exists, $paths,
3078 $globs, $r)) {
3079 if ($gs->rev_db_max >= $r) {
3080 next;
3081 }
3082 next unless $gs->match_paths($paths, $r);
3083 $gs->{logged_rev_props} = $logged;
3084 if (my $last_commit = $gs->last_commit) {
3085 $gs->assert_index_clean($last_commit);
3086 }
3087 my $log_entry = $gs->do_fetch($paths, $r);
3088 if ($log_entry) {
3089 $gs->do_git_commit($log_entry);
3090 }
3091 }
3092 foreach my $g (@$globs) {
3093 my $k = "svn-remote.$g->{remote}." .
3094 "$g->{t}-maxRev";
3095 Git::SVN::tmp_config($k, $r);
3096 }
3097 }
3098 # pre-fill the .rev_db since it'll eventually get filled in
3099 # with '0' x40 if something new gets committed
3100 foreach my $gs (@$gsv) {
3101 next if defined $gs->rev_db_get($max);
3102 $gs->rev_db_set($max, 0 x40);
3103 }
3104 foreach my $g (@$globs) {
3105 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3106 Git::SVN::tmp_config($k, $max);
3107 }
3108 last if $max >= $head;
3109 $min = $max + 1;
3110 $max += $inc;
3111 $max = $head if ($max > $head);
3112 }
3113 }
3114
3115 sub match_globs {
3116 my ($self, $exists, $paths, $globs, $r) = @_;
3117
3118 sub get_dir_check {
3119 my ($self, $exists, $g, $r) = @_;
3120 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3121 return unless scalar @x == 3;
3122 my $dirents = $x[0];
3123 foreach my $de (keys %$dirents) {
3124 next if $dirents->{$de}->kind != $SVN::Node::dir;
3125 my $p = $g->{path}->full_path($de);
3126 next if $exists->{$p};
3127 next if (length $g->{path}->{right} &&
3128 ($self->check_path($p, $r) !=
3129 $SVN::Node::dir));
3130 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3131 $g->{ref}->full_path($de), 1);
3132 }
3133 }
3134 foreach my $g (@$globs) {
3135 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3136 if ($path->{action} =~ /^[AR]$/) {
3137 get_dir_check($self, $exists, $g, $r);
3138 }
3139 }
3140 foreach (keys %$paths) {
3141 if (/$g->{path}->{left_regex}/ &&
3142 !/$g->{path}->{regex}/) {
3143 next if $paths->{$_}->{action} !~ /^[AR]$/;
3144 get_dir_check($self, $exists, $g, $r);
3145 }
3146 next unless /$g->{path}->{regex}/;
3147 my $p = $1;
3148 my $pathname = $g->{path}->full_path($p);
3149 next if $exists->{$pathname};
3150 $exists->{$pathname} = Git::SVN->init(
3151 $self->{url}, $pathname, undef,
3152 $g->{ref}->full_path($p), 1);
3153 }
3154 my $c = '';
3155 foreach (split m#/#, $g->{path}->{left}) {
3156 $c .= "/$_";
3157 next unless ($paths->{$c} &&
3158 ($paths->{$c}->{action} =~ /^[AR]$/));
3159 get_dir_check($self, $exists, $g, $r);
3160 }
3161 }
3162 values %$exists;
3163 }
3164
3165 sub minimize_url {
3166 my ($self) = @_;
3167 return $self->{url} if ($self->{url} eq $self->{repos_root});
3168 my $url = $self->{repos_root};
3169 my @components = split(m!/!, $self->{svn_path});
3170 my $c = '';
3171 do {
3172 $url .= "/$c" if length $c;
3173 eval { (ref $self)->new($url)->get_latest_revnum };
3174 } while ($@ && ($c = shift @components));
3175 $url;
3176 }
3177
3178 sub can_do_switch {
3179 my $self = shift;
3180 unless (defined $can_do_switch) {
3181 my $pool = SVN::Pool->new;
3182 my $rep = eval {
3183 $self->do_switch(1, '', 0, $self->{url},
3184 SVN::Delta::Editor->new, $pool);
3185 };
3186 if ($@) {
3187 $can_do_switch = 0;
3188 } else {
3189 $rep->abort_report($pool);
3190 $can_do_switch = 1;
3191 }
3192 $pool->clear;
3193 }
3194 $can_do_switch;
3195 }
3196
3197 sub skip_unknown_revs {
3198 my ($err) = @_;
3199 my $errno = $err->apr_err();
3200 # Maybe the branch we're tracking didn't
3201 # exist when the repo started, so it's
3202 # not an error if it doesn't, just continue
3203 #
3204 # Wonderfully consistent library, eh?
3205 # 160013 - svn:// and file://
3206 # 175002 - http(s)://
3207 # 175007 - http(s):// (this repo required authorization, too...)
3208 # More codes may be discovered later...
3209 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3210 warn "W: Ignoring error from SVN, path probably ",
3211 "does not exist: ($errno): ",
3212 $err->expanded_message,"\n";
3213 return;
3214 }
3215 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3216 }
3217
3218 # svn_log_changed_path_t objects passed to get_log are likely to be
3219 # overwritten even if only the refs are copied to an external variable,
3220 # so we should dup the structures in their entirety. Using an externally
3221 # passed pool (instead of our temporary and quickly cleared pool in
3222 # Git::SVN::Ra) does not help matters at all...
3223 sub dup_changed_paths {
3224 my ($paths) = @_;
3225 return undef unless $paths;
3226 my %ret;
3227 foreach my $p (keys %$paths) {
3228 my $i = $paths->{$p};
3229 my %s = map { $_ => $i->$_ }
3230 qw/copyfrom_path copyfrom_rev action/;
3231 $ret{$p} = \%s;
3232 }
3233 \%ret;
3234 }
3235
3236 package Git::SVN::Log;
3237 use strict;
3238 use warnings;
3239 use POSIX qw/strftime/;
3240 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3241 %rusers $show_commit $incremental/;
3242 my $l_fmt;
3243
3244 sub cmt_showable {
3245 my ($c) = @_;
3246 return 1 if defined $c->{r};
3247 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3248 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3249 my @log = command(qw/cat-file commit/, $c->{c});
3250 shift @log while ($log[0] ne "\n");
3251 shift @log;
3252 @{$c->{l}} = grep !/^git-svn-id: /, @log;
3253
3254 (undef, $c->{r}, undef) = ::extract_metadata(
3255 (grep(/^git-svn-id: /, @log))[-1]);
3256 }
3257 return defined $c->{r};
3258 }
3259
3260 sub log_use_color {
3261 return 1 if $color;
3262 my ($dc, $dcvar);
3263 $dcvar = 'color.diff';
3264 $dc = `git-config --get $dcvar`;
3265 if ($dc eq '') {
3266 # nothing at all; fallback to "diff.color"
3267 $dcvar = 'diff.color';
3268 $dc = `git-config --get $dcvar`;
3269 }
3270 chomp($dc);
3271 if ($dc eq 'auto') {
3272 my $pc;
3273 $pc = `git-config --get color.pager`;
3274 if ($pc eq '') {
3275 # does not have it -- fallback to pager.color
3276 $pc = `git-config --bool --get pager.color`;
3277 }
3278 else {
3279 $pc = `git-config --bool --get color.pager`;
3280 if ($?) {
3281 $pc = 'false';
3282 }
3283 }
3284 chomp($pc);
3285 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3286 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3287 }
3288 return 0;
3289 }
3290 return 0 if $dc eq 'never';
3291 return 1 if $dc eq 'always';
3292 chomp($dc = `git-config --bool --get $dcvar`);
3293 return ($dc eq 'true');
3294 }
3295
3296 sub git_svn_log_cmd {
3297 my ($r_min, $r_max, @args) = @_;
3298 my $head = 'HEAD';
3299 foreach my $x (@args) {
3300 last if $x eq '--';
3301 next unless ::verify_ref("$x^0");
3302 $head = $x;
3303 last;
3304 }
3305
3306 my $url = (::working_head_info($head))[0];
3307 my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3308 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3309 $gs->refname);
3310 push @cmd, '-r' unless $non_recursive;
3311 push @cmd, qw/--raw --name-status/ if $verbose;
3312 push @cmd, '--color' if log_use_color();
3313 return @cmd unless defined $r_max;
3314 if ($r_max == $r_min) {
3315 push @cmd, '--max-count=1';
3316 if (my $c = $gs->rev_db_get($r_max)) {
3317 push @cmd, $c;
3318 }
3319 } else {
3320 my ($c_min, $c_max);
3321 $c_max = $gs->rev_db_get($r_max);
3322 $c_min = $gs->rev_db_get($r_min);
3323 if (defined $c_min && defined $c_max) {
3324 if ($r_max > $r_max) {
3325 push @cmd, "$c_min..$c_max";
3326 } else {
3327 push @cmd, "$c_max..$c_min";
3328 }
3329 } elsif ($r_max > $r_min) {
3330 push @cmd, $c_max;
3331 } else {
3332 push @cmd, $c_min;
3333 }
3334 }
3335 return @cmd;
3336 }
3337
3338 # adapted from pager.c
3339 sub config_pager {
3340 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3341 if (!defined $pager) {
3342 $pager = 'less';
3343 } elsif (length $pager == 0 || $pager eq 'cat') {
3344 $pager = undef;
3345 }
3346 }
3347
3348 sub run_pager {
3349 return unless -t *STDOUT;
3350 pipe my $rfd, my $wfd or return;
3351 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3352 if (!$pid) {
3353 open STDOUT, '>&', $wfd or
3354 ::fatal "Can't redirect to stdout: $!\n";
3355 return;
3356 }
3357 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3358 $ENV{LESS} ||= 'FRSX';
3359 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3360 }
3361
3362 sub tz_to_s_offset {
3363 my ($tz) = @_;
3364 $tz =~ s/(\d\d)$//;
3365 return ($1 * 60) + ($tz * 3600);
3366 }
3367
3368 sub get_author_info {
3369 my ($dest, $author, $t, $tz) = @_;
3370 $author =~ s/(?:^\s*|\s*$)//g;
3371 $dest->{a_raw} = $author;
3372 my $au;
3373 if ($::_authors) {
3374 $au = $rusers{$author} || undef;
3375 }
3376 if (!$au) {
3377 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3378 }
3379 $dest->{t} = $t;
3380 $dest->{tz} = $tz;
3381 $dest->{a} = $au;
3382 # Date::Parse isn't in the standard Perl distro :(
3383 if ($tz =~ s/^\+//) {
3384 $t += tz_to_s_offset($tz);
3385 } elsif ($tz =~ s/^\-//) {
3386 $t -= tz_to_s_offset($tz);
3387 }
3388 $dest->{t_utc} = $t;
3389 }
3390
3391 sub process_commit {
3392 my ($c, $r_min, $r_max, $defer) = @_;
3393 if (defined $r_min && defined $r_max) {
3394 if ($r_min == $c->{r} && $r_min == $r_max) {
3395 show_commit($c);
3396 return 0;
3397 }
3398 return 1 if $r_min == $r_max;
3399 if ($r_min < $r_max) {
3400 # we need to reverse the print order
3401 return 0 if (defined $limit && --$limit < 0);
3402 push @$defer, $c;
3403 return 1;
3404 }
3405 if ($r_min != $r_max) {
3406 return 1 if ($r_min < $c->{r});
3407 return 1 if ($r_max > $c->{r});
3408 }
3409 }
3410 return 0 if (defined $limit && --$limit < 0);
3411 show_commit($c);
3412 return 1;
3413 }
3414
3415 sub show_commit {
3416 my $c = shift;
3417 if ($oneline) {
3418 my $x = "\n";
3419 if (my $l = $c->{l}) {
3420 while ($l->[0] =~ /^\s*$/) { shift @$l }
3421 $x = $l->[0];
3422 }
3423 $l_fmt ||= 'A' . length($c->{r});
3424 print 'r',pack($l_fmt, $c->{r}),' | ';
3425 print "$c->{c} | " if $show_commit;
3426 print $x;
3427 } else {
3428 show_commit_normal($c);
3429 }
3430 }
3431
3432 sub show_commit_changed_paths {
3433 my ($c) = @_;
3434 return unless $c->{changed};
3435 print "Changed paths:\n", @{$c->{changed}};
3436 }
3437
3438 sub show_commit_normal {
3439 my ($c) = @_;
3440 print '-' x72, "\nr$c->{r} | ";
3441 print "$c->{c} | " if $show_commit;
3442 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3443 localtime($c->{t_utc})), ' | ';
3444 my $nr_line = 0;
3445
3446 if (my $l = $c->{l}) {
3447 while ($l->[$#$l] eq "\n" && $#$l > 0
3448 && $l->[($#$l - 1)] eq "\n") {
3449 pop @$l;
3450 }
3451 $nr_line = scalar @$l;
3452 if (!$nr_line) {
3453 print "1 line\n\n\n";
3454 } else {
3455 if ($nr_line == 1) {
3456 $nr_line = '1 line';
3457 } else {
3458 $nr_line .= ' lines';
3459 }
3460 print $nr_line, "\n";
3461 show_commit_changed_paths($c);
3462 print "\n";
3463 print $_ foreach @$l;
3464 }
3465 } else {
3466 print "1 line\n";
3467 show_commit_changed_paths($c);
3468 print "\n";
3469
3470 }
3471 foreach my $x (qw/raw stat diff/) {
3472 if ($c->{$x}) {
3473 print "\n";
3474 print $_ foreach @{$c->{$x}}
3475 }
3476 }
3477 }
3478
3479 sub cmd_show_log {
3480 my (@args) = @_;
3481 my ($r_min, $r_max);
3482 my $r_last = -1; # prevent dupes
3483 if (defined $TZ) {
3484 $ENV{TZ} = $TZ;
3485 } else {
3486 delete $ENV{TZ};
3487 }
3488 if (defined $::_revision) {
3489 if ($::_revision =~ /^(\d+):(\d+)$/) {
3490 ($r_min, $r_max) = ($1, $2);
3491 } elsif ($::_revision =~ /^\d+$/) {
3492 $r_min = $r_max = $::_revision;
3493 } else {
3494 ::fatal "-r$::_revision is not supported, use ",
3495 "standard \'git log\' arguments instead\n";
3496 }
3497 }
3498
3499 config_pager();
3500 @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3501 my $log = command_output_pipe(@args);
3502 run_pager();
3503 my (@k, $c, $d, $stat);
3504 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3505 while (<$log>) {
3506 if (/^${esc_color}commit ($::sha1_short)/o) {
3507 my $cmt = $1;
3508 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3509 $r_last = $c->{r};
3510 process_commit($c, $r_min, $r_max, \@k) or
3511 goto out;
3512 }
3513 $d = undef;
3514 $c = { c => $cmt };
3515 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3516 get_author_info($c, $1, $2, $3);
3517 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3518 # ignore
3519 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3520 push @{$c->{raw}}, $_;
3521 } elsif (/^${esc_color}[ACRMDT]\t/) {
3522 # we could add $SVN->{svn_path} here, but that requires
3523 # remote access at the moment (repo_path_split)...
3524 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3525 push @{$c->{changed}}, $_;
3526 } elsif (/^${esc_color}diff /o) {
3527 $d = 1;
3528 push @{$c->{diff}}, $_;
3529 } elsif ($d) {
3530 push @{$c->{diff}}, $_;
3531 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3532 $esc_color*[\+\-]*$esc_color$/x) {
3533 $stat = 1;
3534 push @{$c->{stat}}, $_;
3535 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3536 push @{$c->{stat}}, $_;
3537 $stat = undef;
3538 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3539 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3540 } elsif (s/^${esc_color} //o) {
3541 push @{$c->{l}}, $_;
3542 }
3543 }
3544 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3545 $r_last = $c->{r};
3546 process_commit($c, $r_min, $r_max, \@k);
3547 }
3548 if (@k) {
3549 my $swap = $r_max;
3550 $r_max = $r_min;
3551 $r_min = $swap;
3552 process_commit($_, $r_min, $r_max) foreach reverse @k;
3553 }
3554 out:
3555 close $log;
3556 print '-' x72,"\n" unless $incremental || $oneline;
3557 }
3558
3559 package Git::SVN::Migration;
3560 # these version numbers do NOT correspond to actual version numbers
3561 # of git nor git-svn. They are just relative.
3562 #
3563 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3564 #
3565 # v1 layout: .git/$id/info/url, refs/remotes/$id
3566 #
3567 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3568 #
3569 # v3 layout: .git/svn/$id, refs/remotes/$id
3570 # - info/url may remain for backwards compatibility
3571 # - this is what we migrate up to this layout automatically,
3572 # - this will be used by git svn init on single branches
3573 # v3.1 layout (auto migrated):
3574 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3575 # for backwards compatibility
3576 #
3577 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3578 # - this is only created for newly multi-init-ed
3579 # repositories. Similar in spirit to the
3580 # --use-separate-remotes option in git-clone (now default)
3581 # - we do not automatically migrate to this (following
3582 # the example set by core git)
3583 use strict;
3584 use warnings;
3585 use Carp qw/croak/;
3586 use File::Path qw/mkpath/;
3587 use File::Basename qw/dirname basename/;
3588 use vars qw/$_minimize/;
3589
3590 sub migrate_from_v0 {
3591 my $git_dir = $ENV{GIT_DIR};
3592 return undef unless -d $git_dir;
3593 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3594 my $migrated = 0;
3595 while (<$fh>) {
3596 chomp;
3597 my ($id, $orig_ref) = ($_, $_);
3598 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3599 next unless -f "$git_dir/$id/info/url";
3600 my $new_ref = "refs/remotes/$id";
3601 if (::verify_ref("$new_ref^0")) {
3602 print STDERR "W: $orig_ref is probably an old ",
3603 "branch used by an ancient version of ",
3604 "git-svn.\n",
3605 "However, $new_ref also exists.\n",
3606 "We will not be able ",
3607 "to use this branch until this ",
3608 "ambiguity is resolved.\n";
3609 next;
3610 }
3611 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3612 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3613 command_noisy('update-ref', $new_ref, $orig_ref);
3614 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3615 $migrated++;
3616 }
3617 command_close_pipe($fh, $ctx);
3618 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3619 $migrated;
3620 }
3621
3622 sub migrate_from_v1 {
3623 my $git_dir = $ENV{GIT_DIR};
3624 my $migrated = 0;
3625 return $migrated unless -d $git_dir;
3626 my $svn_dir = "$git_dir/svn";
3627
3628 # just in case somebody used 'svn' as their $id at some point...
3629 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3630
3631 print STDERR "Migrating from a git-svn v1 layout...\n";
3632 mkpath([$svn_dir]);
3633 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3634 "$svn_dir\n\t(required for this version ",
3635 "($::VERSION) of git-svn) does not. exist\n";
3636 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3637 while (<$fh>) {
3638 my $x = $_;
3639 next unless $x =~ s#^refs/remotes/##;
3640 chomp $x;
3641 next unless -f "$git_dir/$x/info/url";
3642 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3643 next unless $u;
3644 my $dn = dirname("$git_dir/svn/$x");
3645 mkpath([$dn]) unless -d $dn;
3646 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3647 mkpath(["$git_dir/svn/svn"]);
3648 print STDERR " - $git_dir/$x/info => ",
3649 "$git_dir/svn/$x/info\n";
3650 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3651 croak "$!: $x";
3652 # don't worry too much about these, they probably
3653 # don't exist with repos this old (save for index,
3654 # and we can easily regenerate that)
3655 foreach my $f (qw/unhandled.log index .rev_db/) {
3656 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3657 }
3658 } else {
3659 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3660 rename "$git_dir/$x", "$git_dir/svn/$x" or
3661 croak "$!: $x";
3662 }
3663 $migrated++;
3664 }
3665 command_close_pipe($fh, $ctx);
3666 print STDERR "Done migrating from a git-svn v1 layout\n";
3667 $migrated;
3668 }
3669
3670 sub read_old_urls {
3671 my ($l_map, $pfx, $path) = @_;
3672 my @dir;
3673 foreach (<$path/*>) {
3674 if (-r "$_/info/url") {
3675 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3676 my $ref_id = $pfx . basename $_;
3677 my $url = ::file_to_s("$_/info/url");
3678 $l_map->{$ref_id} = $url;
3679 } elsif (-d $_) {
3680 push @dir, $_;
3681 }
3682 }
3683 foreach (@dir) {
3684 my $x = $_;
3685 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3686 read_old_urls($l_map, $x, $_);
3687 }
3688 }
3689
3690 sub migrate_from_v2 {
3691 my @cfg = command(qw/config -l/);
3692 return if grep /^svn-remote\..+\.url=/, @cfg;
3693 my %l_map;
3694 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3695 my $migrated = 0;
3696
3697 foreach my $ref_id (sort keys %l_map) {
3698 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3699 if ($@) {
3700 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3701 }
3702 $migrated++;
3703 }
3704 $migrated;
3705 }
3706
3707 sub minimize_connections {
3708 my $r = Git::SVN::read_all_remotes();
3709 my $new_urls = {};
3710 my $root_repos = {};
3711 foreach my $repo_id (keys %$r) {
3712 my $url = $r->{$repo_id}->{url} or next;
3713 my $fetch = $r->{$repo_id}->{fetch} or next;
3714 my $ra = Git::SVN::Ra->new($url);
3715
3716 # skip existing cases where we already connect to the root
3717 if (($ra->{url} eq $ra->{repos_root}) ||
3718 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3719 $repo_id)) {
3720 $root_repos->{$ra->{url}} = $repo_id;
3721 next;
3722 }
3723
3724 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3725 my $root_path = $ra->{url};
3726 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3727 foreach my $path (keys %$fetch) {
3728 my $ref_id = $fetch->{$path};
3729 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3730
3731 # make sure we can read when connecting to
3732 # a higher level of a repository
3733 my ($last_rev, undef) = $gs->last_rev_commit;
3734 if (!defined $last_rev) {
3735 $last_rev = eval {
3736 $root_ra->get_latest_revnum;
3737 };
3738 next if $@;
3739 }
3740 my $new = $root_path;
3741 $new .= length $path ? "/$path" : '';
3742 eval {
3743 $root_ra->get_log([$new], $last_rev, $last_rev,
3744 0, 0, 1, sub { });
3745 };
3746 next if $@;
3747 $new_urls->{$ra->{repos_root}}->{$new} =
3748 { ref_id => $ref_id,
3749 old_repo_id => $repo_id,
3750 old_path => $path };
3751 }
3752 }
3753
3754 my @emptied;
3755 foreach my $url (keys %$new_urls) {
3756 # see if we can re-use an existing [svn-remote "repo_id"]
3757 # instead of creating a(n ugly) new section:
3758 my $repo_id = $root_repos->{$url} ||
3759 Git::SVN::sanitize_remote_name($url);
3760
3761 my $fetch = $new_urls->{$url};
3762 foreach my $path (keys %$fetch) {
3763 my $x = $fetch->{$path};
3764 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3765 my $pfx = "svn-remote.$x->{old_repo_id}";
3766
3767 my $old_fetch = quotemeta("$x->{old_path}:".
3768 "refs/remotes/$x->{ref_id}");
3769 command_noisy(qw/config --unset/,
3770 "$pfx.fetch", '^'. $old_fetch . '$');
3771 delete $r->{$x->{old_repo_id}}->
3772 {fetch}->{$x->{old_path}};
3773 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3774 command_noisy(qw/config --unset/,
3775 "$pfx.url");
3776 push @emptied, $x->{old_repo_id}
3777 }
3778 }
3779 }
3780 if (@emptied) {
3781 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3782 "$ENV{GIT_DIR}/config";
3783 print STDERR <<EOF;
3784 The following [svn-remote] sections in your config file ($file) are empty
3785 and can be safely removed:
3786 EOF
3787 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3788 }
3789 }
3790
3791 sub migration_check {
3792 migrate_from_v0();
3793 migrate_from_v1();
3794 migrate_from_v2();
3795 minimize_connections() if $_minimize;
3796 }
3797
3798 package Git::IndexInfo;
3799 use strict;
3800 use warnings;
3801 use Git qw/command_input_pipe command_close_pipe/;
3802
3803 sub new {
3804 my ($class) = @_;
3805 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3806 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3807 }
3808
3809 sub remove {
3810 my ($self, $path) = @_;
3811 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3812 return ++$self->{nr};
3813 }
3814 undef;
3815 }
3816
3817 sub update {
3818 my ($self, $mode, $hash, $path) = @_;
3819 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3820 return ++$self->{nr};
3821 }
3822 undef;
3823 }
3824
3825 sub DESTROY {
3826 my ($self) = @_;
3827 command_close_pipe($self->{gui}, $self->{ctx});
3828 }
3829
3830 package Git::SVN::GlobSpec;
3831 use strict;
3832 use warnings;
3833
3834 sub new {
3835 my ($class, $glob) = @_;
3836 my $re = $glob;
3837 $re =~ s!/+$!!g; # no need for trailing slashes
3838 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3839 my ($left, $right) = ($1, $2);
3840 if ($nr > 1) {
3841 die "Only one '*' wildcard expansion ",
3842 "is supported (got $nr): '$glob'\n";
3843 } elsif ($nr == 0) {
3844 die "One '*' is needed for glob: '$glob'\n";
3845 }
3846 $re = quotemeta($left) . $re . quotemeta($right);
3847 if (length $left && !($left =~ s!/+$!!g)) {
3848 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3849 }
3850 if (length $right && !($right =~ s!^/+!!g)) {
3851 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3852 }
3853 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3854 bless { left => $left, right => $right, left_regex => $left_re,
3855 regex => qr/$re/, glob => $glob }, $class;
3856 }
3857
3858 sub full_path {
3859 my ($self, $path) = @_;
3860 return (length $self->{left} ? "$self->{left}/" : '') .
3861 $path . (length $self->{right} ? "/$self->{right}" : '');
3862 }
3863
3864 __END__
3865
3866 Data structures:
3867
3868
3869 $remotes = { # returned by read_all_remotes()
3870 'svn' => {
3871 # svn-remote.svn.url=https://svn.musicpd.org
3872 url => 'https://svn.musicpd.org',
3873 # svn-remote.svn.fetch=mpd/trunk:trunk
3874 fetch => {
3875 'mpd/trunk' => 'trunk',
3876 },
3877 # svn-remote.svn.tags=mpd/tags/*:tags/*
3878 tags => {
3879 path => {
3880 left => 'mpd/tags',
3881 right => '',
3882 regex => qr!mpd/tags/([^/]+)$!,
3883 glob => 'tags/*',
3884 },
3885 ref => {
3886 left => 'tags',
3887 right => '',
3888 regex => qr!tags/([^/]+)$!,
3889 glob => 'tags/*',
3890 },
3891 }
3892 }
3893 };
3894
3895 $log_entry hashref as returned by libsvn_log_entry()
3896 {
3897 log => 'whitespace-formatted log entry
3898 ', # trailing newline is preserved
3899 revision => '8', # integer
3900 date => '2004-02-24T17:01:44.108345Z', # commit date
3901 author => 'committer name'
3902 };
3903
3904
3905 # this is generated by generate_diff();
3906 @mods = array of diff-index line hashes, each element represents one line
3907 of diff-index output
3908
3909 diff-index line ($m hash)
3910 {
3911 mode_a => first column of diff-index output, no leading ':',
3912 mode_b => second column of diff-index output,
3913 sha1_b => sha1sum of the final blob,
3914 chg => change type [MCRADT],
3915 file_a => original file name of a file (iff chg is 'C' or 'R')
3916 file_b => new/current file name of a file (any chg)
3917 }
3918 ;
3919
3920 # retval of read_url_paths{,_all}();
3921 $l_map = {
3922 # repository root url
3923 'https://svn.musicpd.org' => {
3924 # repository path # GIT_SVN_ID
3925 'mpd/trunk' => 'trunk',
3926 'mpd/tags/0.11.5' => 'tags/0.11.5',
3927 },
3928 }
3929
3930 Notes:
3931 I don't trust the each() function on unless I created %hash myself
3932 because the internal iterator may not have started at base.
Michael J Gruber's git tree