git-svn: replace .rev_db with a more space-efficient .rev_map format
[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
15153451
BS
12# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16
5253dc33 17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
706587fc 18$ENV{GIT_DIR} ||= '.git';
9fa00b65 19$Git::SVN::default_repo_id = 'svn';
8b8fc068 20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
6af1db44 21$Git::SVN::Ra::_log_window_size = 100;
13ccd6d4 22
f8c9d1d2 23$Git::SVN::Log::TZ = $ENV{TZ};
3397f9df 24$ENV{TZ} = 'UTC';
a00439ac 25$| = 1; # unbuffer STDOUT
3397f9df 26
207f1a75 27sub fatal (@) { print STDERR "@_\n"; exit 1 }
b9c85187
EW
28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29require SVN::Ra;
30require SVN::Delta;
31if ($SVN::Core::VERSION lt '1.1.0') {
207f1a75 32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
b9c85187 33}
d81bf827 34push @Git::SVN::Ra::ISA, 'SVN::Ra';
b9c85187
EW
35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
3397f9df 37use Carp qw/croak/;
8d7c4fad 38use Digest::MD5;
3397f9df
EW
39use IO::File qw//;
40use File::Basename qw/dirname basename/;
41use File::Path qw/mkpath/;
512b620b 42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
968bdf1f 43use IPC::Open3;
336f1714 44use Git;
a5e0cedc 45
336f1714 46BEGIN {
c5f71ad0
SV
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
336f1714
EW
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe/) {
c5f71ad0 51 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
8d7c4fad 52 Git::SVN::Migration Git::SVN::Log Git::SVN),
c5f71ad0
SV
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
336f1714 56 }
336f1714
EW
57}
58
b9c85187 59my ($SVN);
83e9940a 60
f8c9d1d2
EW
61$sha1 = qr/[a-f\d]{40}/;
62$sha1_short = qr/[a-f\d]{4,40}/;
44320b9e 63my ($_stdin, $_help, $_edit,
9760adcc 64 $_message, $_file,
d05d72e0 65 $_template, $_shared,
171af110 66 $_version, $_fetch_all, $_no_rebase,
dee41f3e 67 $_merge, $_strategy, $_dry_run, $_local,
8b014d71 68 $_prefix, $_no_checkout, $_url, $_verbose);
0bed5eaa 69$Git::SVN::_follow_parent = 1;
706587fc
EW
70my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
71 'config-dir=s' => \$Git::SVN::Ra::config_dir,
72 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
0bed5eaa 73my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
dc5869c0 74 'authors-file|A=s' => \$_authors,
ecc712dd 75 'repack:i' => \$Git::SVN::_repack,
97ae0911
EW
76 'noMetadata' => \$Git::SVN::_no_metadata,
77 'useSvmProps' => \$Git::SVN::_use_svm_props,
62e349d2 78 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
6af1db44 79 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
1e889ef3 80 'no-checkout' => \$_no_checkout,
80f50749 81 'quiet|q' => \$_q,
ecc712dd
EW
82 'repack-flags|repack-args|repack-opts=s' =>
83 \$Git::SVN::_repack_flags,
70ae04e4 84 'use-log-author' => \$Git::SVN::_use_log_author,
706587fc 85 %remote_opts );
36f5b1f0 86
8f728fb9 87my ($_trunk, $_tags, $_branches, $_stdlayout);
0dfaf0a4 88my %icv;
dadc6d2a
EW
89my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
90 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
91 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
8f728fb9 92 'stdlayout|s' => \$_stdlayout,
4a1bb4c3 93 'minimize-url|m' => \$Git::SVN::_minimize_url,
0dfaf0a4
EW
94 'no-metadata' => sub { $icv{noMetadata} = 1 },
95 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
96 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
97 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
dadc6d2a 98 %remote_opts );
27e9fb8d 99my %cmt_opts = ( 'edit|e' => \$_edit,
24e22aa8
EW
100 'rmdir' => \$SVN::Git::Editor::_rmdir,
101 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
102 'l=i' => \$SVN::Git::Editor::_rename_limit,
103 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
27e9fb8d 104);
9d55b41a 105
3397f9df 106my %cmd = (
2a3240be 107 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
e98671e5 108 { 'revision|r=s' => \$_revision,
905f8b7d 109 'fetch-all|all' => \$_fetch_all,
e98671e5 110 %fc_opts } ],
0425ea90
EW
111 clone => [ \&cmd_clone, "Initialize and fetch revisions",
112 { 'revision|r=s' => \$_revision,
113 %fc_opts, %init_opts } ],
d2866f9e 114 init => [ \&cmd_init, "Initialize a repo for tracking" .
f8ab6b73 115 " (requires URL argument)",
9d55b41a 116 \%init_opts ],
dadc6d2a
EW
117 'multi-init' => [ \&cmd_multi_init,
118 "Deprecated alias for ".
119 "'$0 init -T<trunk> -b<branches> -t<tags>'",
120 \%init_opts ],
d7ad3bed
EW
121 dcommit => [ \&cmd_dcommit,
122 'Commit several diffs to merge with upstream',
3289e86e
EW
123 { 'merge|m|M' => \$_merge,
124 'strategy|s=s' => \$_strategy,
905f8b7d 125 'verbose|v' => \$_verbose,
3289e86e 126 'dry-run|n' => \$_dry_run,
905f8b7d 127 'fetch-all|all' => \$_fetch_all,
171af110 128 'no-rebase' => \$_no_rebase,
4b155223 129 %cmt_opts, %fc_opts } ],
1ce255dc
EW
130 'set-tree' => [ \&cmd_set_tree,
131 "Set an SVN repository to a git tree-ish",
132 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
d05ddec5
BS
133 'create-ignore' => [ \&cmd_create_ignore,
134 'Create a .gitignore per svn:ignore',
135 { 'revision|r=i' => \$_revision
136 } ],
15153451
BS
137 'propget' => [ \&cmd_propget,
138 'Print the value of a property on a file or directory',
139 { 'revision|r=i' => \$_revision } ],
51e057cf
BS
140 'proplist' => [ \&cmd_proplist,
141 'List all properties of a file or directory',
142 { 'revision|r=i' => \$_revision } ],
5969cbe1 143 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
4dbfe2e9 144 { 'revision|r=i' => \$_revision
05b4df31 145 } ],
2d879792
VK
146 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
147 { 'revision|r=i' => \$_revision
148 } ],
1c8443b0 149 'multi-fetch' => [ \&cmd_multi_fetch,
e98671e5
EW
150 "Deprecated alias for $0 fetch --all",
151 { 'revision|r=s' => \$_revision, %fc_opts } ],
706587fc
EW
152 'migrate' => [ sub { },
153 # no-op, we automatically run this anyways,
706587fc
EW
154 'Migrate configuration/metadata/layout from
155 previous versions of git-svn',
a836a0e1
EW
156 { 'minimize' => \$Git::SVN::Migration::_minimize,
157 %remote_opts } ],
f8c9d1d2
EW
158 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
159 { 'limit=i' => \$Git::SVN::Log::limit,
79bb8d88 160 'revision|r=s' => \$_revision,
f8c9d1d2
EW
161 'verbose|v' => \$Git::SVN::Log::verbose,
162 'incremental' => \$Git::SVN::Log::incremental,
163 'oneline' => \$Git::SVN::Log::oneline,
164 'show-commit' => \$Git::SVN::Log::show_commit,
165 'non-recursive' => \$Git::SVN::Log::non_recursive,
79bb8d88 166 'authors-file|A=s' => \$_authors,
f8c9d1d2 167 'color' => \$Git::SVN::Log::color,
4dbfe2e9 168 'pager=s' => \$Git::SVN::Log::pager
79bb8d88 169 } ],
26e60160 170 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
4dbfe2e9 171 {} ],
905f8b7d
EW
172 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
173 { 'merge|m|M' => \$_merge,
174 'verbose|v' => \$_verbose,
175 'strategy|s=s' => \$_strategy,
dee41f3e 176 'local|l' => \$_local,
905f8b7d
EW
177 'fetch-all|all' => \$_fetch_all,
178 %fc_opts } ],
44320b9e
EW
179 'commit-diff' => [ \&cmd_commit_diff,
180 'Commit a diff between two trees',
27e9fb8d
EW
181 { 'message|m=s' => \$_message,
182 'file|F=s' => \$_file,
45bf473a 183 'revision|r=s' => \$_revision,
27e9fb8d 184 %cmt_opts } ],
e6fefa92
DK
185 'info' => [ \&cmd_info,
186 "Show info about the latest SVN revision
187 on the current branch",
8b014d71 188 { 'url' => \$_url, } ],
3397f9df 189);
9d55b41a 190
3397f9df
EW
191my $cmd;
192for (my $i = 0; $i < @ARGV; $i++) {
193 if (defined $cmd{$ARGV[$i]}) {
194 $cmd = $ARGV[$i];
195 splice @ARGV, $i, 1;
196 last;
197 }
198};
199
5253dc33
EW
200# make sure we're always running
201unless ($cmd =~ /(?:clone|init|multi-init)$/) {
202 unless (-d $ENV{GIT_DIR}) {
203 if ($git_dir_user_set) {
204 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
205 "but it is not a directory\n";
206 }
207 my $git_dir = delete $ENV{GIT_DIR};
208 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
209 unless (length $cdup) {
210 die "Already at toplevel, but $git_dir ",
211 "not found '$cdup'\n";
212 }
213 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
214 unless (-d $git_dir) {
215 die "$git_dir still not found after going to ",
216 "'$cdup'\n";
217 }
218 $ENV{GIT_DIR} = $git_dir;
219 }
220}
f4dd334b
GH
221
222my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
223
224read_repo_config(\%opts);
225Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
226my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
227 'minimize-connections' => \$Git::SVN::Migration::_minimize,
228 'id|i=s' => \$Git::SVN::default_ref_id,
229 'svn-remote|remote|R=s' => sub {
230 $Git::SVN::no_reuse_existing = 1;
231 $Git::SVN::default_repo_id = $_[1] });
232exit 1 if (!$rv && $cmd && $cmd ne 'log');
233
234usage(0) if $_help;
235version() if $_version;
236usage(1) unless defined $cmd;
237load_authors() if $_authors;
238
0425ea90 239unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
706587fc
EW
240 Git::SVN::Migration::migration_check();
241}
ecc712dd 242Git::SVN::init_vars();
b805b44a
EW
243eval {
244 Git::SVN::verify_remotes_sanity();
245 $cmd{$cmd}->[0]->(@ARGV);
246};
247fatal $@ if $@;
1e889ef3 248post_fetch_checkout();
3397f9df
EW
249exit 0;
250
251####################### primary functions ######################
252sub usage {
253 my $exit = shift || 0;
254 my $fd = $exit ? \*STDERR : \*STDOUT;
255 print $fd <<"";
256git-svn - bidirectional operations between a single Subversion tree and git
257Usage: $0 <command> [options] [arguments]\n
448c81b4
EW
258
259 print $fd "Available commands:\n" unless $cmd;
3397f9df
EW
260
261 foreach (sort keys %cmd) {
448c81b4 262 next if $cmd && $cmd ne $_;
a836a0e1 263 next if /^multi-/; # don't show deprecated commands
b203b769 264 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
aa807bc2 265 foreach (sort keys %{$cmd{$_}->[2]}) {
512b620b
EW
266 # mixed-case options are for .git/config only
267 next if /[A-Z]/ && /^[a-z]+$/i;
448c81b4 268 # prints out arguments as they should be passed:
b8c92cad 269 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
b203b769 270 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
448c81b4
EW
271 "--$_" : "-$_" }
272 split /\|/,$_)," $x\n";
273 }
3397f9df
EW
274 }
275 print $fd <<"";
448c81b4
EW
276\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
277arbitrary identifier if you're tracking multiple SVN branches/repositories in
278one git repository and want to keep them separate. See git-svn(1) for more
279information.
3397f9df
EW
280
281 exit $exit;
282}
283
551ce28f 284sub version {
7d60ab2c 285 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
551ce28f
EW
286 exit 0;
287}
288
8164b652
EW
289sub do_git_init_db {
290 unless (-d $ENV{GIT_DIR}) {
291 my @init_db = ('init');
292 push @init_db, "--template=$_template" if defined $_template;
dadc6d2a
EW
293 if (defined $_shared) {
294 if ($_shared =~ /[a-z]/) {
295 push @init_db, "--shared=$_shared";
296 } else {
297 push @init_db, "--shared";
298 }
299 }
8164b652
EW
300 command_noisy(@init_db);
301 }
0dfaf0a4
EW
302 my $set;
303 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
304 foreach my $i (keys %icv) {
305 die "'$set' and '$i' cannot both be set\n" if $set;
306 next unless defined $icv{$i};
307 command_noisy('config', "$pfx.$i", $icv{$i});
308 $set = $i;
309 }
8164b652
EW
310}
311
dadc6d2a
EW
312sub init_subdir {
313 my $repo_path = shift or return;
314 mkpath([$repo_path]) unless -d $repo_path;
315 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
f30603fc 316 $ENV{GIT_DIR} = '.git';
dadc6d2a
EW
317}
318
0425ea90
EW
319sub cmd_clone {
320 my ($url, $path) = @_;
321 if (!defined $path &&
8f728fb9 322 (defined $_trunk || defined $_branches || defined $_tags ||
323 defined $_stdlayout) &&
0425ea90
EW
324 $url !~ m#^[a-z\+]+://#) {
325 $path = $url;
326 }
0425ea90 327 $path = basename($url) if !defined $path || !length $path;
f30603fc 328 cmd_init($url, $path);
0425ea90
EW
329 Git::SVN::fetch_all($Git::SVN::default_repo_id);
330}
331
d2866f9e 332sub cmd_init {
8f728fb9 333 if (defined $_stdlayout) {
334 $_trunk = 'trunk' if (!defined $_trunk);
335 $_tags = 'tags' if (!defined $_tags);
336 $_branches = 'branches' if (!defined $_branches);
337 }
dadc6d2a
EW
338 if (defined $_trunk || defined $_branches || defined $_tags) {
339 return cmd_multi_init(@_);
03e0ea87 340 }
dadc6d2a
EW
341 my $url = shift or die "SVN repository location required ",
342 "as a command-line argument\n";
343 init_subdir(@_);
8164b652 344 do_git_init_db();
03e0ea87 345
706587fc 346 Git::SVN->init($url);
3397f9df
EW
347}
348
2a3240be 349sub cmd_fetch {
e98671e5
EW
350 if (grep /^\d+=./, @_) {
351 die "'<rev>=<commit>' fetch arguments are ",
352 "no longer supported.\n";
07a1c950 353 }
e98671e5
EW
354 my ($remote) = @_;
355 if (@_ > 1) {
905f8b7d 356 die "Usage: $0 fetch [--all] [svn-remote]\n";
e98671e5
EW
357 }
358 $remote ||= $Git::SVN::default_repo_id;
359 if ($_fetch_all) {
360 cmd_multi_fetch();
361 } else {
362 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
1c8443b0 363 }
2a3240be
EW
364}
365
1ce255dc 366sub cmd_set_tree {
3397f9df
EW
367 my (@commits) = @_;
368 if ($_stdin || !@commits) {
369 print "Reading from stdin...\n";
370 @commits = ();
371 while (<STDIN>) {
1ca72aef 372 if (/\b($sha1_short)\b/o) {
3397f9df
EW
373 unshift @commits, $1;
374 }
375 }
376 }
377 my @revs;
8de010ad 378 foreach my $c (@commits) {
aef4e921 379 my @tmp = command('rev-parse',$c);
8de010ad
EW
380 if (scalar @tmp == 1) {
381 push @revs, $tmp[0];
382 } elsif (scalar @tmp > 1) {
aef4e921 383 push @revs, reverse(command('rev-list',@tmp));
8de010ad 384 } else {
207f1a75 385 fatal "Failed to rev-parse $c";
8de010ad 386 }
3397f9df 387 }
1ce255dc
EW
388 my $gs = Git::SVN->new;
389 my ($r_last, $cmt_last) = $gs->last_rev_commit;
390 $gs->fetch;
97f6987a 391 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
1ce255dc
EW
392 fatal "There are new revisions that were fetched ",
393 "and need to be merged (or acknowledged) ",
394 "before committing.\nlast rev: $r_last\n",
207f1a75 395 " current: $gs->{last_rev}";
a5e0cedc 396 }
1ce255dc
EW
397 $gs->set_tree($_) foreach @revs;
398 print "Done committing ",scalar @revs," revisions to SVN\n";
a5e0cedc 399}
8f22562c 400
d7ad3bed
EW
401sub cmd_dcommit {
402 my $head = shift;
c8cfa3e4 403 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
826a9339 404 'Cannot dcommit with a dirty index. Commit your changes first, '
c8cfa3e4 405 . "or stash them with `git stash'.\n";
d7ad3bed 406 $head ||= 'HEAD';
a8ae2623 407 my @refs;
13c823fb 408 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
15d54753 409 print "Committing to $url ...\n";
13c823fb 410 unless ($gs) {
a8ae2623 411 die "Unable to determine upstream SVN information from ",
905f8b7d 412 "$head history\n";
a8ae2623 413 }
45bf473a 414 my $last_rev;
733a65aa 415 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
751eb395
EW
416 if ($_no_rebase && scalar(@$linear_refs) > 1) {
417 warn "Attempting to commit more than one change while ",
418 "--no-rebase is enabled.\n",
419 "If these changes depend on each other, re-running ",
420 "without --no-rebase will be required."
421 }
c74d9acf
EW
422 while (1) {
423 my $d = shift @$linear_refs or last;
45bf473a
EW
424 unless (defined $last_rev) {
425 (undef, $last_rev, undef) = cmt_metadata("$d~1");
426 unless (defined $last_rev) {
d7ad3bed 427 fatal "Unable to extract revision information ",
207f1a75 428 "from commit $d~1";
45bf473a
EW
429 }
430 }
b22d4497
EW
431 if ($_dry_run) {
432 print "diff-tree $d~1 $d\n";
433 } else {
751eb395 434 my $cmt_rev;
d7ad3bed 435 my %ed_opts = ( r => $last_rev,
61395354 436 log => get_commit_entry($d)->{log},
645833b5 437 ra => Git::SVN::Ra->new($gs->full_url),
3caf320b
KA
438 config => SVN::Core::config_get_config(
439 $Git::SVN::Ra::config_dir
440 ),
61395354
EW
441 tree_a => "$d~1",
442 tree_b => $d,
443 editor_cb => sub {
444 print "Committed r$_[0]\n";
751eb395
EW
445 $cmt_rev = $_[0];
446 },
a8ae2623 447 svn_path => '');
61395354 448 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
d7ad3bed 449 print "No changes\n$d~1 == $d\n";
733a65aa 450 } elsif ($parents->{$d} && @{$parents->{$d}}) {
751eb395 451 $gs->{inject_parents_dcommit}->{$cmt_rev} =
733a65aa 452 $parents->{$d};
d7ad3bed 453 }
751eb395
EW
454 $_fetch_all ? $gs->fetch_all : $gs->fetch;
455 next if $_no_rebase;
456
457 # we always want to rebase against the current HEAD,
458 # not any head that was passed to us
c74d9acf 459 my @diff = command('diff-tree', $d,
751eb395
EW
460 $gs->refname, '--');
461 my @finish;
462 if (@diff) {
463 @finish = rebase_cmd();
c74d9acf 464 print STDERR "W: $d and ", $gs->refname,
751eb395 465 " differ, using @finish:\n",
c74d9acf 466 join("\n", @diff), "\n";
751eb395
EW
467 } else {
468 print "No changes between current HEAD and ",
469 $gs->refname,
470 "\nResetting to the latest ",
471 $gs->refname, "\n";
472 @finish = qw/reset --mixed/;
473 }
474 command_noisy(@finish, $gs->refname);
c74d9acf
EW
475 if (@diff) {
476 @refs = ();
477 my ($url_, $rev_, $uuid_, $gs_) =
478 working_head_info($head, \@refs);
479 my ($linear_refs_, $parents_) =
480 linearize_history($gs_, \@refs);
481 if (scalar(@$linear_refs) !=
482 scalar(@$linear_refs_)) {
483 fatal "# of revisions changed ",
484 "\nbefore:\n",
485 join("\n", @$linear_refs),
486 "\n\nafter:\n",
487 join("\n", @$linear_refs_), "\n",
488 'If you are attempting to commit ',
489 "merges, try running:\n\t",
490 'git rebase --interactive',
491 '--preserve-merges ',
492 $gs->refname,
493 "\nBefore dcommitting";
494 }
495 if ($url_ ne $url) {
496 fatal "URL mismatch after rebase: ",
497 "$url_ != $url";
498 }
499 if ($uuid_ ne $uuid) {
500 fatal "uuid mismatch after rebase: ",
501 "$uuid_ != $uuid";
502 }
503 # remap parents
504 my (%p, @l, $i);
505 for ($i = 0; $i < scalar @$linear_refs; $i++) {
506 my $new = $linear_refs_->[$i] or next;
507 $p{$new} =
508 $parents->{$linear_refs->[$i]};
509 push @l, $new;
510 }
511 $parents = \%p;
512 $linear_refs = \@l;
513 }
751eb395 514 $last_rev = $cmt_rev;
b22d4497
EW
515 }
516 }
b22d4497
EW
517}
518
26e60160
AR
519sub cmd_find_rev {
520 my $revision_or_hash = shift;
521 my $result;
522 if ($revision_or_hash =~ /^r\d+$/) {
b3cb7e45
AR
523 my $head = shift;
524 $head ||= 'HEAD';
525 my @refs;
526 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
527 unless ($gs) {
528 die "Unable to determine upstream SVN information from ",
529 "$head history\n";
26e60160 530 }
b3cb7e45 531 my $desired_revision = substr($revision_or_hash, 1);
060610c5 532 $result = $gs->rev_map_get($desired_revision);
26e60160
AR
533 } else {
534 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
535 $result = $rev;
536 }
537 print "$result\n" if $result;
538}
539
905f8b7d
EW
540sub cmd_rebase {
541 command_noisy(qw/update-index --refresh/);
13c823fb
EW
542 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
543 unless ($gs) {
905f8b7d
EW
544 die "Unable to determine upstream SVN information from ",
545 "working tree history\n";
546 }
905f8b7d
EW
547 if (command(qw/diff-index HEAD --/)) {
548 print STDERR "Cannot rebase with uncommited changes:\n";
549 command_noisy('status');
550 exit 1;
551 }
dee41f3e 552 unless ($_local) {
cec0d5a3
SG
553 # rebase will checkout for us, so no need to do it explicitly
554 $_no_checkout = 'true';
dee41f3e
EW
555 $_fetch_all ? $gs->fetch_all : $gs->fetch;
556 }
905f8b7d
EW
557 command_noisy(rebase_cmd(), $gs->refname);
558}
559
5969cbe1 560sub cmd_show_ignore {
13c823fb
EW
561 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
562 $gs ||= Git::SVN->new;
5969cbe1 563 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
01bdab84
BS
564 $gs->prop_walk($gs->{path}, $r, sub {
565 my ($gs, $path, $props) = @_;
566 print STDOUT "\n# $path\n";
567 my $s = $props->{'svn:ignore'} or return;
568 $s =~ s/[\r\n]+/\n/g;
569 chomp $s;
570 $s =~ s#^#$path#gm;
571 print STDOUT "$s\n";
572 });
a5e0cedc
EW
573}
574
2d879792
VK
575sub cmd_show_externals {
576 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
577 $gs ||= Git::SVN->new;
578 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
579 $gs->prop_walk($gs->{path}, $r, sub {
580 my ($gs, $path, $props) = @_;
581 print STDOUT "\n# $path\n";
582 my $s = $props->{'svn:externals'} or return;
583 $s =~ s/[\r\n]+/\n/g;
584 chomp $s;
585 $s =~ s#^#$path#gm;
586 print STDOUT "$s\n";
587 });
588}
589
d05ddec5
BS
590sub cmd_create_ignore {
591 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
592 $gs ||= Git::SVN->new;
593 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
594 $gs->prop_walk($gs->{path}, $r, sub {
595 my ($gs, $path, $props) = @_;
596 # $path is of the form /path/to/dir/
597 my $ignore = '.' . $path . '.gitignore';
598 my $s = $props->{'svn:ignore'} or return;
599 open(GITIGNORE, '>', $ignore)
207f1a75 600 or fatal("Failed to open `$ignore' for writing: $!");
d05ddec5
BS
601 $s =~ s/[\r\n]+/\n/g;
602 chomp $s;
603 # Prefix all patterns so that the ignore doesn't apply
604 # to sub-directories.
605 $s =~ s#^#/#gm;
606 print GITIGNORE "$s\n";
607 close(GITIGNORE)
207f1a75 608 or fatal("Failed to close `$ignore': $!");
d05ddec5
BS
609 command_noisy('add', $ignore);
610 });
611}
612
b2b3ada7
DK
613sub canonicalize_path {
614 my ($path) = @_;
e6fefa92
DK
615 my $dot_slash_added = 0;
616 if (substr($path, 0, 1) ne "/") {
617 $path = "./" . $path;
618 $dot_slash_added = 1;
619 }
b2b3ada7
DK
620 # File::Spec->canonpath doesn't collapse x/../y into y (for a
621 # good reason), so let's do this manually.
622 $path =~ s#/+#/#g;
623 $path =~ s#/\.(?:/|$)#/#g;
624 $path =~ s#/[^/]+/\.\.##g;
625 $path =~ s#/$##g;
e6fefa92 626 $path =~ s#^\./## if $dot_slash_added;
b2b3ada7
DK
627 return $path;
628}
629
15153451
BS
630# get_svnprops(PATH)
631# ------------------
51e057cf 632# Helper for cmd_propget and cmd_proplist below.
15153451
BS
633sub get_svnprops {
634 my $path = shift;
635 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
636 $gs ||= Git::SVN->new;
637
638 # prefix THE PATH by the sub-directory from which the user
639 # invoked us.
640 $path = $cmd_dir_prefix . $path;
207f1a75 641 fatal("No such file or directory: $path") unless -e $path;
15153451
BS
642 my $is_dir = -d $path ? 1 : 0;
643 $path = $gs->{path} . '/' . $path;
644
645 # canonicalize the path (otherwise libsvn will abort or fail to
646 # find the file)
b2b3ada7 647 $path = canonicalize_path($path);
15153451
BS
648
649 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
650 my $props;
651 if ($is_dir) {
652 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
653 }
654 else {
655 (undef, $props) = $gs->ra->get_file($path, $r, undef);
656 }
657 return $props;
658}
659
660# cmd_propget (PROP, PATH)
661# ------------------------
662# Print the SVN property PROP for PATH.
663sub cmd_propget {
664 my ($prop, $path) = @_;
665 $path = '.' if not defined $path;
666 usage(1) if not defined $prop;
667 my $props = get_svnprops($path);
668 if (not defined $props->{$prop}) {
207f1a75 669 fatal("`$path' does not have a `$prop' SVN property.");
15153451
BS
670 }
671 print $props->{$prop} . "\n";
672}
673
51e057cf
BS
674# cmd_proplist (PATH)
675# -------------------
676# Print the list of SVN properties for PATH.
677sub cmd_proplist {
678 my $path = shift;
679 $path = '.' if not defined $path;
680 my $props = get_svnprops($path);
681 print "Properties on '$path':\n";
682 foreach (sort keys %{$props}) {
683 print " $_\n";
684 }
685}
686
8164b652 687sub cmd_multi_init {
9d55b41a 688 my $url = shift;
98327e58
EW
689 unless (defined $_trunk || defined $_branches || defined $_tags) {
690 usage(1);
9d55b41a 691 }
dc431666
EW
692
693 # there are currently some bugs that prevent multi-init/multi-fetch
694 # setups from working well without this.
695 $Git::SVN::_minimize_url = 1;
696
8164b652 697 $_prefix = '' unless defined $_prefix;
dadc6d2a
EW
698 if (defined $url) {
699 $url =~ s#/+$##;
700 init_subdir(@_);
701 }
f30603fc 702 do_git_init_db();
98327e58 703 if (defined $_trunk) {
706587fc
EW
704 my $trunk_ref = $_prefix . 'trunk';
705 # try both old-style and new-style lookups:
706 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
8164b652 707 unless ($gs_trunk) {
706587fc
EW
708 my ($trunk_url, $trunk_path) =
709 complete_svn_url($url, $_trunk);
710 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
711 undef, $trunk_ref);
98327e58 712 }
c35b96e7 713 }
706587fc 714 return unless defined $_branches || defined $_tags;
e7db67e6
EW
715 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
716 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
717 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
9d55b41a
EW
718}
719
1c8443b0 720sub cmd_multi_fetch {
0af9c9f9
EW
721 my $remotes = Git::SVN::read_all_remotes();
722 foreach my $repo_id (sort keys %$remotes) {
db03cd24 723 if ($remotes->{$repo_id}->{url}) {
4bb9ed04
EW
724 Git::SVN::fetch_all($repo_id, $remotes);
725 }
9d55b41a 726 }
9d55b41a
EW
727}
728
44320b9e
EW
729# this command is special because it requires no metadata
730sub cmd_commit_diff {
731 my ($ta, $tb, $url) = @_;
732 my $usage = "Usage: $0 commit-diff -r<revision> ".
207f1a75 733 "<tree-ish> <tree-ish> [<URL>]";
44320b9e 734 fatal($usage) if (!defined $ta || !defined $tb);
d3a840dc 735 my $svn_path;
44320b9e
EW
736 if (!defined $url) {
737 my $gs = eval { Git::SVN->new };
738 if (!$gs) {
739 fatal("Needed URL or usable git-svn --id in ",
740 "the command-line\n", $usage);
741 }
742 $url = $gs->{url};
d3a840dc 743 $svn_path = $gs->{path};
44320b9e
EW
744 }
745 unless (defined $_revision) {
746 fatal("-r|--revision is a required argument\n", $usage);
747 }
748 if (defined $_message && defined $_file) {
749 fatal("Both --message/-m and --file/-F specified ",
750 "for the commit message.\n",
207f1a75 751 "I have no idea what you mean");
44320b9e
EW
752 }
753 if (defined $_file) {
754 $_message = file_to_s($_file);
755 } else {
756 $_message ||= get_commit_entry($tb)->{log};
757 }
758 my $ra ||= Git::SVN::Ra->new($url);
d3a840dc 759 $svn_path ||= $ra->{svn_path};
44320b9e
EW
760 my $r = $_revision;
761 if ($r eq 'HEAD') {
762 $r = $ra->get_latest_revnum;
763 } elsif ($r !~ /^\d+$/) {
764 die "revision argument: $r not understood by git-svn\n";
765 }
61395354
EW
766 my %ed_opts = ( r => $r,
767 log => $_message,
768 ra => $ra,
769 tree_a => $ta,
770 tree_b => $tb,
771 editor_cb => sub { print "Committed r$_[0]\n" },
772 svn_path => $svn_path );
773 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
44320b9e
EW
774 print "No changes\n$ta == $tb\n";
775 }
44320b9e
EW
776}
777
e6fefa92
DK
778sub cmd_info {
779 my $path = canonicalize_path(shift or ".");
780 unless (scalar(@_) == 0) {
781 die "Too many arguments specified\n";
782 }
783
784 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
785
786 if (!$file_type && !$diff_status) {
787 print STDERR "$path: (Not a versioned resource)\n\n";
788 return;
789 }
790
791 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
792 unless ($gs) {
793 die "Unable to determine upstream SVN information from ",
794 "working tree history\n";
795 }
796 my $full_url = $url . ($path eq "." ? "" : "/$path");
797
8b014d71
DK
798 if ($_url) {
799 print $full_url, "\n";
800 return;
801 }
802
e6fefa92
DK
803 my $result = "Path: $path\n";
804 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
805 $result .= "URL: " . $full_url . "\n";
806
a5460eb7
EW
807 eval {
808 my $repos_root = $gs->repos_root;
809 Git::SVN::remove_username($repos_root);
810 $result .= "Repository Root: $repos_root\n";
811 };
812 if ($@) {
813 $result .= "Repository Root: (offline)\n";
814 }
e6fefa92
DK
815 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
816 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
817
818 $result .= "Node Kind: " .
819 ($file_type eq "dir" ? "directory" : "file") . "\n";
820
821 my $schedule = $diff_status eq "A"
822 ? "add"
823 : ($diff_status eq "D" ? "delete" : "normal");
824 $result .= "Schedule: $schedule\n";
825
826 if ($diff_status eq "A") {
827 print $result, "\n";
828 return;
829 }
830
831 my ($lc_author, $lc_rev, $lc_date_utc);
832 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
833 my $log = command_output_pipe(@args);
834 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
835 while (<$log>) {
836 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
837 $lc_author = $1;
838 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
839 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
840 (undef, $lc_rev, undef) = ::extract_metadata($1);
841 }
842 }
843 close $log;
844
845 Git::SVN::Log::set_local_timezone();
846
847 $result .= "Last Changed Author: $lc_author\n";
848 $result .= "Last Changed Rev: $lc_rev\n";
849 $result .= "Last Changed Date: " .
850 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
851
852 if ($file_type ne "dir") {
853 my $text_last_updated_date =
854 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
855 $result .=
856 "Text Last Updated: " .
857 Git::SVN::Log::format_svn_date($text_last_updated_date) .
858 "\n";
859 my $checksum;
860 if ($diff_status eq "D") {
861 my ($fh, $ctx) =
862 command_output_pipe(qw(cat-file blob), "HEAD:$path");
863 if ($file_type eq "link") {
864 my $file_name = <$fh>;
8d7c4fad 865 $checksum = md5sum("link $file_name");
e6fefa92 866 } else {
8d7c4fad 867 $checksum = md5sum($fh);
e6fefa92
DK
868 }
869 command_close_pipe($fh, $ctx);
870 } elsif ($file_type eq "link") {
871 my $file_name =
872 command(qw(cat-file blob), "HEAD:$path");
873 $checksum =
8d7c4fad 874 md5sum("link " . $file_name);
e6fefa92
DK
875 } else {
876 open FILE, "<", $path or die $!;
8d7c4fad 877 $checksum = md5sum(\*FILE);
e6fefa92
DK
878 close FILE or die $!;
879 }
880 $result .= "Checksum: " . $checksum . "\n";
881 }
882
883 print $result, "\n";
884}
885
3397f9df
EW
886########################### utility functions #########################
887
905f8b7d
EW
888sub rebase_cmd {
889 my @cmd = qw/rebase/;
890 push @cmd, '-v' if $_verbose;
891 push @cmd, qw/--merge/ if $_merge;
892 push @cmd, "--strategy=$_strategy" if $_strategy;
893 @cmd;
894}
895
1e889ef3
EW
896sub post_fetch_checkout {
897 return if $_no_checkout;
898 my $gs = $Git::SVN::_head or return;
899 return if verify_ref('refs/heads/master^0');
900
901 my $valid_head = verify_ref('HEAD^0');
902 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
903 return if ($valid_head || !verify_ref('HEAD^0'));
904
905 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
906 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
907 return if -f $index;
908
7ae3df8c 909 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1e889ef3
EW
910 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
911 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
912 print STDERR "Checked out HEAD:\n ",
913 $gs->full_url, " r", $gs->last_rev, "\n";
914}
915
98327e58
EW
916sub complete_svn_url {
917 my ($url, $path) = @_;
918 $path =~ s#/+$##;
98327e58 919 if ($path !~ m#^[a-z\+]+://#) {
98327e58
EW
920 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
921 fatal("E: '$path' is not a complete URL ",
207f1a75 922 "and a separate URL is not specified");
98327e58 923 }
706587fc 924 return ($url, $path);
98327e58 925 }
706587fc 926 return ($path, '');
98327e58
EW
927}
928
9d55b41a 929sub complete_url_ls_init {
706587fc
EW
930 my ($ra, $repo_path, $switch, $pfx) = @_;
931 unless ($repo_path) {
9d55b41a
EW
932 print STDERR "W: $switch not specified\n";
933 return;
934 }
706587fc
EW
935 $repo_path =~ s#/+$##;
936 if ($repo_path =~ m#^[a-z\+]+://#) {
937 $ra = Git::SVN::Ra->new($repo_path);
938 $repo_path = '';
e7db67e6 939 } else {
706587fc 940 $repo_path =~ s#^/+##;
e7db67e6 941 unless ($ra) {
706587fc 942 fatal("E: '$repo_path' is not a complete URL ",
207f1a75 943 "and a separate URL is not specified");
8164b652 944 }
e7db67e6 945 }
706587fc 946 my $url = $ra->{url};
b4d57e5e
EW
947 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
948 my $k = "svn-remote.$gs->{repo_id}.url";
949 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
950 if ($orig_url && ($orig_url ne $gs->{url})) {
951 die "$k already set: $orig_url\n",
952 "wanted to set to: $gs->{url}\n";
88cf4107 953 }
b4d57e5e
EW
954 command_oneline('config', $k, $gs->{url}) unless $orig_url;
955 my $remote_path = "$ra->{svn_path}/$repo_path/*";
956 $remote_path =~ s#/+#/#g;
957 $remote_path =~ s#^/##g;
958 my ($n) = ($switch =~ /^--(\w+)/);
959 if (length $pfx && $pfx !~ m#/$#) {
960 die "--prefix='$pfx' must have a trailing slash '/'\n";
9d55b41a 961 }
b4d57e5e
EW
962 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
963 "$remote_path:refs/remotes/$pfx*");
9d55b41a
EW
964}
965
aef4e921
EW
966sub verify_ref {
967 my ($ref) = @_;
2c5c1d53
EW
968 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
969 { STDERR => 0 }); };
aef4e921
EW
970}
971
a5e0cedc 972sub get_tree_from_treeish {
cf52b8f0 973 my ($treeish) = @_;
44320b9e 974 # $treeish can be a symbolic ref, too:
aef4e921 975 my $type = command_oneline(qw/cat-file -t/, $treeish);
cf52b8f0
EW
976 my $expected;
977 while ($type eq 'tag') {
aef4e921 978 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
cf52b8f0
EW
979 }
980 if ($type eq 'commit') {
aef4e921
EW
981 $expected = (grep /^tree /, command(qw/cat-file commit/,
982 $treeish))[0];
44320b9e 983 ($expected) = ($expected =~ /^tree ($sha1)$/o);
cf52b8f0
EW
984 die "Unable to get tree from $treeish\n" unless $expected;
985 } elsif ($type eq 'tree') {
986 $expected = $treeish;
987 } else {
988 die "$treeish is a $type, expected tree, tag or commit\n";
989 }
a5e0cedc
EW
990 return $expected;
991}
992
44320b9e
EW
993sub get_commit_entry {
994 my ($treeish) = shift;
995 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
996 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
997 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
998 open my $log_fh, '>', $commit_editmsg or croak $!;
3397f9df 999
44320b9e 1000 my $type = command_oneline(qw/cat-file -t/, $treeish);
4ad4515d 1001 if ($type eq 'commit' || $type eq 'tag') {
aef4e921 1002 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
44320b9e 1003 $type, $treeish);
3397f9df
EW
1004 my $in_msg = 0;
1005 while (<$msg_fh>) {
1006 if (!$in_msg) {
1007 $in_msg = 1 if (/^\s*$/);
df746c5a 1008 } elsif (/^git-svn-id: /) {
44320b9e
EW
1009 # skip this for now, we regenerate the
1010 # correct one on re-fetch anyways
1011 # TODO: set *:merge properties or like...
3397f9df 1012 } else {
44320b9e 1013 print $log_fh $_ or croak $!;
3397f9df
EW
1014 }
1015 }
aef4e921 1016 command_close_pipe($msg_fh, $ctx);
3397f9df 1017 }
44320b9e 1018 close $log_fh or croak $!;
3397f9df
EW
1019
1020 if ($_edit || ($type eq 'tree')) {
1021 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
44320b9e
EW
1022 # TODO: strip out spaces, comments, like git-commit.sh
1023 system($editor, $commit_editmsg);
3397f9df 1024 }
44320b9e
EW
1025 rename $commit_editmsg, $commit_msg or croak $!;
1026 open $log_fh, '<', $commit_msg or croak $!;
1027 { local $/; chomp($log_entry{log} = <$log_fh>); }
1028 close $log_fh or croak $!;
1029 unlink $commit_msg;
1030 \%log_entry;
a5e0cedc
EW
1031}
1032
3397f9df
EW
1033sub s_to_file {
1034 my ($str, $file, $mode) = @_;
1035 open my $fd,'>',$file or croak $!;
1036 print $fd $str,"\n" or croak $!;
1037 close $fd or croak $!;
1038 chmod ($mode &~ umask, $file) if (defined $mode);
1039}
1040
1041sub file_to_s {
1042 my $file = shift;
1043 open my $fd,'<',$file or croak "$!: file: $file\n";
1044 local $/;
1045 my $ret = <$fd>;
1046 close $fd or croak $!;
1047 $ret =~ s/\s*$//s;
1048 return $ret;
1049}
1050
eeb0abe0
EW
1051# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1052sub load_authors {
1053 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
f8c9d1d2 1054 my $log = $cmd eq 'log';
eeb0abe0
EW
1055 while (<$authors>) {
1056 chomp;
575d025c 1057 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
eeb0abe0 1058 my ($user, $name, $email) = ($1, $2, $3);
f8c9d1d2
EW
1059 if ($log) {
1060 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1061 } else {
1062 $users{$user} = [$name, $email];
1063 }
79bb8d88
EW
1064 }
1065 close $authors or croak $!;
1066}
1067
e0d10e1c 1068# convert GetOpt::Long specs for use by git-config
b8c92cad 1069sub read_repo_config {
706587fc 1070 return unless -d $ENV{GIT_DIR};
b8c92cad 1071 my $opts = shift;
97ae0911 1072 my @config_only;
b8c92cad 1073 foreach my $o (keys %$opts) {
97ae0911
EW
1074 # if we have mixedCase and a long option-only, then
1075 # it's a config-only variable that we don't need for
1076 # the command-line.
1077 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
b8c92cad 1078 my $v = $opts->{$o};
97ae0911 1079 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
b8c92cad 1080 $key =~ s/-//g;
e0d10e1c 1081 my $arg = 'git-config';
b8c92cad
EW
1082 $arg .= ' --int' if ($o =~ /[:=]i$/);
1083 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1084 if (ref $v eq 'ARRAY') {
1085 chomp(my @tmp = `$arg --get-all svn.$key`);
1086 @$v = @tmp if @tmp;
1087 } else {
1088 chomp(my $tmp = `$arg --get svn.$key`);
7774284a 1089 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
b8c92cad
EW
1090 $$v = $tmp;
1091 }
1092 }
1093 }
97ae0911 1094 delete @$opts{@config_only} if @config_only;
b8c92cad
EW
1095}
1096
79bb8d88 1097sub extract_metadata {
c1927a85 1098 my $id = shift or return (undef, undef, undef);
3dfab993 1099 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
79bb8d88 1100 \s([a-f\d\-]+)$/x);
e70dc780 1101 if (!defined $rev || !$uuid || !$url) {
79bb8d88 1102 # some of the original repositories I made had
82e5a82f 1103 # identifiers like this:
3dfab993 1104 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
79bb8d88
EW
1105 }
1106 return ($url, $rev, $uuid);
1107}
1108
c1927a85
EW
1109sub cmt_metadata {
1110 return extract_metadata((grep(/^git-svn-id: /,
aef4e921 1111 command(qw/cat-file commit/, shift)))[-1]);
c1927a85
EW
1112}
1113
905f8b7d
EW
1114sub working_head_info {
1115 my ($head, $refs) = @_;
4dbfe2e9 1116 my @args = ('log', '--no-color', '--first-parent');
05b4df31 1117 my ($fh, $ctx) = command_output_pipe(@args, $head);
3dfab993 1118 my $hash;
40cb8f8f 1119 my %max;
3dfab993
SV
1120 while (<$fh>) {
1121 if ( m{^commit ($::sha1)$} ) {
1122 unshift @$refs, $hash if $hash and $refs;
1123 $hash = $1;
1124 next;
1125 }
1126 next unless s{^\s*(git-svn-id:)}{$1};
1127 my ($url, $rev, $uuid) = extract_metadata($_);
13c823fb 1128 if (defined $url && defined $rev) {
40cb8f8f 1129 next if $max{$url} and $max{$url} < $rev;
13c823fb 1130 if (my $gs = Git::SVN->find_by_url($url)) {
060610c5 1131 my $c = $gs->rev_map_get($rev);
b03c7a63 1132 if ($c && $c eq $hash) {
13c823fb
EW
1133 close $fh; # break the pipe
1134 return ($url, $rev, $uuid, $gs);
40cb8f8f 1135 } else {
060610c5 1136 $max{$url} ||= $gs->rev_map_max;
13c823fb
EW
1137 }
1138 }
1139 }
905f8b7d 1140 }
13c823fb
EW
1141 command_close_pipe($fh, $ctx);
1142 (undef, undef, undef, undef);
905f8b7d
EW
1143}
1144
733a65aa
EW
1145sub read_commit_parents {
1146 my ($parents, $c) = @_;
7b02b85a
EW
1147 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1148 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1149 @{$parents->{$c}} = split(/ /, $p);
733a65aa
EW
1150}
1151
1152sub linearize_history {
1153 my ($gs, $refs) = @_;
1154 my %parents;
1155 foreach my $c (@$refs) {
1156 read_commit_parents(\%parents, $c);
1157 }
1158
1159 my @linear_refs;
1160 my %skip = ();
1161 my $last_svn_commit = $gs->last_commit;
1162 foreach my $c (reverse @$refs) {
1163 next if $c eq $last_svn_commit;
1164 last if $skip{$c};
1165
1166 unshift @linear_refs, $c;
1167 $skip{$c} = 1;
1168
1169 # we only want the first parent to diff against for linear
1170 # history, we save the rest to inject when we finalize the
1171 # svn commit
1172 my $fp_a = verify_ref("$c~1");
1173 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1174 if (!$fp_a || !$fp_b) {
1175 die "Commit $c\n",
1176 "has no parent commit, and therefore ",
1177 "nothing to diff against.\n",
1178 "You should be working from a repository ",
1179 "originally created by git-svn\n";
1180 }
1181 if ($fp_a ne $fp_b) {
1182 die "$c~1 = $fp_a, however parsing commit $c ",
1183 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1184 }
1185
1186 foreach my $p (@{$parents{$c}}) {
1187 $skip{$p} = 1;
1188 }
1189 }
1190 (\@linear_refs, \%parents);
1191}
1192
e6fefa92
DK
1193sub find_file_type_and_diff_status {
1194 my ($path) = @_;
1195 return ('dir', '') if $path eq '.';
1196
1197 my $diff_output =
1198 command_oneline(qw(diff --cached --name-status --), $path) || "";
1199 my $diff_status = (split(' ', $diff_output))[0] || "";
1200
1201 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1202
1203 return (undef, undef) if !$diff_status && !$ls_tree;
1204
1205 if ($diff_status eq "A") {
1206 return ("link", $diff_status) if -l $path;
1207 return ("dir", $diff_status) if -d $path;
1208 return ("file", $diff_status);
1209 }
1210
1211 my $mode = (split(' ', $ls_tree))[0] || "";
1212
1213 return ("link", $diff_status) if $mode eq "120000";
1214 return ("dir", $diff_status) if $mode eq "040000";
1215 return ("file", $diff_status);
1216}
1217
b2b3ada7
DK
1218sub md5sum {
1219 my $arg = shift;
1220 my $ref = ref $arg;
1221 my $md5 = Digest::MD5->new();
1222 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1223 $md5->addfile($arg) or croak $!;
1224 } elsif ($ref eq 'SCALAR') {
1225 $md5->add($$arg) or croak $!;
1226 } elsif (!$ref) {
1227 $md5->add($arg) or croak $!;
1228 } else {
1229 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1230 }
1231 return $md5->hexdigest();
1232}
1233
9b981fc6
EW
1234package Git::SVN;
1235use strict;
1236use warnings;
060610c5
EW
1237use Fcntl qw/:DEFAULT :seek/;
1238use constant rev_map_fmt => 'NH40';
ecc712dd 1239use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
62e349d2 1240 $_repack $_repack_flags $_use_svm_props $_head
70ae04e4
AW
1241 $_use_svnsync_props $no_reuse_existing $_minimize_url
1242 $_use_log_author/;
9b981fc6
EW
1243use Carp qw/croak/;
1244use File::Path qw/mkpath/;
373274f9 1245use File::Copy qw/copy/;
9b981fc6
EW
1246use IPC::Open3;
1247
ecc712dd 1248my $_repack_nr;
9b981fc6
EW
1249# properties that we do not log:
1250my %SKIP_PROP;
1251BEGIN {
1252 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1253 svn:special svn:executable
1254 svn:entry:committed-rev
1255 svn:entry:last-author
1256 svn:entry:uuid
1257 svn:entry:committed-date/;
91b03282
EW
1258
1259 # some options are read globally, but can be overridden locally
1260 # per [svn-remote "..."] section. Command-line options will *NOT*
1261 # override options set in an [svn-remote "..."] section
c5f71ad0
SV
1262 no strict 'refs';
1263 for my $option (qw/follow_parent no_metadata use_svm_props
1264 use_svnsync_props/) {
1265 my $key = $option;
91b03282 1266 $key =~ tr/_//d;
c5f71ad0
SV
1267 my $prop = "-$option";
1268 *$option = sub {
1269 my ($self) = @_;
1270 return $self->{$prop} if exists $self->{$prop};
1271 my $k = "svn-remote.$self->{repo_id}.$key";
1272 eval { command_oneline(qw/config --get/, $k) };
1273 if ($@) {
1274 $self->{$prop} = ${"Git::SVN::_$option"};
91b03282 1275 } else {
c5f71ad0
SV
1276 my $v = command_oneline(qw/config --bool/,$k);
1277 $self->{$prop} = $v eq 'false' ? 0 : 1;
91b03282 1278 }
c5f71ad0
SV
1279 return $self->{$prop};
1280 }
91b03282 1281 }
9b981fc6
EW
1282}
1283
373274f9
EW
1284my %LOCKFILES;
1285END { unlink keys %LOCKFILES if %LOCKFILES }
1286
4bb9ed04
EW
1287sub resolve_local_globs {
1288 my ($url, $fetch, $glob_spec) = @_;
1289 return unless defined $glob_spec;
1290 my $ref = $glob_spec->{ref};
1291 my $path = $glob_spec->{path};
1292 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1293 next unless m#^refs/remotes/$ref->{regex}$#;
1294 my $p = $1;
bf655fd7
RE
1295 my $pathname = desanitize_refname($path->full_path($p));
1296 my $refname = desanitize_refname($ref->full_path($p));
4bb9ed04
EW
1297 if (my $existing = $fetch->{$pathname}) {
1298 if ($existing ne $refname) {
1299 die "Refspec conflict:\n",
1300 "existing: refs/remotes/$existing\n",
1301 " globbed: refs/remotes/$refname\n";
1302 }
1303 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
4e9f6cc7 1304 $u =~ s!^\Q$url\E(/|$)!! or die
4bb9ed04
EW
1305 "refs/remotes/$refname: '$url' not found in '$u'\n";
1306 if ($pathname ne $u) {
1307 warn "W: Refspec glob conflict ",
1308 "(ref: refs/remotes/$refname):\n",
1309 "expected path: $pathname\n",
1310 " real path: $u\n",
1311 "Continuing ahead with $u\n";
1312 next;
1313 }
1314 } else {
4bb9ed04
EW
1315 $fetch->{$pathname} = $refname;
1316 }
1317 }
1318}
1319
e98671e5
EW
1320sub parse_revision_argument {
1321 my ($base, $head) = @_;
1322 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1323 return ($base, $head);
1324 }
1325 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1326 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1327 return ($head, $head) if ($::_revision eq 'HEAD');
1328 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1329 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1330 die "revision argument: $::_revision not understood by git-svn\n";
1331}
1332
0af9c9f9 1333sub fetch_all {
4bb9ed04 1334 my ($repo_id, $remotes) = @_;
905f8b7d
EW
1335 if (ref $repo_id) {
1336 my $gs = $repo_id;
1337 $repo_id = undef;
1338 $repo_id = $gs->{repo_id};
1339 }
1340 $remotes ||= read_all_remotes();
7447b4bc
EW
1341 my $remote = $remotes->{$repo_id} or
1342 die "[svn-remote \"$repo_id\"] unknown\n";
e518192f 1343 my $fetch = $remote->{fetch};
7447b4bc 1344 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
e518192f 1345 my (@gs, @globs);
0af9c9f9 1346 my $ra = Git::SVN::Ra->new($url);
26a62d57 1347 my $uuid = $ra->get_uuid;
0af9c9f9 1348 my $head = $ra->get_latest_revnum;
28710f74 1349 my $base = defined $fetch ? $head : 0;
e518192f
EW
1350
1351 # read the max revs for wildcard expansion (branches/*, tags/*)
1352 foreach my $t (qw/branches tags/) {
1353 defined $remote->{$t} or next;
1354 push @globs, $remote->{$t};
93f2689c
EW
1355 my $max_rev = eval { tmp_config(qw/--int --get/,
1356 "svn-remote.$repo_id.${t}-maxRev") };
1357 if (defined $max_rev && ($max_rev < $base)) {
1358 $base = $max_rev;
d6d3346b
EW
1359 } elsif (!defined $max_rev) {
1360 $base = 0;
e518192f
EW
1361 }
1362 }
1363
db03cd24
EW
1364 if ($fetch) {
1365 foreach my $p (sort keys %$fetch) {
1366 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
060610c5 1367 my $lr = $gs->rev_map_max;
db03cd24
EW
1368 if (defined $lr) {
1369 $base = $lr if ($lr < $base);
1370 }
1371 push @gs, $gs;
0af9c9f9 1372 }
0af9c9f9 1373 }
e98671e5
EW
1374
1375 ($base, $head) = parse_revision_argument($base, $head);
e518192f 1376 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
0af9c9f9
EW
1377}
1378
47e39c55
EW
1379sub read_all_remotes {
1380 my $r = {};
8b8fc068 1381 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
47e39c55 1382 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
46cf98ba
EW
1383 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1384 $local_ref =~ s{^/}{};
1385 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
47e39c55
EW
1386 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1387 $r->{$1}->{url} = $2;
4bb9ed04
EW
1388 } elsif (m!^(.+)\.(branches|tags)=
1389 (.*):refs/remotes/(.+)\s*$/!x) {
1390 my ($p, $g) = ($3, $4);
1391 my $rs = $r->{$1}->{$2} = {
e518192f 1392 t => $2,
93f2689c 1393 remote => $1,
4bb9ed04
EW
1394 path => Git::SVN::GlobSpec->new($p),
1395 ref => Git::SVN::GlobSpec->new($g) };
1396 if (length($rs->{ref}->{right}) != 0) {
1397 die "The '*' glob character must be the last ",
1398 "character of '$g'\n";
1399 }
47e39c55
EW
1400 }
1401 }
1402 $r;
1403}
1404
ecc712dd
EW
1405sub init_vars {
1406 if (defined $_repack) {
1407 $_repack = 1000 if ($_repack <= 0);
1408 $_repack_nr = $_repack;
1409 $_repack_flags ||= '-d';
1410 }
1411}
1412
b805b44a 1413sub verify_remotes_sanity {
536c4b09 1414 return unless -d $ENV{GIT_DIR};
b805b44a
EW
1415 my %seen;
1416 foreach (command(qw/config -l/)) {
1417 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1418 if ($seen{$1}) {
1419 die "Remote ref refs/remote/$1 is tracked by",
1420 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1421 "Please resolve this ambiguity in ",
1422 "your git configuration file before ",
1423 "continuing\n";
1424 }
1425 $seen{$1} = $_;
1426 }
1427 }
1428}
1429
47e39c55 1430# we allow more chars than remotes2config.sh...
706587fc
EW
1431sub sanitize_remote_name {
1432 my ($name) = @_;
47e39c55 1433 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
706587fc
EW
1434 $name;
1435}
1436
e6434f87
EW
1437sub find_existing_remote {
1438 my ($url, $remotes) = @_;
befc9adc 1439 return undef if $no_reuse_existing;
e6434f87
EW
1440 my $existing;
1441 foreach my $repo_id (keys %$remotes) {
1442 my $u = $remotes->{$repo_id}->{url} or next;
1443 next if $u ne $url;
1444 $existing = $repo_id;
1445 last;
1446 }
1447 $existing;
1448}
b805b44a 1449
e6434f87 1450sub init_remote_config {
d8115c51 1451 my ($self, $url, $no_write) = @_;
e6434f87
EW
1452 $url =~ s!/+$!!; # strip trailing slash
1453 my $r = read_all_remotes();
1454 my $existing = find_existing_remote($url, $r);
1455 if ($existing) {
e518192f
EW
1456 unless ($no_write) {
1457 print STDERR "Using existing ",
1458 "[svn-remote \"$existing\"]\n";
1459 }
e6434f87 1460 $self->{repo_id} = $existing;
4a1bb4c3 1461 } elsif ($_minimize_url) {
e6434f87
EW
1462 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1463 $existing = find_existing_remote($min_url, $r);
1464 if ($existing) {
e518192f
EW
1465 unless ($no_write) {
1466 print STDERR "Using existing ",
1467 "[svn-remote \"$existing\"]\n";
1468 }
e6434f87
EW
1469 $self->{repo_id} = $existing;
1470 }
1471 if ($min_url ne $url) {
e518192f
EW
1472 unless ($no_write) {
1473 print STDERR "Using higher level of URL: ",
1474 "$url => $min_url\n";
1475 }
e6434f87
EW
1476 my $old_path = $self->{path};
1477 $self->{path} = $url;
4e9f6cc7 1478 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
e6434f87
EW
1479 if (length $old_path) {
1480 $self->{path} .= "/$old_path";
1481 }
1482 $url = $min_url;
1483 }
1484 }
1485 my $orig_url;
1486 if (!$existing) {
b805b44a 1487 # verify that we aren't overwriting anything:
e6434f87 1488 $orig_url = eval {
706587fc 1489 command_oneline('config', '--get',
e6434f87 1490 "svn-remote.$self->{repo_id}.url")
706587fc 1491 };
b805b44a 1492 if ($orig_url && ($orig_url ne $url)) {
e6434f87 1493 die "svn-remote.$self->{repo_id}.url already set: ",
b805b44a
EW
1494 "$orig_url\nwanted to set to: $url\n";
1495 }
9b981fc6 1496 }
e6434f87
EW
1497 my ($xrepo_id, $xpath) = find_ref($self->refname);
1498 if (defined $xpath) {
1499 die "svn-remote.$xrepo_id.fetch already set to track ",
1500 "$xpath:refs/remotes/", $self->refname, "\n";
1501 }
d8115c51
EW
1502 unless ($no_write) {
1503 command_noisy('config',
1504 "svn-remote.$self->{repo_id}.url", $url);
46cf98ba 1505 $self->{path} =~ s{^/}{};
d8115c51
EW
1506 command_noisy('config', '--add',
1507 "svn-remote.$self->{repo_id}.fetch",
1508 "$self->{path}:".$self->refname);
1509 }
9b981fc6 1510 $self->{url} = $url;
e6434f87
EW
1511}
1512
a8ae2623
EW
1513sub find_by_url { # repos_root and, path are optional
1514 my ($class, $full_url, $repos_root, $path) = @_;
56973d20 1515
1a97a506 1516 return undef unless defined $full_url;
56973d20
AR
1517 remove_username($full_url);
1518 remove_username($repos_root) if defined $repos_root;
a8ae2623
EW
1519 my $remotes = read_all_remotes();
1520 if (defined $full_url && defined $repos_root && !defined $path) {
1521 $path = $full_url;
1522 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1523 }
1524 foreach my $repo_id (keys %$remotes) {
1525 my $u = $remotes->{$repo_id}->{url} or next;
56973d20 1526 remove_username($u);
a8ae2623
EW
1527 next if defined $repos_root && $repos_root ne $u;
1528
1529 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1530 foreach (qw/branches tags/) {
1531 resolve_local_globs($u, $fetch,
1532 $remotes->{$repo_id}->{$_});
1533 }
1534 my $p = $path;
1535 unless (defined $p) {
1536 $p = $full_url;
1537 $p =~ s#^\Q$u\E(?:/|$)## or next;
1538 }
1539 foreach my $f (keys %$fetch) {
1540 next if $f ne $p;
1541 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1542 }
1543 }
1544 undef;
1545}
1546
e6434f87 1547sub init {
d8115c51 1548 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
e6434f87
EW
1549 my $self = _new($class, $repo_id, $ref_id, $path);
1550 if (defined $url) {
d8115c51 1551 $self->init_remote_config($url, $no_write);
e6434f87 1552 }
9b981fc6
EW
1553 $self;
1554}
1555
706587fc
EW
1556sub find_ref {
1557 my ($ref_id) = @_;
1558 foreach (command(qw/config -l/)) {
1559 next unless m!^svn-remote\.(.+)\.fetch=
1560 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1561 my ($repo_id, $path, $ref) = ($1, $2, $3);
1562 if ($ref eq $ref_id) {
1563 $path = '' if ($path =~ m#^\./?#);
1564 return ($repo_id, $path);
1565 }
1566 }
1567 (undef, undef, undef);
1568}
1569
9b981fc6 1570sub new {
706587fc
EW
1571 my ($class, $ref_id, $repo_id, $path) = @_;
1572 if (defined $ref_id && !defined $repo_id && !defined $path) {
1573 ($repo_id, $path) = find_ref($ref_id);
1574 if (!defined $repo_id) {
1575 die "Could not find a \"svn-remote.*.fetch\" key ",
1576 "in the repository configuration matching: ",
1577 "refs/remotes/$ref_id\n";
1578 }
1579 }
1580 my $self = _new($class, $repo_id, $ref_id, $path);
8b8fc068
EW
1581 if (!defined $self->{path} || !length $self->{path}) {
1582 my $fetch = command_oneline('config', '--get',
1583 "svn-remote.$repo_id.fetch",
1584 ":refs/remotes/$ref_id\$") or
1585 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1586 "\":refs/remotes/$ref_id\$\" in config\n";
1587 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1588 }
706587fc
EW
1589 $self->{url} = command_oneline('config', '--get',
1590 "svn-remote.$repo_id.url") or
1591 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
d6d3346b 1592 $self->rebuild;
9b981fc6
EW
1593 $self;
1594}
1595
bf655fd7
RE
1596sub refname {
1597 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1598
1599 # It cannot end with a slash /, we'll throw up on this because
1600 # SVN can't have directories with a slash in their name, either:
1601 if ($refname =~ m{/$}) {
1602 die "ref: '$refname' ends with a trailing slash, this is ",
1603 "not permitted by git nor Subversion\n";
1604 }
1605
1606 # It cannot have ASCII control character space, tilde ~, caret ^,
1607 # colon :, question-mark ?, asterisk *, space, or open bracket [
1608 # anywhere.
1609 #
1610 # Additionally, % must be escaped because it is used for escaping
1611 # and we want our escaped refname to be reversible
1612 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1613
1614 # no slash-separated component can begin with a dot .
1615 # /.* becomes /%2E*
1616 $refname =~ s{/\.}{/%2E}g;
1617
1618 # It cannot have two consecutive dots .. anywhere
1619 # .. becomes %2E%2E
1620 $refname =~ s{\.\.}{%2E%2E}g;
1621
1622 return $refname;
1623}
1624
1625sub desanitize_refname {
1626 my ($refname) = @_;
1627 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1628 return $refname;
1629}
9b981fc6 1630
26a62d57
EW
1631sub svm_uuid {
1632 my ($self) = @_;
1633 return $self->{svm}->{uuid} if $self->svm;
1634 $self->ra;
1635 unless ($self->{svm}) {
1636 die "SVM UUID not cached, and reading remotely failed\n";
1637 }
1638 $self->{svm}->{uuid};
1639}
8a49ee97 1640
26a62d57
EW
1641sub svm {
1642 my ($self) = @_;
1643 return $self->{svm} if $self->{svm};
1644 my $svm;
8a49ee97
EW
1645 # see if we have it in our config, first:
1646 eval {
26a62d57
EW
1647 my $section = "svn-remote.$self->{repo_id}";
1648 $svm = {
93f2689c
EW
1649 source => tmp_config('--get', "$section.svm-source"),
1650 uuid => tmp_config('--get', "$section.svm-uuid"),
befc9adc 1651 replace => tmp_config('--get', "$section.svm-replace"),
8a49ee97
EW
1652 }
1653 };
befc9adc
EW
1654 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1655 $self->{svm} = $svm;
1656 }
26a62d57
EW
1657 $self->{svm};
1658}
1659
1660sub _set_svm_vars {
1661 my ($self, $ra) = @_;
db03cd24
EW
1662 return $ra if $self->svm;
1663
1664 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
befc9adc 1665 "(svm:source, svm:uuid) ",
db03cd24
EW
1666 "from the following URLs:\n" );
1667 sub read_svm_props {
befc9adc
EW
1668 my ($self, $ra, $path, $r) = @_;
1669 my $props = ($ra->get_dir($path, $r))[2];
db03cd24 1670 my $src = $props->{'svm:source'};
db03cd24 1671 my $uuid = $props->{'svm:uuid'};
befc9adc 1672 return undef if (!$src || !$uuid);
26a62d57 1673
befc9adc 1674 chomp($src, $uuid);
26a62d57 1675
db03cd24
EW
1676 $uuid =~ m{^[0-9a-f\-]{30,}$}
1677 or die "doesn't look right - svm:uuid is '$uuid'\n";
befc9adc
EW
1678
1679 # the '!' is used to mark the repos_root!/relative/path
1680 $src =~ s{/?!/?}{/};
db03cd24 1681 $src =~ s{/+$}{}; # no trailing slashes please
befc9adc 1682 # username is of no interest
8a49ee97 1683 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
8a49ee97 1684
befc9adc
EW
1685 my $replace = $ra->{url};
1686 $replace .= "/$path" if length $path;
1687
db03cd24 1688 my $section = "svn-remote.$self->{repo_id}";
befc9adc
EW
1689 tmp_config("$section.svm-source", $src);
1690 tmp_config("$section.svm-replace", $replace);
1691 tmp_config("$section.svm-uuid", $uuid);
1692 $self->{svm} = {
1693 source => $src,
1694 uuid => $uuid,
1695 replace => $replace
1696 };
db03cd24
EW
1697 }
1698
1699 my $r = $ra->get_latest_revnum;
1700 my $path = $self->{path};
befc9adc 1701 my %tried;
db03cd24 1702 while (length $path) {
befc9adc
EW
1703 unless ($tried{"$self->{url}/$path"}) {
1704 return $ra if $self->read_svm_props($ra, $path, $r);
1705 $tried{"$self->{url}/$path"} = 1;
db03cd24 1706 }
befc9adc 1707 $path =~ s#/?[^/]+$##;
8a49ee97 1708 }
befc9adc
EW
1709 die "Path: '$path' should be ''\n" if $path ne '';
1710 return $ra if $self->read_svm_props($ra, $path, $r);
1711 $tried{"$self->{url}/$path"} = 1;
db03cd24
EW
1712
1713 if ($ra->{repos_root} eq $self->{url}) {
befc9adc 1714 die @err, (map { " $_\n" } keys %tried), "\n";
db03cd24
EW
1715 }
1716
1717 # nope, make sure we're connected to the repository root:
1718 my $ok;
1719 my @tried_b;
1720 $path = $ra->{svn_path};
db03cd24
EW
1721 $ra = Git::SVN::Ra->new($ra->{repos_root});
1722 while (length $path) {
befc9adc
EW
1723 unless ($tried{"$ra->{url}/$path"}) {
1724 $ok = $self->read_svm_props($ra, $path, $r);
1725 last if $ok;
1726 $tried{"$ra->{url}/$path"} = 1;
1727 }
1728 $path =~ s#/?[^/]+$##;
db03cd24 1729 }
befc9adc
EW
1730 die "Path: '$path' should be ''\n" if $path ne '';
1731 $ok ||= $self->read_svm_props($ra, $path, $r);
1732 $tried{"$ra->{url}/$path"} = 1;
db03cd24 1733 if (!$ok) {
befc9adc 1734 die @err, (map { " $_\n" } keys %tried), "\n";
db03cd24
EW
1735 }
1736 Git::SVN::Ra->new($self->{url});
8a49ee97
EW
1737}
1738
62e349d2
EW
1739sub svnsync {
1740 my ($self) = @_;
1741 return $self->{svnsync} if $self->{svnsync};
1742
1743 if ($self->no_metadata) {
1744 die "Can't have both 'noMetadata' and ",
1745 "'useSvnsyncProps' options set!\n";
1746 }
1747 if ($self->rewrite_root) {
1748 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1749 "options set!\n";
1750 }
1751
1752 my $svnsync;
1753 # see if we have it in our config, first:
1754 eval {
1755 my $section = "svn-remote.$self->{repo_id}";
1756 $svnsync = {
1757 url => tmp_config('--get', "$section.svnsync-url"),
1758 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1759 }
1760 };
1761 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1762 return $self->{svnsync} = $svnsync;
1763 }
1764
1765 my $err = "useSvnsyncProps set, but failed to read " .
1766 "svnsync property: svn:sync-from-";
1767 my $rp = $self->ra->rev_proplist(0);
1768
1769 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1770 $url =~ m{^[a-z\+]+://} or
1771 die "doesn't look right - svn:sync-from-url is '$url'\n";
1772
1773 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1774 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1775 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1776
1777 my $section = "svn-remote.$self->{repo_id}";
1778 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1779 tmp_config('--add', "$section.svnsync-url", $url);
1780 return $self->{svnsync} = { url => $url, uuid => $uuid };
1781}
1782
26a62d57
EW
1783# this allows us to memoize our SVN::Ra UUID locally and avoid a
1784# remote lookup (useful for 'git svn log').
1785sub ra_uuid {
1786 my ($self) = @_;
1787 unless ($self->{ra_uuid}) {
1788 my $key = "svn-remote.$self->{repo_id}.uuid";
1789 my $uuid = eval { tmp_config('--get', $key) };
1790 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1791 $self->{ra_uuid} = $uuid;
1792 } else {
1793 die "ra_uuid called without URL\n" unless $self->{url};
1794 $self->{ra_uuid} = $self->ra->get_uuid;
1795 tmp_config('--add', $key, $self->{ra_uuid});
1796 }
1797 }
1798 $self->{ra_uuid};
1799}
1800
a5460eb7
EW
1801sub _set_repos_root {
1802 my ($self, $repos_root) = @_;
1803 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1804 $repos_root ||= $self->ra->{repos_root};
1805 tmp_config($k, $repos_root);
1806 $repos_root;
1807}
1808
1809sub repos_root {
1810 my ($self) = @_;
1811 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1812 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1813}
1814
9b981fc6
EW
1815sub ra {
1816 my ($self) = shift;
8a49ee97 1817 my $ra = Git::SVN::Ra->new($self->{url});
a5460eb7 1818 $self->_set_repos_root($ra->{repos_root});
91b03282
EW
1819 if ($self->use_svm_props && !$self->{svm}) {
1820 if ($self->no_metadata) {
97ae0911
EW
1821 die "Can't have both 'noMetadata' and ",
1822 "'useSvmProps' options set!\n";
62e349d2
EW
1823 } elsif ($self->use_svnsync_props) {
1824 die "Can't have both 'useSvnsyncProps' and ",
1825 "'useSvmProps' options set!\n";
91b03282 1826 }
26a62d57 1827 $ra = $self->_set_svm_vars($ra);
8a49ee97
EW
1828 $self->{-want_revprops} = 1;
1829 }
1830 $ra;
9b981fc6
EW
1831}
1832
15710b6f
EW
1833sub rel_path {
1834 my ($self) = @_;
1835 my $repos_root = $self->ra->{repos_root};
1836 return $self->{path} if ($self->{url} eq $repos_root);
0b59451c
EW
1837 my $url = $self->{url} .
1838 (length $self->{path} ? "/$self->{path}" : $self->{path});
1839 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1840 $url;
15710b6f
EW
1841}
1842
01bdab84
BS
1843# prop_walk(PATH, REV, SUB)
1844# -------------------------
1845# Recursively traverse PATH at revision REV and invoke SUB for each
1846# directory that contains a SVN property. SUB will be invoked as
1847# follows: &SUB(gs, path, props); where `gs' is this instance of
1848# Git::SVN, `path' the path to the directory where the properties
1849# `props' were found. The `path' will be relative to point of checkout,
1850# that is, if url://repo/trunk is the current Git branch, and that
1851# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1852# as `path' (note the trailing `/').
1853sub prop_walk {
1854 my ($self, $path, $rev, $sub) = @_;
1855
1856 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1857 $path =~ s#^/*#/#g;
9b981fc6 1858 my $p = $path;
01bdab84
BS
1859 # Strip the irrelevant part of the path.
1860 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1861 # Ensure the path is terminated by a `/'.
1862 $p =~ s#/*$#/#;
1863
1864 # The properties contain all the internal SVN stuff nobody
1865 # (usually) cares about.
1866 my $interesting_props = 0;
1867 foreach (keys %{$props}) {
1868 # If it doesn't start with `svn:', it must be a
1869 # user-defined property.
1870 ++$interesting_props and next if $_ !~ /^svn:/;
1871 # FIXME: Fragile, if SVN adds new public properties,
1872 # this needs to be updated.
1873 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1874 |eol-style|mime-type
1875 |externals|needs-lock)$/x;
1876 }
1877 &$sub($self, $p, $props) if $interesting_props;
1878
9b981fc6 1879 foreach (sort keys %$dirent) {
0dc03d6a 1880 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
01bdab84 1881 $self->prop_walk($path . '/' . $_, $rev, $sub);
9b981fc6
EW
1882 }
1883}
1884
3ebe8df7
EW
1885sub last_rev { ($_[0]->last_rev_commit)[0] }
1886sub last_commit { ($_[0]->last_rev_commit)[1] }
1887
9b981fc6
EW
1888# returns the newest SVN revision number and newest commit SHA1
1889sub last_rev_commit {
1890 my ($self) = @_;
1891 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1892 return ($self->{last_rev}, $self->{last_commit});
1893 }
d2866f9e 1894 my $c = ::verify_ref($self->refname.'^0');
91b03282 1895 if ($c && !$self->use_svm_props && !$self->no_metadata) {
d2866f9e 1896 my $rev = (::cmt_metadata($c))[1];
9b981fc6
EW
1897 if (defined $rev) {
1898 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1899 return ($rev, $c);
1900 }
1901 }
060610c5
EW
1902 my $map_path = $self->map_path;
1903 unless (-e $map_path) {
26a62d57
EW
1904 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1905 return (undef, undef);
1906 }
060610c5
EW
1907 my ($rev, $commit) = $self->rev_map_max;
1908 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1909 return ($rev, $commit);
9b981fc6
EW
1910}
1911
3ebe8df7
EW
1912sub get_fetch_range {
1913 my ($self, $min, $max) = @_;
1914 $max ||= $self->ra->get_latest_revnum;
060610c5 1915 $min ||= $self->rev_map_max;
3ebe8df7 1916 (++$min, $max);
9b981fc6
EW
1917}
1918
8a49ee97 1919sub tmp_config {
93f2689c 1920 my (@args) = @_;
b7e5348c
EW
1921 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1922 my $config = "$ENV{GIT_DIR}/svn/.metadata";
38570a47 1923 if (! -f $config && -f $old_def_config) {
b7e5348c
EW
1924 rename $old_def_config, $config or
1925 die "Failed rename $old_def_config => $config: $!\n";
1926 }
8a49ee97 1927 my $old_config = $ENV{GIT_CONFIG};
93f2689c 1928 $ENV{GIT_CONFIG} = $config;
8a49ee97 1929 $@ = undef;
b4d57e5e
EW
1930 my @ret = eval {
1931 unless (-f $config) {
1932 mkfile($config);
1933 open my $fh, '>', $config or
1934 die "Can't open $config: $!\n";
1935 print $fh "; This file is used internally by ",
1936 "git-svn\n" or die
1937 "Couldn't write to $config: $!\n";
1938 print $fh "; You should not have to edit it\n" or
1939 die "Couldn't write to $config: $!\n";
1940 close $fh or die "Couldn't close $config: $!\n";
1941 }
1942 command('config', @args);
1943 };
8a49ee97
EW
1944 my $err = $@;
1945 if (defined $old_config) {
1946 $ENV{GIT_CONFIG} = $old_config;
1947 } else {
1948 delete $ENV{GIT_CONFIG};
1949 }
1950 die $err if $err;
1951 wantarray ? @ret : $ret[0];
1952}
1953
9b981fc6
EW
1954sub tmp_index_do {
1955 my ($self, $sub) = @_;
1956 my $old_index = $ENV{GIT_INDEX_FILE};
1957 $ENV{GIT_INDEX_FILE} = $self->{index};
8a49ee97 1958 $@ = undef;
b4d57e5e
EW
1959 my @ret = eval {
1960 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1961 mkpath([$dir]) unless -d $dir;
1962 &$sub;
1963 };
8a49ee97
EW
1964 my $err = $@;
1965 if (defined $old_index) {
9b981fc6
EW
1966 $ENV{GIT_INDEX_FILE} = $old_index;
1967 } else {
1968 delete $ENV{GIT_INDEX_FILE};
1969 }
8a49ee97 1970 die $err if $err;
9b981fc6
EW
1971 wantarray ? @ret : $ret[0];
1972}
1973
1974sub assert_index_clean {
1975 my ($self, $treeish) = @_;
1976
1977 $self->tmp_index_do(sub {
1978 command_noisy('read-tree', $treeish) unless -e $self->{index};
1979 my $x = command_oneline('write-tree');
1980 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1981 /^tree ($::sha1)/mo);
e8d120bd
EW
1982 return if $y eq $x;
1983
1984 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1985 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1986 command_noisy('read-tree', $treeish);
9b981fc6
EW
1987 $x = command_oneline('write-tree');
1988 if ($y ne $x) {
1989 ::fatal "trees ($treeish) $y != $x\n",
207f1a75 1990 "Something is seriously wrong...";
9b981fc6
EW
1991 }
1992 });
1993}
1994
1995sub get_commit_parents {
0af9c9f9 1996 my ($self, $log_entry) = @_;
9b981fc6 1997 my (%seen, @ret, @tmp);
0af9c9f9
EW
1998 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1999 if (my $ip = $self->{inject_parents}) {
2000 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2001 push @tmp, $commit;
9b981fc6
EW
2002 }
2003 }
d2866f9e 2004 if (my $cur = ::verify_ref($self->refname.'^0')) {
9b981fc6
EW
2005 push @tmp, $cur;
2006 }
733a65aa
EW
2007 if (my $ipd = $self->{inject_parents_dcommit}) {
2008 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2009 push @tmp, @$commit;
2010 }
2011 }
44320b9e 2012 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
9b981fc6
EW
2013 while (my $p = shift @tmp) {
2014 next if $seen{$p};
2015 $seen{$p} = 1;
2016 push @ret, $p;
2017 # MAXPARENT is defined to 16 in commit-tree.c:
2018 last if @ret >= 16;
2019 }
2020 if (@tmp) {
44320b9e 2021 die "r$log_entry->{revision}: No room for parents:\n\t",
9b981fc6
EW
2022 join("\n\t", @tmp), "\n";
2023 }
2024 @ret;
2025}
2026
aea736cc
EW
2027sub rewrite_root {
2028 my ($self) = @_;
2029 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2030 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2031 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2032 if ($rwr) {
2033 $rwr =~ s#/+$##;
2034 if ($rwr !~ m#^[a-z\+]+://#) {
2035 die "$rwr is not a valid URL (key: $k)\n";
2036 }
2037 }
2038 $self->{-rewrite_root} = $rwr;
2039}
2040
2041sub metadata_url {
2042 my ($self) = @_;
2043 ($self->rewrite_root || $self->{url}) .
2044 (length $self->{path} ? '/' . $self->{path} : '');
2045}
2046
706587fc 2047sub full_url {
9b981fc6 2048 my ($self) = @_;
5d3b7cd5 2049 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
9b981fc6
EW
2050}
2051
2052sub do_git_commit {
0af9c9f9 2053 my ($self, $log_entry) = @_;
8a603774
EW
2054 my $lr = $self->last_rev;
2055 if (defined $lr && $lr >= $log_entry->{revision}) {
2056 die "Last fetched revision of ", $self->refname,
2057 " was r$lr, but we are about to fetch: ",
2058 "r$log_entry->{revision}!\n";
2059 }
060610c5 2060 if (my $c = $self->rev_map_get($log_entry->{revision})) {
44320b9e 2061 croak "$log_entry->{revision} = $c already exists! ",
9b981fc6
EW
2062 "Why are we refetching it?\n";
2063 }
70ae04e4
AW
2064 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2065 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
44320b9e 2066 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
9b981fc6 2067
70ae04e4
AW
2068 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2069 ? $log_entry->{commit_name}
2070 : $log_entry->{name};
2071 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2072 ? $log_entry->{commit_email}
2073 : $log_entry->{email};
2074
44320b9e 2075 my $tree = $log_entry->{tree};
9b981fc6
EW
2076 if (!defined $tree) {
2077 $tree = $self->tmp_index_do(sub {
2078 command_oneline('write-tree') });
2079 }
2080 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2081
2082 my @exec = ('git-commit-tree', $tree);
0af9c9f9 2083 foreach ($self->get_commit_parents($log_entry)) {
9b981fc6
EW
2084 push @exec, '-p', $_;
2085 }
2086 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2087 or croak $!;
44320b9e 2088 print $msg_fh $log_entry->{log} or croak $!;
91b03282 2089 unless ($self->no_metadata) {
8a49ee97
EW
2090 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2091 or croak $!;
9760adcc 2092 }
9b981fc6
EW
2093 $msg_fh->flush == 0 or croak $!;
2094 close $msg_fh or croak $!;
2095 chomp(my $commit = do { local $/; <$out_fh> });
2096 close $out_fh or croak $!;
2097 waitpid $pid, 0;
2098 croak $? if $?;
2099 if ($commit !~ /^$::sha1$/o) {
2100 die "Failed to commit, invalid sha1: $commit\n";
2101 }
2102
060610c5 2103 $self->rev_map_set($log_entry->{revision}, $commit, 1);
9b981fc6 2104
44320b9e 2105 $self->{last_rev} = $log_entry->{revision};
9b981fc6 2106 $self->{last_commit} = $commit;
8a49ee97
EW
2107 print "r$log_entry->{revision}";
2108 if (defined $log_entry->{svm_revision}) {
2109 print " (\@$log_entry->{svm_revision})";
060610c5 2110 $self->rev_map_set($log_entry->{svm_revision}, $commit,
26a62d57 2111 0, $self->svm_uuid);
8a49ee97
EW
2112 }
2113 print " = $commit ($self->{ref_id})\n";
ecc712dd
EW
2114 if (defined $_repack && (--$_repack_nr == 0)) {
2115 $_repack_nr = $_repack;
2116 # repack doesn't use any arguments with spaces in them, does it?
2117 print "Running git repack $_repack_flags ...\n";
2118 command_noisy('repack', split(/\s+/, $_repack_flags));
2119 print "Done repacking\n";
2120 }
9b981fc6
EW
2121 return $commit;
2122}
2123
fbcc1737
EW
2124sub match_paths {
2125 my ($self, $paths, $r) = @_;
4e9f6cc7 2126 return 1 if $self->{path} eq '';
d542aedb
EW
2127 if (my $path = $paths->{"/$self->{path}"}) {
2128 return ($path->{action} eq 'D') ? 0 : 1;
2129 }
4e9f6cc7 2130 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
fbcc1737
EW
2131 if (grep /$self->{path_regex}/, keys %$paths) {
2132 return 1;
2133 }
2134 my $c = '';
2135 foreach (split m#/#, $self->{path}) {
2136 $c .= "/$_";
74a81227
EW
2137 next unless ($paths->{$c} &&
2138 ($paths->{$c}->{action} =~ /^[AR]$/));
e518192f
EW
2139 if ($self->ra->check_path($self->{path}, $r) ==
2140 $SVN::Node::dir) {
fbcc1737
EW
2141 return 1;
2142 }
2143 }
2144 return 0;
2145}
2146
15710b6f
EW
2147sub find_parent_branch {
2148 my ($self, $paths, $rev) = @_;
91b03282 2149 return undef unless $self->follow_parent;
e5a0b240 2150 unless (defined $paths) {
c7eba716
EW
2151 my $err_handler = $SVN::Error::handler;
2152 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
d4eff2bd
EW
2153 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2154 $paths =
2155 Git::SVN::Ra::dup_changed_paths($_[0]) });
c7eba716 2156 $SVN::Error::handler = $err_handler;
e5a0b240
EW
2157 }
2158 return undef unless defined $paths;
15710b6f
EW
2159
2160 # look for a parent from another branch:
7f578c55
EW
2161 my @b_path_components = split m#/#, $self->rel_path;
2162 my @a_path_components;
2163 my $i;
2164 while (@b_path_components) {
2165 $i = $paths->{'/'.join('/', @b_path_components)};
74a81227 2166 last if $i && defined $i->{copyfrom_path};
7f578c55
EW
2167 unshift(@a_path_components, pop(@b_path_components));
2168 }
74a81227
EW
2169 return undef unless defined $i && defined $i->{copyfrom_path};
2170 my $branch_from = $i->{copyfrom_path};
7f578c55
EW
2171 if (@a_path_components) {
2172 print STDERR "branch_from: $branch_from => ";
2173 $branch_from .= '/'.join('/', @a_path_components);
2174 print STDERR $branch_from, "\n";
2175 }
3ebe8df7 2176 my $r = $i->{copyfrom_rev};
15710b6f
EW
2177 my $repos_root = $self->ra->{repos_root};
2178 my $url = $self->ra->{url};
2179 my $new_url = $repos_root . $branch_from;
2180 print STDERR "Found possible branch point: ",
2181 "$new_url => ", $self->full_url, ", $r\n";
2182 $branch_from =~ s#^/##;
a8ae2623 2183 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
15710b6f 2184 unless ($gs) {
ce2a0f2f
EW
2185 my $ref_id = $self->{ref_id};
2186 $ref_id =~ s/\@\d+$//;
2187 $ref_id .= "\@$r";
15710b6f
EW
2188 # just grow a tail if we're not unique enough :x
2189 $ref_id .= '-' while find_ref($ref_id);
ce2a0f2f 2190 print STDERR "Initializing parent: $ref_id\n";
d8115c51 2191 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
15710b6f
EW
2192 }
2193 my ($r0, $parent) = $gs->find_rev_before($r, 1);
91b03282 2194 if (!defined $r0 || !defined $parent) {
d627de6b
EW
2195 my ($base, $head) = parse_revision_argument(0, $r);
2196 if ($base <= $r) {
2197 $gs->fetch($base, $r);
2198 }
15710b6f
EW
2199 ($r0, $parent) = $gs->last_rev_commit;
2200 }
ef70de96 2201 if (defined $r0 && defined $parent) {
15710b6f 2202 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
15710b6f
EW
2203 my $ed;
2204 if ($self->ra->can_do_switch) {
2e5e2480 2205 $self->assert_index_clean($parent);
8b8fc068 2206 print STDERR "Following parent with do_switch\n";
15710b6f 2207 # do_switch works with svn/trunk >= r22312, but that
2b27f6c8 2208 # is not included with SVN 1.4.3 (the latest version
15710b6f
EW
2209 # at the moment), so we can't rely on it
2210 $self->{last_commit} = $parent;
2211 $ed = SVN::Git::Fetcher->new($self);
8a603774 2212 $gs->ra->gs_do_switch($r0, $rev, $gs,
15710b6f
EW
2213 $self->full_url, $ed)
2214 or die "SVN connection failed somewhere...\n";
9ff74e95
SW
2215 } elsif ($self->ra->trees_match($new_url, $r0,
2216 $self->full_url, $rev)) {
2217 print STDERR "Trees match:\n",
2218 " $new_url\@$r0\n",
2219 " ${\$self->full_url}\@$rev\n",
2220 "Following parent with no changes\n";
2221 $self->tmp_index_do(sub {
2222 command_noisy('read-tree', $parent);
2223 });
2224 $self->{last_commit} = $parent;
15710b6f 2225 } else {
8b8fc068 2226 print STDERR "Following parent with do_update\n";
15710b6f 2227 $ed = SVN::Git::Fetcher->new($self);
8a603774 2228 $self->ra->gs_do_update($rev, $rev, $self, $ed)
15710b6f
EW
2229 or die "SVN connection failed somewhere...\n";
2230 }
f7c3fc4a 2231 print STDERR "Successfully followed parent\n";
15710b6f
EW
2232 return $self->make_log_entry($rev, [$parent], $ed);
2233 }
15710b6f
EW
2234 return undef;
2235}
2236
9b981fc6 2237sub do_fetch {
706587fc 2238 my ($self, $paths, $rev) = @_;
15710b6f 2239 my $ed;
9b981fc6 2240 my ($last_rev, @parents);
b9dffd8c
EW
2241 if (my $lc = $self->last_commit) {
2242 # we can have a branch that was deleted, then re-added
2243 # under the same name but copied from another path, in
2244 # which case we'll have multiple parents (we don't
2245 # want to break the original ref, nor lose copypath info):
2246 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2247 push @{$log_entry->{parents}}, $lc;
2248 return $log_entry;
2249 }
15710b6f 2250 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 2251 $last_rev = $self->{last_rev};
b9dffd8c
EW
2252 $ed->{c} = $lc;
2253 @parents = ($lc);
9b981fc6
EW
2254 } else {
2255 $last_rev = $rev;
15710b6f
EW
2256 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2257 return $log_entry;
2258 }
2259 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 2260 }
8a603774 2261 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
9b981fc6
EW
2262 die "SVN connection failed somewhere...\n";
2263 }
2264 $self->make_log_entry($rev, \@parents, $ed);
2265}
2266
97f6987a
EW
2267sub get_untracked {
2268 my ($self, $ed) = @_;
2269 my @out;
2270 my $h = $ed->{empty};
9b981fc6
EW
2271 foreach (sort keys %$h) {
2272 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
97f6987a 2273 push @out, " $act: " . uri_encode($_);
9b981fc6
EW
2274 warn "W: $act: $_\n";
2275 }
2276 foreach my $t (qw/dir_prop file_prop/) {
97f6987a 2277 $h = $ed->{$t} or next;
9b981fc6
EW
2278 foreach my $path (sort keys %$h) {
2279 my $ppath = $path eq '' ? '.' : $path;
2280 foreach my $prop (sort keys %{$h->{$path}}) {
1ce255dc 2281 next if $SKIP_PROP{$prop};
9b981fc6 2282 my $v = $h->{$path}->{$prop};
97f6987a
EW
2283 my $t_ppath_prop = "$t: " .
2284 uri_encode($ppath) . ' ' .
2285 uri_encode($prop);
9b981fc6 2286 if (defined $v) {
97f6987a
EW
2287 push @out, " +$t_ppath_prop " .
2288 uri_encode($v);
9b981fc6 2289 } else {
97f6987a 2290 push @out, " -$t_ppath_prop";
9b981fc6
EW
2291 }
2292 }
2293 }
2294 }
2295 foreach my $t (qw/absent_file absent_directory/) {
97f6987a 2296 $h = $ed->{$t} or next;
9b981fc6
EW
2297 foreach my $parent (sort keys %$h) {
2298 foreach my $path (sort @{$h->{$parent}}) {
97f6987a
EW
2299 push @out, " $t: " .
2300 uri_encode("$parent/$path");
9b981fc6
EW
2301 warn "W: $t: $parent/$path ",
2302 "Insufficient permissions?\n";
2303 }
2304 }
2305 }
97f6987a 2306 \@out;
9b981fc6
EW
2307}
2308
1c8443b0
EW
2309sub parse_svn_date {
2310 my $date = shift || return '+0000 1970-01-01 00:00:00';
2311 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2312 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2313 croak "Unable to parse date: $date\n";
2314 "+0000 $Y-$m-$d $H:$M:$S";
2315}
2316
2317sub check_author {
2318 my ($author) = @_;
2319 if (!defined $author || length $author == 0) {
2320 $author = '(no author)';
2321 }
2322 if (defined $::_authors && ! defined $::users{$author}) {
2323 die "Author: $author not defined in $::_authors file\n";
2324 }
2325 $author;
2326}
2327
9b981fc6 2328sub make_log_entry {
97f6987a
EW
2329 my ($self, $rev, $parents, $ed) = @_;
2330 my $untracked = $self->get_untracked($ed);
2331
9b981fc6 2332 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
97f6987a
EW
2333 print $un "r$rev\n" or croak $!;
2334 print $un $_, "\n" foreach @$untracked;
2335 my %log_entry = ( parents => $parents || [], revision => $rev,
2336 log => '');
fbcc1737 2337
8a49ee97 2338 my $headrev;
fbcc1737 2339 my $logged = delete $self->{logged_rev_props};
8a49ee97 2340 if (!$logged || $self->{-want_revprops}) {
fbcc1737
EW
2341 my $rp = $self->ra->rev_proplist($rev);
2342 foreach (sort keys %$rp) {
2343 my $v = $rp->{$_};
2344 if (/^svn:(author|date|log)$/) {
2345 $log_entry{$1} = $v;
8a49ee97
EW
2346 } elsif ($_ eq 'svm:headrev') {
2347 $headrev = $v;
fbcc1737
EW
2348 } else {
2349 print $un " rev_prop: ", uri_encode($_), ' ',
2350 uri_encode($v), "\n";
2351 }
9b981fc6 2352 }
fbcc1737
EW
2353 } else {
2354 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
9b981fc6
EW
2355 }
2356 close $un or croak $!;
97f6987a 2357
9b981fc6 2358 $log_entry{date} = parse_svn_date($log_entry{date});
9b981fc6 2359 $log_entry{log} .= "\n";
db03cd24
EW
2360 my $author = $log_entry{author} = check_author($log_entry{author});
2361 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
70ae04e4
AW
2362 : ($author, undef);
2363
2364 my ($commit_name, $commit_email) = ($name, $email);
2365 if ($_use_log_author) {
2366 if ($log_entry{log} =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
2367 ($name, $email) = ($1, $2);
2368 } elsif ($log_entry{log} =~
2369 /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
2370 ($name, $email) = ($1, $2);
2371 }
2372 }
91b03282 2373 if (defined $headrev && $self->use_svm_props) {
aea736cc
EW
2374 if ($self->rewrite_root) {
2375 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2376 "options set!\n";
2377 }
8a49ee97 2378 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
befc9adc
EW
2379 # we don't want "SVM: initializing mirror for junk" ...
2380 return undef if $r == 0;
2381 my $svm = $self->svm;
2382 if ($uuid ne $svm->{uuid}) {
8a49ee97 2383 die "UUID mismatch on SVM path:\n",
befc9adc 2384 "expected: $svm->{uuid}\n",
8a49ee97
EW
2385 " got: $uuid\n";
2386 }
befc9adc
EW
2387 my $full_url = $self->full_url;
2388 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2389 die "Failed to replace '$svm->{replace}' with ",
2390 "'$svm->{source}' in $full_url\n";
18ea92bd
SV
2391 # throw away username for storing in records
2392 remove_username($full_url);
8a49ee97
EW
2393 $log_entry{metadata} = "$full_url\@$r $uuid";
2394 $log_entry{svm_revision} = $r;
70ae04e4
AW
2395 $email ||= "$author\@$uuid";
2396 $commit_email ||= "$author\@$uuid";
62e349d2
EW
2397 } elsif ($self->use_svnsync_props) {
2398 my $full_url = $self->svnsync->{url};
2399 $full_url .= "/$self->{path}" if length $self->{path};
ce118739 2400 remove_username($full_url);
62e349d2
EW
2401 my $uuid = $self->svnsync->{uuid};
2402 $log_entry{metadata} = "$full_url\@$rev $uuid";
70ae04e4
AW
2403 $email ||= "$author\@$uuid";
2404 $commit_email ||= "$author\@$uuid";
8a49ee97 2405 } else {
ce118739
AR
2406 my $url = $self->metadata_url;
2407 remove_username($url);
2408 $log_entry{metadata} = "$url\@$rev " .
26a62d57 2409 $self->ra->get_uuid;
db03cd24 2410 $email ||= "$author\@" . $self->ra->get_uuid;
70ae04e4 2411 $commit_email ||= "$author\@" . $self->ra->get_uuid;
8a49ee97 2412 }
db03cd24
EW
2413 $log_entry{name} = $name;
2414 $log_entry{email} = $email;
70ae04e4
AW
2415 $log_entry{commit_name} = $commit_name;
2416 $log_entry{commit_email} = $commit_email;
9b981fc6
EW
2417 \%log_entry;
2418}
2419
2420sub fetch {
3ebe8df7 2421 my ($self, $min_rev, $max_rev, @parents) = @_;
9b981fc6 2422 my ($last_rev, $last_commit) = $self->last_rev_commit;
3ebe8df7 2423 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
e518192f 2424 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
9b981fc6
EW
2425}
2426
2427sub set_tree_cb {
2428 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
490f49ea
EW
2429 $self->{inject_parents} = { $rev => $tree };
2430 $self->fetch(undef, undef);
9b981fc6
EW
2431}
2432
2433sub set_tree {
2434 my ($self, $tree) = (shift, shift);
1ce255dc 2435 my $log_entry = ::get_commit_entry($tree);
9b981fc6 2436 unless ($self->{last_rev}) {
207f1a75 2437 fatal("Must have an existing revision to commit");
9b981fc6 2438 }
61395354
EW
2439 my %ed_opts = ( r => $self->{last_rev},
2440 log => $log_entry->{log},
2441 ra => $self->ra,
2442 tree_a => $self->{last_commit},
2443 tree_b => $tree,
2444 editor_cb => sub {
2445 $self->set_tree_cb($log_entry, $tree, @_) },
2446 svn_path => $self->{path} );
2447 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
9b981fc6
EW
2448 print "No changes\nr$self->{last_rev} = $tree\n";
2449 }
9b981fc6
EW
2450}
2451
060610c5
EW
2452sub rebuild_from_rev_db {
2453 my ($self, $path) = @_;
2454 my $r = -1;
2455 open my $fh, '<', $path or croak "open: $!";
2456 while (<$fh>) {
2457 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2458 chomp($_);
2459 ++$r;
2460 next if $_ eq ('0' x 40);
2461 $self->rev_map_set($r, $_);
2462 print "r$r = $_\n";
2463 }
2464 close $fh or croak "close: $!";
2465 unlink $path or croak "unlink: $!";
2466}
2467
f0ecca10
EW
2468sub rebuild {
2469 my ($self) = @_;
060610c5
EW
2470 my $map_path = $self->map_path;
2471 return if (-e $map_path && ! -z $map_path);
d6d3346b 2472 return unless ::verify_ref($self->refname.'^0');
060610c5
EW
2473 if ($self->use_svm_props || $self->no_metadata) {
2474 my $rev_db = $self->rev_db_path;
2475 $self->rebuild_from_rev_db($rev_db);
2476 if ($self->use_svm_props) {
2477 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2478 $self->rebuild_from_rev_db($svm_rev_db);
2479 }
2480 $self->unlink_rev_db_symlink;
26a62d57
EW
2481 return;
2482 }
060610c5
EW
2483 print "Rebuilding $map_path ...\n";
2484 my ($log, $ctx) =
2485 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2486 $self->refname, '--');
f0ecca10 2487 my $full_url = $self->full_url;
18ea92bd 2488 remove_username($full_url);
060610c5 2489 my $svn_uuid = $self->ra_uuid;
3dfab993
SV
2490 my $c;
2491 while (<$log>) {
2492 if ( m{^commit ($::sha1)$} ) {
2493 $c = $1;
2494 next;
2495 }
2496 next unless s{^\s*(git-svn-id:)}{$1};
2497 my ($url, $rev, $uuid) = ::extract_metadata($_);
18ea92bd 2498 remove_username($url);
f0ecca10
EW
2499
2500 # ignore merges (from set-tree)
2501 next if (!defined $rev || !$uuid);
2502
2503 # if we merged or otherwise started elsewhere, this is
2504 # how we break out of it
060610c5 2505 if (($uuid ne $svn_uuid) ||
f0ecca10
EW
2506 ($full_url && $url && ($url ne $full_url))) {
2507 next;
2508 }
f0ecca10 2509
060610c5 2510 $self->rev_map_set($rev, $c);
f0ecca10
EW
2511 print "r$rev = $c\n";
2512 }
3dfab993 2513 command_close_pipe($log, $ctx);
060610c5
EW
2514 print "Done rebuilding $map_path\n";
2515 my $rev_db_path = $self->rev_db_path;
2516 if (-f $self->rev_db_path) {
2517 unlink $self->rev_db_path or croak "unlink: $!";
2518 }
2519 $self->unlink_rev_db_symlink;
f0ecca10
EW
2520}
2521
060610c5 2522# rev_map:
9b981fc6
EW
2523# Tie::File seems to be prone to offset errors if revisions get sparse,
2524# it's not that fast, either. Tie::File is also not in Perl 5.6. So
2525# one of my favorite modules is out :< Next up would be one of the DBM
060610c5
EW
2526# modules, but I'm not sure which is most portable...
2527#
2528# This is the replacement for the rev_db format, which was too big
2529# and inefficient for large repositories with a lot of sparse history
2530# (mainly tags)
2531#
2532# The format is this:
2533# - 24 bytes for every record,
2534# * 4 bytes for the integer representing an SVN revision number
2535# * 20 bytes representing the sha1 of a git commit
2536# - No empty padding records like the old format
2537# - new records are written append-only since SVN revision numbers
2538# increase monotonically
2539# - lookups on SVN revision number are done via a binary search
2540# - Piping the file to xxd(1) -c24 is a good way of dumping it for
2541# viewing or editing, should the need ever arise.
2542#
97ae0911 2543# These files are disposable unless noMetadata or useSvmProps is set
9b981fc6 2544
060610c5 2545sub _rev_map_set {
26a62d57 2546 my ($fh, $rev, $commit) = @_;
060610c5
EW
2547
2548 my $size = (stat($fh))[7];
2549 ($size % 24) == 0 or croak "inconsistent size: $size";
2550
2551 if ($size > 0) {
2552 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2553 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2554 $read == 24 or croak "read only $read bytes (!= 24)";
2555 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2556 if ($last_rev >= $rev) {
2557 croak "last_rev is higher!: $last_rev >= $rev";
26a62d57
EW
2558 }
2559 }
060610c5
EW
2560 sysseek($fh, 0, SEEK_END) or croak "seek: $!";
2561 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2562 croak "write: $!";
26a62d57
EW
2563}
2564
2565sub mkfile {
2566 my ($path) = @_;
2567 unless (-e $path) {
2568 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2569 mkpath([$dir]) unless -d $dir;
2570 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2571 close $fh or die "Couldn't close (create) $path: $!\n";
2572 }
2573}
2574
060610c5 2575sub rev_map_set {
26a62d57
EW
2576 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2577 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
060610c5 2578 my $db = $self->map_path($uuid);
26a62d57 2579 my $db_lock = "$db.lock";
373274f9
EW
2580 my $sig;
2581 if ($update_ref) {
2582 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2583 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2584 }
26a62d57
EW
2585 mkfile($db);
2586
373274f9 2587 $LOCKFILES{$db_lock} = 1;
97ae0911 2588 my $sync;
97ae0911
EW
2589 # both of these options make our .rev_db file very, very important
2590 # and we can't afford to lose it because rebuild() won't work
2591 if ($self->use_svm_props || $self->no_metadata) {
2592 $sync = 1;
060610c5 2593 copy($db, $db_lock) or die "rev_map_set(@_): ",
26a62d57 2594 "Failed to copy: ",
373274f9
EW
2595 "$db => $db_lock ($!)\n";
2596 } else {
060610c5 2597 rename $db, $db_lock or die "rev_map_set(@_): ",
26a62d57 2598 "Failed to rename: ",
373274f9
EW
2599 "$db => $db_lock ($!)\n";
2600 }
060610c5
EW
2601
2602 sysopen(my $fh, $db_lock, O_RDWR | O_APPEND | O_CREAT)
2603 or croak "Couldn't open $db_lock: $!\n";
2604 _rev_map_set($fh, $rev, $commit);
97ae0911
EW
2605 if ($sync) {
2606 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2607 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2608 }
9b981fc6 2609 close $fh or croak $!;
373274f9 2610 if ($update_ref) {
1e889ef3 2611 $_head = $self;
373274f9
EW
2612 command_noisy('update-ref', '-m', "r$rev",
2613 $self->refname, $commit);
2614 }
060610c5 2615 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
373274f9
EW
2616 "$db_lock => $db ($!)\n";
2617 delete $LOCKFILES{$db_lock};
2618 if ($update_ref) {
2619 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2620 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2621 kill $sig, $$ if defined $sig;
2622 }
9b981fc6
EW
2623}
2624
060610c5 2625sub rev_map_max {
9c93fee5 2626 my ($self) = @_;
d6d3346b 2627 $self->rebuild;
060610c5
EW
2628 my $map_path = $self->map_path;
2629 stat $map_path or return wantarray ? (0, undef) : 0;
2630 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2631 my $size = (stat($fh))[7];
2632 ($size % 24) == 0 or croak "inconsistent size: $size";
2633
2634 if ($size == 0) {
2635 close $fh or croak "close: $!";
2636 return wantarray ? (0, undef) : 0;
2637 }
2638
2639 sysseek($fh, -24, SEEK_END);
2640 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2641 close $fh or croak "close: $!";
2642 my ($r, $c) = unpack(rev_map_fmt, $buf);
2643 wantarray ? ($r, $c) : $r;
9c93fee5
EW
2644}
2645
060610c5 2646sub rev_map_get {
26a62d57 2647 my ($self, $rev, $uuid) = @_;
060610c5
EW
2648 my $map_path = $self->map_path($uuid);
2649 return undef unless -e $map_path;
2650
2651 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2652 my $size = (stat($fh))[7];
2653 ($size % 24) == 0 or croak "inconsistent size: $size";
2654
2655 if ($size == 0) {
2656 close $fh or croak "close: $fh";
2657 return undef;
2658 }
2659
2660 my ($l, $u) = (0, $size - 24);
2661 my ($r, $c, $buf);
2662
2663 while ($l <= $u) {
2664 my $i = int(($l/24 + $u/24) / 2) * 24;
2665 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2666 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2667 my ($r, $c) = unpack('NH40', $buf);
2668
2669 if ($r < $rev) {
2670 $l = $i + 24;
2671 } elsif ($r > $rev) {
2672 $u = $i - 24;
2673 } else { # $r == $rev
2674 close($fh) or croak "close: $!";
2675 return $c;
2676 }
9b981fc6 2677 }
060610c5
EW
2678 close($fh) or croak "close: $!";
2679 undef;
9b981fc6
EW
2680}
2681
111947ef
DK
2682# Finds the first svn revision that exists on (if $eq_ok is true) or
2683# before $rev for the current branch. It will not search any lower
2684# than $min_rev. Returns the git commit hash and svn revision number
2685# if found, else (undef, undef).
15710b6f 2686sub find_rev_before {
111947ef 2687 my ($self, $rev, $eq_ok, $min_rev) = @_;
15710b6f 2688 --$rev unless $eq_ok;
111947ef
DK
2689 $min_rev ||= 1;
2690 while ($rev >= $min_rev) {
060610c5 2691 if (my $c = $self->rev_map_get($rev)) {
15710b6f
EW
2692 return ($rev, $c);
2693 }
2694 --$rev;
2695 }
2696 return (undef, undef);
2697}
2698
111947ef
DK
2699# Finds the first svn revision that exists on (if $eq_ok is true) or
2700# after $rev for the current branch. It will not search any higher
2701# than $max_rev. Returns the git commit hash and svn revision number
2702# if found, else (undef, undef).
2703sub find_rev_after {
2704 my ($self, $rev, $eq_ok, $max_rev) = @_;
2705 ++$rev unless $eq_ok;
060610c5 2706 $max_rev ||= $self->rev_map_max;
111947ef 2707 while ($rev <= $max_rev) {
060610c5 2708 if (my $c = $self->rev_map_get($rev)) {
111947ef
DK
2709 return ($rev, $c);
2710 }
2711 ++$rev;
2712 }
2713 return (undef, undef);
2714}
2715
9b981fc6 2716sub _new {
706587fc
EW
2717 my ($class, $repo_id, $ref_id, $path) = @_;
2718 unless (defined $repo_id && length $repo_id) {
2719 $repo_id = $Git::SVN::default_repo_id;
2720 }
2721 unless (defined $ref_id && length $ref_id) {
8b8fc068 2722 $_[2] = $ref_id = $Git::SVN::default_ref_id;
706587fc
EW
2723 }
2724 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2725 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2726 $_[3] = $path = '' unless (defined $path);
b4d57e5e 2727 mkpath(["$ENV{GIT_DIR}/svn"]);
26a62d57
EW
2728 bless {
2729 ref_id => $ref_id, dir => $dir, index => "$dir/index",
8a49ee97 2730 path => $path, config => "$ENV{GIT_DIR}/svn/config",
060610c5 2731 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
26a62d57
EW
2732}
2733
060610c5
EW
2734# for read-only access of old .rev_db formats
2735sub unlink_rev_db_symlink {
2736 my ($self) = @_;
2737 my $link = $self->rev_db_path;
2738 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2739 if (-l $link) {
2740 unlink $link or croak "unlink: $link failed!";
2741 }
2742}
2743
2744sub rev_db_path {
2745 my ($self, $uuid) = @_;
2746 my $db_path = $self->map_path($uuid);
2747 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2748 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2749 $db_path;
2750}
2751
2752# the new replacement for .rev_db
2753sub map_path {
26a62d57
EW
2754 my ($self, $uuid) = @_;
2755 $uuid ||= $self->ra_uuid;
060610c5 2756 "$self->{map_root}.$uuid";
9b981fc6
EW
2757}
2758
1c8443b0
EW
2759sub uri_encode {
2760 my ($f) = @_;
2761 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2762 $f
2763}
9b981fc6 2764
18ea92bd
SV
2765sub remove_username {
2766 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2767}
2768
d976acfd
EW
2769package Git::SVN::Prompt;
2770use strict;
2771use warnings;
2772require SVN::Core;
2773use vars qw/$_no_auth_cache $_username/;
2774
2775sub simple {
30d055aa
EW
2776 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2777 $may_save = undef if $_no_auth_cache;
2778 $default_username = $_username if defined $_username;
2779 if (defined $default_username && length $default_username) {
2780 if (defined $realm && length $realm) {
6f729591
EW
2781 print STDERR "Authentication realm: $realm\n";
2782 STDERR->flush;
30d055aa
EW
2783 }
2784 $cred->username($default_username);
2785 } else {
d976acfd 2786 username($cred, $realm, $may_save, $pool);
30d055aa
EW
2787 }
2788 $cred->password(_read_password("Password for '" .
2789 $cred->username . "': ", $realm));
2790 $cred->may_save($may_save);
2791 $SVN::_Core::SVN_NO_ERROR;
2792}
2793
d976acfd 2794sub ssl_server_trust {
30d055aa
EW
2795 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2796 $may_save = undef if $_no_auth_cache;
6f729591 2797 print STDERR "Error validating server certificate for '$realm':\n";
fd499bcc
ER
2798 {
2799 no warnings 'once';
2800 # All variables SVN::Auth::SSL::* are used only once,
2801 # so we're shutting up Perl warnings about this.
2802 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2803 print STDERR " - The certificate is not issued ",
2804 "by a trusted authority. Use the\n",
2805 " fingerprint to validate ",
2806 "the certificate manually!\n";
2807 }
2808 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2809 print STDERR " - The certificate hostname ",
2810 "does not match.\n";
2811 }
2812 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2813 print STDERR " - The certificate is not yet valid.\n";
2814 }
2815 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2816 print STDERR " - The certificate has expired.\n";
2817 }
2818 if ($failures & $SVN::Auth::SSL::OTHER) {
2819 print STDERR " - The certificate has ",
2820 "an unknown error.\n";
2821 }
2822 } # no warnings 'once'
6f729591
EW
2823 printf STDERR
2824 "Certificate information:\n".
30d055aa
EW
2825 " - Hostname: %s\n".
2826 " - Valid: from %s until %s\n".
2827 " - Issuer: %s\n".
2828 " - Fingerprint: %s\n",
2829 map $cert_info->$_, qw(hostname valid_from valid_until
6f729591 2830 issuer_dname fingerprint);
30d055aa
EW
2831 my $choice;
2832prompt:
6f729591 2833 print STDERR $may_save ?
30d055aa
EW
2834 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2835 "(R)eject or accept (t)emporarily? ";
6f729591 2836 STDERR->flush;
30d055aa
EW
2837 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2838 if ($choice =~ /^t$/i) {
2839 $cred->may_save(undef);
2840 } elsif ($choice =~ /^r$/i) {
2841 return -1;
2842 } elsif ($may_save && $choice =~ /^p$/i) {
2843 $cred->may_save($may_save);
2844 } else {
2845 goto prompt;
2846 }
2847 $cred->accepted_failures($failures);
2848 $SVN::_Core::SVN_NO_ERROR;
2849}
2850
d976acfd 2851sub ssl_client_cert {
30d055aa
EW
2852 my ($cred, $realm, $may_save, $pool) = @_;
2853 $may_save = undef if $_no_auth_cache;
6f729591
EW
2854 print STDERR "Client certificate filename: ";
2855 STDERR->flush;
30d055aa
EW
2856 chomp(my $filename = <STDIN>);
2857 $cred->cert_file($filename);
2858 $cred->may_save($may_save);
2859 $SVN::_Core::SVN_NO_ERROR;
2860}
2861
d976acfd 2862sub ssl_client_cert_pw {
30d055aa
EW
2863 my ($cred, $realm, $may_save, $pool) = @_;
2864 $may_save = undef if $_no_auth_cache;
2865 $cred->password(_read_password("Password: ", $realm));
2866 $cred->may_save($may_save);
2867 $SVN::_Core::SVN_NO_ERROR;
2868}
2869
d976acfd 2870sub username {
30d055aa
EW
2871 my ($cred, $realm, $may_save, $pool) = @_;
2872 $may_save = undef if $_no_auth_cache;
2873 if (defined $realm && length $realm) {
6f729591 2874 print STDERR "Authentication realm: $realm\n";
30d055aa
EW
2875 }
2876 my $username;
2877 if (defined $_username) {
2878 $username = $_username;
2879 } else {
6f729591
EW
2880 print STDERR "Username: ";
2881 STDERR->flush;
30d055aa
EW
2882 chomp($username = <STDIN>);
2883 }
2884 $cred->username($username);
2885 $cred->may_save($may_save);
2886 $SVN::_Core::SVN_NO_ERROR;
2887}
2888
2889sub _read_password {
2890 my ($prompt, $realm) = @_;
6f729591
EW
2891 print STDERR $prompt;
2892 STDERR->flush;
30d055aa
EW
2893 require Term::ReadKey;
2894 Term::ReadKey::ReadMode('noecho');
2895 my $password = '';
2896 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2897 last if $key =~ /[\012\015]/; # \n\r
2898 $password .= $key;
2899 }
2900 Term::ReadKey::ReadMode('restore');
6f729591
EW
2901 print STDERR "\n";
2902 STDERR->flush;
30d055aa
EW
2903 $password;
2904}
2905
27a1a801
EW
2906package SVN::Git::Fetcher;
2907use vars qw/@ISA/;
2908use strict;
2909use warnings;
2910use Carp qw/croak/;
2911use IO::File qw//;
2912
2913# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2914sub new {
2915 my ($class, $git_svn) = @_;
2916 my $self = SVN::Delta::Editor->new;
2917 bless $self, $class;
1c8443b0 2918 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
d2a9a87b
EW
2919 $self->{empty} = {};
2920 $self->{dir_prop} = {};
2921 $self->{file_prop} = {};
2922 $self->{absent_dir} = {};
2923 $self->{absent_file} = {};
ef3cfaad 2924 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
27a1a801
EW
2925 $self;
2926}
2927
8b8fc068
EW
2928sub set_path_strip {
2929 my ($self, $path) = @_;
4e9f6cc7 2930 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
8b8fc068
EW
2931}
2932
d2a9a87b
EW
2933sub open_root {
2934 { path => '' };
2935}
2936
2937sub open_directory {
2938 my ($self, $path, $pb, $rev) = @_;
2939 { path => $path };
2940}
2941
706587fc
EW
2942sub git_path {
2943 my ($self, $path) = @_;
2b27f6c8
EW
2944 if ($self->{path_strip}) {
2945 $path =~ s!$self->{path_strip}!! or
2946 die "Failed to strip path '$path' ($self->{path_strip})\n";
2947 }
706587fc
EW
2948 $path;
2949}
2950
27a1a801
EW
2951sub delete_entry {
2952 my ($self, $path, $rev, $pb) = @_;
4a87db0e 2953
706587fc 2954 my $gpath = $self->git_path($path);
8a603774
EW
2955 return undef if ($gpath eq '');
2956
4a87db0e 2957 # remove entire directories.
706587fc 2958 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
4a87db0e
EW
2959 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2960 -r --name-only -z/,
706587fc 2961 $self->{c}, '--', $gpath);
4a87db0e
EW
2962 local $/ = "\0";
2963 while (<$ls>) {
ef3cfaad
EW
2964 chomp;
2965 $self->{gii}->remove($_);
9e3cdbd4 2966 print "\tD\t$_\n" unless $::_q;
4a87db0e 2967 }
9e3cdbd4 2968 print "\tD\t$gpath/\n" unless $::_q;
4a87db0e
EW
2969 command_close_pipe($ls, $ctx);
2970 $self->{empty}->{$path} = 0
2971 } else {
ef3cfaad 2972 $self->{gii}->remove($gpath);
9e3cdbd4 2973 print "\tD\t$gpath\n" unless $::_q;
4a87db0e 2974 }
27a1a801
EW
2975 undef;
2976}
2977
2978sub open_file {
2979 my ($self, $path, $pb, $rev) = @_;
706587fc
EW
2980 my $gpath = $self->git_path($path);
2981 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
27a1a801 2982 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
006ede5e
EW
2983 unless (defined $mode && defined $blob) {
2984 die "$path was not found in commit $self->{c} (r$rev)\n";
2985 }
27a1a801 2986 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
0864e3ba 2987 pool => SVN::Pool->new, action => 'M' };
27a1a801
EW
2988}
2989
2990sub add_file {
2991 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
d2a9a87b
EW
2992 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2993 delete $self->{empty}->{$dir};
27a1a801 2994 { path => $path, mode_a => 100644, mode_b => 100644,
0864e3ba 2995 pool => SVN::Pool->new, action => 'A' };
27a1a801
EW
2996}
2997
d2a9a87b
EW
2998sub add_directory {
2999 my ($self, $path, $cp_path, $cp_rev) = @_;
3000 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3001 delete $self->{empty}->{$dir};
3002 $self->{empty}->{$path} = 1;
3003 { path => $path };
3004}
3005
3006sub change_dir_prop {
3007 my ($self, $db, $prop, $value) = @_;
3008 $self->{dir_prop}->{$db->{path}} ||= {};
3009 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3010 undef;
3011}
3012
3013sub absent_directory {
3014 my ($self, $path, $pb) = @_;
3015 $self->{absent_dir}->{$pb->{path}} ||= [];
3016 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3017 undef;
3018}
3019
3020sub absent_file {
3021 my ($self, $path, $pb) = @_;
3022 $self->{absent_file}->{$pb->{path}} ||= [];
3023 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3024 undef;
3025}
3026
27a1a801
EW
3027sub change_file_prop {
3028 my ($self, $fb, $prop, $value) = @_;
3029 if ($prop eq 'svn:executable') {
3030 if ($fb->{mode_b} != 120000) {
3031 $fb->{mode_b} = defined $value ? 100755 : 100644;
3032 }
3033 } elsif ($prop eq 'svn:special') {
3034 $fb->{mode_b} = defined $value ? 120000 : 100644;
d2a9a87b
EW
3035 } else {
3036 $self->{file_prop}->{$fb->{path}} ||= {};
3037 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
27a1a801
EW
3038 }
3039 undef;
3040}
3041
3042sub apply_textdelta {
3043 my ($self, $fb, $exp) = @_;
3044 my $fh = IO::File->new_tmpfile;
3045 $fh->autoflush(1);
3046 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3047 # (but $base does not,) so dup() it for reading in close_file
3048 open my $dup, '<&', $fh or croak $!;
3049 my $base = IO::File->new_tmpfile;
3050 $base->autoflush(1);
3051 if ($fb->{blob}) {
3052 defined (my $pid = fork) or croak $!;
3053 if (!$pid) {
3054 open STDOUT, '>&', $base or croak $!;
3055 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3056 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3057 }
3058 waitpid $pid, 0;
3059 croak $? if $?;
3060
3061 if (defined $exp) {
3062 seek $base, 0, 0 or croak $!;
8d7c4fad 3063 my $got = ::md5sum($base);
27a1a801
EW
3064 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3065 "expected: $exp\n",
3066 " got: $got\n" if ($got ne $exp);
3067 }
3068 }
3069 seek $base, 0, 0 or croak $!;
3070 $fb->{fh} = $dup;
3071 $fb->{base} = $base;
3072 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3073}
3074
3075sub close_file {
3076 my ($self, $fb, $exp) = @_;
3077 my $hash;
706587fc 3078 my $path = $self->git_path($fb->{path});
27a1a801 3079 if (my $fh = $fb->{fh}) {
7faf0686
EW
3080 if (defined $exp) {
3081 seek($fh, 0, 0) or croak $!;
8d7c4fad 3082 my $got = ::md5sum($fh);
7faf0686
EW
3083 if ($got ne $exp) {
3084 die "Checksum mismatch: $path\n",
3085 "expected: $exp\n got: $got\n";
3086 }
3087 }
bcd8ee5b 3088 sysseek($fh, 0, 0) or croak $!;
27a1a801 3089 if ($fb->{mode_b} == 120000) {
bcd8ee5b 3090 sysread($fh, my $buf, 5) == 5 or croak $!;
27a1a801
EW
3091 $buf eq 'link ' or die "$path has mode 120000",
3092 "but is not a link\n";
3093 }
3094 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3095 if (!$pid) {
3096 open STDIN, '<&', $fh or croak $!;
3097 exec qw/git-hash-object -w --stdin/ or croak $!;
3098 }
3099 chomp($hash = do { local $/; <$out> });
3100 close $out or croak $!;
3101 close $fh or croak $!;
3102 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3103 close $fb->{base} or croak $!;
3104 } else {
3105 $hash = $fb->{blob} or die "no blob information\n";
3106 }
3107 $fb->{pool}->clear;
ef3cfaad 3108 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
9e3cdbd4 3109 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
27a1a801
EW
3110 undef;
3111}
3112
3113sub abort_edit {
3114 my $self = shift;
ef3cfaad
EW
3115 $self->{nr} = $self->{gii}->{nr};
3116 delete $self->{gii};
27a1a801
EW
3117 $self->SUPER::abort_edit(@_);
3118}
3119
3120sub close_edit {
3121 my $self = shift;
dad73c0b 3122 $self->{git_commit_ok} = 1;
ef3cfaad
EW
3123 $self->{nr} = $self->{gii}->{nr};
3124 delete $self->{gii};
27a1a801
EW
3125 $self->SUPER::close_edit(@_);
3126}
1a82e793 3127
a5e0cedc 3128package SVN::Git::Editor;
24e22aa8 3129use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
a5e0cedc
EW
3130use strict;
3131use warnings;
3132use Carp qw/croak/;
3133use IO::File;
3134
3135sub new {
61395354
EW
3136 my ($class, $opts) = @_;
3137 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3138 die "$_ required!\n" unless (defined $opts->{$_});
3139 }
3140
3141 my $pool = SVN::Pool->new;
3142 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3143 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3144 $opts->{r}, $mods);
3145
3146 # $opts->{ra} functions should not be used after this:
3147 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3148 $opts->{editor_cb}, $pool);
3149 my $self = SVN::Delta::Editor->new(@ce, $pool);
a5e0cedc 3150 bless $self, $class;
61395354
EW
3151 foreach (qw/svn_path r tree_a tree_b/) {
3152 $self->{$_} = $opts->{$_};
a5e0cedc 3153 }
61395354
EW
3154 $self->{url} = $opts->{ra}->{url};
3155 $self->{mods} = $mods;
3156 $self->{types} = $types;
3157 $self->{pool} = $pool;
a5e0cedc
EW
3158 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3159 $self->{rm} = { };
d3a840dc
EW
3160 $self->{path_prefix} = length $self->{svn_path} ?
3161 "$self->{svn_path}/" : '';
a5e0cedc
EW
3162 return $self;
3163}
3164
61395354
EW
3165sub generate_diff {
3166 my ($tree_a, $tree_b) = @_;
3167 my @diff_tree = qw(diff-tree -z -r);
24e22aa8
EW
3168 if ($_cp_similarity) {
3169 push @diff_tree, "-C$_cp_similarity";
61395354
EW
3170 } else {
3171 push @diff_tree, '-C';
3172 }
24e22aa8
EW
3173 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3174 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
61395354
EW
3175 push @diff_tree, $tree_a, $tree_b;
3176 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3177 local $/ = "\0";
3178 my $state = 'meta';
3179 my @mods;
3180 while (<$diff_fh>) {
3181 chomp $_; # this gets rid of the trailing "\0"
3182 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3183 $::sha1\s($::sha1)\s
3184 ([MTCRAD])\d*$/xo) {
3185 push @mods, { mode_a => $1, mode_b => $2,
3186 sha1_b => $3, chg => $4 };
3187 if ($4 =~ /^(?:C|R)$/) {
3188 $state = 'file_a';
3189 } else {
3190 $state = 'file_b';
3191 }
3192 } elsif ($state eq 'file_a') {
3193 my $x = $mods[$#mods] or croak "Empty array\n";
3194 if ($x->{chg} !~ /^(?:C|R)$/) {
3195 croak "Error parsing $_, $x->{chg}\n";
3196 }
3197 $x->{file_a} = $_;
3198 $state = 'file_b';
3199 } elsif ($state eq 'file_b') {
3200 my $x = $mods[$#mods] or croak "Empty array\n";
3201 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3202 croak "Error parsing $_, $x->{chg}\n";
3203 }
3204 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3205 croak "Error parsing $_, $x->{chg}\n";
3206 }
3207 $x->{file_b} = $_;
3208 $state = 'meta';
3209 } else {
3210 croak "Error parsing $_\n";
3211 }
3212 }
3213 command_close_pipe($diff_fh, $ctx);
3214 \@mods;
3215}
3216
3217sub check_diff_paths {
3218 my ($ra, $pfx, $rev, $mods) = @_;
3219 my %types;
3220 $pfx .= '/' if length $pfx;
3221
3222 sub type_diff_paths {
3223 my ($ra, $types, $path, $rev) = @_;
3224 my @p = split m#/+#, $path;
3225 my $c = shift @p;
3226 unless (defined $types->{$c}) {
3227 $types->{$c} = $ra->check_path($c, $rev);
3228 }
3229 while (@p) {
3230 $c .= '/' . shift @p;
3231 next if defined $types->{$c};
3232 $types->{$c} = $ra->check_path($c, $rev);
3233 }
3234 }
3235
3236 foreach my $m (@$mods) {
3237 foreach my $f (qw/file_a file_b/) {
3238 next unless defined $m->{$f};
3239 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3240 if (length $pfx.$dir && ! defined $types{$dir}) {
3241 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3242 }
3243 }
3244 }
3245 \%types;
3246}
3247
a5e0cedc
EW
3248sub split_path {
3249 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3250}
3251
3252sub repo_path {
d3a840dc
EW
3253 my ($self, $path) = @_;
3254 $self->{path_prefix}.(defined $path ? $path : '');
a5e0cedc
EW
3255}
3256
3257sub url_path {
3258 my ($self, $path) = @_;
29633bb9
EW
3259 if ($self->{url} =~ m#^https?://#) {
3260 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3261 }
6e8548cc 3262 $self->{url} . '/' . $self->repo_path($path);
a5e0cedc
EW
3263}
3264
3265sub rmdirs {
61395354 3266 my ($self) = @_;
a5e0cedc
EW
3267 my $rm = $self->{rm};
3268 delete $rm->{''}; # we never delete the url we're tracking
3269 return unless %$rm;
3270
3271 foreach (keys %$rm) {
3272 my @d = split m#/#, $_;
3273 my $c = shift @d;
3274 $rm->{$c} = 1;
3275 while (@d) {
3276 $c .= '/' . shift @d;
3277 $rm->{$c} = 1;
3278 }
3279 }
3280 delete $rm->{$self->{svn_path}};
3281 delete $rm->{''}; # we never delete the url we're tracking
3282 return unless %$rm;
3283
61395354
EW
3284 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3285 $self->{tree_b});
a5e0cedc
EW
3286 local $/ = "\0";
3287 while (<$fh>) {
3288 chomp;
747fa12c 3289 my @dn = split m#/#, $_;
c07eee1f
EW
3290 while (pop @dn) {
3291 delete $rm->{join '/', @dn};
3292 }
3293 unless (%$rm) {
22600a25 3294 close $fh;
c07eee1f
EW
3295 return;
3296 }
a5e0cedc 3297 }
aef4e921 3298 command_close_pipe($fh, $ctx);
c07eee1f 3299
a5e0cedc
EW
3300 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3301 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3302 $self->close_directory($bat->{$d}, $p);
3303 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
44320b9e 3304 print "\tD+\t$d/\n" unless $::_q;
a5e0cedc
EW
3305 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3306 delete $bat->{$d};
3307 }
3308}
3309
3310sub open_or_add_dir {
3311 my ($self, $full_path, $baton) = @_;
6e8548cc
EW
3312 my $t = $self->{types}->{$full_path};
3313 if (!defined $t) {
3314 die "$full_path not known in r$self->{r} or we have a bug!\n";
3315 }
fd499bcc
ER
3316 {
3317 no warnings 'once';
3318 # SVN::Node::none and SVN::Node::file are used only once,
3319 # so we're shutting up Perl's warnings about them.
3320 if ($t == $SVN::Node::none) {
3321 return $self->add_directory($full_path, $baton,
3322 undef, -1, $self->{pool});
3323 } elsif ($t == $SVN::Node::dir) {
3324 return $self->open_directory($full_path, $baton,
3325 $self->{r}, $self->{pool});
3326 } # no warnings 'once'
3327 print STDERR "$full_path already exists in repository at ",
3328 "r$self->{r} and it is not a directory (",
3329 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3330 } # no warnings 'once'
a5e0cedc
EW
3331 exit 1;
3332}
3333
3334sub ensure_path {
3335 my ($self, $path) = @_;
3336 my $bat = $self->{bat};
6e8548cc
EW
3337 my $repo_path = $self->repo_path($path);
3338 return $bat->{''} unless (length $repo_path);
3339 my @p = split m#/+#, $repo_path;
a5e0cedc
EW
3340 my $c = shift @p;
3341 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3342 while (@p) {
3343 my $c0 = $c;
3344 $c .= '/' . shift @p;
3345 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3346 }
3347 return $bat->{$c};
3348}
3349
3350sub A {
44320b9e 3351 my ($self, $m) = @_;
a5e0cedc
EW
3352 my ($dir, $file) = split_path($m->{file_b});
3353 my $pbat = $self->ensure_path($dir);
3354 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3355 undef, -1);
44320b9e 3356 print "\tA\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3357 $self->chg_file($fbat, $m);
3358 $self->close_file($fbat,undef,$self->{pool});
3359}
3360
3361sub C {
44320b9e 3362 my ($self, $m) = @_;
a5e0cedc
EW
3363 my ($dir, $file) = split_path($m->{file_b});
3364 my $pbat = $self->ensure_path($dir);
3365 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3366 $self->url_path($m->{file_a}), $self->{r});
44320b9e 3367 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3368 $self->chg_file($fbat, $m);
3369 $self->close_file($fbat,undef,$self->{pool});
3370}
3371
3372sub delete_entry {
3373 my ($self, $path, $pbat) = @_;
3374 my $rpath = $self->repo_path($path);
3375 my ($dir, $file) = split_path($rpath);
3376 $self->{rm}->{$dir} = 1;
3377 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3378}
3379
3380sub R {
44320b9e 3381 my ($self, $m) = @_;
a5e0cedc
EW
3382 my ($dir, $file) = split_path($m->{file_b});
3383 my $pbat = $self->ensure_path($dir);
3384 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3385 $self->url_path($m->{file_a}), $self->{r});
44320b9e 3386 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3387 $self->chg_file($fbat, $m);
3388 $self->close_file($fbat,undef,$self->{pool});
3389
3390 ($dir, $file) = split_path($m->{file_a});
3391 $pbat = $self->ensure_path($dir);
3392 $self->delete_entry($m->{file_a}, $pbat);
3393}
3394
3395sub M {
44320b9e 3396 my ($self, $m) = @_;
a5e0cedc
EW
3397 my ($dir, $file) = split_path($m->{file_b});
3398 my $pbat = $self->ensure_path($dir);
3399 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3400 $pbat,$self->{r},$self->{pool});
44320b9e 3401 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3402 $self->chg_file($fbat, $m);
3403 $self->close_file($fbat,undef,$self->{pool});
3404}
3405
3406sub T { shift->M(@_) }
3407
3408sub change_file_prop {
3409 my ($self, $fbat, $pname, $pval) = @_;
3410 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3411}
3412
3413sub chg_file {
3414 my ($self, $fbat, $m) = @_;
3415 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3416 $self->change_file_prop($fbat,'svn:executable','*');
3417 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3418 $self->change_file_prop($fbat,'svn:executable',undef);
3419 }
3420 my $fh = IO::File->new_tmpfile or croak $!;
3421 if ($m->{mode_b} =~ /^120/) {
3422 print $fh 'link ' or croak $!;
3423 $self->change_file_prop($fbat,'svn:special','*');
3424 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3425 $self->change_file_prop($fbat,'svn:special',undef);
3426 }
3427 defined(my $pid = fork) or croak $!;
3428 if (!$pid) {
3429 open STDOUT, '>&', $fh or croak $!;
3430 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3431 }
3432 waitpid $pid, 0;
3433 croak $? if $?;
3434 $fh->flush == 0 or croak $!;
3435 seek $fh, 0, 0 or croak $!;
3436
8d7c4fad 3437 my $exp = ::md5sum($fh);
a5e0cedc
EW
3438 seek $fh, 0, 0 or croak $!;
3439
f7197dff
EW
3440 my $pool = SVN::Pool->new;
3441 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3442 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
a5e0cedc 3443 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
f7197dff 3444 $pool->clear;
a5e0cedc
EW
3445
3446 close $fh or croak $!;
3447}
3448
3449sub D {
44320b9e 3450 my ($self, $m) = @_;
a5e0cedc
EW
3451 my ($dir, $file) = split_path($m->{file_b});
3452 my $pbat = $self->ensure_path($dir);
44320b9e 3453 print "\tD\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3454 $self->delete_entry($m->{file_b}, $pbat);
3455}
3456
3457sub close_edit {
3458 my ($self) = @_;
3459 my ($p,$bat) = ($self->{pool}, $self->{bat});
3460 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
6442754d 3461 next if $_ eq '';
a5e0cedc
EW
3462 $self->close_directory($bat->{$_}, $p);
3463 }
6442754d 3464 $self->close_directory($bat->{''}, $p);
a5e0cedc
EW
3465 $self->SUPER::close_edit($p);
3466 $p->clear;
3467}
3468
3469sub abort_edit {
3470 my ($self) = @_;
3471 $self->SUPER::abort_edit($self->{pool});
61395354
EW
3472}
3473
3474sub DESTROY {
3475 my $self = shift;
3476 $self->SUPER::DESTROY(@_);
a5e0cedc
EW
3477 $self->{pool}->clear;
3478}
3479
44320b9e
EW
3480# this drives the editor
3481sub apply_diff {
61395354
EW
3482 my ($self) = @_;
3483 my $mods = $self->{mods};
44320b9e 3484 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
6e8548cc 3485 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
44320b9e
EW
3486 my $f = $m->{chg};
3487 if (defined $o{$f}) {
3488 $self->$f($m);
3489 } else {
207f1a75 3490 fatal("Invalid change type: $f");
44320b9e
EW
3491 }
3492 }
24e22aa8 3493 $self->rmdirs if $_rmdir;
6e8548cc 3494 if (@$mods == 0) {
44320b9e
EW
3495 $self->abort_edit;
3496 } else {
3497 $self->close_edit;
3498 }
6e8548cc 3499 return scalar @$mods;
44320b9e
EW
3500}
3501
d81bf827 3502package Git::SVN::Ra;
6af1db44 3503use vars qw/@ISA $config_dir $_log_window_size/;
d81bf827
EW
3504use strict;
3505use warnings;
a51cdb0c 3506my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
d81bf827
EW
3507
3508BEGIN {
3509 # enforce temporary pool usage for some simple functions
c5f71ad0
SV
3510 no strict 'refs';
3511 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3512 my $SUPER = "SUPER::$f";
3513 *$f = sub {
3514 my $self = shift;
3515 my $pool = SVN::Pool->new;
3516 my @ret = $self->$SUPER(@_,$pool);
3517 $pool->clear;
3518 wantarray ? @ret : $ret[0];
3519 };
d81bf827 3520 }
d81bf827
EW
3521}
3522
9ff74e95
SW
3523sub _auth_providers () {
3524 [
3525 SVN::Client::get_simple_provider(),
3526 SVN::Client::get_ssl_server_trust_file_provider(),
3527 SVN::Client::get_simple_prompt_provider(
3528 \&Git::SVN::Prompt::simple, 2),
3529 SVN::Client::get_ssl_client_cert_file_provider(),
3530 SVN::Client::get_ssl_client_cert_prompt_provider(
3531 \&Git::SVN::Prompt::ssl_client_cert, 2),
3532 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3533 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3534 SVN::Client::get_username_provider(),
3535 SVN::Client::get_ssl_server_trust_prompt_provider(
3536 \&Git::SVN::Prompt::ssl_server_trust),
3537 SVN::Client::get_username_prompt_provider(
3538 \&Git::SVN::Prompt::username, 2)
3539 ]
3540}
3541
cfbe7ab3
EW
3542sub escape_uri_only {
3543 my ($uri) = @_;
3544 my @tmp;
3545 foreach (split m{/}, $uri) {
3546 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3547 push @tmp, $_;
3548 }
3549 join('/', @tmp);
3550}
3551
3552sub escape_url {
3553 my ($url) = @_;
3554 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3555 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3556 $url = "$scheme://$domain$uri";
3557 }
3558 $url;
3559}
3560
d81bf827
EW
3561sub new {
3562 my ($class, $url) = @_;
f6f09876 3563 $url =~ s!/+$!!;
5d3b7cd5 3564 return $RA if ($RA && $RA->{url} eq $url);
f6f09876 3565
d81bf827