git-svn: extra safety for noMetadata and useSvmProps users
[git/git.git] / git-svn.perl
CommitLineData
3397f9df 1#!/usr/bin/env perl
551ce28f
EW
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
3397f9df
EW
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
9760adcc
EW
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
3397f9df 9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
60d02ccc 10$VERSION = '@@GIT_VERSION@@';
13ccd6d4 11
706587fc 12$ENV{GIT_DIR} ||= '.git';
9fa00b65 13$Git::SVN::default_repo_id = 'svn';
8b8fc068 14$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
13ccd6d4 15
f8c9d1d2 16$Git::SVN::Log::TZ = $ENV{TZ};
3397f9df 17$ENV{TZ} = 'UTC';
a00439ac 18$| = 1; # unbuffer STDOUT
3397f9df 19
6fda05ae 20sub fatal (@) { print STDERR @_; exit 1 }
b9c85187
EW
21require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
22require SVN::Ra;
23require SVN::Delta;
24if ($SVN::Core::VERSION lt '1.1.0') {
25 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
26}
d81bf827 27push @Git::SVN::Ra::ISA, 'SVN::Ra';
b9c85187
EW
28push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
29push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
3397f9df
EW
30use Carp qw/croak/;
31use IO::File qw//;
32use File::Basename qw/dirname basename/;
33use File::Path qw/mkpath/;
79bb8d88 34use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
968bdf1f 35use IPC::Open3;
336f1714 36use Git;
a5e0cedc 37
336f1714
EW
38BEGIN {
39 my $s;
40 foreach (qw/command command_oneline command_noisy command_output_pipe
41 command_input_pipe command_close_pipe/) {
42 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
706587fc 43 "*Git::SVN::Migration::$_ = ".
f8c9d1d2 44 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
336f1714
EW
45 }
46 eval $s;
47}
48
b9c85187 49my ($SVN);
83e9940a 50
f8c9d1d2
EW
51$sha1 = qr/[a-f\d]{40}/;
52$sha1_short = qr/[a-f\d]{4,40}/;
44320b9e 53my ($_stdin, $_help, $_edit,
9760adcc 54 $_message, $_file,
d05d72e0 55 $_template, $_shared,
9760adcc 56 $_version,
f8c9d1d2
EW
57 $_merge, $_strategy, $_dry_run,
58 $_prefix);
0bed5eaa 59$Git::SVN::_follow_parent = 1;
706587fc
EW
60my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
61 'config-dir=s' => \$Git::SVN::Ra::config_dir,
62 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
0bed5eaa 63my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
dc5869c0 64 'authors-file|A=s' => \$_authors,
ecc712dd 65 'repack:i' => \$Git::SVN::_repack,
97ae0911
EW
66 'noMetadata' => \$Git::SVN::_no_metadata,
67 'useSvmProps' => \$Git::SVN::_use_svm_props,
80f50749 68 'quiet|q' => \$_q,
ecc712dd
EW
69 'repack-flags|repack-args|repack-opts=s' =>
70 \$Git::SVN::_repack_flags,
706587fc 71 %remote_opts );
36f5b1f0 72
9d55b41a
EW
73my ($_trunk, $_tags, $_branches);
74my %multi_opts = ( 'trunk|T=s' => \$_trunk,
75 'tags|t=s' => \$_tags,
76 'branches|b=s' => \$_branches );
77my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
27e9fb8d 78my %cmt_opts = ( 'edit|e' => \$_edit,
24e22aa8
EW
79 'rmdir' => \$SVN::Git::Editor::_rmdir,
80 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
81 'l=i' => \$SVN::Git::Editor::_rename_limit,
82 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
27e9fb8d 83);
9d55b41a 84
3397f9df 85my %cmd = (
2a3240be 86 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
eeb0abe0 87 { 'revision|r=s' => \$_revision, %fc_opts } ],
d2866f9e 88 init => [ \&cmd_init, "Initialize a repo for tracking" .
f8ab6b73 89 " (requires URL argument)",
9d55b41a 90 \%init_opts ],
d7ad3bed
EW
91 dcommit => [ \&cmd_dcommit,
92 'Commit several diffs to merge with upstream',
3289e86e
EW
93 { 'merge|m|M' => \$_merge,
94 'strategy|s=s' => \$_strategy,
95 'dry-run|n' => \$_dry_run,
4b155223 96 %cmt_opts, %fc_opts } ],
1ce255dc
EW
97 'set-tree' => [ \&cmd_set_tree,
98 "Set an SVN repository to a git tree-ish",
99 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
5969cbe1 100 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
a5e0cedc 101 { 'revision|r=i' => \$_revision } ],
8164b652 102 'multi-init' => [ \&cmd_multi_init,
9d55b41a 103 'Initialize multiple trees (like git-svnimport)',
706587fc 104 { %multi_opts, %init_opts, %remote_opts,
1ca7558d 105 'revision|r=i' => \$_revision,
ae410987 106 'prefix=s' => \$_prefix,
1ca7558d 107 } ],
1c8443b0 108 'multi-fetch' => [ \&cmd_multi_fetch,
9d55b41a
EW
109 'Fetch multiple trees (like git-svnimport)',
110 \%fc_opts ],
706587fc
EW
111 'migrate' => [ sub { },
112 # no-op, we automatically run this anyways,
706587fc
EW
113 'Migrate configuration/metadata/layout from
114 previous versions of git-svn',
115 \%remote_opts ],
f8c9d1d2
EW
116 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
117 { 'limit=i' => \$Git::SVN::Log::limit,
79bb8d88 118 'revision|r=s' => \$_revision,
f8c9d1d2
EW
119 'verbose|v' => \$Git::SVN::Log::verbose,
120 'incremental' => \$Git::SVN::Log::incremental,
121 'oneline' => \$Git::SVN::Log::oneline,
122 'show-commit' => \$Git::SVN::Log::show_commit,
123 'non-recursive' => \$Git::SVN::Log::non_recursive,
79bb8d88 124 'authors-file|A=s' => \$_authors,
f8c9d1d2
EW
125 'color' => \$Git::SVN::Log::color,
126 'pager=s' => \$Git::SVN::Log::pager,
79bb8d88 127 } ],
44320b9e
EW
128 'commit-diff' => [ \&cmd_commit_diff,
129 'Commit a diff between two trees',
27e9fb8d
EW
130 { 'message|m=s' => \$_message,
131 'file|F=s' => \$_file,
45bf473a 132 'revision|r=s' => \$_revision,
27e9fb8d 133 %cmt_opts } ],
3397f9df 134);
9d55b41a 135
3397f9df
EW
136my $cmd;
137for (my $i = 0; $i < @ARGV; $i++) {
138 if (defined $cmd{$ARGV[$i]}) {
139 $cmd = $ARGV[$i];
140 splice @ARGV, $i, 1;
141 last;
142 }
143};
144
448c81b4 145my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
a9612be2 146
b8c92cad 147read_repo_config(\%opts);
9760adcc
EW
148my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
149 'minimize-connections' => \$Git::SVN::Migration::_minimize,
150 'id|i=s' => \$Git::SVN::default_ref_id,
151 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
79bb8d88 152exit 1 if (!$rv && $cmd ne 'log');
6f0783cf 153
3397f9df 154usage(0) if $_help;
551ce28f 155version() if $_version;
eeb0abe0
EW
156usage(1) unless defined $cmd;
157load_authors() if $_authors;
f0ecca10 158unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
706587fc
EW
159 Git::SVN::Migration::migration_check();
160}
ecc712dd 161Git::SVN::init_vars();
b805b44a
EW
162eval {
163 Git::SVN::verify_remotes_sanity();
164 $cmd{$cmd}->[0]->(@ARGV);
165};
166fatal $@ if $@;
3397f9df
EW
167exit 0;
168
169####################### primary functions ######################
170sub usage {
171 my $exit = shift || 0;
172 my $fd = $exit ? \*STDERR : \*STDOUT;
173 print $fd <<"";
174git-svn - bidirectional operations between a single Subversion tree and git
175Usage: $0 <command> [options] [arguments]\n
448c81b4
EW
176
177 print $fd "Available commands:\n" unless $cmd;
3397f9df
EW
178
179 foreach (sort keys %cmd) {
448c81b4 180 next if $cmd && $cmd ne $_;
b203b769 181 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
448c81b4
EW
182 foreach (keys %{$cmd{$_}->[2]}) {
183 # prints out arguments as they should be passed:
b8c92cad 184 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
b203b769 185 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
448c81b4
EW
186 "--$_" : "-$_" }
187 split /\|/,$_)," $x\n";
188 }
3397f9df
EW
189 }
190 print $fd <<"";
448c81b4
EW
191\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
192arbitrary identifier if you're tracking multiple SVN branches/repositories in
193one git repository and want to keep them separate. See git-svn(1) for more
194information.
3397f9df
EW
195
196 exit $exit;
197}
198
551ce28f 199sub version {
7d60ab2c 200 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
551ce28f
EW
201 exit 0;
202}
203
8164b652
EW
204sub do_git_init_db {
205 unless (-d $ENV{GIT_DIR}) {
206 my @init_db = ('init');
207 push @init_db, "--template=$_template" if defined $_template;
208 push @init_db, "--shared" if defined $_shared;
209 command_noisy(@init_db);
210 }
211}
212
d2866f9e 213sub cmd_init {
03e0ea87 214 my $url = shift or die "SVN repository location required " .
81c5a0e6 215 "as a command-line argument\n";
03e0ea87
EW
216 if (my $repo_path = shift) {
217 unless (-d $repo_path) {
218 mkpath([$repo_path]);
219 }
d2866f9e
EW
220 chdir $repo_path or croak $!;
221 $ENV{GIT_DIR} = $repo_path . "/.git";
03e0ea87 222 }
8164b652 223 do_git_init_db();
03e0ea87 224
706587fc 225 Git::SVN->init($url);
3397f9df
EW
226}
227
2a3240be 228sub cmd_fetch {
07a1c950
EW
229 if (@_) {
230 die "Additional fetch arguments are no longer supported.\n",
231 "Use --follow-parent if you have moved/copied directories
232 instead.\n";
233 }
1c8443b0 234 my $gs = Git::SVN->new;
3ebe8df7 235 $gs->fetch(parse_revision_argument());
1c8443b0
EW
236 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
237 command_noisy(qw(update-ref refs/heads/master),
238 $gs->{last_commit});
239 }
2a3240be
EW
240}
241
1ce255dc 242sub cmd_set_tree {
3397f9df
EW
243 my (@commits) = @_;
244 if ($_stdin || !@commits) {
245 print "Reading from stdin...\n";
246 @commits = ();
247 while (<STDIN>) {
1ca72aef 248 if (/\b($sha1_short)\b/o) {
3397f9df
EW
249 unshift @commits, $1;
250 }
251 }
252 }
253 my @revs;
8de010ad 254 foreach my $c (@commits) {
aef4e921 255 my @tmp = command('rev-parse',$c);
8de010ad
EW
256 if (scalar @tmp == 1) {
257 push @revs, $tmp[0];
258 } elsif (scalar @tmp > 1) {
aef4e921 259 push @revs, reverse(command('rev-list',@tmp));
8de010ad 260 } else {
1ce255dc 261 fatal "Failed to rev-parse $c\n";
8de010ad 262 }
3397f9df 263 }
1ce255dc
EW
264 my $gs = Git::SVN->new;
265 my ($r_last, $cmt_last) = $gs->last_rev_commit;
266 $gs->fetch;
97f6987a 267 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
1ce255dc
EW
268 fatal "There are new revisions that were fetched ",
269 "and need to be merged (or acknowledged) ",
270 "before committing.\nlast rev: $r_last\n",
271 " current: $gs->{last_rev}\n";
a5e0cedc 272 }
1ce255dc
EW
273 $gs->set_tree($_) foreach @revs;
274 print "Done committing ",scalar @revs," revisions to SVN\n";
a5e0cedc 275}
8f22562c 276
d7ad3bed
EW
277sub cmd_dcommit {
278 my $head = shift;
279 my $gs = Git::SVN->new;
280 $head ||= 'HEAD';
281 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
45bf473a 282 my $last_rev;
b22d4497 283 foreach my $d (reverse @refs) {
aef4e921 284 if (!verify_ref("$d~1")) {
d7ad3bed
EW
285 fatal "Commit $d\n",
286 "has no parent commit, and therefore ",
287 "nothing to diff against.\n",
288 "You should be working from a repository ",
289 "originally created by git-svn\n";
48d044b5 290 }
45bf473a
EW
291 unless (defined $last_rev) {
292 (undef, $last_rev, undef) = cmt_metadata("$d~1");
293 unless (defined $last_rev) {
d7ad3bed
EW
294 fatal "Unable to extract revision information ",
295 "from commit $d~1\n";
45bf473a
EW
296 }
297 }
b22d4497
EW
298 if ($_dry_run) {
299 print "diff-tree $d~1 $d\n";
300 } else {
d7ad3bed 301 my %ed_opts = ( r => $last_rev,
61395354
EW
302 log => get_commit_entry($d)->{log},
303 ra => $gs->ra,
304 tree_a => "$d~1",
305 tree_b => $d,
306 editor_cb => sub {
307 print "Committed r$_[0]\n";
308 $last_rev = $_[0]; },
d3a840dc 309 svn_path => $gs->{path} );
61395354 310 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
d7ad3bed
EW
311 print "No changes\n$d~1 == $d\n";
312 }
b22d4497
EW
313 }
314 }
315 return if $_dry_run;
d7ad3bed
EW
316 $gs->fetch;
317 # we always want to rebase against the current HEAD, not any
318 # head that was passed to us
319 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
b22d4497
EW
320 my @finish;
321 if (@diff) {
322 @finish = qw/rebase/;
323 push @finish, qw/--merge/ if $_merge;
324 push @finish, "--strategy=$_strategy" if $_strategy;
d7ad3bed
EW
325 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
326 "using @finish:\n", "@diff";
b22d4497 327 } else {
d7ad3bed
EW
328 print "No changes between current HEAD and ",
329 $gs->refname, "\nResetting to the latest ",
330 $gs->refname, "\n";
4769489a 331 @finish = qw/reset --mixed/;
b22d4497 332 }
d7ad3bed 333 command_noisy(@finish, $gs->refname);
b22d4497
EW
334}
335
5969cbe1
EW
336sub cmd_show_ignore {
337 my $gs = Git::SVN->new;
338 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
339 $gs->traverse_ignore(\*STDOUT, '', $r);
a5e0cedc
EW
340}
341
8164b652 342sub cmd_multi_init {
9d55b41a 343 my $url = shift;
98327e58
EW
344 unless (defined $_trunk || defined $_branches || defined $_tags) {
345 usage(1);
9d55b41a 346 }
8164b652
EW
347 do_git_init_db();
348 $_prefix = '' unless defined $_prefix;
706587fc 349 $url =~ s#/+$## if defined $url;
98327e58 350 if (defined $_trunk) {
706587fc
EW
351 my $trunk_ref = $_prefix . 'trunk';
352 # try both old-style and new-style lookups:
353 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
8164b652 354 unless ($gs_trunk) {
706587fc
EW
355 my ($trunk_url, $trunk_path) =
356 complete_svn_url($url, $_trunk);
357 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
358 undef, $trunk_ref);
98327e58 359 }
c35b96e7 360 }
706587fc 361 return unless defined $_branches || defined $_tags;
e7db67e6
EW
362 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
363 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
364 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
9d55b41a
EW
365}
366
1c8443b0 367sub cmd_multi_fetch {
0af9c9f9
EW
368 my $remotes = Git::SVN::read_all_remotes();
369 foreach my $repo_id (sort keys %$remotes) {
4bb9ed04
EW
370 if ($remotes->{$repo_id}->{url} &&
371 $remotes->{$repo_id}->{fetch}) {
372 Git::SVN::fetch_all($repo_id, $remotes);
373 }
9d55b41a 374 }
9d55b41a
EW
375}
376
44320b9e
EW
377# this command is special because it requires no metadata
378sub cmd_commit_diff {
379 my ($ta, $tb, $url) = @_;
380 my $usage = "Usage: $0 commit-diff -r<revision> ".
381 "<tree-ish> <tree-ish> [<URL>]\n";
382 fatal($usage) if (!defined $ta || !defined $tb);
d3a840dc 383 my $svn_path;
44320b9e
EW
384 if (!defined $url) {
385 my $gs = eval { Git::SVN->new };
386 if (!$gs) {
387 fatal("Needed URL or usable git-svn --id in ",
388 "the command-line\n", $usage);
389 }
390 $url = $gs->{url};
d3a840dc 391 $svn_path = $gs->{path};
44320b9e
EW
392 }
393 unless (defined $_revision) {
394 fatal("-r|--revision is a required argument\n", $usage);
395 }
396 if (defined $_message && defined $_file) {
397 fatal("Both --message/-m and --file/-F specified ",
398 "for the commit message.\n",
399 "I have no idea what you mean\n");
400 }
401 if (defined $_file) {
402 $_message = file_to_s($_file);
403 } else {
404 $_message ||= get_commit_entry($tb)->{log};
405 }
406 my $ra ||= Git::SVN::Ra->new($url);
d3a840dc 407 $svn_path ||= $ra->{svn_path};
44320b9e
EW
408 my $r = $_revision;
409 if ($r eq 'HEAD') {
410 $r = $ra->get_latest_revnum;
411 } elsif ($r !~ /^\d+$/) {
412 die "revision argument: $r not understood by git-svn\n";
413 }
61395354
EW
414 my %ed_opts = ( r => $r,
415 log => $_message,
416 ra => $ra,
417 tree_a => $ta,
418 tree_b => $tb,
419 editor_cb => sub { print "Committed r$_[0]\n" },
420 svn_path => $svn_path );
421 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
44320b9e
EW
422 print "No changes\n$ta == $tb\n";
423 }
44320b9e
EW
424}
425
3397f9df
EW
426########################### utility functions #########################
427
3ebe8df7
EW
428sub parse_revision_argument {
429 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
430 return (undef, undef);
431 }
432 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
433 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
434 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
435 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
436 die "revision argument: $_revision not understood by git-svn\n",
437 "Try using the command-line svn client instead\n";
438}
439
98327e58
EW
440sub complete_svn_url {
441 my ($url, $path) = @_;
442 $path =~ s#/+$##;
98327e58 443 if ($path !~ m#^[a-z\+]+://#) {
98327e58
EW
444 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
445 fatal("E: '$path' is not a complete URL ",
446 "and a separate URL is not specified\n");
447 }
706587fc 448 return ($url, $path);
98327e58 449 }
706587fc 450 return ($path, '');
98327e58
EW
451}
452
9d55b41a 453sub complete_url_ls_init {
706587fc
EW
454 my ($ra, $repo_path, $switch, $pfx) = @_;
455 unless ($repo_path) {
9d55b41a
EW
456 print STDERR "W: $switch not specified\n";
457 return;
458 }
706587fc
EW
459 $repo_path =~ s#/+$##;
460 if ($repo_path =~ m#^[a-z\+]+://#) {
461 $ra = Git::SVN::Ra->new($repo_path);
462 $repo_path = '';
e7db67e6 463 } else {
706587fc 464 $repo_path =~ s#^/+##;
e7db67e6 465 unless ($ra) {
706587fc 466 fatal("E: '$repo_path' is not a complete URL ",
e7db67e6 467 "and a separate URL is not specified\n");
8164b652 468 }
e7db67e6
EW
469 }
470 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
706587fc
EW
471 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
472 my $url = $ra->{url};
88cf4107
EW
473 my $remote_id;
474 my $remote_path;
e7db67e6
EW
475 foreach my $d (sort keys %$dirent) {
476 next if ($dirent->{$d}->kind != $SVN::Node::dir);
706587fc
EW
477 my $path = "$repo_path/$d";
478 my $ref = "$pfx$d";
479 my $gs = eval { Git::SVN->new($ref) };
e7db67e6 480 # don't try to init already existing refs
8164b652 481 unless ($gs) {
706587fc 482 print "init $url/$path => $ref\n";
4bb9ed04
EW
483 $gs = Git::SVN->init($url, $path, undef, $ref, 1);
484 }
485 if ($gs) {
486 $remote_id = $gs->{repo_id};
487 last;
9d55b41a 488 }
88cf4107
EW
489 }
490 if (defined $remote_id) {
491 $remote_path = "$ra->{svn_path}/$repo_path/*";
492 $remote_path =~ s#/+#/#g;
493 $remote_path =~ s#^/##g;
494 my ($n) = ($switch =~ /^--(\w+)/);
4e9f6cc7
EW
495 if (length $pfx && $pfx !~ m#/$#) {
496 die "--prefix='$pfx' must have a trailing slash '/'\n";
497 }
88cf4107 498 command_noisy('config', "svn-remote.$remote_id.$n",
4bb9ed04 499 "$remote_path:refs/remotes/$pfx*");
9d55b41a 500 }
9d55b41a
EW
501}
502
aef4e921
EW
503sub verify_ref {
504 my ($ref) = @_;
2c5c1d53
EW
505 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
506 { STDERR => 0 }); };
aef4e921
EW
507}
508
a5e0cedc 509sub get_tree_from_treeish {
cf52b8f0 510 my ($treeish) = @_;
44320b9e 511 # $treeish can be a symbolic ref, too:
aef4e921 512 my $type = command_oneline(qw/cat-file -t/, $treeish);
cf52b8f0
EW
513 my $expected;
514 while ($type eq 'tag') {
aef4e921 515 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
cf52b8f0
EW
516 }
517 if ($type eq 'commit') {
aef4e921
EW
518 $expected = (grep /^tree /, command(qw/cat-file commit/,
519 $treeish))[0];
44320b9e 520 ($expected) = ($expected =~ /^tree ($sha1)$/o);
cf52b8f0
EW
521 die "Unable to get tree from $treeish\n" unless $expected;
522 } elsif ($type eq 'tree') {
523 $expected = $treeish;
524 } else {
525 die "$treeish is a $type, expected tree, tag or commit\n";
526 }
a5e0cedc
EW
527 return $expected;
528}
529
44320b9e
EW
530sub get_commit_entry {
531 my ($treeish) = shift;
532 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
533 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
534 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
535 open my $log_fh, '>', $commit_editmsg or croak $!;
3397f9df 536
44320b9e 537 my $type = command_oneline(qw/cat-file -t/, $treeish);
4ad4515d 538 if ($type eq 'commit' || $type eq 'tag') {
aef4e921 539 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
44320b9e 540 $type, $treeish);
3397f9df
EW
541 my $in_msg = 0;
542 while (<$msg_fh>) {
543 if (!$in_msg) {
544 $in_msg = 1 if (/^\s*$/);
df746c5a 545 } elsif (/^git-svn-id: /) {
44320b9e
EW
546 # skip this for now, we regenerate the
547 # correct one on re-fetch anyways
548 # TODO: set *:merge properties or like...
3397f9df 549 } else {
44320b9e 550 print $log_fh $_ or croak $!;
3397f9df
EW
551 }
552 }
aef4e921 553 command_close_pipe($msg_fh, $ctx);
3397f9df 554 }
44320b9e 555 close $log_fh or croak $!;
3397f9df
EW
556
557 if ($_edit || ($type eq 'tree')) {
558 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
44320b9e
EW
559 # TODO: strip out spaces, comments, like git-commit.sh
560 system($editor, $commit_editmsg);
3397f9df 561 }
44320b9e
EW
562 rename $commit_editmsg, $commit_msg or croak $!;
563 open $log_fh, '<', $commit_msg or croak $!;
564 { local $/; chomp($log_entry{log} = <$log_fh>); }
565 close $log_fh or croak $!;
566 unlink $commit_msg;
567 \%log_entry;
a5e0cedc
EW
568}
569
3397f9df
EW
570sub s_to_file {
571 my ($str, $file, $mode) = @_;
572 open my $fd,'>',$file or croak $!;
573 print $fd $str,"\n" or croak $!;
574 close $fd or croak $!;
575 chmod ($mode &~ umask, $file) if (defined $mode);
576}
577
578sub file_to_s {
579 my $file = shift;
580 open my $fd,'<',$file or croak "$!: file: $file\n";
581 local $/;
582 my $ret = <$fd>;
583 close $fd or croak $!;
584 $ret =~ s/\s*$//s;
585 return $ret;
586}
587
eeb0abe0
EW
588# '<svn username> = real-name <email address>' mapping based on git-svnimport:
589sub load_authors {
590 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
f8c9d1d2 591 my $log = $cmd eq 'log';
eeb0abe0
EW
592 while (<$authors>) {
593 chomp;
8815788e 594 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
eeb0abe0 595 my ($user, $name, $email) = ($1, $2, $3);
f8c9d1d2
EW
596 if ($log) {
597 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
598 } else {
599 $users{$user} = [$name, $email];
600 }
79bb8d88
EW
601 }
602 close $authors or croak $!;
603}
604
e0d10e1c 605# convert GetOpt::Long specs for use by git-config
b8c92cad 606sub read_repo_config {
706587fc 607 return unless -d $ENV{GIT_DIR};
b8c92cad 608 my $opts = shift;
97ae0911 609 my @config_only;
b8c92cad 610 foreach my $o (keys %$opts) {
97ae0911
EW
611 # if we have mixedCase and a long option-only, then
612 # it's a config-only variable that we don't need for
613 # the command-line.
614 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
b8c92cad 615 my $v = $opts->{$o};
97ae0911 616 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
b8c92cad 617 $key =~ s/-//g;
e0d10e1c 618 my $arg = 'git-config';
b8c92cad
EW
619 $arg .= ' --int' if ($o =~ /[:=]i$/);
620 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
621 if (ref $v eq 'ARRAY') {
622 chomp(my @tmp = `$arg --get-all svn.$key`);
623 @$v = @tmp if @tmp;
624 } else {
625 chomp(my $tmp = `$arg --get svn.$key`);
7774284a 626 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
b8c92cad
EW
627 $$v = $tmp;
628 }
629 }
630 }
97ae0911 631 delete @$opts{@config_only} if @config_only;
b8c92cad
EW
632}
633
79bb8d88 634sub extract_metadata {
c1927a85 635 my $id = shift or return (undef, undef, undef);
79bb8d88
EW
636 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
637 \s([a-f\d\-]+)$/x);
e70dc780 638 if (!defined $rev || !$uuid || !$url) {
79bb8d88 639 # some of the original repositories I made had
82e5a82f 640 # identifiers like this:
79bb8d88
EW
641 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
642 }
643 return ($url, $rev, $uuid);
644}
645
c1927a85
EW
646sub cmt_metadata {
647 return extract_metadata((grep(/^git-svn-id: /,
aef4e921 648 command(qw/cat-file commit/, shift)))[-1]);
c1927a85
EW
649}
650
9b981fc6
EW
651package Git::SVN;
652use strict;
653use warnings;
ecc712dd 654use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
8a49ee97 655 $_repack $_repack_flags $_use_svm_props/;
9b981fc6
EW
656use Carp qw/croak/;
657use File::Path qw/mkpath/;
373274f9 658use File::Copy qw/copy/;
9b981fc6
EW
659use IPC::Open3;
660
ecc712dd 661my $_repack_nr;
9b981fc6
EW
662# properties that we do not log:
663my %SKIP_PROP;
664BEGIN {
665 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
666 svn:special svn:executable
667 svn:entry:committed-rev
668 svn:entry:last-author
669 svn:entry:uuid
670 svn:entry:committed-date/;
91b03282
EW
671
672 # some options are read globally, but can be overridden locally
673 # per [svn-remote "..."] section. Command-line options will *NOT*
674 # override options set in an [svn-remote "..."] section
675 my $e;
676 foreach (qw/follow_parent no_metadata use_svm_props/) {
677 my $key = $_;
678 $key =~ tr/_//d;
679 $e .= "sub $_ {
680 my (\$self) = \@_;
681 return \$self->{-$_} if exists \$self->{-$_};
682 my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
683 eval { command_oneline(qw/config --get/, \$k) };
684 if (\$@) {
685 \$self->{-$_} = \$Git::SVN::_$_;
686 } else {
687 my \$v = command_oneline(qw/config --bool/,\$k);
688 \$self->{-$_} = \$v eq 'false' ? 0 : 1;
689 }
690 return \$self->{-$_} }\n";
691 }
692 $e .= "1;\n";
693 eval $e or die $@;
9b981fc6
EW
694}
695
373274f9
EW
696my %LOCKFILES;
697END { unlink keys %LOCKFILES if %LOCKFILES }
698
4bb9ed04
EW
699sub resolve_local_globs {
700 my ($url, $fetch, $glob_spec) = @_;
701 return unless defined $glob_spec;
702 my $ref = $glob_spec->{ref};
703 my $path = $glob_spec->{path};
704 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
705 next unless m#^refs/remotes/$ref->{regex}$#;
706 my $p = $1;
707 my $pathname = $path->full_path($p);
708 my $refname = $ref->full_path($p);
709 if (my $existing = $fetch->{$pathname}) {
710 if ($existing ne $refname) {
711 die "Refspec conflict:\n",
712 "existing: refs/remotes/$existing\n",
713 " globbed: refs/remotes/$refname\n";
714 }
715 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
4e9f6cc7 716 $u =~ s!^\Q$url\E(/|$)!! or die
4bb9ed04
EW
717 "refs/remotes/$refname: '$url' not found in '$u'\n";
718 if ($pathname ne $u) {
719 warn "W: Refspec glob conflict ",
720 "(ref: refs/remotes/$refname):\n",
721 "expected path: $pathname\n",
722 " real path: $u\n",
723 "Continuing ahead with $u\n";
724 next;
725 }
726 } else {
4bb9ed04
EW
727 $fetch->{$pathname} = $refname;
728 }
729 }
730}
731
0af9c9f9 732sub fetch_all {
4bb9ed04 733 my ($repo_id, $remotes) = @_;
e518192f
EW
734 my $remote = $remotes->{$repo_id};
735 my $fetch = $remote->{fetch};
736 my $url = $remote->{url};
737 my (@gs, @globs);
0af9c9f9 738 my $ra = Git::SVN::Ra->new($url);
e518192f 739 my $uuid = $ra->uuid;
0af9c9f9
EW
740 my $head = $ra->get_latest_revnum;
741 my $base = $head;
e518192f
EW
742
743 # read the max revs for wildcard expansion (branches/*, tags/*)
744 foreach my $t (qw/branches tags/) {
745 defined $remote->{$t} or next;
746 push @globs, $remote->{$t};
93f2689c
EW
747 my $max_rev = eval { tmp_config(qw/--int --get/,
748 "svn-remote.$repo_id.${t}-maxRev") };
749 if (defined $max_rev && ($max_rev < $base)) {
750 $base = $max_rev;
e518192f
EW
751 }
752 }
753
0af9c9f9
EW
754 foreach my $p (sort keys %$fetch) {
755 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
9c93fee5 756 my $lr = $gs->rev_db_max;
0af9c9f9
EW
757 if (defined $lr) {
758 $base = $lr if ($lr < $base);
0af9c9f9
EW
759 }
760 push @gs, $gs;
761 }
e518192f 762 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
0af9c9f9
EW
763}
764
47e39c55
EW
765sub read_all_remotes {
766 my $r = {};
8b8fc068 767 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
47e39c55
EW
768 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
769 $r->{$1}->{fetch}->{$2} = $3;
770 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
771 $r->{$1}->{url} = $2;
4bb9ed04
EW
772 } elsif (m!^(.+)\.(branches|tags)=
773 (.*):refs/remotes/(.+)\s*$/!x) {
774 my ($p, $g) = ($3, $4);
775 my $rs = $r->{$1}->{$2} = {
e518192f 776 t => $2,
93f2689c 777 remote => $1,
4bb9ed04
EW
778 path => Git::SVN::GlobSpec->new($p),
779 ref => Git::SVN::GlobSpec->new($g) };
780 if (length($rs->{ref}->{right}) != 0) {
781 die "The '*' glob character must be the last ",
782 "character of '$g'\n";
783 }
47e39c55
EW
784 }
785 }
786 $r;
787}
788
ecc712dd
EW
789sub init_vars {
790 if (defined $_repack) {
791 $_repack = 1000 if ($_repack <= 0);
792 $_repack_nr = $_repack;
793 $_repack_flags ||= '-d';
794 }
795}
796
b805b44a 797sub verify_remotes_sanity {
536c4b09 798 return unless -d $ENV{GIT_DIR};
b805b44a
EW
799 my %seen;
800 foreach (command(qw/config -l/)) {
801 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
802 if ($seen{$1}) {
803 die "Remote ref refs/remote/$1 is tracked by",
804 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
805 "Please resolve this ambiguity in ",
806 "your git configuration file before ",
807 "continuing\n";
808 }
809 $seen{$1} = $_;
810 }
811 }
812}
813
47e39c55 814# we allow more chars than remotes2config.sh...
706587fc
EW
815sub sanitize_remote_name {
816 my ($name) = @_;
47e39c55 817 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
706587fc
EW
818 $name;
819}
820
e6434f87
EW
821sub find_existing_remote {
822 my ($url, $remotes) = @_;
823 my $existing;
824 foreach my $repo_id (keys %$remotes) {
825 my $u = $remotes->{$repo_id}->{url} or next;
826 next if $u ne $url;
827 $existing = $repo_id;
828 last;
829 }
830 $existing;
831}
b805b44a 832
e6434f87 833sub init_remote_config {
d8115c51 834 my ($self, $url, $no_write) = @_;
e6434f87
EW
835 $url =~ s!/+$!!; # strip trailing slash
836 my $r = read_all_remotes();
837 my $existing = find_existing_remote($url, $r);
838 if ($existing) {
e518192f
EW
839 unless ($no_write) {
840 print STDERR "Using existing ",
841 "[svn-remote \"$existing\"]\n";
842 }
e6434f87
EW
843 $self->{repo_id} = $existing;
844 } else {
845 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
846 $existing = find_existing_remote($min_url, $r);
847 if ($existing) {
e518192f
EW
848 unless ($no_write) {
849 print STDERR "Using existing ",
850 "[svn-remote \"$existing\"]\n";
851 }
e6434f87
EW
852 $self->{repo_id} = $existing;
853 }
854 if ($min_url ne $url) {
e518192f
EW
855 unless ($no_write) {
856 print STDERR "Using higher level of URL: ",
857 "$url => $min_url\n";
858 }
e6434f87
EW
859 my $old_path = $self->{path};
860 $self->{path} = $url;
4e9f6cc7 861 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
e6434f87
EW
862 if (length $old_path) {
863 $self->{path} .= "/$old_path";
864 }
865 $url = $min_url;
866 }
867 }
868 my $orig_url;
869 if (!$existing) {
b805b44a 870 # verify that we aren't overwriting anything:
e6434f87 871 $orig_url = eval {
706587fc 872 command_oneline('config', '--get',
e6434f87 873 "svn-remote.$self->{repo_id}.url")
706587fc 874 };
b805b44a 875 if ($orig_url && ($orig_url ne $url)) {
e6434f87 876 die "svn-remote.$self->{repo_id}.url already set: ",
b805b44a
EW
877 "$orig_url\nwanted to set to: $url\n";
878 }
9b981fc6 879 }
e6434f87
EW
880 my ($xrepo_id, $xpath) = find_ref($self->refname);
881 if (defined $xpath) {
882 die "svn-remote.$xrepo_id.fetch already set to track ",
883 "$xpath:refs/remotes/", $self->refname, "\n";
884 }
d8115c51
EW
885 unless ($no_write) {
886 command_noisy('config',
887 "svn-remote.$self->{repo_id}.url", $url);
888 command_noisy('config', '--add',
889 "svn-remote.$self->{repo_id}.fetch",
890 "$self->{path}:".$self->refname);
891 }
9b981fc6 892 $self->{url} = $url;
e6434f87
EW
893}
894
895sub init {
d8115c51 896 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
e6434f87
EW
897 my $self = _new($class, $repo_id, $ref_id, $path);
898 if (defined $url) {
d8115c51 899 $self->init_remote_config($url, $no_write);
e6434f87 900 }
9b981fc6
EW
901 $self;
902}
903
706587fc
EW
904sub find_ref {
905 my ($ref_id) = @_;
906 foreach (command(qw/config -l/)) {
907 next unless m!^svn-remote\.(.+)\.fetch=
908 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
909 my ($repo_id, $path, $ref) = ($1, $2, $3);
910 if ($ref eq $ref_id) {
911 $path = '' if ($path =~ m#^\./?#);
912 return ($repo_id, $path);
913 }
914 }
915 (undef, undef, undef);
916}
917
9b981fc6 918sub new {
706587fc
EW
919 my ($class, $ref_id, $repo_id, $path) = @_;
920 if (defined $ref_id && !defined $repo_id && !defined $path) {
921 ($repo_id, $path) = find_ref($ref_id);
922 if (!defined $repo_id) {
923 die "Could not find a \"svn-remote.*.fetch\" key ",
924 "in the repository configuration matching: ",
925 "refs/remotes/$ref_id\n";
926 }
927 }
928 my $self = _new($class, $repo_id, $ref_id, $path);
8b8fc068
EW
929 if (!defined $self->{path} || !length $self->{path}) {
930 my $fetch = command_oneline('config', '--get',
931 "svn-remote.$repo_id.fetch",
932 ":refs/remotes/$ref_id\$") or
933 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
934 "\":refs/remotes/$ref_id\$\" in config\n";
935 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
936 }
706587fc
EW
937 $self->{url} = command_oneline('config', '--get',
938 "svn-remote.$repo_id.url") or
939 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
f0ecca10
EW
940 if (-z $self->{db_path} && ::verify_ref($self->refname.'^0')) {
941 $self->rebuild;
942 }
9b981fc6
EW
943 $self;
944}
945
706587fc 946sub refname { "refs/remotes/$_[0]->{ref_id}" }
9b981fc6 947
8a49ee97
EW
948sub set_svm_vars {
949 my ($self, $ra) = @_;
950 my $section = "svn-remote.$self->{repo_id}";
951
952 # see if we have it in our config, first:
953 eval {
954 $self->{svm} = {
93f2689c
EW
955 source => tmp_config('--get', "$section.svm-source"),
956 uuid => tmp_config('--get', "$section.svm-uuid"),
8a49ee97
EW
957 }
958 };
959 return $ra if ($self->{svm}->{source} && $self->{svm}->{uuid});
960
961 # nope, make sure we're connected to the repository root:
962 if ($ra->{repos_root} ne $self->{url}) {
963 $ra = Git::SVN::Ra->new($ra->{repos_root});
964 }
965 my $r = $ra->get_latest_revnum;
966 my ($props) = ($ra->get_dir('', $r))[2];
967 if (my $src = $props->{'svm:source'}) {
968 # don't know what a '!' is there for, also the
969 # username is of no interest
970 $src =~ s{!$}{};
971 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
93f2689c 972 tmp_config('--add', "$section.svm-source", $src);
8a49ee97
EW
973
974 my $uuid = $props->{'svm:uuid'};
975 $uuid =~ m{^[0-9a-f\-]{30,}$}
976 or die "doesn't look right - svm:uuid is '$uuid'\n";
93f2689c 977 tmp_config('--add', "$section.svm-uuid", $uuid);
8a49ee97
EW
978
979 $self->{svm} = { source => $src , uuid => $uuid };
980 }
981 if ($ra->{repos_root} ne $self->{url}) {
982 $ra = Git::SVN::Ra->new($self->{url});
983 }
984 $ra;
985}
986
9b981fc6
EW
987sub ra {
988 my ($self) = shift;
8a49ee97 989 my $ra = Git::SVN::Ra->new($self->{url});
91b03282
EW
990 if ($self->use_svm_props && !$self->{svm}) {
991 if ($self->no_metadata) {
97ae0911
EW
992 die "Can't have both 'noMetadata' and ",
993 "'useSvmProps' options set!\n";
91b03282 994 }
8a49ee97
EW
995 $ra = $self->set_svm_vars($ra);
996 $self->{-want_revprops} = 1;
997 }
998 $ra;
9b981fc6
EW
999}
1000
15710b6f
EW
1001sub rel_path {
1002 my ($self) = @_;
1003 my $repos_root = $self->ra->{repos_root};
1004 return $self->{path} if ($self->{url} eq $repos_root);
4e9f6cc7
EW
1005 die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1006 $self->ra->{url}, " path: $self->{path}, URL: $self->{url}\n";
15710b6f
EW
1007}
1008
9b981fc6
EW
1009sub traverse_ignore {
1010 my ($self, $fh, $path, $r) = @_;
1011 $path =~ s#^/+##g;
5d3b7cd5
EW
1012 my $ra = $self->ra;
1013 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
9b981fc6 1014 my $p = $path;
5d3b7cd5 1015 $p =~ s#^\Q$ra->{svn_path}\E/##;
9b981fc6
EW
1016 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1017 if (my $s = $props->{'svn:ignore'}) {
1018 $s =~ s/[\r\n]+/\n/g;
1019 chomp $s;
1020 if (length $p == 0) {
1021 $s =~ s#\n#\n/$p#g;
1022 print $fh "/$s\n";
1023 } else {
1024 $s =~ s#\n#\n/$p/#g;
1025 print $fh "/$p/$s\n";
1026 }
1027 }
1028 foreach (sort keys %$dirent) {
1029 next if $dirent->{$_}->kind != $SVN::Node::dir;
1030 $self->traverse_ignore($fh, "$path/$_", $r);
1031 }
1032}
1033
3ebe8df7
EW
1034sub last_rev { ($_[0]->last_rev_commit)[0] }
1035sub last_commit { ($_[0]->last_rev_commit)[1] }
1036
9b981fc6
EW
1037# returns the newest SVN revision number and newest commit SHA1
1038sub last_rev_commit {
1039 my ($self) = @_;
1040 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1041 return ($self->{last_rev}, $self->{last_commit});
1042 }
d2866f9e 1043 my $c = ::verify_ref($self->refname.'^0');
91b03282 1044 if ($c && !$self->use_svm_props && !$self->no_metadata) {
d2866f9e 1045 my $rev = (::cmt_metadata($c))[1];
9b981fc6
EW
1046 if (defined $rev) {
1047 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1048 return ($rev, $c);
1049 }
1050 }
1051 my $offset = -41; # from tail
1052 my $rl;
1053 open my $fh, '<', $self->{db_path} or
1054 croak "$self->{db_path} not readable: $!\n";
ce4b4af7
EW
1055 sysseek($fh, $offset, 2); # don't care for errors
1056 sysread($fh, $rl, 41) == 41 or return (undef, undef);
9b981fc6 1057 chomp $rl;
ce4b4af7 1058 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
9b981fc6 1059 $offset -= 41;
ce4b4af7
EW
1060 sysseek($fh, $offset, 2); # don't care for errors
1061 sysread($fh, $rl, 41) == 41 or return (undef, undef);
9b981fc6
EW
1062 chomp $rl;
1063 }
91b03282 1064 if ($c && $c ne $rl) {
9c93fee5
EW
1065 die "$self->{db_path} and ", $self->refname,
1066 " inconsistent!:\n$c != $rl\n";
1067 }
ce4b4af7 1068 my $rev = sysseek($fh, 0, 1) or croak $!;
9b981fc6
EW
1069 $rev = ($rev - 41) / 41;
1070 close $fh or croak $!;
1071 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1072 return ($rev, $c);
1073}
1074
3ebe8df7
EW
1075sub get_fetch_range {
1076 my ($self, $min, $max) = @_;
1077 $max ||= $self->ra->get_latest_revnum;
9c93fee5 1078 $min ||= $self->rev_db_max;
3ebe8df7 1079 (++$min, $max);
9b981fc6
EW
1080}
1081
8a49ee97 1082sub tmp_config {
93f2689c
EW
1083 my (@args) = @_;
1084 my $config = "$ENV{GIT_DIR}/svn/config";
1085 unless (-f $config) {
1086 open my $fh, '>', $config or
1087 die "Can't open $config: $!\n";
8a49ee97 1088 print $fh "; This file is used internally by git-svn\n" or
93f2689c 1089 die "Couldn't write to $config: $!\n";
8a49ee97 1090 print $fh "; You should not have to edit it\n" or
93f2689c
EW
1091 die "Couldn't write to $config: $!\n";
1092 close $fh or die "Couldn't close $config: $!\n";
8a49ee97
EW
1093 }
1094 my $old_config = $ENV{GIT_CONFIG};
93f2689c 1095 $ENV{GIT_CONFIG} = $config;
8a49ee97
EW
1096 $@ = undef;
1097 my @ret = eval { command('config', @args) };
1098 my $err = $@;
1099 if (defined $old_config) {
1100 $ENV{GIT_CONFIG} = $old_config;
1101 } else {
1102 delete $ENV{GIT_CONFIG};
1103 }
1104 die $err if $err;
1105 wantarray ? @ret : $ret[0];
1106}
1107
9b981fc6
EW
1108sub tmp_index_do {
1109 my ($self, $sub) = @_;
1110 my $old_index = $ENV{GIT_INDEX_FILE};
1111 $ENV{GIT_INDEX_FILE} = $self->{index};
8a49ee97
EW
1112 $@ = undef;
1113 my @ret = eval { &$sub };
1114 my $err = $@;
1115 if (defined $old_index) {
9b981fc6
EW
1116 $ENV{GIT_INDEX_FILE} = $old_index;
1117 } else {
1118 delete $ENV{GIT_INDEX_FILE};
1119 }
8a49ee97 1120 die $err if $err;
9b981fc6
EW
1121 wantarray ? @ret : $ret[0];
1122}
1123
1124sub assert_index_clean {
1125 my ($self, $treeish) = @_;
1126
1127 $self->tmp_index_do(sub {
1128 command_noisy('read-tree', $treeish) unless -e $self->{index};
1129 my $x = command_oneline('write-tree');
1130 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1131 /^tree ($::sha1)/mo);
1132 if ($y ne $x) {
1133 unlink $self->{index} or croak $!;
1134 command_noisy('read-tree', $treeish);
1135 }
1136 $x = command_oneline('write-tree');
1137 if ($y ne $x) {
1138 ::fatal "trees ($treeish) $y != $x\n",
1139 "Something is seriously wrong...\n";
1140 }
1141 });
1142}
1143
1144sub get_commit_parents {
0af9c9f9 1145 my ($self, $log_entry) = @_;
9b981fc6 1146 my (%seen, @ret, @tmp);
0af9c9f9
EW
1147 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1148 if (my $ip = $self->{inject_parents}) {
1149 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1150 push @tmp, $commit;
9b981fc6
EW
1151 }
1152 }
d2866f9e 1153 if (my $cur = ::verify_ref($self->refname.'^0')) {
9b981fc6
EW
1154 push @tmp, $cur;
1155 }
44320b9e 1156 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
9b981fc6
EW
1157 while (my $p = shift @tmp) {
1158 next if $seen{$p};
1159 $seen{$p} = 1;
1160 push @ret, $p;
1161 # MAXPARENT is defined to 16 in commit-tree.c:
1162 last if @ret >= 16;
1163 }
1164 if (@tmp) {
44320b9e 1165 die "r$log_entry->{revision}: No room for parents:\n\t",
9b981fc6
EW
1166 join("\n\t", @tmp), "\n";
1167 }
1168 @ret;
1169}
1170
706587fc 1171sub full_url {
9b981fc6 1172 my ($self) = @_;
5d3b7cd5 1173 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
9b981fc6
EW
1174}
1175
1176sub do_git_commit {
0af9c9f9 1177 my ($self, $log_entry) = @_;
8a603774
EW
1178 my $lr = $self->last_rev;
1179 if (defined $lr && $lr >= $log_entry->{revision}) {
1180 die "Last fetched revision of ", $self->refname,
1181 " was r$lr, but we are about to fetch: ",
1182 "r$log_entry->{revision}!\n";
1183 }
44320b9e
EW
1184 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1185 croak "$log_entry->{revision} = $c already exists! ",
9b981fc6
EW
1186 "Why are we refetching it?\n";
1187 }
1c8443b0
EW
1188 my $author = $log_entry->{author};
1189 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1190 : ($author, "$author\@".$self->ra->uuid));
9b981fc6
EW
1191 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1192 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
44320b9e 1193 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
9b981fc6 1194
44320b9e 1195 my $tree = $log_entry->{tree};
9b981fc6
EW
1196 if (!defined $tree) {
1197 $tree = $self->tmp_index_do(sub {
1198 command_oneline('write-tree') });
1199 }
1200 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1201
1202 my @exec = ('git-commit-tree', $tree);
0af9c9f9 1203 foreach ($self->get_commit_parents($log_entry)) {
9b981fc6
EW
1204 push @exec, '-p', $_;
1205 }
1206 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1207 or croak $!;
44320b9e 1208 print $msg_fh $log_entry->{log} or croak $!;
91b03282 1209 unless ($self->no_metadata) {
8a49ee97
EW
1210 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1211 or croak $!;
9760adcc 1212 }
9b981fc6
EW
1213 $msg_fh->flush == 0 or croak $!;
1214 close $msg_fh or croak $!;
1215 chomp(my $commit = do { local $/; <$out_fh> });
1216 close $out_fh or croak $!;
1217 waitpid $pid, 0;
1218 croak $? if $?;
1219 if ($commit !~ /^$::sha1$/o) {
1220 die "Failed to commit, invalid sha1: $commit\n";
1221 }
1222
373274f9 1223 $self->rev_db_set($log_entry->{revision}, $commit, 1);
9b981fc6 1224
44320b9e 1225 $self->{last_rev} = $log_entry->{revision};
9b981fc6 1226 $self->{last_commit} = $commit;
8a49ee97
EW
1227 print "r$log_entry->{revision}";
1228 if (defined $log_entry->{svm_revision}) {
1229 print " (\@$log_entry->{svm_revision})";
1230 }
1231 print " = $commit ($self->{ref_id})\n";
ecc712dd
EW
1232 if (defined $_repack && (--$_repack_nr == 0)) {
1233 $_repack_nr = $_repack;
1234 # repack doesn't use any arguments with spaces in them, does it?
1235 print "Running git repack $_repack_flags ...\n";
1236 command_noisy('repack', split(/\s+/, $_repack_flags));
1237 print "Done repacking\n";
1238 }
9b981fc6
EW
1239 return $commit;
1240}
1241
fbcc1737
EW
1242sub match_paths {
1243 my ($self, $paths, $r) = @_;
4e9f6cc7 1244 return 1 if $self->{path} eq '';
d542aedb
EW
1245 if (my $path = $paths->{"/$self->{path}"}) {
1246 return ($path->{action} eq 'D') ? 0 : 1;
1247 }
4e9f6cc7 1248 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
fbcc1737
EW
1249 if (grep /$self->{path_regex}/, keys %$paths) {
1250 return 1;
1251 }
1252 my $c = '';
1253 foreach (split m#/#, $self->{path}) {
1254 $c .= "/$_";
74a81227
EW
1255 next unless ($paths->{$c} &&
1256 ($paths->{$c}->{action} =~ /^[AR]$/));
e518192f
EW
1257 if ($self->ra->check_path($self->{path}, $r) ==
1258 $SVN::Node::dir) {
fbcc1737
EW
1259 return 1;
1260 }
1261 }
1262 return 0;
1263}
1264
15710b6f
EW
1265sub find_parent_branch {
1266 my ($self, $paths, $rev) = @_;
91b03282 1267 return undef unless $self->follow_parent;
e5a0b240 1268 unless (defined $paths) {
c7eba716
EW
1269 my $err_handler = $SVN::Error::handler;
1270 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
d4eff2bd
EW
1271 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1272 $paths =
1273 Git::SVN::Ra::dup_changed_paths($_[0]) });
c7eba716 1274 $SVN::Error::handler = $err_handler;
e5a0b240
EW
1275 }
1276 return undef unless defined $paths;
15710b6f
EW
1277
1278 # look for a parent from another branch:
7f578c55
EW
1279 my @b_path_components = split m#/#, $self->rel_path;
1280 my @a_path_components;
1281 my $i;
1282 while (@b_path_components) {
1283 $i = $paths->{'/'.join('/', @b_path_components)};
74a81227 1284 last if $i && defined $i->{copyfrom_path};
7f578c55
EW
1285 unshift(@a_path_components, pop(@b_path_components));
1286 }
74a81227
EW
1287 return undef unless defined $i && defined $i->{copyfrom_path};
1288 my $branch_from = $i->{copyfrom_path};
7f578c55
EW
1289 if (@a_path_components) {
1290 print STDERR "branch_from: $branch_from => ";
1291 $branch_from .= '/'.join('/', @a_path_components);
1292 print STDERR $branch_from, "\n";
1293 }
3ebe8df7 1294 my $r = $i->{copyfrom_rev};
15710b6f
EW
1295 my $repos_root = $self->ra->{repos_root};
1296 my $url = $self->ra->{url};
1297 my $new_url = $repos_root . $branch_from;
1298 print STDERR "Found possible branch point: ",
1299 "$new_url => ", $self->full_url, ", $r\n";
1300 $branch_from =~ s#^/##;
1301 my $remotes = read_all_remotes();
1302 my $gs;
1303 foreach my $repo_id (keys %$remotes) {
1304 my $u = $remotes->{$repo_id}->{url} or next;
1305 next if $url ne $u;
1306 my $fetch = $remotes->{$repo_id}->{fetch};
e518192f
EW
1307 foreach (qw/branches tags/) {
1308 resolve_local_globs($url, $fetch,
1309 $remotes->{$repo_id}->{$_});
1310 }
15710b6f
EW
1311 foreach my $f (keys %$fetch) {
1312 next if $f ne $branch_from;
1313 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1314 last;
1315 }
1316 last if $gs;
1317 }
1318 unless ($gs) {
ce2a0f2f
EW
1319 my $ref_id = $self->{ref_id};
1320 $ref_id =~ s/\@\d+$//;
1321 $ref_id .= "\@$r";
15710b6f
EW
1322 # just grow a tail if we're not unique enough :x
1323 $ref_id .= '-' while find_ref($ref_id);
ce2a0f2f 1324 print STDERR "Initializing parent: $ref_id\n";
d8115c51 1325 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
15710b6f
EW
1326 }
1327 my ($r0, $parent) = $gs->find_rev_before($r, 1);
91b03282 1328 if (!defined $r0 || !defined $parent) {
3ebe8df7 1329 $gs->fetch(0, $r);
15710b6f
EW
1330 ($r0, $parent) = $gs->last_rev_commit;
1331 }
ef70de96 1332 if (defined $r0 && defined $parent) {
15710b6f 1333 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
a2003abc 1334 $self->assert_index_clean($parent);
15710b6f
EW
1335 my $ed;
1336 if ($self->ra->can_do_switch) {
8b8fc068 1337 print STDERR "Following parent with do_switch\n";
15710b6f 1338 # do_switch works with svn/trunk >= r22312, but that
2b27f6c8 1339 # is not included with SVN 1.4.3 (the latest version
15710b6f
EW
1340 # at the moment), so we can't rely on it
1341 $self->{last_commit} = $parent;
1342 $ed = SVN::Git::Fetcher->new($self);
8a603774 1343 $gs->ra->gs_do_switch($r0, $rev, $gs,
15710b6f
EW
1344 $self->full_url, $ed)
1345 or die "SVN connection failed somewhere...\n";
1346 } else {
8b8fc068 1347 print STDERR "Following parent with do_update\n";
15710b6f 1348 $ed = SVN::Git::Fetcher->new($self);
8a603774 1349 $self->ra->gs_do_update($rev, $rev, $self, $ed)
15710b6f
EW
1350 or die "SVN connection failed somewhere...\n";
1351 }
f7c3fc4a 1352 print STDERR "Successfully followed parent\n";
15710b6f
EW
1353 return $self->make_log_entry($rev, [$parent], $ed);
1354 }
15710b6f
EW
1355 return undef;
1356}
1357
9b981fc6 1358sub do_fetch {
706587fc 1359 my ($self, $paths, $rev) = @_;
15710b6f 1360 my $ed;
9b981fc6 1361 my ($last_rev, @parents);
b9dffd8c
EW
1362 if (my $lc = $self->last_commit) {
1363 # we can have a branch that was deleted, then re-added
1364 # under the same name but copied from another path, in
1365 # which case we'll have multiple parents (we don't
1366 # want to break the original ref, nor lose copypath info):
1367 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1368 push @{$log_entry->{parents}}, $lc;
1369 return $log_entry;
1370 }
15710b6f 1371 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 1372 $last_rev = $self->{last_rev};
b9dffd8c
EW
1373 $ed->{c} = $lc;
1374 @parents = ($lc);
9b981fc6
EW
1375 } else {
1376 $last_rev = $rev;
15710b6f
EW
1377 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1378 return $log_entry;
1379 }
1380 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 1381 }
8a603774 1382 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
9b981fc6
EW
1383 die "SVN connection failed somewhere...\n";
1384 }
1385 $self->make_log_entry($rev, \@parents, $ed);
1386}
1387
97f6987a
EW
1388sub get_untracked {
1389 my ($self, $ed) = @_;
1390 my @out;
1391 my $h = $ed->{empty};
9b981fc6
EW
1392 foreach (sort keys %$h) {
1393 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
97f6987a 1394 push @out, " $act: " . uri_encode($_);
9b981fc6
EW
1395 warn "W: $act: $_\n";
1396 }
1397 foreach my $t (qw/dir_prop file_prop/) {
97f6987a 1398 $h = $ed->{$t} or next;
9b981fc6
EW
1399 foreach my $path (sort keys %$h) {
1400 my $ppath = $path eq '' ? '.' : $path;
1401 foreach my $prop (sort keys %{$h->{$path}}) {
1ce255dc 1402 next if $SKIP_PROP{$prop};
9b981fc6 1403 my $v = $h->{$path}->{$prop};
97f6987a
EW
1404 my $t_ppath_prop = "$t: " .
1405 uri_encode($ppath) . ' ' .
1406 uri_encode($prop);
9b981fc6 1407 if (defined $v) {
97f6987a
EW
1408 push @out, " +$t_ppath_prop " .
1409 uri_encode($v);
9b981fc6 1410 } else {
97f6987a 1411 push @out, " -$t_ppath_prop";
9b981fc6
EW
1412 }
1413 }
1414 }
1415 }
1416 foreach my $t (qw/absent_file absent_directory/) {
97f6987a 1417 $h = $ed->{$t} or next;
9b981fc6
EW
1418 foreach my $parent (sort keys %$h) {
1419 foreach my $path (sort @{$h->{$parent}}) {
97f6987a
EW
1420 push @out, " $t: " .
1421 uri_encode("$parent/$path");
9b981fc6
EW
1422 warn "W: $t: $parent/$path ",
1423 "Insufficient permissions?\n";
1424 }
1425 }
1426 }
97f6987a 1427 \@out;
9b981fc6
EW
1428}
1429
1c8443b0
EW
1430sub parse_svn_date {
1431 my $date = shift || return '+0000 1970-01-01 00:00:00';
1432 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1433 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1434 croak "Unable to parse date: $date\n";
1435 "+0000 $Y-$m-$d $H:$M:$S";
1436}
1437
1438sub check_author {
1439 my ($author) = @_;
1440 if (!defined $author || length $author == 0) {
1441 $author = '(no author)';
1442 }
1443 if (defined $::_authors && ! defined $::users{$author}) {
1444 die "Author: $author not defined in $::_authors file\n";
1445 }
1446 $author;
1447}
1448
9b981fc6 1449sub make_log_entry {
97f6987a
EW
1450 my ($self, $rev, $parents, $ed) = @_;
1451 my $untracked = $self->get_untracked($ed);
1452
9b981fc6 1453 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
97f6987a
EW
1454 print $un "r$rev\n" or croak $!;
1455 print $un $_, "\n" foreach @$untracked;
1456 my %log_entry = ( parents => $parents || [], revision => $rev,
1457 log => '');
fbcc1737 1458
8a49ee97 1459 my $headrev;
fbcc1737 1460 my $logged = delete $self->{logged_rev_props};
8a49ee97 1461 if (!$logged || $self->{-want_revprops}) {
fbcc1737
EW
1462 my $rp = $self->ra->rev_proplist($rev);
1463 foreach (sort keys %$rp) {
1464 my $v = $rp->{$_};
1465 if (/^svn:(author|date|log)$/) {
1466 $log_entry{$1} = $v;
8a49ee97
EW
1467 } elsif ($_ eq 'svm:headrev') {
1468 $headrev = $v;
fbcc1737
EW
1469 } else {
1470 print $un " rev_prop: ", uri_encode($_), ' ',
1471 uri_encode($v), "\n";
1472 }
9b981fc6 1473 }
fbcc1737
EW
1474 } else {
1475 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
9b981fc6
EW
1476 }
1477 close $un or croak $!;
97f6987a 1478
9b981fc6
EW
1479 $log_entry{date} = parse_svn_date($log_entry{date});
1480 $log_entry{author} = check_author($log_entry{author});
1481 $log_entry{log} .= "\n";
91b03282 1482 if (defined $headrev && $self->use_svm_props) {
8a49ee97
EW
1483 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1484 if ($uuid ne $self->{svm}->{uuid}) {
1485 die "UUID mismatch on SVM path:\n",
1486 "expected: $self->{svm}->{uuid}\n",
1487 " got: $uuid\n";
1488 }
1489 my $full_url = $self->{svm}->{source};
1490 $full_url .= "/$self->{path}" if length $self->{path};
1491 $log_entry{metadata} = "$full_url\@$r $uuid";
1492 $log_entry{svm_revision} = $r;
1493 } else {
1494 $log_entry{metadata} = $self->full_url . "\@$rev " .
1495 $self->ra->uuid;
1496 }
9b981fc6
EW
1497 \%log_entry;
1498}
1499
1500sub fetch {
3ebe8df7 1501 my ($self, $min_rev, $max_rev, @parents) = @_;
9b981fc6 1502 my ($last_rev, $last_commit) = $self->last_rev_commit;
3ebe8df7 1503 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
e518192f 1504 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
9b981fc6
EW
1505}
1506
1507sub set_tree_cb {
1508 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
490f49ea
EW
1509 $self->{inject_parents} = { $rev => $tree };
1510 $self->fetch(undef, undef);
9b981fc6
EW
1511}
1512
1513sub set_tree {
1514 my ($self, $tree) = (shift, shift);
1ce255dc 1515 my $log_entry = ::get_commit_entry($tree);
9b981fc6
EW
1516 unless ($self->{last_rev}) {
1517 fatal("Must have an existing revision to commit\n");
1518 }
61395354
EW
1519 my %ed_opts = ( r => $self->{last_rev},
1520 log => $log_entry->{log},
1521 ra => $self->ra,
1522 tree_a => $self->{last_commit},
1523 tree_b => $tree,
1524 editor_cb => sub {
1525 $self->set_tree_cb($log_entry, $tree, @_) },
1526 svn_path => $self->{path} );
1527 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
9b981fc6
EW
1528 print "No changes\nr$self->{last_rev} = $tree\n";
1529 }
9b981fc6
EW
1530}
1531
f0ecca10
EW
1532sub rebuild {
1533 my ($self) = @_;
1534 print "Rebuilding $self->{db_path} ...\n";
1535 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1536 my $latest;
1537 my $full_url = $self->full_url;
1538 my $svn_uuid;
1539 while (<$rev_list>) {
1540 chomp;
1541 my $c = $_;
1542 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1543 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1544
1545 # ignore merges (from set-tree)
1546 next if (!defined $rev || !$uuid);
1547
1548 # if we merged or otherwise started elsewhere, this is
1549 # how we break out of it
1550 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1551 ($full_url && $url && ($url ne $full_url))) {
1552 next;
1553 }
1554 $latest ||= $rev;
1555 $svn_uuid ||= $uuid;
1556
1557 $self->rev_db_set($rev, $c);
1558 print "r$rev = $c\n";
1559 }
1560 command_close_pipe($rev_list, $ctx);
1561 print "Done rebuilding $self->{db_path}\n";
1562}
1563
9b981fc6
EW
1564# rev_db:
1565# Tie::File seems to be prone to offset errors if revisions get sparse,
1566# it's not that fast, either. Tie::File is also not in Perl 5.6. So
1567# one of my favorite modules is out :< Next up would be one of the DBM
1568# modules, but I'm not sure which is most portable... So I'll just
1569# go with something that's plain-text, but still capable of
1570# being randomly accessed. So here's my ultra-simple fixed-width
1571# database. All records are 40 characters + "\n", so it's easy to seek
1572# to a revision: (41 * rev) is the byte offset.
1573# A record of 40 0s denotes an empty revision.
1574# And yes, it's still pretty fast (faster than Tie::File).
97ae0911 1575# These files are disposable unless noMetadata or useSvmProps is set
9b981fc6
EW
1576
1577sub rev_db_set {
373274f9 1578 my ($self, $rev, $commit, $update_ref) = @_;
9b981fc6 1579 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
373274f9
EW
1580 my ($db, $db_lock) = ($self->{db_path}, "$self->{db_path}.lock");
1581 my $sig;
1582 if ($update_ref) {
1583 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1584 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1585 }
1586 $LOCKFILES{$db_lock} = 1;
97ae0911
EW
1587 my $sync;
1588
1589 # both of these options make our .rev_db file very, very important
1590 # and we can't afford to lose it because rebuild() won't work
1591 if ($self->use_svm_props || $self->no_metadata) {
1592 $sync = 1;
373274f9
EW
1593 copy($db, $db_lock) or die "rev_db_set(@_): ",
1594 "Failed to copy: ",
1595 "$db => $db_lock ($!)\n";
1596 } else {
1597 rename $db, $db_lock or die "rev_db_set(@_): ",
1598 "Failed to rename: ",
1599 "$db => $db_lock ($!)\n";
1600 }
1601 open my $fh, '+<', $db_lock or croak $!;
9b981fc6
EW
1602 my $offset = $rev * 41;
1603 # assume that append is the common case:
1604 seek $fh, 0, 2 or croak $!;
1605 my $pos = tell $fh;
1606 if ($pos < $offset) {
47a0b75e
EW
1607 for (1 .. (($offset - $pos) / 41)) {
1608 print $fh (('0' x 40),"\n") or croak $!;
1609 }
9b981fc6
EW
1610 }
1611 seek $fh, $offset, 0 or croak $!;
1612 print $fh $commit,"\n" or croak $!;
97ae0911
EW
1613 if ($sync) {
1614 $fh->flush or die "Couldn't flush $db_lock: $!\n";
1615 $fh->sync or die "Couldn't sync $db_lock: $!\n";
1616 }
9b981fc6 1617 close $fh or croak $!;
373274f9
EW
1618 if ($update_ref) {
1619 command_noisy('update-ref', '-m', "r$rev",
1620 $self->refname, $commit);
1621 }
1622 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1623 "$db_lock => $db ($!)\n";
1624 delete $LOCKFILES{$db_lock};
1625 if ($update_ref) {
1626 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1627 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1628 kill $sig, $$ if defined $sig;
1629 }
9b981fc6
EW
1630}
1631
9c93fee5
EW
1632sub rev_db_max {
1633 my ($self) = @_;
1634 my @stat = stat $self->{db_path} or
1635 die "Couldn't stat $self->{db_path}: $!\n";
1636 ($stat[7] % 41) == 0 or
1637 die "$self->{db_path} inconsistent size:$stat[7]\n";
1638 my $max = $stat[7] / 41;
1639 (($max > 0) ? $max - 1 : 0);
1640}
1641
9b981fc6
EW
1642sub rev_db_get {
1643 my ($self, $rev) = @_;
1644 my $ret;
1645 my $offset = $rev * 41;
1646 open my $fh, '<', $self->{db_path} or croak $!;
ce4b4af7
EW
1647 if (sysseek($fh, $offset, 0) == $offset) {
1648 my $read = sysread($fh, $ret, 40);
1649 $ret = undef if ($read != 40 || $ret eq ('0'x40));
9b981fc6
EW
1650 }
1651 close $fh or croak $!;
1652 $ret;
1653}
1654
15710b6f
EW
1655sub find_rev_before {
1656 my ($self, $rev, $eq_ok) = @_;
1657 --$rev unless $eq_ok;
1658 while ($rev > 0) {
1659 if (my $c = $self->rev_db_get($rev)) {
1660 return ($rev, $c);
1661 }
1662 --$rev;
1663 }
1664 return (undef, undef);
1665}
1666
9b981fc6 1667sub _new {
706587fc
EW
1668 my ($class, $repo_id, $ref_id, $path) = @_;
1669 unless (defined $repo_id && length $repo_id) {
1670 $repo_id = $Git::SVN::default_repo_id;
1671 }
1672 unless (defined $ref_id && length $ref_id) {
8b8fc068 1673 $_[2] = $ref_id = $Git::SVN::default_ref_id;
706587fc
EW
1674 }
1675 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1676 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1677 $_[3] = $path = '' unless (defined $path);
8b8fc068
EW
1678 mkpath([$dir]);
1679 unless (-f "$dir/.rev_db") {
1680 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1681 close $fh or croak $!;
1682 }
706587fc 1683 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
8a49ee97 1684 path => $path, config => "$ENV{GIT_DIR}/svn/config",
706587fc 1685 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
9b981fc6
EW
1686}
1687
1c8443b0
EW
1688sub uri_encode {
1689 my ($f) = @_;
1690 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1691 $f
1692}
9b981fc6 1693
d976acfd
EW
1694package Git::SVN::Prompt;
1695use strict;
1696use warnings;
1697require SVN::Core;
1698use vars qw/$_no_auth_cache $_username/;
1699
1700sub simple {
30d055aa
EW
1701 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1702 $may_save = undef if $_no_auth_cache;
1703 $default_username = $_username if defined $_username;
1704 if (defined $default_username && length $default_username) {
1705 if (defined $realm && length $realm) {
6f729591
EW
1706 print STDERR "Authentication realm: $realm\n";
1707 STDERR->flush;
30d055aa
EW
1708 }
1709 $cred->username($default_username);
1710 } else {
d976acfd 1711 username($cred, $realm, $may_save, $pool);
30d055aa
EW
1712 }
1713 $cred->password(_read_password("Password for '" .
1714 $cred->username . "': ", $realm));
1715 $cred->may_save($may_save);
1716 $SVN::_Core::SVN_NO_ERROR;
1717}
1718
d976acfd 1719sub ssl_server_trust {
30d055aa
EW
1720 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1721 $may_save = undef if $_no_auth_cache;
6f729591 1722 print STDERR "Error validating server certificate for '$realm':\n";
30d055aa 1723 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
6f729591 1724 print STDERR " - The certificate is not issued by a trusted ",
30d055aa
EW
1725 "authority. Use the\n",
1726 " fingerprint to validate the certificate manually!\n";
1727 }
1728 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
6f729591 1729 print STDERR " - The certificate hostname does not match.\n";
30d055aa
EW
1730 }
1731 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
6f729591 1732 print STDERR " - The certificate is not yet valid.\n";
30d055aa
EW
1733 }
1734 if ($failures & $SVN::Auth::SSL::EXPIRED) {
6f729591 1735 print STDERR " - The certificate has expired.\n";
30d055aa
EW
1736 }
1737 if ($failures & $SVN::Auth::SSL::OTHER) {
6f729591 1738 print STDERR " - The certificate has an unknown error.\n";
30d055aa 1739 }
6f729591
EW
1740 printf STDERR
1741 "Certificate information:\n".
30d055aa
EW
1742 " - Hostname: %s\n".
1743 " - Valid: from %s until %s\n".
1744 " - Issuer: %s\n".
1745 " - Fingerprint: %s\n",
1746 map $cert_info->$_, qw(hostname valid_from valid_until
6f729591 1747 issuer_dname fingerprint);
30d055aa
EW
1748 my $choice;
1749prompt:
6f729591 1750 print STDERR $may_save ?
30d055aa
EW
1751 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1752 "(R)eject or accept (t)emporarily? ";
6f729591 1753 STDERR->flush;
30d055aa
EW
1754 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1755 if ($choice =~ /^t$/i) {
1756 $cred->may_save(undef);
1757 } elsif ($choice =~ /^r$/i) {
1758 return -1;
1759 } elsif ($may_save && $choice =~ /^p$/i) {
1760 $cred->may_save($may_save);
1761 } else {
1762 goto prompt;
1763 }
1764 $cred->accepted_failures($failures);
1765 $SVN::_Core::SVN_NO_ERROR;
1766}
1767
d976acfd 1768sub ssl_client_cert {
30d055aa
EW
1769 my ($cred, $realm, $may_save, $pool) = @_;
1770 $may_save = undef if $_no_auth_cache;
6f729591
EW
1771 print STDERR "Client certificate filename: ";
1772 STDERR->flush;
30d055aa
EW
1773 chomp(my $filename = <STDIN>);
1774 $cred->cert_file($filename);
1775 $cred->may_save($may_save);
1776 $SVN::_Core::SVN_NO_ERROR;
1777}
1778
d976acfd 1779sub ssl_client_cert_pw {
30d055aa
EW
1780 my ($cred, $realm, $may_save, $pool) = @_;
1781 $may_save = undef if $_no_auth_cache;
1782 $cred->password(_read_password("Password: ", $realm));
1783 $cred->may_save($may_save);
1784 $SVN::_Core::SVN_NO_ERROR;
1785}
1786
d976acfd 1787sub username {
30d055aa
EW
1788 my ($cred, $realm, $may_save, $pool) = @_;
1789 $may_save = undef if $_no_auth_cache;
1790 if (defined $realm && length $realm) {
6f729591 1791 print STDERR "Authentication realm: $realm\n";
30d055aa
EW
1792 }
1793 my $username;
1794 if (defined $_username) {
1795 $username = $_username;
1796 } else {
6f729591
EW
1797 print STDERR "Username: ";
1798 STDERR->flush;
30d055aa
EW
1799 chomp($username = <STDIN>);
1800 }
1801 $cred->username($username);
1802 $cred->may_save($may_save);
1803 $SVN::_Core::SVN_NO_ERROR;
1804}
1805
1806sub _read_password {
1807 my ($prompt, $realm) = @_;
6f729591
EW
1808 print STDERR $prompt;
1809 STDERR->flush;
30d055aa
EW
1810 require Term::ReadKey;
1811 Term::ReadKey::ReadMode('noecho');
1812 my $password = '';
1813 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1814 last if $key =~ /[\012\015]/; # \n\r
1815 $password .= $key;
1816 }
1817 Term::ReadKey::ReadMode('restore');
6f729591
EW
1818 print STDERR "\n";
1819 STDERR->flush;
30d055aa
EW
1820 $password;
1821}
1822
d976acfd
EW
1823package main;
1824
b9c85187
EW
1825{
1826 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1827 $SVN::Node::dir.$SVN::Node::unknown.
1828 $SVN::Node::none.$SVN::Node::file.
1829 $SVN::Node::dir.$SVN::Node::unknown.
1830 $SVN::Auth::SSL::CNMISMATCH.
1831 $SVN::Auth::SSL::NOTYETVALID.
1832 $SVN::Auth::SSL::EXPIRED.
1833 $SVN::Auth::SSL::UNKNOWNCA.
1834 $SVN::Auth::SSL::OTHER;
1835}
1836
27a1a801
EW
1837package SVN::Git::Fetcher;
1838use vars qw/@ISA/;
1839use strict;
1840use warnings;
1841use Carp qw/croak/;
1842use IO::File qw//;
90c1b15d 1843use Digest::MD5;
27a1a801
EW
1844
1845# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1846sub new {
1847 my ($class, $git_svn) = @_;
1848 my $self = SVN::Delta::Editor->new;
1849 bless $self, $class;
1c8443b0 1850 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
d2a9a87b
EW
1851 $self->{empty} = {};
1852 $self->{dir_prop} = {};
1853 $self->{file_prop} = {};
1854 $self->{absent_dir} = {};
1855 $self->{absent_file} = {};
ef3cfaad 1856 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
27a1a801
EW
1857 $self;
1858}
1859
8b8fc068
EW
1860sub set_path_strip {
1861 my ($self, $path) = @_;
4e9f6cc7 1862 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
8b8fc068
EW
1863}
1864
d2a9a87b
EW
1865sub open_root {
1866 { path => '' };
1867}
1868
1869sub open_directory {
1870 my ($self, $path, $pb, $rev) = @_;
1871 { path => $path };
1872}
1873
706587fc
EW
1874sub git_path {
1875 my ($self, $path) = @_;
2b27f6c8
EW
1876 if ($self->{path_strip}) {
1877 $path =~ s!$self->{path_strip}!! or
1878 die "Failed to strip path '$path' ($self->{path_strip})\n";
1879 }
706587fc
EW
1880 $path;
1881}
1882
27a1a801
EW
1883sub delete_entry {
1884 my ($self, $path, $rev, $pb) = @_;
4a87db0e 1885
706587fc 1886 my $gpath = $self->git_path($path);
8a603774
EW
1887 return undef if ($gpath eq '');
1888
4a87db0e 1889 # remove entire directories.
706587fc 1890 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
4a87db0e
EW
1891 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1892 -r --name-only -z/,
706587fc 1893 $self->{c}, '--', $gpath);
4a87db0e
EW
1894 local $/ = "\0";
1895 while (<$ls>) {
ef3cfaad
EW
1896 chomp;
1897 $self->{gii}->remove($_);
9e3cdbd4 1898 print "\tD\t$_\n" unless $::_q;
4a87db0e 1899 }
9e3cdbd4 1900 print "\tD\t$gpath/\n" unless $::_q;
4a87db0e
EW
1901 command_close_pipe($ls, $ctx);
1902 $self->{empty}->{$path} = 0
1903 } else {
ef3cfaad 1904 $self->{gii}->remove($gpath);
9e3cdbd4 1905 print "\tD\t$gpath\n" unless $::_q;
4a87db0e 1906 }
27a1a801
EW
1907 undef;
1908}
1909
1910sub open_file {
1911 my ($self, $path, $pb, $rev) = @_;
706587fc
EW
1912 my $gpath = $self->git_path($path);
1913 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
27a1a801 1914 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
006ede5e
EW
1915 unless (defined $mode && defined $blob) {
1916 die "$path was not found in commit $self->{c} (r$rev)\n";
1917 }
27a1a801 1918 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
0864e3ba 1919 pool => SVN::Pool->new, action => 'M' };
27a1a801
EW
1920}
1921
1922sub add_file {
1923 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
d2a9a87b
EW
1924 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1925 delete $self->{empty}->{$dir};
27a1a801 1926 { path => $path, mode_a => 100644, mode_b => 100644,
0864e3ba 1927 pool => SVN::Pool->new, action => 'A' };
27a1a801
EW
1928}
1929
d2a9a87b
EW
1930sub add_directory {
1931 my ($self, $path, $cp_path, $cp_rev) = @_;
1932 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1933 delete $self->{empty}->{$dir};
1934 $self->{empty}->{$path} = 1;
1935 { path => $path };
1936}
1937
1938sub change_dir_prop {
1939 my ($self, $db, $prop, $value) = @_;
1940 $self->{dir_prop}->{$db->{path}} ||= {};
1941 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1942 undef;
1943}
1944
1945sub absent_directory {
1946 my ($self, $path, $pb) = @_;
1947 $self->{absent_dir}->{$pb->{path}} ||= [];
1948 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1949 undef;
1950}
1951
1952sub absent_file {
1953 my ($self, $path, $pb) = @_;
1954 $self->{absent_file}->{$pb->{path}} ||= [];
1955 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1956 undef;
1957}
1958
27a1a801
EW
1959sub change_file_prop {
1960 my ($self, $fb, $prop, $value) = @_;
1961 if ($prop eq 'svn:executable') {
1962 if ($fb->{mode_b} != 120000) {
1963 $fb->{mode_b} = defined $value ? 100755 : 100644;
1964 }
1965 } elsif ($prop eq 'svn:special') {
1966 $fb->{mode_b} = defined $value ? 120000 : 100644;
d2a9a87b
EW
1967 } else {
1968 $self->{file_prop}->{$fb->{path}} ||= {};
1969 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
27a1a801
EW
1970 }
1971 undef;
1972}
1973
1974sub apply_textdelta {
1975 my ($self, $fb, $exp) = @_;
1976 my $fh = IO::File->new_tmpfile;
1977 $fh->autoflush(1);
1978 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1979 # (but $base does not,) so dup() it for reading in close_file
1980 open my $dup, '<&', $fh or croak $!;
1981 my $base = IO::File->new_tmpfile;
1982 $base->autoflush(1);
1983 if ($fb->{blob}) {
1984 defined (my $pid = fork) or croak $!;
1985 if (!$pid) {
1986 open STDOUT, '>&', $base or croak $!;
1987 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1988 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1989 }
1990 waitpid $pid, 0;
1991 croak $? if $?;
1992
1993 if (defined $exp) {
1994 seek $base, 0, 0 or croak $!;
1995 my $md5 = Digest::MD5->new;
1996 $md5->addfile($base);
1997 my $got = $md5->hexdigest;
1998 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1999 "expected: $exp\n",
2000 " got: $got\n" if ($got ne $exp);
2001 }
2002 }
2003 seek $base, 0, 0 or croak $!;
2004 $fb->{fh} = $dup;
2005 $fb->{base} = $base;
2006 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2007}
2008
2009sub close_file {
2010 my ($self, $fb, $exp) = @_;
2011 my $hash;
706587fc 2012 my $path = $self->git_path($fb->{path});
27a1a801
EW
2013 if (my $fh = $fb->{fh}) {
2014 seek($fh, 0, 0) or croak $!;
2015 my $md5 = Digest::MD5->new;
2016 $md5->addfile($fh);
2017 my $got = $md5->hexdigest;
2018 die "Checksum mismatch: $path\n",
2019 "expected: $exp\n got: $got\n" if ($got ne $exp);
2020 seek($fh, 0, 0) or croak $!;
2021 if ($fb->{mode_b} == 120000) {
2022 read($fh, my $buf, 5) == 5 or croak $!;
2023 $buf eq 'link ' or die "$path has mode 120000",
2024 "but is not a link\n";
2025 }
2026 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2027 if (!$pid) {
2028 open STDIN, '<&', $fh or croak $!;
2029 exec qw/git-hash-object -w --stdin/ or croak $!;
2030 }
2031 chomp($hash = do { local $/; <$out> });
2032 close $out or croak $!;
2033 close $fh or croak $!;
2034 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2035 close $fb->{base} or croak $!;
2036 } else {
2037 $hash = $fb->{blob} or die "no blob information\n";
2038 }
2039 $fb->{pool}->clear;
ef3cfaad 2040 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
9e3cdbd4 2041 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
27a1a801
EW
2042 undef;
2043}
2044
2045sub abort_edit {
2046 my $self = shift;
ef3cfaad
EW
2047 $self->{nr} = $self->{gii}->{nr};
2048 delete $self->{gii};
27a1a801
EW
2049 $self->SUPER::abort_edit(@_);
2050}
2051
2052sub close_edit {
2053 my $self = shift;
dad73c0b 2054 $self->{git_commit_ok} = 1;
ef3cfaad
EW
2055 $self->{nr} = $self->{gii}->{nr};
2056 delete $self->{gii};
27a1a801
EW
2057 $self->SUPER::close_edit(@_);
2058}
1a82e793 2059
a5e0cedc 2060package SVN::Git::Editor;
24e22aa8 2061use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
a5e0cedc
EW
2062use strict;
2063use warnings;
2064use Carp qw/croak/;
2065use IO::File;
90c1b15d 2066use Digest::MD5;
a5e0cedc
EW
2067
2068sub new {
61395354
EW
2069 my ($class, $opts) = @_;
2070 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2071 die "$_ required!\n" unless (defined $opts->{$_});
2072 }
2073
2074 my $pool = SVN::Pool->new;
2075 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2076 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2077 $opts->{r}, $mods);
2078
2079 # $opts->{ra} functions should not be used after this:
2080 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2081 $opts->{editor_cb}, $pool);
2082 my $self = SVN::Delta::Editor->new(@ce, $pool);
a5e0cedc 2083 bless $self, $class;
61395354
EW
2084 foreach (qw/svn_path r tree_a tree_b/) {
2085 $self->{$_} = $opts->{$_};
a5e0cedc 2086 }
61395354
EW
2087 $self->{url} = $opts->{ra}->{url};
2088 $self->{mods} = $mods;
2089 $self->{types} = $types;
2090 $self->{pool} = $pool;
a5e0cedc
EW
2091 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2092 $self->{rm} = { };
d3a840dc
EW
2093 $self->{path_prefix} = length $self->{svn_path} ?
2094 "$self->{svn_path}/" : '';
a5e0cedc
EW
2095 return $self;
2096}
2097
61395354
EW
2098sub generate_diff {
2099 my ($tree_a, $tree_b) = @_;
2100 my @diff_tree = qw(diff-tree -z -r);
24e22aa8
EW
2101 if ($_cp_similarity) {
2102 push @diff_tree, "-C$_cp_similarity";
61395354
EW
2103 } else {
2104 push @diff_tree, '-C';
2105 }
24e22aa8
EW
2106 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2107 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
61395354
EW
2108 push @diff_tree, $tree_a, $tree_b;
2109 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2110 local $/ = "\0";
2111 my $state = 'meta';
2112 my @mods;
2113 while (<$diff_fh>) {
2114 chomp $_; # this gets rid of the trailing "\0"
2115 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2116 $::sha1\s($::sha1)\s
2117 ([MTCRAD])\d*$/xo) {
2118 push @mods, { mode_a => $1, mode_b => $2,
2119 sha1_b => $3, chg => $4 };
2120 if ($4 =~ /^(?:C|R)$/) {
2121 $state = 'file_a';
2122 } else {
2123 $state = 'file_b';
2124 }
2125 } elsif ($state eq 'file_a') {
2126 my $x = $mods[$#mods] or croak "Empty array\n";
2127 if ($x->{chg} !~ /^(?:C|R)$/) {
2128 croak "Error parsing $_, $x->{chg}\n";
2129 }
2130 $x->{file_a} = $_;
2131 $state = 'file_b';
2132 } elsif ($state eq 'file_b') {
2133 my $x = $mods[$#mods] or croak "Empty array\n";
2134 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2135 croak "Error parsing $_, $x->{chg}\n";
2136 }
2137 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2138 croak "Error parsing $_, $x->{chg}\n";
2139 }
2140 $x->{file_b} = $_;
2141 $state = 'meta';
2142 } else {
2143 croak "Error parsing $_\n";
2144 }
2145 }
2146 command_close_pipe($diff_fh, $ctx);
2147 \@mods;
2148}
2149
2150sub check_diff_paths {
2151 my ($ra, $pfx, $rev, $mods) = @_;
2152 my %types;
2153 $pfx .= '/' if length $pfx;
2154
2155 sub type_diff_paths {
2156 my ($ra, $types, $path, $rev) = @_;
2157 my @p = split m#/+#, $path;
2158 my $c = shift @p;
2159 unless (defined $types->{$c}) {
2160 $types->{$c} = $ra->check_path($c, $rev);
2161 }
2162 while (@p) {
2163 $c .= '/' . shift @p;
2164 next if defined $types->{$c};
2165 $types->{$c} = $ra->check_path($c, $rev);
2166 }
2167 }
2168
2169 foreach my $m (@$mods) {
2170 foreach my $f (qw/file_a file_b/) {
2171 next unless defined $m->{$f};
2172 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2173 if (length $pfx.$dir && ! defined $types{$dir}) {
2174 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2175 }
2176 }
2177 }
2178 \%types;
2179}
2180
a5e0cedc
EW
2181sub split_path {
2182 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2183}
2184
2185sub repo_path {
d3a840dc
EW
2186 my ($self, $path) = @_;
2187 $self->{path_prefix}.(defined $path ? $path : '');
a5e0cedc
EW
2188}
2189
2190sub url_path {
2191 my ($self, $path) = @_;
6e8548cc 2192 $self->{url} . '/' . $self->repo_path($path);
a5e0cedc
EW
2193}
2194
2195sub rmdirs {
61395354 2196 my ($self) = @_;
a5e0cedc
EW
2197 my $rm = $self->{rm};
2198 delete $rm->{''}; # we never delete the url we're tracking
2199 return unless %$rm;
2200
2201 foreach (keys %$rm) {
2202 my @d = split m#/#, $_;
2203 my $c = shift @d;
2204 $rm->{$c} = 1;
2205 while (@d) {
2206 $c .= '/' . shift @d;
2207 $rm->{$c} = 1;
2208 }
2209 }
2210 delete $rm->{$self->{svn_path}};
2211 delete $rm->{''}; # we never delete the url we're tracking
2212 return unless %$rm;
2213
61395354
EW
2214 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2215 $self->{tree_b});
a5e0cedc
EW
2216 local $/ = "\0";
2217 while (<$fh>) {
2218 chomp;
747fa12c 2219 my @dn = split m#/#, $_;
c07eee1f
EW
2220 while (pop @dn) {
2221 delete $rm->{join '/', @dn};
2222 }
2223 unless (%$rm) {
22600a25 2224 close $fh;
c07eee1f
EW
2225 return;
2226 }
a5e0cedc 2227 }
aef4e921 2228 command_close_pipe($fh, $ctx);
c07eee1f 2229
a5e0cedc
EW
2230 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2231 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2232 $self->close_directory($bat->{$d}, $p);
2233 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
44320b9e 2234 print "\tD+\t$d/\n" unless $::_q;
a5e0cedc
EW
2235 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2236 delete $bat->{$d};
2237 }
2238}
2239
2240sub open_or_add_dir {
2241 my ($self, $full_path, $baton) = @_;
6e8548cc
EW
2242 my $t = $self->{types}->{$full_path};
2243 if (!defined $t) {
2244 die "$full_path not known in r$self->{r} or we have a bug!\n";
2245 }
a5e0cedc
EW
2246 if ($t == $SVN::Node::none) {
2247 return $self->add_directory($full_path, $baton,
2248 undef, -1, $self->{pool});
2249 } elsif ($t == $SVN::Node::dir) {
2250 return $self->open_directory($full_path, $baton,
2251 $self->{r}, $self->{pool});
2252 }
2253 print STDERR "$full_path already exists in repository at ",
2254 "r$self->{r} and it is not a directory (",
2255 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2256 exit 1;
2257}
2258
2259sub ensure_path {
2260 my ($self, $path) = @_;
2261 my $bat = $self->{bat};
6e8548cc
EW
2262 my $repo_path = $self->repo_path($path);
2263 return $bat->{''} unless (length $repo_path);
2264 my @p = split m#/+#, $repo_path;
a5e0cedc
EW
2265 my $c = shift @p;
2266 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2267 while (@p) {
2268 my $c0 = $c;
2269 $c .= '/' . shift @p;
2270 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2271 }
2272 return $bat->{$c};
2273}
2274
2275sub A {
44320b9e 2276 my ($self, $m) = @_;
a5e0cedc
EW
2277 my ($dir, $file) = split_path($m->{file_b});
2278 my $pbat = $self->ensure_path($dir);
2279 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2280 undef, -1);
44320b9e 2281 print "\tA\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
2282 $self->chg_file($fbat, $m);
2283 $self->close_file($fbat,undef,$self->{pool});
2284}
2285
2286sub C {
44320b9e 2287 my ($self, $m) = @_;
a5e0cedc
EW
2288 my ($dir, $file) = split_path($m->{file_b});
2289 my $pbat = $self->ensure_path($dir);
2290 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2291 $self->url_path($m->{file_a}), $self->{r});
44320b9e 2292 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
2293 $self->chg_file($fbat, $m);
2294 $self->close_file($fbat,undef,$self->{pool});
2295}
2296
2297sub delete_entry {
2298 my ($self, $path, $pbat) = @_;
2299 my $rpath = $self->repo_path($path);
2300 my ($dir, $file) = split_path($rpath);
2301 $self->{rm}->{$dir} = 1;
2302 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2303}
2304
2305sub R {
44320b9e 2306 my ($self, $m) = @_;
a5e0cedc
EW
2307 my ($dir, $file) = split_path($m->{file_b});
2308 my $pbat = $self->ensure_path($dir);
2309 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2310 $self->url_path($m->{file_a}), $self->{r});
44320b9e 2311 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
2312 $self->chg_file($fbat, $m);
2313 $self->close_file($fbat,undef,$self->{pool});
2314
2315 ($dir, $file) = split_path($m->{file_a});
2316 $pbat = $self->ensure_path($dir);
2317 $self->delete_entry($m->{file_a}, $pbat);
2318}
2319
2320sub M {
44320b9e 2321 my ($self, $m) = @_;
a5e0cedc
EW
2322 my ($dir, $file) = split_path($m->{file_b});
2323 my $pbat = $self->ensure_path($dir);
2324 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2325 $pbat,$self->{r},$self->{pool});
44320b9e 2326 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
2327 $self->chg_file($fbat, $m);
2328 $self->close_file($fbat,undef,$self->{pool});
2329}
2330
2331sub T { shift->M(@_) }
2332
2333sub change_file_prop {
2334 my ($self, $fbat, $pname, $pval) = @_;
2335 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2336}
2337
2338sub chg_file {
2339 my ($self, $fbat, $m) = @_;
2340 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2341 $self->change_file_prop($fbat,'svn:executable','*');
2342 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2343 $self->change_file_prop($fbat,'svn:executable',undef);
2344 }
2345 my $fh = IO::File->new_tmpfile or croak $!;
2346 if ($m->{mode_b} =~ /^120/) {
2347 print $fh 'link ' or croak $!;
2348 $self->change_file_prop($fbat,'svn:special','*');
2349 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2350 $self->change_file_prop($fbat,'svn:special',undef);
2351 }
2352 defined(my $pid = fork) or croak $!;
2353 if (!$pid) {
2354 open STDOUT, '>&', $fh or croak $!;
2355 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2356 }
2357 waitpid $pid, 0;
2358 croak $? if $?;
2359 $fh->flush == 0 or croak $!;
2360 seek $fh, 0, 0 or croak $!;
2361
2362 my $md5 = Digest::MD5->new;
2363 $md5->addfile($fh) or croak $!;
2364 seek $fh, 0, 0 or croak $!;
2365
2366 my $exp = $md5->hexdigest;
f7197dff
EW
2367 my $pool = SVN::Pool->new;
2368 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2369 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
a5e0cedc 2370 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
f7197dff 2371 $pool->clear;
a5e0cedc
EW
2372
2373 close $fh or croak $!;
2374}
2375
2376sub D {
44320b9e 2377 my ($self, $m) = @_;
a5e0cedc
EW
2378 my ($dir, $file) = split_path($m->{file_b});
2379 my $pbat = $self->ensure_path($dir);
44320b9e 2380 print "\tD\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
2381 $self->delete_entry($m->{file_b}, $pbat);
2382}
2383
2384sub close_edit {
2385 my ($self) = @_;
2386 my ($p,$bat) = ($self->{pool}, $self->{bat});
2387 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2388 $self->close_directory($bat->{$_}, $p);
2389 }
2390 $self->SUPER::close_edit($p);
2391 $p->clear;
2392}
2393
2394sub abort_edit {
2395 my ($self) = @_;
2396 $self->SUPER::abort_edit($self->{pool});
61395354
EW
2397}
2398
2399sub DESTROY {
2400 my $self = shift;
2401 $self->SUPER::DESTROY(@_);
a5e0cedc
EW
2402 $self->{pool}->clear;
2403}
2404
44320b9e
EW
2405# this drives the editor
2406sub apply_diff {
61395354
EW
2407 my ($self) = @_;
2408 my $mods = $self->{mods};
44320b9e 2409 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
6e8548cc 2410 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
44320b9e
EW
2411 my $f = $m->{chg};
2412 if (defined $o{$f}) {
2413 $self->$f($m);
2414 } else {
2415 fatal("Invalid change type: $f\n");
2416 }
2417 }
24e22aa8 2418 $self->rmdirs if $_rmdir;
6e8548cc 2419 if (@$mods == 0) {
44320b9e
EW
2420 $self->abort_edit;
2421 } else {
2422 $self->close_edit;
2423 }
6e8548cc 2424 return scalar @$mods;
44320b9e
EW
2425}
2426
d81bf827
EW
2427package Git::SVN::Ra;
2428use vars qw/@ISA $config_dir/;
2429use strict;
2430use warnings;
2431my ($can_do_switch);
5d3b7cd5 2432my $RA;
d81bf827
EW
2433
2434BEGIN {
2435 # enforce temporary pool usage for some simple functions
2436 my $e;
74a81227 2437 foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
d81bf827
EW
2438 $e .= "sub $_ {
2439 my \$self = shift;
2440 my \$pool = SVN::Pool->new;
2441 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2442 \$pool->clear;
2443 wantarray ? \@ret : \$ret[0]; }\n";
2444 }
74a81227
EW
2445
2446 # get_dir needs $pool held in cache for dirents to work,
2447 # check_path is cacheable and rev_proplist is close enough
2448 # for our purposes.
2449 foreach (qw/check_path get_dir rev_proplist/) {
2450 $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2451 my \$self = shift;
2452 my \$r = pop;
2453 my \$k = join(\"\\0\", \@_);
2454 if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2455 return wantarray ? \@\$x : \$x->[0];
2456 }
2457 my \$pool = SVN::Pool->new;
2458 my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2459 if (\$r != \$${_}_rev) {
2460 \%${_}_cache = ( pool => [] );
2461 \$${_}_rev = \$r;
2462 }
2463 \$${_}_cache{\$r}->{\$k} = \\\@ret;
2464 push \@{\$${_}_cache{pool}}, \$pool;
2465 wantarray ? \@ret : \$ret[0]; }\n";
2466 }
2467 $e .= "\n1;";
2468 eval $e or die $@;
d81bf827
EW
2469}
2470
2471sub new {
2472 my ($class, $url) = @_;
f6f09876 2473 $url =~ s!/+$!!;
5d3b7cd5 2474 return $RA if ($RA && $RA->{url} eq $url);
f6f09876 2475
d81bf827
EW
2476 SVN::_Core::svn_config_ensure($config_dir, undef);
2477 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2478 SVN::Client::get_simple_provider(),
2479 SVN::Client::get_ssl_server_trust_file_provider(),
2480 SVN::Client::get_simple_prompt_provider(
2481 \&Git::SVN::Prompt::simple, 2),
2482 SVN::Client::get_ssl_client_cert_prompt_provider(
2483 \&Git::SVN::Prompt::ssl_client_cert, 2),
2484 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2485 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2486 SVN::Client::get_username_provider(),
2487 SVN::Client::get_ssl_server_trust_prompt_provider(
2488 \&Git::SVN::Prompt::ssl_server_trust),
2489 SVN::Client::get_username_prompt_provider(
2490 \&Git::SVN::Prompt::username, 2),
2491 ]);
2492 my $config = SVN::Core::config_get_config($config_dir);
2493 my $self = SVN::Ra->new(url => $url, auth => $baton,
2494 config => $config,
2495 pool => SVN::Pool->new,
2496 auth_provider_callbacks => $callbacks);
2497 $self->{svn_path} = $url;
2498 $self->{repos_root} = $self->get_repos_root;
4e9f6cc7 2499 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
5d3b7cd5 2500 $RA = bless $self, $class;
d81bf827
EW
2501}
2502
2503sub DESTROY {
5d3b7cd5 2504 # do not call the real DESTROY since we store ourselves in $RA
d81bf827
EW
2505}
2506
d81bf827
EW
2507sub get_log {
2508 my ($self, @args) = @_;
2509 my $pool = SVN::Pool->new;
d81bf827
EW
2510 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2511 my $ret = $self->SUPER::get_log(@args, $pool);
2512 $pool->clear;
2513 $ret;
2514}
2515
2516sub get_commit_editor {
44320b9e 2517 my ($self, $log, $cb, $pool) = @_;
d81bf827 2518 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
44320b9e 2519 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
d81bf827
EW
2520}
2521
2522sub uuid {
2523 my ($self) = @_;
2524 $self->{uuid} ||= $self->get_uuid;
2525}
2526
2527sub gs_do_update {
8a603774
EW
2528 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2529 my $new = ($rev_a == $rev_b);
2530 my $path = $gs->{path};
2531
d81bf827 2532 my $pool = SVN::Pool->new;
8b8fc068 2533 $editor->set_path_strip($path);
2b27f6c8
EW
2534 my (@pc) = split m#/#, $path;
2535 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
8a603774 2536 1, $editor, $pool);
d81bf827 2537 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2b27f6c8
EW
2538
2539 # Since we can't rely on svn_ra_reparent being available, we'll
2540 # just have to do some magic with set_path to make it so
2541 # we only want a partial path.
2542 my $sp = '';
2543 my $final = join('/', @pc);
2544 while (@pc) {
2545 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2546 $sp .= '/' if length $sp;
2547 $sp .= shift @pc;
2548 }
2549 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2550
2b27f6c8
EW
2551 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2552
d81bf827
EW
2553 $reporter->finish_report($pool);
2554 $pool->clear;
2555 $editor->{git_commit_ok};
2556}
2557
2b27f6c8
EW
2558# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2559# svn_ra_reparent didn't work before 1.4)
d81bf827 2560sub gs_do_switch {
8a603774
EW
2561 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2562 my $path = $gs->{path};
d81bf827 2563 my $pool = SVN::Pool->new;
2b27f6c8
EW
2564
2565 my $full_url = $self->{url};
2566 my $old_url = $full_url;
2567 $full_url .= "/$path" if length $path;
5d3b7cd5
EW
2568 my ($ra, $reparented);
2569 if ($old_url ne $full_url) {
2570 if ($old_url !~ m#^svn(\+ssh)?://#) {
2571 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2572 $pool);
2573 $self->{url} = $full_url;
2574 $reparented = 1;
2575 } else {
2576 $ra = Git::SVN::Ra->new($full_url);
2577 }
2578 }
2579 $ra ||= $self;
8a603774 2580 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
d81bf827 2581 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
8b8fc068 2582 $reporter->set_path('', $rev_a, 0, @lock, $pool);
d81bf827 2583 $reporter->finish_report($pool);
2b27f6c8 2584
5d3b7cd5
EW
2585 if ($reparented) {
2586 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2587 $self->{url} = $old_url;
2588 }
2b27f6c8 2589
d81bf827
EW
2590 $pool->clear;
2591 $editor->{git_commit_ok};
2592}
2593
0af9c9f9 2594sub gs_fetch_loop_common {
e518192f
EW
2595 my ($self, $base, $head, $gsv, $globs) = @_;
2596 return if ($base > $head);
0af9c9f9
EW
2597 my $inc = 1000;
2598 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
d2ae1434 2599 my %common;
e518192f
EW
2600 my $common_max = scalar @$gsv;
2601
2602 foreach my $gs (@$gsv) {
0af9c9f9
EW
2603 if (my $last_commit = $gs->last_commit) {
2604 $gs->assert_index_clean($last_commit);
2605 }
d2ae1434
EW
2606 my @tmp = split m#/#, $gs->{path};
2607 my $p = '';
2608 foreach (@tmp) {
2609 $p .= length($p) ? "/$_" : $_;
2610 $common{$p} ||= 0;
2611 $common{$p}++;
2612 }
2613 }
e518192f
EW
2614 $globs ||= [];
2615 $common_max += scalar @$globs;
2616 foreach my $glob (@$globs) {
2617 my @tmp = split m#/#, $glob->{path}->{left};
2618 my $p = '';
2619 foreach (@tmp) {
2620 $p .= length($p) ? "/$_" : $_;
2621 $common{$p} ||= 0;
2622 $common{$p}++;
2623 }
2624 }
2625
d2ae1434
EW
2626 my $longest_path = '';
2627 foreach (sort {length $b <=> length $a} keys %common) {
e518192f 2628 if ($common{$_} == $common_max) {
d2ae1434
EW
2629 $longest_path = $_;
2630 last;
2631 }
0af9c9f9
EW
2632 }
2633 while (1) {
d4eff2bd 2634 my %revs;
d2ae1434 2635 my $err;
f7c3fc4a 2636 my $err_handler = $SVN::Error::handler;
d2ae1434
EW
2637 $SVN::Error::handler = sub {
2638 ($err) = @_;
2639 skip_unknown_revs($err);
2640 };
2641 sub _cb {
2642 my ($paths, $r, $author, $date, $log) = @_;
2643 [ dup_changed_paths($paths),
2644 { author => $author, date => $date, log => $log } ];
2645 }
2646 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2647 sub { $revs{$_[1]} = _cb(@_) });
2648 if ($err && $max >= $head) {
2649 print STDERR "Path '$longest_path' ",
2650 "was probably deleted:\n",
2651 $err->expanded_message,
2652 "\nWill attempt to follow ",
2653 "revisions r$min .. r$max ",
2654 "committed before the deletion\n";
2655 my $hi = $max;
2656 while (--$hi >= $min) {
2657 my $ok;
2658 $self->get_log([$longest_path], $min, $hi,
2659 0, 1, 1, sub {
2660 $ok ||= $_[1];
2661 $revs{$_[1]} = _cb(@_) });
2662 if ($ok) {
2663 print STDERR "r$min .. r$ok OK\n";
2664 last;
2665 }
2666 }
2667 }
d4eff2bd 2668 $SVN::Error::handler = $err_handler;
fbcc1737 2669
e518192f 2670 my %exists = map { $_->{path} => $_ } @$gsv;
d4eff2bd 2671 foreach my $r (sort {$a <=> $b} keys %revs) {
fbcc1737 2672 my ($paths, $logged) = @{$revs{$r}};
e518192f
EW
2673
2674 foreach my $gs ($self->match_globs(\%exists, $paths,
2675 $globs, $r)) {
fbcc1737
EW
2676 if ($gs->rev_db_max >= $r) {
2677 next;
2678 }
2679 next unless $gs->match_paths($paths, $r);
2680 $gs->{logged_rev_props} = $logged;
2681 my $log_entry = $gs->do_fetch($paths, $r);
2682 if ($log_entry) {
0af9c9f9
EW
2683 $gs->do_git_commit($log_entry);
2684 }
2685 }
e518192f 2686 foreach my $g (@$globs) {
93f2689c
EW
2687 my $k = "svn-remote.$g->{remote}." .
2688 "$g->{t}-maxRev";
2689 Git::SVN::tmp_config($k, $r);
e518192f 2690 }
0af9c9f9 2691 }
9c93fee5
EW
2692 # pre-fill the .rev_db since it'll eventually get filled in
2693 # with '0' x40 if something new gets committed
e518192f 2694 foreach my $gs (@$gsv) {
9c93fee5
EW
2695 next if defined $gs->rev_db_get($max);
2696 $gs->rev_db_set($max, 0 x40);
2697 }
0af9c9f9
EW
2698 last if $max >= $head;
2699 $min = $max + 1;
2700 $max += $inc;
2701 $max = $head if ($max > $head);
2702 }
0af9c9f9
EW
2703}
2704
e518192f
EW
2705sub match_globs {
2706 my ($self, $exists, $paths, $globs, $r) = @_;
74a81227
EW
2707
2708 sub get_dir_check {
2709 my ($self, $exists, $g, $r) = @_;
2710 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2711 return unless scalar @x == 3;
2712 my $dirents = $x[0];
2713 foreach my $de (keys %$dirents) {
2714 next if $dirents->{$de}->kind != $SVN::Node::dir;
2715 my $p = $g->{path}->full_path($de);
2716 next if $exists->{$p};
2717 next if (length $g->{path}->{right} &&
2718 ($self->check_path($p, $r) !=
2719 $SVN::Node::dir));
2720 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2721 $g->{ref}->full_path($de), 1);
2722 }
2723 }
e518192f 2724 foreach my $g (@$globs) {
74a81227
EW
2725 if (my $path = $paths->{"/$g->{path}->{left}"}) {
2726 if ($path->{action} =~ /^[AR]$/) {
2727 get_dir_check($self, $exists, $g, $r);
2728 }
2729 }
e518192f 2730 foreach (keys %$paths) {
74a81227
EW
2731 if (/$g->{path}->{left_regex}/) {
2732 next if $paths->{$_}->{action} !~ /^[AR]$/;
2733 get_dir_check($self, $exists, $g, $r);
2734 }
e518192f
EW
2735 next unless /$g->{path}->{regex}/;
2736 my $p = $1;
2737 my $pathname = $g->{path}->full_path($p);
2738 next if $exists->{$pathname};
2739 $exists->{$pathname} = Git::SVN->init(
2740 $self->{url}, $pathname, undef,
2741 $g->{ref}->full_path($p), 1);
2742 }
2743 my $c = '';
2744 foreach (split m#/#, $g->{path}->{left}) {
2745 $c .= "/$_";
2746 next unless ($paths->{$c} &&
74a81227
EW
2747 ($paths->{$c}->{action} =~ /^[AR]$/));
2748 get_dir_check($self, $exists, $g, $r);
e518192f
EW
2749 }
2750 }
2751 values %$exists;
2752}
2753
e6434f87
EW
2754sub minimize_url {
2755 my ($self) = @_;
2756 return $self->{url} if ($self->{url} eq $self->{repos_root});
2757 my $url = $self->{repos_root};
2758 my @components = split(m!/!, $self->{svn_path});
2759 my $c = '';
2760 do {
2761 $url .= "/$c" if length $c;
2762 eval { (ref $self)->new($url)->get_latest_revnum };
2763 } while ($@ && ($c = shift @components));
2764 $url;
2765}
2766
d81bf827
EW
2767sub can_do_switch {
2768 my $self = shift;
2769 unless (defined $can_do_switch) {
2770 my $pool = SVN::Pool->new;
2771 my $rep = eval {
2772 $self->do_switch(1, '', 0, $self->{url},
2773 SVN::Delta::Editor->new, $pool);
2774 };
2775 if ($@) {
2776 $can_do_switch = 0;
2777 } else {
2778 $rep->abort_report($pool);
2779 $can_do_switch = 1;
2780 }
2781 $pool->clear;
2782 }
2783 $can_do_switch;
2784}
2785
0af9c9f9
EW
2786sub skip_unknown_revs {
2787 my ($err) = @_;
2788 my $errno = $err->apr_err();
2789 # Maybe the branch we're tracking didn't
2790 # exist when the repo started, so it's
2791 # not an error if it doesn't, just continue
2792 #
2793 # Wonderfully consistent library, eh?
2794 # 160013 - svn:// and file://
2795 # 175002 - http(s)://
2796 # 175007 - http(s):// (this repo required authorization, too...)
2797 # More codes may be discovered later...
2798 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2799 return;
2800 }
2801 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2802}
2803
2804# svn_log_changed_path_t objects passed to get_log are likely to be
2805# overwritten even if only the refs are copied to an external variable,
2806# so we should dup the structures in their entirety. Using an externally
2807# passed pool (instead of our temporary and quickly cleared pool in
2808# Git::SVN::Ra) does not help matters at all...
2809sub dup_changed_paths {
2810 my ($paths) = @_;
2811 return undef unless $paths;
2812 my %ret;
2813 foreach my $p (keys %$paths) {
2814 my $i = $paths->{$p};
2815 my %s = map { $_ => $i->$_ }
2816 qw/copyfrom_path copyfrom_rev action/;
2817 $ret{$p} = \%s;
2818 }
2819 \%ret;
2820}
2821
f8c9d1d2
EW
2822package Git::SVN::Log;
2823use strict;
2824use warnings;
2825use POSIX qw/strftime/;
2826use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2827 %rusers $show_commit $incremental/;
2828my $l_fmt;
2829
2830sub cmt_showable {
2831 my ($c) = @_;
2832 return 1 if defined $c->{r};
2833 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2834 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
44320b9e
EW
2835 my @log = command(qw/cat-file commit/, $c->{c});
2836 shift @log while ($log[0] ne "\n");
2837 shift @log;
2838 @{$c->{l}} = grep !/^git-svn-id: /, @log;
f8c9d1d2
EW
2839
2840 (undef, $c->{r}, undef) = ::extract_metadata(
44320b9e 2841 (grep(/^git-svn-id: /, @log))[-1]);
f8c9d1d2
EW
2842 }
2843 return defined $c->{r};
2844}
2845
2846sub log_use_color {
2847 return 1 if $color;
2848 my ($dc, $dcvar);
2849 $dcvar = 'color.diff';
2850 $dc = `git-config --get $dcvar`;
2851 if ($dc eq '') {
2852 # nothing at all; fallback to "diff.color"
2853 $dcvar = 'diff.color';
2854 $dc = `git-config --get $dcvar`;
2855 }
2856 chomp($dc);
2857 if ($dc eq 'auto') {
2858 my $pc;
2859 $pc = `git-config --get color.pager`;
2860 if ($pc eq '') {
2861 # does not have it -- fallback to pager.color
2862 $pc = `git-config --bool --get pager.color`;
2863 }
2864 else {
2865 $pc = `git-config --bool --get color.pager`;
2866 if ($?) {
2867 $pc = 'false';
2868 }
2869 }
2870 chomp($pc);
2871 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2872 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2873 }
2874 return 0;
2875 }
2876 return 0 if $dc eq 'never';
2877 return 1 if $dc eq 'always';
2878 chomp($dc = `git-config --bool --get $dcvar`);
2879 return ($dc eq 'true');
2880}
2881
2882sub git_svn_log_cmd {
2883 my ($r_min, $r_max) = @_;
2884 my $gs = Git::SVN->_new;
2885 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2886 $gs->refname);
2887 push @cmd, '-r' unless $non_recursive;
2888 push @cmd, qw/--raw --name-status/ if $verbose;
2889 push @cmd, '--color' if log_use_color();
2890 return @cmd unless defined $r_max;
2891 if ($r_max == $r_min) {
2892 push @cmd, '--max-count=1';
2893 if (my $c = $gs->rev_db_get($r_max)) {
2894 push @cmd, $c;
2895 }
2896 } else {
2897 my ($c_min, $c_max);
2898 $c_max = $gs->rev_db_get($r_max);
2899 $c_min = $gs->rev_db_get($r_min);
2900 if (defined $c_min && defined $c_max) {
2901 if ($r_max > $r_max) {
2902 push @cmd, "$c_min..$c_max";
2903 } else {
2904 push @cmd, "$c_max..$c_min";
2905 }
2906 } elsif ($r_max > $r_min) {
2907 push @cmd, $c_max;
2908 } else {
2909 push @cmd, $c_min;
2910 }
2911 }
2912 return @cmd;
2913}
2914
2915# adapted from pager.c
2916sub config_pager {
2917 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2918 if (!defined $pager) {
2919 $pager = 'less';
2920 } elsif (length $pager == 0 || $pager eq 'cat') {
2921 $pager = undef;
2922 }
2923}
2924
2925sub run_pager {
2926 return unless -t *STDOUT;
2927 pipe my $rfd, my $wfd or return;
2928 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2929 if (!$pid) {
2930 open STDOUT, '>&', $wfd or
2931 ::fatal "Can't redirect to stdout: $!\n";
2932 return;
2933 }
2934 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2935 $ENV{LESS} ||= 'FRSX';
2936 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2937}
2938
21819a37
EW
2939sub tz_to_s_offset {
2940 my ($tz) = @_;
2941 $tz =~ s/(\d\d)$//;
2942 return ($1 * 60) + ($tz * 3600);
2943}
2944
f8c9d1d2
EW
2945sub get_author_info {
2946 my ($dest, $author, $t, $tz) = @_;
2947 $author =~ s/(?:^\s*|\s*$)//g;
2948 $dest->{a_raw} = $author;
2949 my $au;
1c8443b0 2950 if ($::_authors) {
f8c9d1d2
EW
2951 $au = $rusers{$author} || undef;
2952 }
2953 if (!$au) {
2954 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2955 }
2956 $dest->{t} = $t;
2957 $dest->{tz} = $tz;
2958 $dest->{a} = $au;
2959 # Date::Parse isn't in the standard Perl distro :(
2960 if ($tz =~ s/^\+//) {
21819a37 2961 $t += tz_to_s_offset($tz);
f8c9d1d2 2962 } elsif ($tz =~ s/^\-//) {
21819a37 2963 $t -= tz_to_s_offset($tz);
f8c9d1d2
EW
2964 }
2965 $dest->{t_utc} = $t;
2966}
2967
2968sub process_commit {
2969 my ($c, $r_min, $r_max, $defer) = @_;
2970 if (defined $r_min && defined $r_max) {
2971 if ($r_min == $c->{r} && $r_min == $r_max) {
2972 show_commit($c);
2973 return 0;
2974 }
2975 return 1 if $r_min == $r_max;
2976 if ($r_min < $r_max) {
2977 # we need to reverse the print order
2978 return 0 if (defined $limit && --$limit < 0);
2979 push @$defer, $c;
2980 return 1;
2981 }
2982 if ($r_min != $r_max) {
2983 return 1 if ($r_min < $c->{r});
2984 return 1 if ($r_max > $c->{r});
2985 }
2986 }
2987 return 0 if (defined $limit && --$limit < 0);
2988 show_commit($c);
2989 return 1;
2990}
2991
2992sub show_commit {
2993 my $c = shift;
2994 if ($oneline) {
2995 my $x = "\n";
2996 if (my $l = $c->{l}) {
2997 while ($l->[0] =~ /^\s*$/) { shift @$l }
2998 $x = $l->[0];
2999 }
3000 $l_fmt ||= 'A' . length($c->{r});
3001 print 'r',pack($l_fmt, $c->{r}),' | ';
3002 print "$c->{c} | " if $show_commit;
3003 print $x;
3004 } else {
3005 show_commit_normal($c);
3006 }
3007}
3008
3009sub show_commit_changed_paths {
3010 my ($c) = @_;
3011 return unless $c->{changed};
3012 print "Changed paths:\n", @{$c->{changed}};
3013}
3014
3015sub show_commit_normal {
3016 my ($c) = @_;
3017 print '-' x72, "\nr$c->{r} | ";
3018 print "$c->{c} | " if $show_commit;
3019 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3020 localtime($c->{t_utc})), ' | ';
3021 my $nr_line = 0;
3022
3023 if (my $l = $c->{l}) {
3024 while ($l->[$#$l] eq "\n" && $#$l > 0
3025 && $l->[($#$l - 1)] eq "\n") {
3026 pop @$l;
3027 }
3028 $nr_line = scalar @$l;
3029 if (!$nr_line) {
3030 print "1 line\n\n\n";
3031 } else {
3032 if ($nr_line == 1) {
3033 $nr_line = '1 line';
3034 } else {
3035 $nr_line .= ' lines';
3036 }
3037 print $nr_line, "\n";
3038 show_commit_changed_paths($c);
3039 print "\n";
3040 print $_ foreach @$l;
3041 }
3042 } else {
3043 print "1 line\n";
3044 show_commit_changed_paths($c);
3045 print "\n";
3046
3047 }
3048 foreach my $x (qw/raw diff/) {
3049 if ($c->{$x}) {
3050 print "\n";
3051 print $_ foreach @{$c->{$x}}
3052 }
3053 }
3054}
3055
3056sub cmd_show_log {
3057 my (@args) = @_;
3058 my ($r_min, $r_max);
3059 my $r_last = -1; # prevent dupes
3060 if (defined $TZ) {
3061 $ENV{TZ} = $TZ;
3062 } else {
3063 delete $ENV{TZ};
3064 }
3065 if (defined $::_revision) {
3066 if ($::_revision =~ /^(\d+):(\d+)$/) {
3067 ($r_min, $r_max) = ($1, $2);
3068 } elsif ($::_revision =~ /^\d+$/) {
3069 $r_min = $r_max = $::_revision;
3070 } else {
3071 ::fatal "-r$::_revision is not supported, use ",
3072 "standard \'git log\' arguments instead\n";
3073 }
3074 }
3075
3076 config_pager();
3077 @args = (git_svn_log_cmd($r_min, $r_max), @args);
3078 my $log = command_output_pipe(@args);
3079 run_pager();
3080 my (@k, $c, $d);
3081 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3082 while (<$log>) {
3083 if (/^${esc_color}commit ($::sha1_short)/o) {
3084 my $cmt = $1;
3085 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3086 $r_last = $c->{r};
3087 process_commit($c, $r_min, $r_max, \@k) or
3088 goto out;
3089 }
3090 $d = undef;
3091 $c = { c => $cmt };
3092 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3093 get_author_info($c, $1, $2, $3);
3094 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3095 # ignore
3096 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3097 push @{$c->{raw}}, $_;
3098 } elsif (/^${esc_color}[ACRMDT]\t/) {
3099 # we could add $SVN->{svn_path} here, but that requires
3100 # remote access at the moment (repo_path_split)...
3101 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3102 push @{$c->{changed}}, $_;
3103 } elsif (/^${esc_color}diff /o) {
3104 $d = 1;
3105 push @{$c->{diff}}, $_;
3106 } elsif ($d) {
3107 push @{$c->{diff}}, $_;
3108 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3109 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3110 } elsif (s/^${esc_color} //o) {
3111 push @{$c->{l}}, $_;
3112 }
3113 }
3114 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3115 $r_last = $c->{r};
3116 process_commit($c, $r_min, $r_max, \@k);
3117 }
3118 if (@k) {
3119 my $swap = $r_max;
3120 $r_max = $r_min;
3121 $r_min = $swap;
3122 process_commit($_, $r_min, $r_max) foreach reverse @k;
3123 }
3124out:
c843c464 3125 close $log;
f8c9d1d2
EW
3126 print '-' x72,"\n" unless $incremental || $oneline;
3127}
3128
706587fc
EW
3129package Git::SVN::Migration;
3130# these version numbers do NOT correspond to actual version numbers
3131# of git nor git-svn. They are just relative.
3132#
3133# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3134#
3135# v1 layout: .git/$id/info/url, refs/remotes/$id
3136#
3137# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3138#
3139# v3 layout: .git/svn/$id, refs/remotes/$id
3140# - info/url may remain for backwards compatibility
3141# - this is what we migrate up to this layout automatically,
3142# - this will be used by git svn init on single branches
3143#
3144# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3145# - this is only created for newly multi-init-ed
3146# repositories. Similar in spirit to the
3147# --use-separate-remotes option in git-clone (now default)
3148# - we do not automatically migrate to this (following
3149# the example set by core git)
3150use strict;
3151use warnings;
3152use Carp qw/croak/;
3153use File::Path qw/mkpath/;
47e39c55
EW
3154use File::Basename qw/dirname basename/;
3155use vars qw/$_minimize/;
706587fc
EW
3156
3157sub migrate_from_v0 {
3158 my $git_dir = $ENV{GIT_DIR};
3159 return undef unless -d $git_dir;
3160 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3161 my $migrated = 0;
3162 while (<$fh>) {
3163 chomp;
3164 my ($id, $orig_ref) = ($_, $_);
3165 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3166 next unless -f "$git_dir/$id/info/url";
3167 my $new_ref = "refs/remotes/$id";
3168 if (::verify_ref("$new_ref^0")) {
3169 print STDERR "W: $orig_ref is probably an old ",
3170 "branch used by an ancient version of ",
3171 "git-svn.\n",
3172 "However, $new_ref also exists.\n",
3173 "We will not be able ",
3174 "to use this branch until this ",
3175 "ambiguity is resolved.\n";
3176 next;
3177 }
3178 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3179 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3180 command_noisy('update-ref', $new_ref, $orig_ref);
3181 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3182 $migrated++;
3183 }
3184 command_close_pipe($fh, $ctx);
3185 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3186 $migrated;
3187}
3188
3189sub migrate_from_v1 {
3190 my $git_dir = $ENV{GIT_DIR};
3191 my $migrated = 0;
3192 return $migrated unless -d $git_dir;
3193 my $svn_dir = "$git_dir/svn";
3194
3195 # just in case somebody used 'svn' as their $id at some point...
3196 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3197
3198 print STDERR "Migrating from a git-svn v1 layout...\n";
3199 mkpath([$svn_dir]);
3200 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3201 "$svn_dir\n\t(required for this version ",
3202 "($::VERSION) of git-svn) does not. exist\n";
3203 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3204 while (<$fh>) {
3205 my $x = $_;
3206 next unless $x =~ s#^refs/remotes/##;
3207 chomp $x;
3208 next unless -f "$git_dir/$x/info/url";
3209 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3210 next unless $u;
3211 my $dn = dirname("$git_dir/svn/$x");
3212 mkpath([$dn]) unless -d $dn;
3213 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3214 mkpath(["$git_dir/svn/svn"]);
3215 print STDERR " - $git_dir/$x/info => ",
3216 "$git_dir/svn/$x/info\n";
3217 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3218 croak "$!: $x";
3219 # don't worry too much about these, they probably
3220 # don't exist with repos this old (save for index,
3221 # and we can easily regenerate that)
3222 foreach my $f (qw/unhandled.log index .rev_db/) {
3223 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3224 }
3225 } else {
3226 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3227 rename "$git_dir/$x", "$git_dir/svn/$x" or
3228 croak "$!: $x";
3229 }
3230 $migrated++;
3231 }
3232 command_close_pipe($fh, $ctx);
3233 print STDERR "Done migrating from a git-svn v1 layout\n";
3234 $migrated;
3235}
3236
3237sub read_old_urls {
3238 my ($l_map, $pfx, $path) = @_;
3239 my @dir;
3240 foreach (<$path/*>) {
3241 if (-r "$_/info/url") {
3242 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3243 my $ref_id = $pfx . basename $_;
3244 my $url = ::file_to_s("$_/info/url");
3245 $l_map->{$ref_id} = $url;
3246 } elsif (-d $_) {
3247 push @dir, $_;
3248 }
3249 }
3250 foreach (@dir) {
3251 my $x = $_;
3252 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3253 read_old_urls($l_map, $x, $_);
3254 }
3255}
3256
3257sub migrate_from_v2 {
3258 my @cfg = command(qw/config -l/);
3259 return if grep /^svn-remote\..+\.url=/, @cfg;
3260 my %l_map;
3261 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3262 my $migrated = 0;
3263
3264 foreach my $ref_id (sort keys %l_map) {
471bc000
EW
3265 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3266 if ($@) {
3267 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3268 }
706587fc
EW
3269 $migrated++;
3270 }
3271 $migrated;
3272}
3273
47e39c55
EW
3274sub minimize_connections {
3275 my $r = Git::SVN::read_all_remotes();
3276 my $new_urls = {};
3277 my $root_repos = {};
3278 foreach my $repo_id (keys %$r) {
3279 my $url = $r->{$repo_id}->{url} or next;
3280 my $fetch = $r->{$repo_id}->{fetch} or next;
3281 my $ra = Git::SVN::Ra->new($url);
3282
3283 # skip existing cases where we already connect to the root
3284 if (($ra->{url} eq $ra->{repos_root}) ||
3285 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3286 $repo_id)) {
3287 $root_repos->{$ra->{url}} = $repo_id;
3288 next;
3289 }
3290
3291 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3292 my $root_path = $ra->{url};
4e9f6cc7 3293 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
47e39c55
EW
3294 foreach my $path (keys %$fetch) {
3295 my $ref_id = $fetch->{$path};
3296 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3297
3298 # make sure we can read when connecting to
3299 # a higher level of a repository
3300 my ($last_rev, undef) = $gs->last_rev_commit;
3301 if (!defined $last_rev) {
3302 $last_rev = eval {
3303 $root_ra->get_latest_revnum;
3304 };
3305 next if $@;
3306 }
3307 my $new = $root_path;
3308 $new .= length $path ? "/$path" : '';
3309 eval {
3310 $root_ra->get_log([$new], $last_rev, $last_rev,
3311 0, 0, 1, sub { });
3312 };
3313 next if $@;
3314 $new_urls->{$ra->{repos_root}}->{$new} =
3315 { ref_id => $ref_id,
3316 old_repo_id => $repo_id,
3317 old_path => $path };
3318 }
3319 }
3320
3321 my @emptied;
3322 foreach my $url (keys %$new_urls) {
3323 # see if we can re-use an existing [svn-remote "repo_id"]
3324 # instead of creating a(n ugly) new section:
3325 my $repo_id = $root_repos->{$url} ||
3326 Git::SVN::sanitize_remote_name($url);
3327
3328 my $fetch = $new_urls->{$url};
3329 foreach my $path (keys %$fetch) {
3330 my $x = $fetch->{$path};
3331 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3332 my $pfx = "svn-remote.$x->{old_repo_id}";
3333
3334 my $old_fetch = quotemeta("$x->{old_path}:".
3335 "refs/remotes/$x->{ref_id}");
8b8fc068 3336 command_noisy(qw/config --unset/,
47e39c55
EW
3337 "$pfx.fetch", '^'. $old_fetch . '$');
3338 delete $r->{$x->{old_repo_id}}->
3339 {fetch}->{$x->{old_path}};
3340 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
8b8fc068 3341 command_noisy(qw/config --unset/,
47e39c55
EW
3342 "$pfx.url");
3343 push @emptied, $x->{old_repo_id}
3344 }
3345 }
3346 }
3347 if (@emptied) {
3348 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3349 "$ENV{GIT_DIR}/config";
3350 print STDERR <<EOF;
3351The following [svn-remote] sections in your config file ($file) are empty
3352and can be safely removed:
3353EOF
3354 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3355 }
3356}
3357
706587fc
EW
3358sub migration_check {
3359 migrate_from_v0();
3360 migrate_from_v1();
3361 migrate_from_v2();
47e39c55 3362 minimize_connections() if $_minimize;
706587fc
EW
3363}
3364
ef3cfaad
EW
3365package Git::IndexInfo;
3366use strict;
3367use warnings;
3368use Git qw/command_input_pipe command_close_pipe/;
3369
3370sub new {
3371 my ($class) = @_;
3372 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3373 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3374}
3375
3376sub remove {
3377 my ($self, $path) = @_;
3378 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3379 return ++$self->{nr};
3380 }
3381 undef;
3382}
3383
3384sub update {
3385 my ($self, $mode, $hash, $path) = @_;
3386 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3387 return ++$self->{nr};
3388 }
3389 undef;
3390}
3391
3392sub DESTROY {
3393 my ($self) = @_;
3394 command_close_pipe($self->{gui}, $self->{ctx});
3395}
3396
4bb9ed04
EW
3397package Git::SVN::GlobSpec;
3398use strict;
3399use warnings;
3400
3401sub new {
3402 my ($class, $glob) = @_;
4bb9ed04
EW
3403 my $re = $glob;
3404 $re =~ s!/+$!!g; # no need for trailing slashes
4e9f6cc7 3405 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4bb9ed04
EW
3406 my ($left, $right) = ($1, $2);
3407 if ($nr > 1) {
e518192f
EW
3408 die "Only one '*' wildcard expansion ",
3409 "is supported (got $nr): '$glob'\n";
4bb9ed04 3410 } elsif ($nr == 0) {
e518192f 3411 die "One '*' is needed for glob: '$glob'\n";
4bb9ed04
EW
3412 }
3413 $re = quotemeta($left) . $re . quotemeta($right);
4e9f6cc7
EW
3414 if (length $left && !($left =~ s!/+$!!g)) {
3415 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3416 }
3417 if (length $right && !($right =~ s!^/+!!g)) {
3418 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3419 }
74a81227
EW
3420 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3421 bless { left => $left, right => $right, left_regex => $left_re,
4bb9ed04
EW
3422 regex => qr/$re/, glob => $glob }, $class;
3423}
3424
3425sub full_path {
3426 my ($self, $path) = @_;
3427 return (length $self->{left} ? "$self->{left}/" : '') .
3428 $path . (length $self->{right} ? "/$self->{right}" : '');
3429}
3430
3397f9df
EW
3431__END__
3432
3433Data structures:
3434
4bb9ed04
EW
3435
3436$remotes = { # returned by read_all_remotes()
3437 'svn' => {
3438 # svn-remote.svn.url=https://svn.musicpd.org
3439 url => 'https://svn.musicpd.org',
3440 # svn-remote.svn.fetch=mpd/trunk:trunk
3441 fetch => {
3442 'mpd/trunk' => 'trunk',
3443 },
3444 # svn-remote.svn.tags=mpd/tags/*:tags/*
3445 tags => {
3446 path => {
3447 left => 'mpd/tags',
3448 right => '',
3449 regex => qr!mpd/tags/([^/]+)$!,
3450 glob => 'tags/*',
3451 },
3452 ref => {
3453 left => 'tags',
3454 right => '',
3455 regex => qr!tags/([^/]+)$!,
3456 glob => 'tags/*',
3457 },
3458 }
3459 }
3460};
3461
44320b9e 3462$log_entry hashref as returned by libsvn_log_entry()
3397f9df 3463{
44320b9e 3464 log => 'whitespace-formatted log entry
3397f9df
EW
3465', # trailing newline is preserved
3466 revision => '8', # integer
3467 date => '2004-02-24T17:01:44.108345Z', # commit date
3468 author => 'committer name'
3469};
3470
6e8548cc
EW
3471
3472# this is generated by generate_diff();
3397f9df
EW
3473@mods = array of diff-index line hashes, each element represents one line
3474 of diff-index output
3475
3476diff-index line ($m hash)
3477{
3478 mode_a => first column of diff-index output, no leading ':',
3479 mode_b => second column of diff-index output,
3480 sha1_b => sha1sum of the final blob,
ac8e0b91 3481 chg => change type [MCRADT],
3397f9df
EW
3482 file_a => original file name of a file (iff chg is 'C' or 'R')
3483 file_b => new/current file name of a file (any chg)
3484}
3485;
a5e0cedc 3486
a00439ac
EW
3487# retval of read_url_paths{,_all}();
3488$l_map = {
3489 # repository root url
3490 'https://svn.musicpd.org' => {
3491 # repository path # GIT_SVN_ID
3492 'mpd/trunk' => 'trunk',
3493 'mpd/tags/0.11.5' => 'tags/0.11.5',
3494 },
3495}
3496
a5e0cedc
EW
3497Notes:
3498 I don't trust the each() function on unless I created %hash myself
3499 because the internal iterator may not have started at base.