mergetool: Make ECMerge use the settings as specified by the user in the GUI
[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, } ],
6fb5375e
TS
189 'blame' => [ \&Git::SVN::Log::cmd_blame,
190 "Show what revision and author last modified each line of a file",
191 {} ],
3397f9df 192);
9d55b41a 193
3397f9df
EW
194my $cmd;
195for (my $i = 0; $i < @ARGV; $i++) {
196 if (defined $cmd{$ARGV[$i]}) {
197 $cmd = $ARGV[$i];
198 splice @ARGV, $i, 1;
199 last;
200 }
201};
202
540424b2
EW
203# make sure we're always running at the top-level working directory
204unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
5253dc33
EW
205 unless (-d $ENV{GIT_DIR}) {
206 if ($git_dir_user_set) {
207 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
208 "but it is not a directory\n";
209 }
210 my $git_dir = delete $ENV{GIT_DIR};
211 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
212 unless (length $cdup) {
213 die "Already at toplevel, but $git_dir ",
214 "not found '$cdup'\n";
215 }
216 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
217 unless (-d $git_dir) {
218 die "$git_dir still not found after going to ",
219 "'$cdup'\n";
220 }
221 $ENV{GIT_DIR} = $git_dir;
222 }
223}
f4dd334b
GH
224
225my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
226
227read_repo_config(\%opts);
228Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
229my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
230 'minimize-connections' => \$Git::SVN::Migration::_minimize,
231 'id|i=s' => \$Git::SVN::default_ref_id,
232 'svn-remote|remote|R=s' => sub {
233 $Git::SVN::no_reuse_existing = 1;
234 $Git::SVN::default_repo_id = $_[1] });
235exit 1 if (!$rv && $cmd && $cmd ne 'log');
236
237usage(0) if $_help;
238version() if $_version;
239usage(1) unless defined $cmd;
240load_authors() if $_authors;
241
0425ea90 242unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
706587fc
EW
243 Git::SVN::Migration::migration_check();
244}
ecc712dd 245Git::SVN::init_vars();
b805b44a
EW
246eval {
247 Git::SVN::verify_remotes_sanity();
248 $cmd{$cmd}->[0]->(@ARGV);
249};
250fatal $@ if $@;
1e889ef3 251post_fetch_checkout();
3397f9df
EW
252exit 0;
253
254####################### primary functions ######################
255sub usage {
256 my $exit = shift || 0;
257 my $fd = $exit ? \*STDERR : \*STDOUT;
258 print $fd <<"";
259git-svn - bidirectional operations between a single Subversion tree and git
260Usage: $0 <command> [options] [arguments]\n
448c81b4
EW
261
262 print $fd "Available commands:\n" unless $cmd;
3397f9df
EW
263
264 foreach (sort keys %cmd) {
448c81b4 265 next if $cmd && $cmd ne $_;
a836a0e1 266 next if /^multi-/; # don't show deprecated commands
b203b769 267 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
aa807bc2 268 foreach (sort keys %{$cmd{$_}->[2]}) {
512b620b
EW
269 # mixed-case options are for .git/config only
270 next if /[A-Z]/ && /^[a-z]+$/i;
448c81b4 271 # prints out arguments as they should be passed:
b8c92cad 272 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
b203b769 273 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
448c81b4
EW
274 "--$_" : "-$_" }
275 split /\|/,$_)," $x\n";
276 }
3397f9df
EW
277 }
278 print $fd <<"";
448c81b4
EW
279\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
280arbitrary identifier if you're tracking multiple SVN branches/repositories in
281one git repository and want to keep them separate. See git-svn(1) for more
282information.
3397f9df
EW
283
284 exit $exit;
285}
286
551ce28f 287sub version {
7d60ab2c 288 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
551ce28f
EW
289 exit 0;
290}
291
8164b652
EW
292sub do_git_init_db {
293 unless (-d $ENV{GIT_DIR}) {
294 my @init_db = ('init');
295 push @init_db, "--template=$_template" if defined $_template;
dadc6d2a
EW
296 if (defined $_shared) {
297 if ($_shared =~ /[a-z]/) {
298 push @init_db, "--shared=$_shared";
299 } else {
300 push @init_db, "--shared";
301 }
302 }
8164b652
EW
303 command_noisy(@init_db);
304 }
0dfaf0a4
EW
305 my $set;
306 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
307 foreach my $i (keys %icv) {
308 die "'$set' and '$i' cannot both be set\n" if $set;
309 next unless defined $icv{$i};
310 command_noisy('config', "$pfx.$i", $icv{$i});
311 $set = $i;
312 }
8164b652
EW
313}
314
dadc6d2a
EW
315sub init_subdir {
316 my $repo_path = shift or return;
317 mkpath([$repo_path]) unless -d $repo_path;
318 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
f30603fc 319 $ENV{GIT_DIR} = '.git';
dadc6d2a
EW
320}
321
0425ea90
EW
322sub cmd_clone {
323 my ($url, $path) = @_;
324 if (!defined $path &&
8f728fb9 325 (defined $_trunk || defined $_branches || defined $_tags ||
326 defined $_stdlayout) &&
0425ea90
EW
327 $url !~ m#^[a-z\+]+://#) {
328 $path = $url;
329 }
0425ea90 330 $path = basename($url) if !defined $path || !length $path;
f30603fc 331 cmd_init($url, $path);
0425ea90
EW
332 Git::SVN::fetch_all($Git::SVN::default_repo_id);
333}
334
d2866f9e 335sub cmd_init {
8f728fb9 336 if (defined $_stdlayout) {
337 $_trunk = 'trunk' if (!defined $_trunk);
338 $_tags = 'tags' if (!defined $_tags);
339 $_branches = 'branches' if (!defined $_branches);
340 }
dadc6d2a
EW
341 if (defined $_trunk || defined $_branches || defined $_tags) {
342 return cmd_multi_init(@_);
03e0ea87 343 }
dadc6d2a
EW
344 my $url = shift or die "SVN repository location required ",
345 "as a command-line argument\n";
346 init_subdir(@_);
8164b652 347 do_git_init_db();
03e0ea87 348
706587fc 349 Git::SVN->init($url);
3397f9df
EW
350}
351
2a3240be 352sub cmd_fetch {
e98671e5
EW
353 if (grep /^\d+=./, @_) {
354 die "'<rev>=<commit>' fetch arguments are ",
355 "no longer supported.\n";
07a1c950 356 }
e98671e5
EW
357 my ($remote) = @_;
358 if (@_ > 1) {
905f8b7d 359 die "Usage: $0 fetch [--all] [svn-remote]\n";
e98671e5
EW
360 }
361 $remote ||= $Git::SVN::default_repo_id;
362 if ($_fetch_all) {
363 cmd_multi_fetch();
364 } else {
365 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
1c8443b0 366 }
2a3240be
EW
367}
368
1ce255dc 369sub cmd_set_tree {
3397f9df
EW
370 my (@commits) = @_;
371 if ($_stdin || !@commits) {
372 print "Reading from stdin...\n";
373 @commits = ();
374 while (<STDIN>) {
1ca72aef 375 if (/\b($sha1_short)\b/o) {
3397f9df
EW
376 unshift @commits, $1;
377 }
378 }
379 }
380 my @revs;
8de010ad 381 foreach my $c (@commits) {
aef4e921 382 my @tmp = command('rev-parse',$c);
8de010ad
EW
383 if (scalar @tmp == 1) {
384 push @revs, $tmp[0];
385 } elsif (scalar @tmp > 1) {
aef4e921 386 push @revs, reverse(command('rev-list',@tmp));
8de010ad 387 } else {
207f1a75 388 fatal "Failed to rev-parse $c";
8de010ad 389 }
3397f9df 390 }
1ce255dc
EW
391 my $gs = Git::SVN->new;
392 my ($r_last, $cmt_last) = $gs->last_rev_commit;
393 $gs->fetch;
97f6987a 394 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
1ce255dc
EW
395 fatal "There are new revisions that were fetched ",
396 "and need to be merged (or acknowledged) ",
397 "before committing.\nlast rev: $r_last\n",
207f1a75 398 " current: $gs->{last_rev}";
a5e0cedc 399 }
1ce255dc
EW
400 $gs->set_tree($_) foreach @revs;
401 print "Done committing ",scalar @revs," revisions to SVN\n";
3157dd9e 402 unlink $gs->{index};
a5e0cedc 403}
8f22562c 404
d7ad3bed
EW
405sub cmd_dcommit {
406 my $head = shift;
c8cfa3e4 407 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
826a9339 408 'Cannot dcommit with a dirty index. Commit your changes first, '
c8cfa3e4 409 . "or stash them with `git stash'.\n";
d7ad3bed 410 $head ||= 'HEAD';
a8ae2623 411 my @refs;
13c823fb 412 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
59b0c24d
MM
413 if ($url) {
414 print "Committing to $url ...\n";
415 }
13c823fb 416 unless ($gs) {
a8ae2623 417 die "Unable to determine upstream SVN information from ",
59b0c24d 418 "$head history.\nPerhaps the repository is empty.";
a8ae2623 419 }
45bf473a 420 my $last_rev;
733a65aa 421 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
751eb395
EW
422 if ($_no_rebase && scalar(@$linear_refs) > 1) {
423 warn "Attempting to commit more than one change while ",
424 "--no-rebase is enabled.\n",
425 "If these changes depend on each other, re-running ",
7dfa16b9 426 "without --no-rebase may be required."
751eb395 427 }
c74d9acf
EW
428 while (1) {
429 my $d = shift @$linear_refs or last;
45bf473a
EW
430 unless (defined $last_rev) {
431 (undef, $last_rev, undef) = cmt_metadata("$d~1");
432 unless (defined $last_rev) {
d7ad3bed 433 fatal "Unable to extract revision information ",
207f1a75 434 "from commit $d~1";
45bf473a
EW
435 }
436 }
b22d4497
EW
437 if ($_dry_run) {
438 print "diff-tree $d~1 $d\n";
439 } else {
751eb395 440 my $cmt_rev;
d7ad3bed 441 my %ed_opts = ( r => $last_rev,
61395354 442 log => get_commit_entry($d)->{log},
645833b5 443 ra => Git::SVN::Ra->new($gs->full_url),
3caf320b
KA
444 config => SVN::Core::config_get_config(
445 $Git::SVN::Ra::config_dir
446 ),
61395354
EW
447 tree_a => "$d~1",
448 tree_b => $d,
449 editor_cb => sub {
450 print "Committed r$_[0]\n";
751eb395
EW
451 $cmt_rev = $_[0];
452 },
a8ae2623 453 svn_path => '');
61395354 454 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
d7ad3bed 455 print "No changes\n$d~1 == $d\n";
733a65aa 456 } elsif ($parents->{$d} && @{$parents->{$d}}) {
751eb395 457 $gs->{inject_parents_dcommit}->{$cmt_rev} =
733a65aa 458 $parents->{$d};
d7ad3bed 459 }
751eb395 460 $_fetch_all ? $gs->fetch_all : $gs->fetch;
7dfa16b9 461 $last_rev = $cmt_rev;
751eb395
EW
462 next if $_no_rebase;
463
464 # we always want to rebase against the current HEAD,
465 # not any head that was passed to us
c74d9acf 466 my @diff = command('diff-tree', $d,
751eb395
EW
467 $gs->refname, '--');
468 my @finish;
469 if (@diff) {
470 @finish = rebase_cmd();
c74d9acf 471 print STDERR "W: $d and ", $gs->refname,
751eb395 472 " differ, using @finish:\n",
c74d9acf 473 join("\n", @diff), "\n";
751eb395
EW
474 } else {
475 print "No changes between current HEAD and ",
476 $gs->refname,
477 "\nResetting to the latest ",
478 $gs->refname, "\n";
479 @finish = qw/reset --mixed/;
480 }
481 command_noisy(@finish, $gs->refname);
c74d9acf
EW
482 if (@diff) {
483 @refs = ();
484 my ($url_, $rev_, $uuid_, $gs_) =
485 working_head_info($head, \@refs);
486 my ($linear_refs_, $parents_) =
487 linearize_history($gs_, \@refs);
488 if (scalar(@$linear_refs) !=
489 scalar(@$linear_refs_)) {
490 fatal "# of revisions changed ",
491 "\nbefore:\n",
492 join("\n", @$linear_refs),
493 "\n\nafter:\n",
494 join("\n", @$linear_refs_), "\n",
495 'If you are attempting to commit ',
496 "merges, try running:\n\t",
497 'git rebase --interactive',
498 '--preserve-merges ',
499 $gs->refname,
500 "\nBefore dcommitting";
501 }
502 if ($url_ ne $url) {
503 fatal "URL mismatch after rebase: ",
504 "$url_ != $url";
505 }
506 if ($uuid_ ne $uuid) {
507 fatal "uuid mismatch after rebase: ",
508 "$uuid_ != $uuid";
509 }
510 # remap parents
511 my (%p, @l, $i);
512 for ($i = 0; $i < scalar @$linear_refs; $i++) {
513 my $new = $linear_refs_->[$i] or next;
514 $p{$new} =
515 $parents->{$linear_refs->[$i]};
516 push @l, $new;
517 }
518 $parents = \%p;
519 $linear_refs = \@l;
520 }
b22d4497
EW
521 }
522 }
3157dd9e 523 unlink $gs->{index};
b22d4497
EW
524}
525
26e60160 526sub cmd_find_rev {
ea14e6c5
MAL
527 my $revision_or_hash = shift or die "SVN or git revision required ",
528 "as a command-line argument\n";
26e60160
AR
529 my $result;
530 if ($revision_or_hash =~ /^r\d+$/) {
b3cb7e45
AR
531 my $head = shift;
532 $head ||= 'HEAD';
533 my @refs;
534 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
535 unless ($gs) {
536 die "Unable to determine upstream SVN information from ",
537 "$head history\n";
26e60160 538 }
b3cb7e45 539 my $desired_revision = substr($revision_or_hash, 1);
060610c5 540 $result = $gs->rev_map_get($desired_revision);
26e60160
AR
541 } else {
542 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
543 $result = $rev;
544 }
545 print "$result\n" if $result;
546}
547
905f8b7d
EW
548sub cmd_rebase {
549 command_noisy(qw/update-index --refresh/);
13c823fb
EW
550 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
551 unless ($gs) {
905f8b7d
EW
552 die "Unable to determine upstream SVN information from ",
553 "working tree history\n";
554 }
905f8b7d
EW
555 if (command(qw/diff-index HEAD --/)) {
556 print STDERR "Cannot rebase with uncommited changes:\n";
557 command_noisy('status');
558 exit 1;
559 }
dee41f3e 560 unless ($_local) {
cec0d5a3
SG
561 # rebase will checkout for us, so no need to do it explicitly
562 $_no_checkout = 'true';
dee41f3e
EW
563 $_fetch_all ? $gs->fetch_all : $gs->fetch;
564 }
905f8b7d
EW
565 command_noisy(rebase_cmd(), $gs->refname);
566}
567
5969cbe1 568sub cmd_show_ignore {
13c823fb
EW
569 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
570 $gs ||= Git::SVN->new;
5969cbe1 571 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
01bdab84
BS
572 $gs->prop_walk($gs->{path}, $r, sub {
573 my ($gs, $path, $props) = @_;
574 print STDOUT "\n# $path\n";
575 my $s = $props->{'svn:ignore'} or return;
576 $s =~ s/[\r\n]+/\n/g;
577 chomp $s;
578 $s =~ s#^#$path#gm;
579 print STDOUT "$s\n";
580 });
a5e0cedc
EW
581}
582
2d879792
VK
583sub cmd_show_externals {
584 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
585 $gs ||= Git::SVN->new;
586 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
587 $gs->prop_walk($gs->{path}, $r, sub {
588 my ($gs, $path, $props) = @_;
589 print STDOUT "\n# $path\n";
590 my $s = $props->{'svn:externals'} or return;
591 $s =~ s/[\r\n]+/\n/g;
592 chomp $s;
593 $s =~ s#^#$path#gm;
594 print STDOUT "$s\n";
595 });
596}
597
d05ddec5
BS
598sub cmd_create_ignore {
599 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
600 $gs ||= Git::SVN->new;
601 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
602 $gs->prop_walk($gs->{path}, $r, sub {
603 my ($gs, $path, $props) = @_;
604 # $path is of the form /path/to/dir/
605 my $ignore = '.' . $path . '.gitignore';
606 my $s = $props->{'svn:ignore'} or return;
607 open(GITIGNORE, '>', $ignore)
207f1a75 608 or fatal("Failed to open `$ignore' for writing: $!");
d05ddec5
BS
609 $s =~ s/[\r\n]+/\n/g;
610 chomp $s;
611 # Prefix all patterns so that the ignore doesn't apply
612 # to sub-directories.
613 $s =~ s#^#/#gm;
614 print GITIGNORE "$s\n";
615 close(GITIGNORE)
207f1a75 616 or fatal("Failed to close `$ignore': $!");
c4c66b26 617 command_noisy('add', '-f', $ignore);
d05ddec5
BS
618 });
619}
620
b2b3ada7
DK
621sub canonicalize_path {
622 my ($path) = @_;
e6fefa92
DK
623 my $dot_slash_added = 0;
624 if (substr($path, 0, 1) ne "/") {
625 $path = "./" . $path;
626 $dot_slash_added = 1;
627 }
b2b3ada7
DK
628 # File::Spec->canonpath doesn't collapse x/../y into y (for a
629 # good reason), so let's do this manually.
630 $path =~ s#/+#/#g;
631 $path =~ s#/\.(?:/|$)#/#g;
632 $path =~ s#/[^/]+/\.\.##g;
633 $path =~ s#/$##g;
e6fefa92 634 $path =~ s#^\./## if $dot_slash_added;
b2b3ada7
DK
635 return $path;
636}
637
15153451
BS
638# get_svnprops(PATH)
639# ------------------
51e057cf 640# Helper for cmd_propget and cmd_proplist below.
15153451
BS
641sub get_svnprops {
642 my $path = shift;
643 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
644 $gs ||= Git::SVN->new;
645
646 # prefix THE PATH by the sub-directory from which the user
647 # invoked us.
648 $path = $cmd_dir_prefix . $path;
207f1a75 649 fatal("No such file or directory: $path") unless -e $path;
15153451
BS
650 my $is_dir = -d $path ? 1 : 0;
651 $path = $gs->{path} . '/' . $path;
652
653 # canonicalize the path (otherwise libsvn will abort or fail to
654 # find the file)
b2b3ada7 655 $path = canonicalize_path($path);
15153451
BS
656
657 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
658 my $props;
659 if ($is_dir) {
660 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
661 }
662 else {
663 (undef, $props) = $gs->ra->get_file($path, $r, undef);
664 }
665 return $props;
666}
667
668# cmd_propget (PROP, PATH)
669# ------------------------
670# Print the SVN property PROP for PATH.
671sub cmd_propget {
672 my ($prop, $path) = @_;
673 $path = '.' if not defined $path;
674 usage(1) if not defined $prop;
675 my $props = get_svnprops($path);
676 if (not defined $props->{$prop}) {
207f1a75 677 fatal("`$path' does not have a `$prop' SVN property.");
15153451
BS
678 }
679 print $props->{$prop} . "\n";
680}
681
51e057cf
BS
682# cmd_proplist (PATH)
683# -------------------
684# Print the list of SVN properties for PATH.
685sub cmd_proplist {
686 my $path = shift;
687 $path = '.' if not defined $path;
688 my $props = get_svnprops($path);
689 print "Properties on '$path':\n";
690 foreach (sort keys %{$props}) {
691 print " $_\n";
692 }
693}
694
8164b652 695sub cmd_multi_init {
9d55b41a 696 my $url = shift;
98327e58
EW
697 unless (defined $_trunk || defined $_branches || defined $_tags) {
698 usage(1);
9d55b41a 699 }
dc431666
EW
700
701 # there are currently some bugs that prevent multi-init/multi-fetch
702 # setups from working well without this.
703 $Git::SVN::_minimize_url = 1;
704
8164b652 705 $_prefix = '' unless defined $_prefix;
dadc6d2a
EW
706 if (defined $url) {
707 $url =~ s#/+$##;
708 init_subdir(@_);
709 }
f30603fc 710 do_git_init_db();
98327e58 711 if (defined $_trunk) {
706587fc
EW
712 my $trunk_ref = $_prefix . 'trunk';
713 # try both old-style and new-style lookups:
714 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
8164b652 715 unless ($gs_trunk) {
706587fc
EW
716 my ($trunk_url, $trunk_path) =
717 complete_svn_url($url, $_trunk);
718 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
719 undef, $trunk_ref);
98327e58 720 }
c35b96e7 721 }
706587fc 722 return unless defined $_branches || defined $_tags;
e7db67e6
EW
723 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
724 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
725 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
9d55b41a
EW
726}
727
1c8443b0 728sub cmd_multi_fetch {
0af9c9f9
EW
729 my $remotes = Git::SVN::read_all_remotes();
730 foreach my $repo_id (sort keys %$remotes) {
db03cd24 731 if ($remotes->{$repo_id}->{url}) {
4bb9ed04
EW
732 Git::SVN::fetch_all($repo_id, $remotes);
733 }
9d55b41a 734 }
9d55b41a
EW
735}
736
44320b9e
EW
737# this command is special because it requires no metadata
738sub cmd_commit_diff {
739 my ($ta, $tb, $url) = @_;
740 my $usage = "Usage: $0 commit-diff -r<revision> ".
207f1a75 741 "<tree-ish> <tree-ish> [<URL>]";
44320b9e 742 fatal($usage) if (!defined $ta || !defined $tb);
d3a840dc 743 my $svn_path;
44320b9e
EW
744 if (!defined $url) {
745 my $gs = eval { Git::SVN->new };
746 if (!$gs) {
747 fatal("Needed URL or usable git-svn --id in ",
748 "the command-line\n", $usage);
749 }
750 $url = $gs->{url};
d3a840dc 751 $svn_path = $gs->{path};
44320b9e
EW
752 }
753 unless (defined $_revision) {
754 fatal("-r|--revision is a required argument\n", $usage);
755 }
756 if (defined $_message && defined $_file) {
757 fatal("Both --message/-m and --file/-F specified ",
758 "for the commit message.\n",
207f1a75 759 "I have no idea what you mean");
44320b9e
EW
760 }
761 if (defined $_file) {
762 $_message = file_to_s($_file);
763 } else {
764 $_message ||= get_commit_entry($tb)->{log};
765 }
766 my $ra ||= Git::SVN::Ra->new($url);
d3a840dc 767 $svn_path ||= $ra->{svn_path};
44320b9e
EW
768 my $r = $_revision;
769 if ($r eq 'HEAD') {
770 $r = $ra->get_latest_revnum;
771 } elsif ($r !~ /^\d+$/) {
772 die "revision argument: $r not understood by git-svn\n";
773 }
61395354
EW
774 my %ed_opts = ( r => $r,
775 log => $_message,
776 ra => $ra,
777 tree_a => $ta,
778 tree_b => $tb,
779 editor_cb => sub { print "Committed r$_[0]\n" },
780 svn_path => $svn_path );
781 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
44320b9e
EW
782 print "No changes\n$ta == $tb\n";
783 }
44320b9e
EW
784}
785
e6fefa92
DK
786sub cmd_info {
787 my $path = canonicalize_path(shift or ".");
788 unless (scalar(@_) == 0) {
789 die "Too many arguments specified\n";
790 }
791
792 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
793
794 if (!$file_type && !$diff_status) {
795 print STDERR "$path: (Not a versioned resource)\n\n";
796 return;
797 }
798
799 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
800 unless ($gs) {
801 die "Unable to determine upstream SVN information from ",
802 "working tree history\n";
803 }
804 my $full_url = $url . ($path eq "." ? "" : "/$path");
805
8b014d71
DK
806 if ($_url) {
807 print $full_url, "\n";
808 return;
809 }
810
e6fefa92
DK
811 my $result = "Path: $path\n";
812 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
813 $result .= "URL: " . $full_url . "\n";
814
a5460eb7
EW
815 eval {
816 my $repos_root = $gs->repos_root;
817 Git::SVN::remove_username($repos_root);
818 $result .= "Repository Root: $repos_root\n";
819 };
820 if ($@) {
821 $result .= "Repository Root: (offline)\n";
822 }
e6fefa92
DK
823 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
824 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
825
826 $result .= "Node Kind: " .
827 ($file_type eq "dir" ? "directory" : "file") . "\n";
828
829 my $schedule = $diff_status eq "A"
830 ? "add"
831 : ($diff_status eq "D" ? "delete" : "normal");
832 $result .= "Schedule: $schedule\n";
833
834 if ($diff_status eq "A") {
835 print $result, "\n";
836 return;
837 }
838
839 my ($lc_author, $lc_rev, $lc_date_utc);
840 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
841 my $log = command_output_pipe(@args);
842 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
843 while (<$log>) {
844 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
845 $lc_author = $1;
846 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
847 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
848 (undef, $lc_rev, undef) = ::extract_metadata($1);
849 }
850 }
851 close $log;
852
853 Git::SVN::Log::set_local_timezone();
854
855 $result .= "Last Changed Author: $lc_author\n";
856 $result .= "Last Changed Rev: $lc_rev\n";
857 $result .= "Last Changed Date: " .
858 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
859
860 if ($file_type ne "dir") {
861 my $text_last_updated_date =
862 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
863 $result .=
864 "Text Last Updated: " .
865 Git::SVN::Log::format_svn_date($text_last_updated_date) .
866 "\n";
867 my $checksum;
868 if ($diff_status eq "D") {
869 my ($fh, $ctx) =
870 command_output_pipe(qw(cat-file blob), "HEAD:$path");
871 if ($file_type eq "link") {
872 my $file_name = <$fh>;
8d7c4fad 873 $checksum = md5sum("link $file_name");
e6fefa92 874 } else {
8d7c4fad 875 $checksum = md5sum($fh);
e6fefa92
DK
876 }
877 command_close_pipe($fh, $ctx);
878 } elsif ($file_type eq "link") {
879 my $file_name =
880 command(qw(cat-file blob), "HEAD:$path");
881 $checksum =
8d7c4fad 882 md5sum("link " . $file_name);
e6fefa92
DK
883 } else {
884 open FILE, "<", $path or die $!;
8d7c4fad 885 $checksum = md5sum(\*FILE);
e6fefa92
DK
886 close FILE or die $!;
887 }
888 $result .= "Checksum: " . $checksum . "\n";
889 }
890
891 print $result, "\n";
892}
893
3397f9df
EW
894########################### utility functions #########################
895
905f8b7d
EW
896sub rebase_cmd {
897 my @cmd = qw/rebase/;
898 push @cmd, '-v' if $_verbose;
899 push @cmd, qw/--merge/ if $_merge;
900 push @cmd, "--strategy=$_strategy" if $_strategy;
901 @cmd;
902}
903
1e889ef3
EW
904sub post_fetch_checkout {
905 return if $_no_checkout;
906 my $gs = $Git::SVN::_head or return;
907 return if verify_ref('refs/heads/master^0');
908
909 my $valid_head = verify_ref('HEAD^0');
910 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
911 return if ($valid_head || !verify_ref('HEAD^0'));
912
913 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
914 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
915 return if -f $index;
916
7ae3df8c 917 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1e889ef3
EW
918 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
919 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
920 print STDERR "Checked out HEAD:\n ",
921 $gs->full_url, " r", $gs->last_rev, "\n";
922}
923
98327e58
EW
924sub complete_svn_url {
925 my ($url, $path) = @_;
926 $path =~ s#/+$##;
98327e58 927 if ($path !~ m#^[a-z\+]+://#) {
98327e58
EW
928 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
929 fatal("E: '$path' is not a complete URL ",
207f1a75 930 "and a separate URL is not specified");
98327e58 931 }
706587fc 932 return ($url, $path);
98327e58 933 }
706587fc 934 return ($path, '');
98327e58
EW
935}
936
9d55b41a 937sub complete_url_ls_init {
706587fc
EW
938 my ($ra, $repo_path, $switch, $pfx) = @_;
939 unless ($repo_path) {
9d55b41a
EW
940 print STDERR "W: $switch not specified\n";
941 return;
942 }
706587fc
EW
943 $repo_path =~ s#/+$##;
944 if ($repo_path =~ m#^[a-z\+]+://#) {
945 $ra = Git::SVN::Ra->new($repo_path);
946 $repo_path = '';
e7db67e6 947 } else {
706587fc 948 $repo_path =~ s#^/+##;
e7db67e6 949 unless ($ra) {
706587fc 950 fatal("E: '$repo_path' is not a complete URL ",
207f1a75 951 "and a separate URL is not specified");
8164b652 952 }
e7db67e6 953 }
706587fc 954 my $url = $ra->{url};
b4d57e5e
EW
955 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
956 my $k = "svn-remote.$gs->{repo_id}.url";
957 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
958 if ($orig_url && ($orig_url ne $gs->{url})) {
959 die "$k already set: $orig_url\n",
960 "wanted to set to: $gs->{url}\n";
88cf4107 961 }
b4d57e5e 962 command_oneline('config', $k, $gs->{url}) unless $orig_url;
ed0b9d43 963 my $remote_path = "$ra->{svn_path}/$repo_path";
b4d57e5e
EW
964 $remote_path =~ s#/+#/#g;
965 $remote_path =~ s#^/##g;
ed0b9d43 966 $remote_path .= "/*" if $remote_path !~ /\*/;
b4d57e5e
EW
967 my ($n) = ($switch =~ /^--(\w+)/);
968 if (length $pfx && $pfx !~ m#/$#) {
969 die "--prefix='$pfx' must have a trailing slash '/'\n";
9d55b41a 970 }
b4d57e5e
EW
971 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
972 "$remote_path:refs/remotes/$pfx*");
9d55b41a
EW
973}
974
aef4e921
EW
975sub verify_ref {
976 my ($ref) = @_;
2c5c1d53
EW
977 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
978 { STDERR => 0 }); };
aef4e921
EW
979}
980
a5e0cedc 981sub get_tree_from_treeish {
cf52b8f0 982 my ($treeish) = @_;
44320b9e 983 # $treeish can be a symbolic ref, too:
aef4e921 984 my $type = command_oneline(qw/cat-file -t/, $treeish);
cf52b8f0
EW
985 my $expected;
986 while ($type eq 'tag') {
aef4e921 987 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
cf52b8f0
EW
988 }
989 if ($type eq 'commit') {
aef4e921
EW
990 $expected = (grep /^tree /, command(qw/cat-file commit/,
991 $treeish))[0];
44320b9e 992 ($expected) = ($expected =~ /^tree ($sha1)$/o);
cf52b8f0
EW
993 die "Unable to get tree from $treeish\n" unless $expected;
994 } elsif ($type eq 'tree') {
995 $expected = $treeish;
996 } else {
997 die "$treeish is a $type, expected tree, tag or commit\n";
998 }
a5e0cedc
EW
999 return $expected;
1000}
1001
44320b9e
EW
1002sub get_commit_entry {
1003 my ($treeish) = shift;
1004 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1005 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1006 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1007 open my $log_fh, '>', $commit_editmsg or croak $!;
3397f9df 1008
44320b9e 1009 my $type = command_oneline(qw/cat-file -t/, $treeish);
4ad4515d 1010 if ($type eq 'commit' || $type eq 'tag') {
aef4e921 1011 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
44320b9e 1012 $type, $treeish);
3397f9df
EW
1013 my $in_msg = 0;
1014 while (<$msg_fh>) {
1015 if (!$in_msg) {
1016 $in_msg = 1 if (/^\s*$/);
df746c5a 1017 } elsif (/^git-svn-id: /) {
44320b9e
EW
1018 # skip this for now, we regenerate the
1019 # correct one on re-fetch anyways
1020 # TODO: set *:merge properties or like...
3397f9df 1021 } else {
44320b9e 1022 print $log_fh $_ or croak $!;
3397f9df
EW
1023 }
1024 }
aef4e921 1025 command_close_pipe($msg_fh, $ctx);
3397f9df 1026 }
44320b9e 1027 close $log_fh or croak $!;
3397f9df
EW
1028
1029 if ($_edit || ($type eq 'tree')) {
1030 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
44320b9e
EW
1031 # TODO: strip out spaces, comments, like git-commit.sh
1032 system($editor, $commit_editmsg);
3397f9df 1033 }
44320b9e
EW
1034 rename $commit_editmsg, $commit_msg or croak $!;
1035 open $log_fh, '<', $commit_msg or croak $!;
1036 { local $/; chomp($log_entry{log} = <$log_fh>); }
1037 close $log_fh or croak $!;
1038 unlink $commit_msg;
1039 \%log_entry;
a5e0cedc
EW
1040}
1041
3397f9df
EW
1042sub s_to_file {
1043 my ($str, $file, $mode) = @_;
1044 open my $fd,'>',$file or croak $!;
1045 print $fd $str,"\n" or croak $!;
1046 close $fd or croak $!;
1047 chmod ($mode &~ umask, $file) if (defined $mode);
1048}
1049
1050sub file_to_s {
1051 my $file = shift;
1052 open my $fd,'<',$file or croak "$!: file: $file\n";
1053 local $/;
1054 my $ret = <$fd>;
1055 close $fd or croak $!;
1056 $ret =~ s/\s*$//s;
1057 return $ret;
1058}
1059
eeb0abe0
EW
1060# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1061sub load_authors {
1062 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
f8c9d1d2 1063 my $log = $cmd eq 'log';
eeb0abe0
EW
1064 while (<$authors>) {
1065 chomp;
575d025c 1066 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
eeb0abe0 1067 my ($user, $name, $email) = ($1, $2, $3);
f8c9d1d2
EW
1068 if ($log) {
1069 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1070 } else {
1071 $users{$user} = [$name, $email];
1072 }
79bb8d88
EW
1073 }
1074 close $authors or croak $!;
1075}
1076
e0d10e1c 1077# convert GetOpt::Long specs for use by git-config
b8c92cad 1078sub read_repo_config {
706587fc 1079 return unless -d $ENV{GIT_DIR};
b8c92cad 1080 my $opts = shift;
97ae0911 1081 my @config_only;
b8c92cad 1082 foreach my $o (keys %$opts) {
97ae0911
EW
1083 # if we have mixedCase and a long option-only, then
1084 # it's a config-only variable that we don't need for
1085 # the command-line.
1086 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
b8c92cad 1087 my $v = $opts->{$o};
97ae0911 1088 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
b8c92cad 1089 $key =~ s/-//g;
e0d10e1c 1090 my $arg = 'git-config';
b8c92cad
EW
1091 $arg .= ' --int' if ($o =~ /[:=]i$/);
1092 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1093 if (ref $v eq 'ARRAY') {
1094 chomp(my @tmp = `$arg --get-all svn.$key`);
1095 @$v = @tmp if @tmp;
1096 } else {
1097 chomp(my $tmp = `$arg --get svn.$key`);
7774284a 1098 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
b8c92cad
EW
1099 $$v = $tmp;
1100 }
1101 }
1102 }
97ae0911 1103 delete @$opts{@config_only} if @config_only;
b8c92cad
EW
1104}
1105
79bb8d88 1106sub extract_metadata {
c1927a85 1107 my $id = shift or return (undef, undef, undef);
3dfab993 1108 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
79bb8d88 1109 \s([a-f\d\-]+)$/x);
e70dc780 1110 if (!defined $rev || !$uuid || !$url) {
79bb8d88 1111 # some of the original repositories I made had
82e5a82f 1112 # identifiers like this:
3dfab993 1113 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
79bb8d88
EW
1114 }
1115 return ($url, $rev, $uuid);
1116}
1117
c1927a85
EW
1118sub cmt_metadata {
1119 return extract_metadata((grep(/^git-svn-id: /,
aef4e921 1120 command(qw/cat-file commit/, shift)))[-1]);
c1927a85
EW
1121}
1122
905f8b7d
EW
1123sub working_head_info {
1124 my ($head, $refs) = @_;
b6309ac2 1125 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
05b4df31 1126 my ($fh, $ctx) = command_output_pipe(@args, $head);
3dfab993 1127 my $hash;
40cb8f8f 1128 my %max;
3dfab993
SV
1129 while (<$fh>) {
1130 if ( m{^commit ($::sha1)$} ) {
1131 unshift @$refs, $hash if $hash and $refs;
1132 $hash = $1;
1133 next;
1134 }
1135 next unless s{^\s*(git-svn-id:)}{$1};
1136 my ($url, $rev, $uuid) = extract_metadata($_);
13c823fb 1137 if (defined $url && defined $rev) {
40cb8f8f 1138 next if $max{$url} and $max{$url} < $rev;
13c823fb 1139 if (my $gs = Git::SVN->find_by_url($url)) {
060610c5 1140 my $c = $gs->rev_map_get($rev);
b03c7a63 1141 if ($c && $c eq $hash) {
13c823fb
EW
1142 close $fh; # break the pipe
1143 return ($url, $rev, $uuid, $gs);
40cb8f8f 1144 } else {
060610c5 1145 $max{$url} ||= $gs->rev_map_max;
13c823fb
EW
1146 }
1147 }
1148 }
905f8b7d 1149 }
13c823fb
EW
1150 command_close_pipe($fh, $ctx);
1151 (undef, undef, undef, undef);
905f8b7d
EW
1152}
1153
733a65aa
EW
1154sub read_commit_parents {
1155 my ($parents, $c) = @_;
7b02b85a
EW
1156 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1157 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1158 @{$parents->{$c}} = split(/ /, $p);
733a65aa
EW
1159}
1160
1161sub linearize_history {
1162 my ($gs, $refs) = @_;
1163 my %parents;
1164 foreach my $c (@$refs) {
1165 read_commit_parents(\%parents, $c);
1166 }
1167
1168 my @linear_refs;
1169 my %skip = ();
1170 my $last_svn_commit = $gs->last_commit;
1171 foreach my $c (reverse @$refs) {
1172 next if $c eq $last_svn_commit;
1173 last if $skip{$c};
1174
1175 unshift @linear_refs, $c;
1176 $skip{$c} = 1;
1177
1178 # we only want the first parent to diff against for linear
1179 # history, we save the rest to inject when we finalize the
1180 # svn commit
1181 my $fp_a = verify_ref("$c~1");
1182 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1183 if (!$fp_a || !$fp_b) {
1184 die "Commit $c\n",
1185 "has no parent commit, and therefore ",
1186 "nothing to diff against.\n",
1187 "You should be working from a repository ",
1188 "originally created by git-svn\n";
1189 }
1190 if ($fp_a ne $fp_b) {
1191 die "$c~1 = $fp_a, however parsing commit $c ",
1192 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1193 }
1194
1195 foreach my $p (@{$parents{$c}}) {
1196 $skip{$p} = 1;
1197 }
1198 }
1199 (\@linear_refs, \%parents);
1200}
1201
e6fefa92
DK
1202sub find_file_type_and_diff_status {
1203 my ($path) = @_;
1204 return ('dir', '') if $path eq '.';
1205
1206 my $diff_output =
1207 command_oneline(qw(diff --cached --name-status --), $path) || "";
1208 my $diff_status = (split(' ', $diff_output))[0] || "";
1209
1210 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1211
1212 return (undef, undef) if !$diff_status && !$ls_tree;
1213
1214 if ($diff_status eq "A") {
1215 return ("link", $diff_status) if -l $path;
1216 return ("dir", $diff_status) if -d $path;
1217 return ("file", $diff_status);
1218 }
1219
1220 my $mode = (split(' ', $ls_tree))[0] || "";
1221
1222 return ("link", $diff_status) if $mode eq "120000";
1223 return ("dir", $diff_status) if $mode eq "040000";
1224 return ("file", $diff_status);
1225}
1226
b2b3ada7
DK
1227sub md5sum {
1228 my $arg = shift;
1229 my $ref = ref $arg;
1230 my $md5 = Digest::MD5->new();
1231 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1232 $md5->addfile($arg) or croak $!;
1233 } elsif ($ref eq 'SCALAR') {
1234 $md5->add($$arg) or croak $!;
1235 } elsif (!$ref) {
1236 $md5->add($arg) or croak $!;
1237 } else {
1238 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1239 }
1240 return $md5->hexdigest();
1241}
1242
9b981fc6
EW
1243package Git::SVN;
1244use strict;
1245use warnings;
060610c5
EW
1246use Fcntl qw/:DEFAULT :seek/;
1247use constant rev_map_fmt => 'NH40';
ecc712dd 1248use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
62e349d2 1249 $_repack $_repack_flags $_use_svm_props $_head
70ae04e4
AW
1250 $_use_svnsync_props $no_reuse_existing $_minimize_url
1251 $_use_log_author/;
9b981fc6
EW
1252use Carp qw/croak/;
1253use File::Path qw/mkpath/;
373274f9 1254use File::Copy qw/copy/;
9b981fc6
EW
1255use IPC::Open3;
1256
94bc914c
KW
1257my ($_gc_nr, $_gc_period);
1258
9b981fc6
EW
1259# properties that we do not log:
1260my %SKIP_PROP;
1261BEGIN {
1262 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1263 svn:special svn:executable
1264 svn:entry:committed-rev
1265 svn:entry:last-author
1266 svn:entry:uuid
1267 svn:entry:committed-date/;
91b03282
EW
1268
1269 # some options are read globally, but can be overridden locally
1270 # per [svn-remote "..."] section. Command-line options will *NOT*
1271 # override options set in an [svn-remote "..."] section
c5f71ad0
SV
1272 no strict 'refs';
1273 for my $option (qw/follow_parent no_metadata use_svm_props
1274 use_svnsync_props/) {
1275 my $key = $option;
91b03282 1276 $key =~ tr/_//d;
c5f71ad0
SV
1277 my $prop = "-$option";
1278 *$option = sub {
1279 my ($self) = @_;
1280 return $self->{$prop} if exists $self->{$prop};
1281 my $k = "svn-remote.$self->{repo_id}.$key";
1282 eval { command_oneline(qw/config --get/, $k) };
1283 if ($@) {
1284 $self->{$prop} = ${"Git::SVN::_$option"};
91b03282 1285 } else {
c5f71ad0
SV
1286 my $v = command_oneline(qw/config --bool/,$k);
1287 $self->{$prop} = $v eq 'false' ? 0 : 1;
91b03282 1288 }
c5f71ad0
SV
1289 return $self->{$prop};
1290 }
91b03282 1291 }
9b981fc6
EW
1292}
1293
321b1842
EW
1294my (%LOCKFILES, %INDEX_FILES);
1295END {
1296 unlink keys %LOCKFILES if %LOCKFILES;
1297 unlink keys %INDEX_FILES if %INDEX_FILES;
1298}
373274f9 1299
4bb9ed04
EW
1300sub resolve_local_globs {
1301 my ($url, $fetch, $glob_spec) = @_;
1302 return unless defined $glob_spec;
1303 my $ref = $glob_spec->{ref};
1304 my $path = $glob_spec->{path};
1305 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1306 next unless m#^refs/remotes/$ref->{regex}$#;
1307 my $p = $1;
bf655fd7
RE
1308 my $pathname = desanitize_refname($path->full_path($p));
1309 my $refname = desanitize_refname($ref->full_path($p));
4bb9ed04
EW
1310 if (my $existing = $fetch->{$pathname}) {
1311 if ($existing ne $refname) {
1312 die "Refspec conflict:\n",
1313 "existing: refs/remotes/$existing\n",
1314 " globbed: refs/remotes/$refname\n";
1315 }
1316 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
4e9f6cc7 1317 $u =~ s!^\Q$url\E(/|$)!! or die
4bb9ed04
EW
1318 "refs/remotes/$refname: '$url' not found in '$u'\n";
1319 if ($pathname ne $u) {
1320 warn "W: Refspec glob conflict ",
1321 "(ref: refs/remotes/$refname):\n",
1322 "expected path: $pathname\n",
1323 " real path: $u\n",
1324 "Continuing ahead with $u\n";
1325 next;
1326 }
1327 } else {
4bb9ed04
EW
1328 $fetch->{$pathname} = $refname;
1329 }
1330 }
1331}
1332
e98671e5
EW
1333sub parse_revision_argument {
1334 my ($base, $head) = @_;
1335 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1336 return ($base, $head);
1337 }
1338 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1339 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1340 return ($head, $head) if ($::_revision eq 'HEAD');
1341 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1342 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1343 die "revision argument: $::_revision not understood by git-svn\n";
1344}
1345
0af9c9f9 1346sub fetch_all {
4bb9ed04 1347 my ($repo_id, $remotes) = @_;
905f8b7d
EW
1348 if (ref $repo_id) {
1349 my $gs = $repo_id;
1350 $repo_id = undef;
1351 $repo_id = $gs->{repo_id};
1352 }
1353 $remotes ||= read_all_remotes();
7447b4bc
EW
1354 my $remote = $remotes->{$repo_id} or
1355 die "[svn-remote \"$repo_id\"] unknown\n";
e518192f 1356 my $fetch = $remote->{fetch};
7447b4bc 1357 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
e518192f 1358 my (@gs, @globs);
0af9c9f9 1359 my $ra = Git::SVN::Ra->new($url);
26a62d57 1360 my $uuid = $ra->get_uuid;
0af9c9f9 1361 my $head = $ra->get_latest_revnum;
28710f74 1362 my $base = defined $fetch ? $head : 0;
e518192f
EW
1363
1364 # read the max revs for wildcard expansion (branches/*, tags/*)
1365 foreach my $t (qw/branches tags/) {
1366 defined $remote->{$t} or next;
1367 push @globs, $remote->{$t};
93f2689c
EW
1368 my $max_rev = eval { tmp_config(qw/--int --get/,
1369 "svn-remote.$repo_id.${t}-maxRev") };
1370 if (defined $max_rev && ($max_rev < $base)) {
1371 $base = $max_rev;
d6d3346b
EW
1372 } elsif (!defined $max_rev) {
1373 $base = 0;
e518192f
EW
1374 }
1375 }
1376
db03cd24
EW
1377 if ($fetch) {
1378 foreach my $p (sort keys %$fetch) {
1379 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
060610c5 1380 my $lr = $gs->rev_map_max;
db03cd24
EW
1381 if (defined $lr) {
1382 $base = $lr if ($lr < $base);
1383 }
1384 push @gs, $gs;
0af9c9f9 1385 }
0af9c9f9 1386 }
e98671e5
EW
1387
1388 ($base, $head) = parse_revision_argument($base, $head);
e518192f 1389 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
0af9c9f9
EW
1390}
1391
47e39c55
EW
1392sub read_all_remotes {
1393 my $r = {};
8b8fc068 1394 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
47e39c55 1395 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
46cf98ba
EW
1396 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1397 $local_ref =~ s{^/}{};
1398 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
47e39c55
EW
1399 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1400 $r->{$1}->{url} = $2;
4bb9ed04
EW
1401 } elsif (m!^(.+)\.(branches|tags)=
1402 (.*):refs/remotes/(.+)\s*$/!x) {
1403 my ($p, $g) = ($3, $4);
1404 my $rs = $r->{$1}->{$2} = {
e518192f 1405 t => $2,
93f2689c 1406 remote => $1,
4bb9ed04
EW
1407 path => Git::SVN::GlobSpec->new($p),
1408 ref => Git::SVN::GlobSpec->new($g) };
1409 if (length($rs->{ref}->{right}) != 0) {
1410 die "The '*' glob character must be the last ",
1411 "character of '$g'\n";
1412 }
47e39c55
EW
1413 }
1414 }
1415 $r;
1416}
1417
ecc712dd 1418sub init_vars {
94bc914c 1419 $_gc_nr = $_gc_period = 1000;
af788a6e
KW
1420 if (defined $_repack || defined $_repack_flags) {
1421 warn "Repack options are obsolete; they have no effect.\n";
1422 }
ecc712dd
EW
1423}
1424
b805b44a 1425sub verify_remotes_sanity {
536c4b09 1426 return unless -d $ENV{GIT_DIR};
b805b44a
EW
1427 my %seen;
1428 foreach (command(qw/config -l/)) {
1429 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1430 if ($seen{$1}) {
1431 die "Remote ref refs/remote/$1 is tracked by",
1432 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1433 "Please resolve this ambiguity in ",
1434 "your git configuration file before ",
1435 "continuing\n";
1436 }
1437 $seen{$1} = $_;
1438 }
1439 }
1440}
1441
47e39c55 1442# we allow more chars than remotes2config.sh...
706587fc
EW
1443sub sanitize_remote_name {
1444 my ($name) = @_;
47e39c55 1445 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
706587fc
EW
1446 $name;
1447}
1448
e6434f87
EW
1449sub find_existing_remote {
1450 my ($url, $remotes) = @_;
befc9adc 1451 return undef if $no_reuse_existing;
e6434f87
EW
1452 my $existing;
1453 foreach my $repo_id (keys %$remotes) {
1454 my $u = $remotes->{$repo_id}->{url} or next;
1455 next if $u ne $url;
1456 $existing = $repo_id;
1457 last;
1458 }
1459 $existing;
1460}
b805b44a 1461
e6434f87 1462sub init_remote_config {
d8115c51 1463 my ($self, $url, $no_write) = @_;
e6434f87
EW
1464 $url =~ s!/+$!!; # strip trailing slash
1465 my $r = read_all_remotes();
1466 my $existing = find_existing_remote($url, $r);
1467 if ($existing) {
e518192f
EW
1468 unless ($no_write) {
1469 print STDERR "Using existing ",
1470 "[svn-remote \"$existing\"]\n";
1471 }
e6434f87 1472 $self->{repo_id} = $existing;
4a1bb4c3 1473 } elsif ($_minimize_url) {
e6434f87
EW
1474 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1475 $existing = find_existing_remote($min_url, $r);
1476 if ($existing) {
e518192f
EW
1477 unless ($no_write) {
1478 print STDERR "Using existing ",
1479 "[svn-remote \"$existing\"]\n";
1480 }
e6434f87
EW
1481 $self->{repo_id} = $existing;
1482 }
1483 if ($min_url ne $url) {
e518192f
EW
1484 unless ($no_write) {
1485 print STDERR "Using higher level of URL: ",
1486 "$url => $min_url\n";
1487 }
e6434f87
EW
1488 my $old_path = $self->{path};
1489 $self->{path} = $url;
4e9f6cc7 1490 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
e6434f87
EW
1491 if (length $old_path) {
1492 $self->{path} .= "/$old_path";
1493 }
1494 $url = $min_url;
1495 }
1496 }
1497 my $orig_url;
1498 if (!$existing) {
b805b44a 1499 # verify that we aren't overwriting anything:
e6434f87 1500 $orig_url = eval {
706587fc 1501 command_oneline('config', '--get',
e6434f87 1502 "svn-remote.$self->{repo_id}.url")
706587fc 1503 };
b805b44a 1504 if ($orig_url && ($orig_url ne $url)) {
e6434f87 1505 die "svn-remote.$self->{repo_id}.url already set: ",
b805b44a
EW
1506 "$orig_url\nwanted to set to: $url\n";
1507 }
9b981fc6 1508 }
e6434f87
EW
1509 my ($xrepo_id, $xpath) = find_ref($self->refname);
1510 if (defined $xpath) {
1511 die "svn-remote.$xrepo_id.fetch already set to track ",
1512 "$xpath:refs/remotes/", $self->refname, "\n";
1513 }
d8115c51
EW
1514 unless ($no_write) {
1515 command_noisy('config',
1516 "svn-remote.$self->{repo_id}.url", $url);
46cf98ba 1517 $self->{path} =~ s{^/}{};
d8115c51
EW
1518 command_noisy('config', '--add',
1519 "svn-remote.$self->{repo_id}.fetch",
1520 "$self->{path}:".$self->refname);
1521 }
9b981fc6 1522 $self->{url} = $url;
e6434f87
EW
1523}
1524
a8ae2623
EW
1525sub find_by_url { # repos_root and, path are optional
1526 my ($class, $full_url, $repos_root, $path) = @_;
56973d20 1527
1a97a506 1528 return undef unless defined $full_url;
56973d20
AR
1529 remove_username($full_url);
1530 remove_username($repos_root) if defined $repos_root;
a8ae2623
EW
1531 my $remotes = read_all_remotes();
1532 if (defined $full_url && defined $repos_root && !defined $path) {
1533 $path = $full_url;
1534 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1535 }
1536 foreach my $repo_id (keys %$remotes) {
1537 my $u = $remotes->{$repo_id}->{url} or next;
56973d20 1538 remove_username($u);
a8ae2623
EW
1539 next if defined $repos_root && $repos_root ne $u;
1540
1541 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1542 foreach (qw/branches tags/) {
1543 resolve_local_globs($u, $fetch,
1544 $remotes->{$repo_id}->{$_});
1545 }
1546 my $p = $path;
0bb91d9a 1547 my $rwr = rewrite_root({repo_id => $repo_id});
a8ae2623
EW
1548 unless (defined $p) {
1549 $p = $full_url;
0bb91d9a
JG
1550 my $z = $u;
1551 if ($rwr) {
1552 $z = $rwr;
1553 }
1554 $p =~ s#^\Q$z\E(?:/|$)## or next;
a8ae2623
EW
1555 }
1556 foreach my $f (keys %$fetch) {
1557 next if $f ne $p;
1558 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1559 }
1560 }
1561 undef;
1562}
1563
e6434f87 1564sub init {
d8115c51 1565 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
e6434f87
EW
1566 my $self = _new($class, $repo_id, $ref_id, $path);
1567 if (defined $url) {
d8115c51 1568 $self->init_remote_config($url, $no_write);
e6434f87 1569 }
9b981fc6
EW
1570 $self;
1571}
1572
706587fc
EW
1573sub find_ref {
1574 my ($ref_id) = @_;
1575 foreach (command(qw/config -l/)) {
1576 next unless m!^svn-remote\.(.+)\.fetch=
1577 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1578 my ($repo_id, $path, $ref) = ($1, $2, $3);
1579 if ($ref eq $ref_id) {
1580 $path = '' if ($path =~ m#^\./?#);
1581 return ($repo_id, $path);
1582 }
1583 }
1584 (undef, undef, undef);
1585}
1586
9b981fc6 1587sub new {
706587fc
EW
1588 my ($class, $ref_id, $repo_id, $path) = @_;
1589 if (defined $ref_id && !defined $repo_id && !defined $path) {
1590 ($repo_id, $path) = find_ref($ref_id);
1591 if (!defined $repo_id) {
1592 die "Could not find a \"svn-remote.*.fetch\" key ",
1593 "in the repository configuration matching: ",
1594 "refs/remotes/$ref_id\n";
1595 }
1596 }
1597 my $self = _new($class, $repo_id, $ref_id, $path);
8b8fc068
EW
1598 if (!defined $self->{path} || !length $self->{path}) {
1599 my $fetch = command_oneline('config', '--get',
1600 "svn-remote.$repo_id.fetch",
1601 ":refs/remotes/$ref_id\$") or
1602 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1603 "\":refs/remotes/$ref_id\$\" in config\n";
1604 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1605 }
706587fc
EW
1606 $self->{url} = command_oneline('config', '--get',
1607 "svn-remote.$repo_id.url") or
1608 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
d6d3346b 1609 $self->rebuild;
9b981fc6
EW
1610 $self;
1611}
1612
bf655fd7
RE
1613sub refname {
1614 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1615
1616 # It cannot end with a slash /, we'll throw up on this because
1617 # SVN can't have directories with a slash in their name, either:
1618 if ($refname =~ m{/$}) {
1619 die "ref: '$refname' ends with a trailing slash, this is ",
1620 "not permitted by git nor Subversion\n";
1621 }
1622
1623 # It cannot have ASCII control character space, tilde ~, caret ^,
1624 # colon :, question-mark ?, asterisk *, space, or open bracket [
1625 # anywhere.
1626 #
1627 # Additionally, % must be escaped because it is used for escaping
1628 # and we want our escaped refname to be reversible
1629 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1630
1631 # no slash-separated component can begin with a dot .
1632 # /.* becomes /%2E*
1633 $refname =~ s{/\.}{/%2E}g;
1634
1635 # It cannot have two consecutive dots .. anywhere
1636 # .. becomes %2E%2E
1637 $refname =~ s{\.\.}{%2E%2E}g;
1638
1639 return $refname;
1640}
1641
1642sub desanitize_refname {
1643 my ($refname) = @_;
1644 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1645 return $refname;
1646}
9b981fc6 1647
26a62d57
EW
1648sub svm_uuid {
1649 my ($self) = @_;
1650 return $self->{svm}->{uuid} if $self->svm;
1651 $self->ra;
1652 unless ($self->{svm}) {
1653 die "SVM UUID not cached, and reading remotely failed\n";
1654 }
1655 $self->{svm}->{uuid};
1656}
8a49ee97 1657
26a62d57
EW
1658sub svm {
1659 my ($self) = @_;
1660 return $self->{svm} if $self->{svm};
1661 my $svm;
8a49ee97
EW
1662 # see if we have it in our config, first:
1663 eval {
26a62d57
EW
1664 my $section = "svn-remote.$self->{repo_id}";
1665 $svm = {
93f2689c
EW
1666 source => tmp_config('--get', "$section.svm-source"),
1667 uuid => tmp_config('--get', "$section.svm-uuid"),
befc9adc 1668 replace => tmp_config('--get', "$section.svm-replace"),
8a49ee97
EW
1669 }
1670 };
befc9adc
EW
1671 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1672 $self->{svm} = $svm;
1673 }
26a62d57
EW
1674 $self->{svm};
1675}
1676
1677sub _set_svm_vars {
1678 my ($self, $ra) = @_;
db03cd24
EW
1679 return $ra if $self->svm;
1680
1681 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
befc9adc 1682 "(svm:source, svm:uuid) ",
db03cd24
EW
1683 "from the following URLs:\n" );
1684 sub read_svm_props {
befc9adc
EW
1685 my ($self, $ra, $path, $r) = @_;
1686 my $props = ($ra->get_dir($path, $r))[2];
db03cd24 1687 my $src = $props->{'svm:source'};
db03cd24 1688 my $uuid = $props->{'svm:uuid'};
befc9adc 1689 return undef if (!$src || !$uuid);
26a62d57 1690
befc9adc 1691 chomp($src, $uuid);
26a62d57 1692
db03cd24
EW
1693 $uuid =~ m{^[0-9a-f\-]{30,}$}
1694 or die "doesn't look right - svm:uuid is '$uuid'\n";
befc9adc
EW
1695
1696 # the '!' is used to mark the repos_root!/relative/path
1697 $src =~ s{/?!/?}{/};
db03cd24 1698 $src =~ s{/+$}{}; # no trailing slashes please
befc9adc 1699 # username is of no interest
8a49ee97 1700 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
8a49ee97 1701
befc9adc
EW
1702 my $replace = $ra->{url};
1703 $replace .= "/$path" if length $path;
1704
db03cd24 1705 my $section = "svn-remote.$self->{repo_id}";
befc9adc
EW
1706 tmp_config("$section.svm-source", $src);
1707 tmp_config("$section.svm-replace", $replace);
1708 tmp_config("$section.svm-uuid", $uuid);
1709 $self->{svm} = {
1710 source => $src,
1711 uuid => $uuid,
1712 replace => $replace
1713 };
db03cd24
EW
1714 }
1715
1716 my $r = $ra->get_latest_revnum;
1717 my $path = $self->{path};
befc9adc 1718 my %tried;
db03cd24 1719 while (length $path) {
befc9adc
EW
1720 unless ($tried{"$self->{url}/$path"}) {
1721 return $ra if $self->read_svm_props($ra, $path, $r);
1722 $tried{"$self->{url}/$path"} = 1;
db03cd24 1723 }
befc9adc 1724 $path =~ s#/?[^/]+$##;
8a49ee97 1725 }
befc9adc
EW
1726 die "Path: '$path' should be ''\n" if $path ne '';
1727 return $ra if $self->read_svm_props($ra, $path, $r);
1728 $tried{"$self->{url}/$path"} = 1;
db03cd24
EW
1729
1730 if ($ra->{repos_root} eq $self->{url}) {
befc9adc 1731 die @err, (map { " $_\n" } keys %tried), "\n";
db03cd24
EW
1732 }
1733
1734 # nope, make sure we're connected to the repository root:
1735 my $ok;
1736 my @tried_b;
1737 $path = $ra->{svn_path};
db03cd24
EW
1738 $ra = Git::SVN::Ra->new($ra->{repos_root});
1739 while (length $path) {
befc9adc
EW
1740 unless ($tried{"$ra->{url}/$path"}) {
1741 $ok = $self->read_svm_props($ra, $path, $r);
1742 last if $ok;
1743 $tried{"$ra->{url}/$path"} = 1;
1744 }
1745 $path =~ s#/?[^/]+$##;
db03cd24 1746 }
befc9adc
EW
1747 die "Path: '$path' should be ''\n" if $path ne '';
1748 $ok ||= $self->read_svm_props($ra, $path, $r);
1749 $tried{"$ra->{url}/$path"} = 1;
db03cd24 1750 if (!$ok) {
befc9adc 1751 die @err, (map { " $_\n" } keys %tried), "\n";
db03cd24
EW
1752 }
1753 Git::SVN::Ra->new($self->{url});
8a49ee97
EW
1754}
1755
62e349d2
EW
1756sub svnsync {
1757 my ($self) = @_;
1758 return $self->{svnsync} if $self->{svnsync};
1759
1760 if ($self->no_metadata) {
1761 die "Can't have both 'noMetadata' and ",
1762 "'useSvnsyncProps' options set!\n";
1763 }
1764 if ($self->rewrite_root) {
1765 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1766 "options set!\n";
1767 }
1768
1769 my $svnsync;
1770 # see if we have it in our config, first:
1771 eval {
1772 my $section = "svn-remote.$self->{repo_id}";
98fa5b68
EW
1773
1774 my $url = tmp_config('--get', "$section.svnsync-url");
1775 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1776 die "doesn't look right - svn:sync-from-url is '$url'\n";
1777
1778 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1779 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1780 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1781
1782 $svnsync = { url => $url, uuid => $uuid }
62e349d2
EW
1783 };
1784 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1785 return $self->{svnsync} = $svnsync;
1786 }
1787
1788 my $err = "useSvnsyncProps set, but failed to read " .
1789 "svnsync property: svn:sync-from-";
1790 my $rp = $self->ra->rev_proplist(0);
1791
1792 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
98fa5b68 1793 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
62e349d2
EW
1794 die "doesn't look right - svn:sync-from-url is '$url'\n";
1795
1796 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
98fa5b68 1797 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
62e349d2
EW
1798 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1799
1800 my $section = "svn-remote.$self->{repo_id}";
1801 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1802 tmp_config('--add', "$section.svnsync-url", $url);
1803 return $self->{svnsync} = { url => $url, uuid => $uuid };
1804}
1805
26a62d57
EW
1806# this allows us to memoize our SVN::Ra UUID locally and avoid a
1807# remote lookup (useful for 'git svn log').
1808sub ra_uuid {
1809 my ($self) = @_;
1810 unless ($self->{ra_uuid}) {
1811 my $key = "svn-remote.$self->{repo_id}.uuid";
1812 my $uuid = eval { tmp_config('--get', $key) };
1813 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1814 $self->{ra_uuid} = $uuid;
1815 } else {
1816 die "ra_uuid called without URL\n" unless $self->{url};
1817 $self->{ra_uuid} = $self->ra->get_uuid;
1818 tmp_config('--add', $key, $self->{ra_uuid});
1819 }
1820 }
1821 $self->{ra_uuid};
1822}
1823
a5460eb7
EW
1824sub _set_repos_root {
1825 my ($self, $repos_root) = @_;
1826 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1827 $repos_root ||= $self->ra->{repos_root};
1828 tmp_config($k, $repos_root);
1829 $repos_root;
1830}
1831
1832sub repos_root {
1833 my ($self) = @_;
1834 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1835 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1836}
1837
9b981fc6
EW
1838sub ra {
1839 my ($self) = shift;
8a49ee97 1840 my $ra = Git::SVN::Ra->new($self->{url});
a5460eb7 1841 $self->_set_repos_root($ra->{repos_root});
91b03282
EW
1842 if ($self->use_svm_props && !$self->{svm}) {
1843 if ($self->no_metadata) {
97ae0911
EW
1844 die "Can't have both 'noMetadata' and ",
1845 "'useSvmProps' options set!\n";
62e349d2
EW
1846 } elsif ($self->use_svnsync_props) {
1847 die "Can't have both 'useSvnsyncProps' and ",
1848 "'useSvmProps' options set!\n";
91b03282 1849 }
26a62d57 1850 $ra = $self->_set_svm_vars($ra);
8a49ee97
EW
1851 $self->{-want_revprops} = 1;
1852 }
1853 $ra;
9b981fc6
EW
1854}
1855
15710b6f
EW
1856sub rel_path {
1857 my ($self) = @_;
1858 my $repos_root = $self->ra->{repos_root};
1859 return $self->{path} if ($self->{url} eq $repos_root);
0b59451c
EW
1860 my $url = $self->{url} .
1861 (length $self->{path} ? "/$self->{path}" : $self->{path});
1862 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1863 $url;
15710b6f
EW
1864}
1865
01bdab84
BS
1866# prop_walk(PATH, REV, SUB)
1867# -------------------------
1868# Recursively traverse PATH at revision REV and invoke SUB for each
1869# directory that contains a SVN property. SUB will be invoked as
1870# follows: &SUB(gs, path, props); where `gs' is this instance of
1871# Git::SVN, `path' the path to the directory where the properties
1872# `props' were found. The `path' will be relative to point of checkout,
1873# that is, if url://repo/trunk is the current Git branch, and that
1874# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1875# as `path' (note the trailing `/').
1876sub prop_walk {
1877 my ($self, $path, $rev, $sub) = @_;
1878
35cda061 1879 $path =~ s#^/##;
01bdab84
BS
1880 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1881 $path =~ s#^/*#/#g;
9b981fc6 1882 my $p = $path;
01bdab84
BS
1883 # Strip the irrelevant part of the path.
1884 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1885 # Ensure the path is terminated by a `/'.
1886 $p =~ s#/*$#/#;
1887
1888 # The properties contain all the internal SVN stuff nobody
1889 # (usually) cares about.
1890 my $interesting_props = 0;
1891 foreach (keys %{$props}) {
1892 # If it doesn't start with `svn:', it must be a
1893 # user-defined property.
1894 ++$interesting_props and next if $_ !~ /^svn:/;
1895 # FIXME: Fragile, if SVN adds new public properties,
1896 # this needs to be updated.
1897 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1898 |eol-style|mime-type
1899 |externals|needs-lock)$/x;
1900 }
1901 &$sub($self, $p, $props) if $interesting_props;
1902
9b981fc6 1903 foreach (sort keys %$dirent) {
0dc03d6a 1904 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
67dac28b 1905 $self->prop_walk($p . $_, $rev, $sub);
9b981fc6
EW
1906 }
1907}
1908
3ebe8df7
EW
1909sub last_rev { ($_[0]->last_rev_commit)[0] }
1910sub last_commit { ($_[0]->last_rev_commit)[1] }
1911
9b981fc6
EW
1912# returns the newest SVN revision number and newest commit SHA1
1913sub last_rev_commit {
1914 my ($self) = @_;
1915 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1916 return ($self->{last_rev}, $self->{last_commit});
1917 }
d2866f9e 1918 my $c = ::verify_ref($self->refname.'^0');
91b03282 1919 if ($c && !$self->use_svm_props && !$self->no_metadata) {
d2866f9e 1920 my $rev = (::cmt_metadata($c))[1];
9b981fc6
EW
1921 if (defined $rev) {
1922 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1923 return ($rev, $c);
1924 }
1925 }
060610c5
EW
1926 my $map_path = $self->map_path;
1927 unless (-e $map_path) {
26a62d57
EW
1928 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1929 return (undef, undef);
1930 }
66ab84b9 1931 my ($rev, $commit) = $self->rev_map_max(1);
060610c5
EW
1932 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1933 return ($rev, $commit);
9b981fc6
EW
1934}
1935
3ebe8df7
EW
1936sub get_fetch_range {
1937 my ($self, $min, $max) = @_;
1938 $max ||= $self->ra->get_latest_revnum;
060610c5 1939 $min ||= $self->rev_map_max;
3ebe8df7 1940 (++$min, $max);
9b981fc6
EW
1941}
1942
8a49ee97 1943sub tmp_config {
93f2689c 1944 my (@args) = @_;
b7e5348c
EW
1945 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1946 my $config = "$ENV{GIT_DIR}/svn/.metadata";
38570a47 1947 if (! -f $config && -f $old_def_config) {
b7e5348c
EW
1948 rename $old_def_config, $config or
1949 die "Failed rename $old_def_config => $config: $!\n";
1950 }
8a49ee97 1951 my $old_config = $ENV{GIT_CONFIG};
93f2689c 1952 $ENV{GIT_CONFIG} = $config;
8a49ee97 1953 $@ = undef;
b4d57e5e
EW
1954 my @ret = eval {
1955 unless (-f $config) {
1956 mkfile($config);
1957 open my $fh, '>', $config or
1958 die "Can't open $config: $!\n";
1959 print $fh "; This file is used internally by ",
1960 "git-svn\n" or die
1961 "Couldn't write to $config: $!\n";
1962 print $fh "; You should not have to edit it\n" or
1963 die "Couldn't write to $config: $!\n";
1964 close $fh or die "Couldn't close $config: $!\n";
1965 }
1966 command('config', @args);
1967 };
8a49ee97
EW
1968 my $err = $@;
1969 if (defined $old_config) {
1970 $ENV{GIT_CONFIG} = $old_config;
1971 } else {
1972 delete $ENV{GIT_CONFIG};
1973 }
1974 die $err if $err;
1975 wantarray ? @ret : $ret[0];
1976}
1977
9b981fc6
EW
1978sub tmp_index_do {
1979 my ($self, $sub) = @_;
1980 my $old_index = $ENV{GIT_INDEX_FILE};
1981 $ENV{GIT_INDEX_FILE} = $self->{index};
8a49ee97 1982 $@ = undef;
b4d57e5e
EW
1983 my @ret = eval {
1984 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1985 mkpath([$dir]) unless -d $dir;
1986 &$sub;
1987 };
8a49ee97
EW
1988 my $err = $@;
1989 if (defined $old_index) {
9b981fc6
EW
1990 $ENV{GIT_INDEX_FILE} = $old_index;
1991 } else {
1992 delete $ENV{GIT_INDEX_FILE};
1993 }
8a49ee97 1994 die $err if $err;
9b981fc6
EW
1995 wantarray ? @ret : $ret[0];
1996}
1997
1998sub assert_index_clean {
1999 my ($self, $treeish) = @_;
2000
2001 $self->tmp_index_do(sub {
2002 command_noisy('read-tree', $treeish) unless -e $self->{index};
2003 my $x = command_oneline('write-tree');
2004 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2005 /^tree ($::sha1)/mo);
e8d120bd
EW
2006 return if $y eq $x;
2007
2008 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2009 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2010 command_noisy('read-tree', $treeish);
9b981fc6
EW
2011 $x = command_oneline('write-tree');
2012 if ($y ne $x) {
2013 ::fatal "trees ($treeish) $y != $x\n",
207f1a75 2014 "Something is seriously wrong...";
9b981fc6
EW
2015 }
2016 });
2017}
2018
2019sub get_commit_parents {
0af9c9f9 2020 my ($self, $log_entry) = @_;
9b981fc6 2021 my (%seen, @ret, @tmp);
0af9c9f9
EW
2022 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2023 if (my $ip = $self->{inject_parents}) {
2024 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2025 push @tmp, $commit;
9b981fc6
EW
2026 }
2027 }
d2866f9e 2028 if (my $cur = ::verify_ref($self->refname.'^0')) {
9b981fc6
EW
2029 push @tmp, $cur;
2030 }
733a65aa
EW
2031 if (my $ipd = $self->{inject_parents_dcommit}) {
2032 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2033 push @tmp, @$commit;
2034 }
2035 }
44320b9e 2036 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
9b981fc6
EW
2037 while (my $p = shift @tmp) {
2038 next if $seen{$p};
2039 $seen{$p} = 1;
2040 push @ret, $p;
2041 # MAXPARENT is defined to 16 in commit-tree.c:
2042 last if @ret >= 16;
2043 }
2044 if (@tmp) {
44320b9e 2045 die "r$log_entry->{revision}: No room for parents:\n\t",
9b981fc6
EW
2046 join("\n\t", @tmp), "\n";
2047 }
2048 @ret;
2049}
2050
aea736cc
EW
2051sub rewrite_root {
2052 my ($self) = @_;
2053 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2054 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2055 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2056 if ($rwr) {
2057 $rwr =~ s#/+$##;
2058 if ($rwr !~ m#^[a-z\+]+://#) {
2059 die "$rwr is not a valid URL (key: $k)\n";
2060 }
2061 }
2062 $self->{-rewrite_root} = $rwr;
2063}
2064
2065sub metadata_url {
2066 my ($self) = @_;
2067 ($self->rewrite_root || $self->{url}) .
2068 (length $self->{path} ? '/' . $self->{path} : '');
2069}
2070
706587fc 2071sub full_url {
9b981fc6 2072 my ($self) = @_;
5d3b7cd5 2073 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
9b981fc6
EW
2074}
2075
ad94802a
EW
2076
2077sub set_commit_header_env {
2078 my ($log_entry) = @_;
2079 my %env;
2080 foreach my $ned (qw/NAME EMAIL DATE/) {
2081 foreach my $ac (qw/AUTHOR COMMITTER/) {
2082 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2083 }
9b981fc6 2084 }
ad94802a 2085
70ae04e4
AW
2086 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2087 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
44320b9e 2088 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
9b981fc6 2089
70ae04e4
AW
2090 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2091 ? $log_entry->{commit_name}
2092 : $log_entry->{name};
2093 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2094 ? $log_entry->{commit_email}
2095 : $log_entry->{email};
ad94802a
EW
2096 \%env;
2097}
70ae04e4 2098
ad94802a
EW
2099sub restore_commit_header_env {
2100 my ($env) = @_;
2101 foreach my $ned (qw/NAME EMAIL DATE/) {
2102 foreach my $ac (qw/AUTHOR COMMITTER/) {
2103 my $k = "GIT_${ac}_${ned}";
2104 if (defined $env->{$k}) {
2105 $ENV{$k} = $env->{$k};
2106 } else {
2107 delete $ENV{$k};
2108 }
2109 }
2110 }
2111}
2112
94bc914c
KW
2113sub gc {
2114 command_noisy('gc', '--auto');
2115};
2116
ad94802a
EW
2117sub do_git_commit {
2118 my ($self, $log_entry) = @_;
2119 my $lr = $self->last_rev;
2120 if (defined $lr && $lr >= $log_entry->{revision}) {
2121 die "Last fetched revision of ", $self->refname,
2122 " was r$lr, but we are about to fetch: ",
2123 "r$log_entry->{revision}!\n";
2124 }
2125 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2126 croak "$log_entry->{revision} = $c already exists! ",
2127 "Why are we refetching it?\n";
2128 }
2129 my $old_env = set_commit_header_env($log_entry);
44320b9e 2130 my $tree = $log_entry->{tree};
9b981fc6
EW
2131 if (!defined $tree) {
2132 $tree = $self->tmp_index_do(sub {
2133 command_oneline('write-tree') });
2134 }
2135 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2136
2137 my @exec = ('git-commit-tree', $tree);
0af9c9f9 2138 foreach ($self->get_commit_parents($log_entry)) {
9b981fc6
EW
2139 push @exec, '-p', $_;
2140 }
2141 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2142 or croak $!;
44320b9e 2143 print $msg_fh $log_entry->{log} or croak $!;
ad94802a 2144 restore_commit_header_env($old_env);
91b03282 2145 unless ($self->no_metadata) {
8a49ee97
EW
2146 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2147 or croak $!;
9760adcc 2148 }
9b981fc6
EW
2149 $msg_fh->flush == 0 or croak $!;
2150 close $msg_fh or croak $!;
2151 chomp(my $commit = do { local $/; <$out_fh> });
2152 close $out_fh or croak $!;
2153 waitpid $pid, 0;
2154 croak $? if $?;
2155 if ($commit !~ /^$::sha1$/o) {
2156 die "Failed to commit, invalid sha1: $commit\n";
2157 }
2158
060610c5 2159 $self->rev_map_set($log_entry->{revision}, $commit, 1);
9b981fc6 2160
44320b9e 2161 $self->{last_rev} = $log_entry->{revision};
9b981fc6 2162 $self->{last_commit} = $commit;
8a49ee97
EW
2163 print "r$log_entry->{revision}";
2164 if (defined $log_entry->{svm_revision}) {
2165 print " (\@$log_entry->{svm_revision})";
060610c5 2166 $self->rev_map_set($log_entry->{svm_revision}, $commit,
26a62d57 2167 0, $self->svm_uuid);
8a49ee97
EW
2168 }
2169 print " = $commit ($self->{ref_id})\n";
94bc914c
KW
2170 if (--$_gc_nr == 0) {
2171 $_gc_nr = $_gc_period;
2172 gc();
2173 }
9b981fc6
EW
2174 return $commit;
2175}
2176
fbcc1737
EW
2177sub match_paths {
2178 my ($self, $paths, $r) = @_;
4e9f6cc7 2179 return 1 if $self->{path} eq '';
d542aedb
EW
2180 if (my $path = $paths->{"/$self->{path}"}) {
2181 return ($path->{action} eq 'D') ? 0 : 1;
2182 }
4e9f6cc7 2183 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
fbcc1737
EW
2184 if (grep /$self->{path_regex}/, keys %$paths) {
2185 return 1;
2186 }
2187 my $c = '';
2188 foreach (split m#/#, $self->{path}) {
2189 $c .= "/$_";
74a81227
EW
2190 next unless ($paths->{$c} &&
2191 ($paths->{$c}->{action} =~ /^[AR]$/));
e518192f
EW
2192 if ($self->ra->check_path($self->{path}, $r) ==
2193 $SVN::Node::dir) {
fbcc1737
EW
2194 return 1;
2195 }
2196 }
2197 return 0;
2198}
2199
15710b6f
EW
2200sub find_parent_branch {
2201 my ($self, $paths, $rev) = @_;
91b03282 2202 return undef unless $self->follow_parent;
e5a0b240 2203 unless (defined $paths) {
c7eba716
EW
2204 my $err_handler = $SVN::Error::handler;
2205 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
d4eff2bd
EW
2206 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2207 $paths =
2208 Git::SVN::Ra::dup_changed_paths($_[0]) });
c7eba716 2209 $SVN::Error::handler = $err_handler;
e5a0b240
EW
2210 }
2211 return undef unless defined $paths;
15710b6f
EW
2212
2213 # look for a parent from another branch:
7f578c55
EW
2214 my @b_path_components = split m#/#, $self->rel_path;
2215 my @a_path_components;
2216 my $i;
2217 while (@b_path_components) {
2218 $i = $paths->{'/'.join('/', @b_path_components)};
74a81227 2219 last if $i && defined $i->{copyfrom_path};
7f578c55
EW
2220 unshift(@a_path_components, pop(@b_path_components));
2221 }
74a81227
EW
2222 return undef unless defined $i && defined $i->{copyfrom_path};
2223 my $branch_from = $i->{copyfrom_path};
7f578c55
EW
2224 if (@a_path_components) {
2225 print STDERR "branch_from: $branch_from => ";
2226 $branch_from .= '/'.join('/', @a_path_components);
2227 print STDERR $branch_from, "\n";
2228 }
3ebe8df7 2229 my $r = $i->{copyfrom_rev};
15710b6f
EW
2230 my $repos_root = $self->ra->{repos_root};
2231 my $url = $self->ra->{url};
2232 my $new_url = $repos_root . $branch_from;
2233 print STDERR "Found possible branch point: ",
2234 "$new_url => ", $self->full_url, ", $r\n";
2235 $branch_from =~ s#^/##;
a8ae2623 2236 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
15710b6f 2237 unless ($gs) {
ce2a0f2f
EW
2238 my $ref_id = $self->{ref_id};
2239 $ref_id =~ s/\@\d+$//;
2240 $ref_id .= "\@$r";
15710b6f
EW
2241 # just grow a tail if we're not unique enough :x
2242 $ref_id .= '-' while find_ref($ref_id);
ce2a0f2f 2243 print STDERR "Initializing parent: $ref_id\n";
f61cc48d 2244 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
c586879c
EW
2245 if ($u =~ s#^\Q$url\E(/|$)##) {
2246 $p = $u;
2247 $u = $url;
f61cc48d 2248 $repo_id = $self->{repo_id};
c586879c 2249 }
f61cc48d 2250 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
15710b6f
EW
2251 }
2252 my ($r0, $parent) = $gs->find_rev_before($r, 1);
91b03282 2253 if (!defined $r0 || !defined $parent) {
d627de6b
EW
2254 my ($base, $head) = parse_revision_argument(0, $r);
2255 if ($base <= $r) {
2256 $gs->fetch($base, $r);
2257 }
15710b6f
EW
2258 ($r0, $parent) = $gs->last_rev_commit;
2259 }
ef70de96 2260 if (defined $r0 && defined $parent) {
15710b6f 2261 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
15710b6f
EW
2262 my $ed;
2263 if ($self->ra->can_do_switch) {
2e5e2480 2264 $self->assert_index_clean($parent);
8b8fc068 2265 print STDERR "Following parent with do_switch\n";
15710b6f 2266 # do_switch works with svn/trunk >= r22312, but that
2b27f6c8 2267 # is not included with SVN 1.4.3 (the latest version
15710b6f
EW
2268 # at the moment), so we can't rely on it
2269 $self->{last_commit} = $parent;
2270 $ed = SVN::Git::Fetcher->new($self);
8a603774 2271 $gs->ra->gs_do_switch($r0, $rev, $gs,
15710b6f
EW
2272 $self->full_url, $ed)
2273 or die "SVN connection failed somewhere...\n";
9ff74e95
SW
2274 } elsif ($self->ra->trees_match($new_url, $r0,
2275 $self->full_url, $rev)) {
2276 print STDERR "Trees match:\n",
2277 " $new_url\@$r0\n",
2278 " ${\$self->full_url}\@$rev\n",
2279 "Following parent with no changes\n";
2280 $self->tmp_index_do(sub {
2281 command_noisy('read-tree', $parent);
2282 });
2283 $self->{last_commit} = $parent;
15710b6f 2284 } else {
8b8fc068 2285 print STDERR "Following parent with do_update\n";
15710b6f 2286 $ed = SVN::Git::Fetcher->new($self);
8a603774 2287 $self->ra->gs_do_update($rev, $rev, $self, $ed)
15710b6f
EW
2288 or die "SVN connection failed somewhere...\n";
2289 }
f7c3fc4a 2290 print STDERR "Successfully followed parent\n";
15710b6f
EW
2291 return $self->make_log_entry($rev, [$parent], $ed);
2292 }
15710b6f
EW
2293 return undef;
2294}
2295
9b981fc6 2296sub do_fetch {
706587fc 2297 my ($self, $paths, $rev) = @_;
15710b6f 2298 my $ed;
9b981fc6 2299 my ($last_rev, @parents);
b9dffd8c
EW
2300 if (my $lc = $self->last_commit) {
2301 # we can have a branch that was deleted, then re-added
2302 # under the same name but copied from another path, in
2303 # which case we'll have multiple parents (we don't
2304 # want to break the original ref, nor lose copypath info):
2305 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2306 push @{$log_entry->{parents}}, $lc;
2307 return $log_entry;
2308 }
15710b6f 2309 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 2310 $last_rev = $self->{last_rev};
b9dffd8c
EW
2311 $ed->{c} = $lc;
2312 @parents = ($lc);
9b981fc6
EW
2313 } else {
2314 $last_rev = $rev;
15710b6f
EW
2315 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2316 return $log_entry;
2317 }
2318 $ed = SVN::Git::Fetcher->new($self);
9b981fc6 2319 }
8a603774 2320 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
9b981fc6
EW
2321 die "SVN connection failed somewhere...\n";
2322 }
2323 $self->make_log_entry($rev, \@parents, $ed);
2324}
2325
97f6987a
EW
2326sub get_untracked {
2327 my ($self, $ed) = @_;
2328 my @out;
2329 my $h = $ed->{empty};
9b981fc6
EW
2330 foreach (sort keys %$h) {
2331 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
97f6987a 2332 push @out, " $act: " . uri_encode($_);
9b981fc6
EW
2333 warn "W: $act: $_\n";
2334 }
2335 foreach my $t (qw/dir_prop file_prop/) {
97f6987a 2336 $h = $ed->{$t} or next;
9b981fc6
EW
2337 foreach my $path (sort keys %$h) {
2338 my $ppath = $path eq '' ? '.' : $path;
2339 foreach my $prop (sort keys %{$h->{$path}}) {
1ce255dc 2340 next if $SKIP_PROP{$prop};
9b981fc6 2341 my $v = $h->{$path}->{$prop};
97f6987a
EW
2342 my $t_ppath_prop = "$t: " .
2343 uri_encode($ppath) . ' ' .
2344 uri_encode($prop);
9b981fc6 2345 if (defined $v) {
97f6987a
EW
2346 push @out, " +$t_ppath_prop " .
2347 uri_encode($v);
9b981fc6 2348 } else {
97f6987a 2349 push @out, " -$t_ppath_prop";
9b981fc6
EW
2350 }
2351 }
2352 }
2353 }
2354 foreach my $t (qw/absent_file absent_directory/) {
97f6987a 2355 $h = $ed->{$t} or next;
9b981fc6
EW
2356 foreach my $parent (sort keys %$h) {
2357 foreach my $path (sort @{$h->{$parent}}) {
97f6987a
EW
2358 push @out, " $t: " .
2359 uri_encode("$parent/$path");
9b981fc6
EW
2360 warn "W: $t: $parent/$path ",
2361 "Insufficient permissions?\n";
2362 }
2363 }
2364 }
97f6987a 2365 \@out;
9b981fc6
EW
2366}
2367
1c8443b0
EW
2368sub parse_svn_date {
2369 my $date = shift || return '+0000 1970-01-01 00:00:00';
2370 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2371 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2372 croak "Unable to parse date: $date\n";
2373 "+0000 $Y-$m-$d $H:$M:$S";
2374}
2375
2376sub check_author {
2377 my ($author) = @_;
2378 if (!defined $author || length $author == 0) {
2379 $author = '(no author)';
9231f500 2380 } elsif (defined $::_authors && ! defined $::users{$author}) {
1c8443b0
EW
2381 die "Author: $author not defined in $::_authors file\n";
2382 }
2383 $author;
2384}
2385
9b981fc6 2386sub make_log_entry {
97f6987a
EW
2387 my ($self, $rev, $parents, $ed) = @_;
2388 my $untracked = $self->get_untracked($ed);
2389
9b981fc6 2390 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
97f6987a
EW
2391 print $un "r$rev\n" or croak $!;
2392 print $un $_, "\n" foreach @$untracked;
2393 my %log_entry = ( parents => $parents || [], revision => $rev,
2394 log => '');
fbcc1737 2395
8a49ee97 2396 my $headrev;
fbcc1737 2397 my $logged = delete $self->{logged_rev_props};
8a49ee97 2398 if (!$logged || $self->{-want_revprops}) {
fbcc1737
EW
2399 my $rp = $self->ra->rev_proplist($rev);
2400 foreach (sort keys %$rp) {
2401 my $v = $rp->{$_};
2402 if (/^svn:(author|date|log)$/) {
2403 $log_entry{$1} = $v;
8a49ee97
EW
2404 } elsif ($_ eq 'svm:headrev') {
2405 $headrev = $v;
fbcc1737
EW
2406 } else {
2407 print $un " rev_prop: ", uri_encode($_), ' ',
2408 uri_encode($v), "\n";
2409 }
9b981fc6 2410 }
fbcc1737
EW
2411 } else {
2412 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
9b981fc6
EW
2413 }
2414 close $un or croak $!;
97f6987a 2415
9b981fc6 2416 $log_entry{date} = parse_svn_date($log_entry{date});
9b981fc6 2417 $log_entry{log} .= "\n";
db03cd24
EW
2418 my $author = $log_entry{author} = check_author($log_entry{author});
2419 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
70ae04e4
AW
2420 : ($author, undef);
2421
2422 my ($commit_name, $commit_email) = ($name, $email);
2423 if ($_use_log_author) {
5ff6aae8
AW
2424 my $name_field;
2425 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2426 $name_field = $1;
2427 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2428 $name_field = $1;
2429 }
2430 if (!defined $name_field) {
abfa533d
SB
2431 if (!defined $email) {
2432 $email = $name;
2433 }
5ff6aae8 2434 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
70ae04e4 2435 ($name, $email) = ($1, $2);
5ff6aae8
AW
2436 } elsif ($name_field =~ /(.*)@/) {
2437 ($name, $email) = ($1, $name_field);
2438 } else {
abfa533d 2439 ($name, $email) = ($name_field, $name_field);
70ae04e4
AW
2440 }
2441 }
91b03282 2442 if (defined $headrev && $self->use_svm_props) {
aea736cc
EW
2443 if ($self->rewrite_root) {
2444 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2445 "options set!\n";
2446 }
8a49ee97 2447 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
befc9adc
EW
2448 # we don't want "SVM: initializing mirror for junk" ...
2449 return undef if $r == 0;
2450 my $svm = $self->svm;
2451 if ($uuid ne $svm->{uuid}) {
8a49ee97 2452 die "UUID mismatch on SVM path:\n",
befc9adc 2453 "expected: $svm->{uuid}\n",
8a49ee97
EW
2454 " got: $uuid\n";
2455 }
befc9adc
EW
2456 my $full_url = $self->full_url;
2457 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2458 die "Failed to replace '$svm->{replace}' with ",
2459 "'$svm->{source}' in $full_url\n";
18ea92bd
SV
2460 # throw away username for storing in records
2461 remove_username($full_url);
8a49ee97
EW
2462 $log_entry{metadata} = "$full_url\@$r $uuid";
2463 $log_entry{svm_revision} = $r;
70ae04e4
AW
2464 $email ||= "$author\@$uuid";
2465 $commit_email ||= "$author\@$uuid";
62e349d2
EW
2466 } elsif ($self->use_svnsync_props) {
2467 my $full_url = $self->svnsync->{url};
2468 $full_url .= "/$self->{path}" if length $self->{path};
ce118739 2469 remove_username($full_url);
62e349d2
EW
2470 my $uuid = $self->svnsync->{uuid};
2471 $log_entry{metadata} = "$full_url\@$rev $uuid";
70ae04e4
AW
2472 $email ||= "$author\@$uuid";
2473 $commit_email ||= "$author\@$uuid";
8a49ee97 2474 } else {
ce118739
AR
2475 my $url = $self->metadata_url;
2476 remove_username($url);
2477 $log_entry{metadata} = "$url\@$rev " .
26a62d57 2478 $self->ra->get_uuid;
db03cd24 2479 $email ||= "$author\@" . $self->ra->get_uuid;
70ae04e4 2480 $commit_email ||= "$author\@" . $self->ra->get_uuid;
8a49ee97 2481 }
db03cd24
EW
2482 $log_entry{name} = $name;
2483 $log_entry{email} = $email;
70ae04e4
AW
2484 $log_entry{commit_name} = $commit_name;
2485 $log_entry{commit_email} = $commit_email;
9b981fc6
EW
2486 \%log_entry;
2487}
2488
2489sub fetch {
3ebe8df7 2490 my ($self, $min_rev, $max_rev, @parents) = @_;
9b981fc6 2491 my ($last_rev, $last_commit) = $self->last_rev_commit;
3ebe8df7 2492 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
e518192f 2493 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
9b981fc6
EW
2494}
2495
2496sub set_tree_cb {
2497 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
490f49ea
EW
2498 $self->{inject_parents} = { $rev => $tree };
2499 $self->fetch(undef, undef);
9b981fc6
EW
2500}
2501
2502sub set_tree {
2503 my ($self, $tree) = (shift, shift);
1ce255dc 2504 my $log_entry = ::get_commit_entry($tree);
9b981fc6 2505 unless ($self->{last_rev}) {
207f1a75 2506 fatal("Must have an existing revision to commit");
9b981fc6 2507 }
61395354
EW
2508 my %ed_opts = ( r => $self->{last_rev},
2509 log => $log_entry->{log},
2510 ra => $self->ra,
2511 tree_a => $self->{last_commit},
2512 tree_b => $tree,
2513 editor_cb => sub {
2514 $self->set_tree_cb($log_entry, $tree, @_) },
2515 svn_path => $self->{path} );
2516 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
9b981fc6
EW
2517 print "No changes\nr$self->{last_rev} = $tree\n";
2518 }
9b981fc6
EW
2519}
2520
060610c5
EW
2521sub rebuild_from_rev_db {
2522 my ($self, $path) = @_;
2523 my $r = -1;
2524 open my $fh, '<', $path or croak "open: $!";
4f7ec797 2525 binmode $fh or croak "binmode: $!";
060610c5
EW
2526 while (<$fh>) {
2527 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2528 chomp($_);
2529 ++$r;
2530 next if $_ eq ('0' x 40);
2531 $self->rev_map_set($r, $_);
2532 print "r$r = $_\n";
2533 }
2534 close $fh or croak "close: $!";
2535 unlink $path or croak "unlink: $!";
2536}
2537
f0ecca10
EW
2538sub rebuild {
2539 my ($self) = @_;
060610c5
EW
2540 my $map_path = $self->map_path;
2541 return if (-e $map_path && ! -z $map_path);
d6d3346b 2542 return unless ::verify_ref($self->refname.'^0');
060610c5
EW
2543 if ($self->use_svm_props || $self->no_metadata) {
2544 my $rev_db = $self->rev_db_path;
2545 $self->rebuild_from_rev_db($rev_db);
2546 if ($self->use_svm_props) {
2547 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2548 $self->rebuild_from_rev_db($svm_rev_db);
2549 }
2550 $self->unlink_rev_db_symlink;
26a62d57
EW
2551 return;
2552 }
060610c5
EW
2553 print "Rebuilding $map_path ...\n";
2554 my ($log, $ctx) =
2555 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2556 $self->refname, '--');
f0ecca10 2557 my $full_url = $self->full_url;
18ea92bd 2558 remove_username($full_url);
060610c5 2559 my $svn_uuid = $self->ra_uuid;
3dfab993
SV
2560 my $c;
2561 while (<$log>) {
2562 if ( m{^commit ($::sha1)$} ) {
2563 $c = $1;
2564 next;
2565 }
2566 next unless s{^\s*(git-svn-id:)}{$1};
2567 my ($url, $rev, $uuid) = ::extract_metadata($_);
18ea92bd 2568 remove_username($url);
f0ecca10
EW
2569
2570 # ignore merges (from set-tree)
2571 next if (!defined $rev || !$uuid);
2572
2573 # if we merged or otherwise started elsewhere, this is
2574 # how we break out of it
060610c5 2575 if (($uuid ne $svn_uuid) ||
f0ecca10
EW
2576 ($full_url && $url && ($url ne $full_url))) {
2577 next;
2578 }
f0ecca10 2579
060610c5 2580 $self->rev_map_set($rev, $c);
f0ecca10
EW
2581 print "r$rev = $c\n";
2582 }
3dfab993 2583 command_close_pipe($log, $ctx);
060610c5
EW
2584 print "Done rebuilding $map_path\n";
2585 my $rev_db_path = $self->rev_db_path;
2586 if (-f $self->rev_db_path) {
2587 unlink $self->rev_db_path or croak "unlink: $!";
2588 }
2589 $self->unlink_rev_db_symlink;
f0ecca10
EW
2590}
2591
060610c5 2592# rev_map:
9b981fc6
EW
2593# Tie::File seems to be prone to offset errors if revisions get sparse,
2594# it's not that fast, either. Tie::File is also not in Perl 5.6. So
2595# one of my favorite modules is out :< Next up would be one of the DBM
060610c5
EW
2596# modules, but I'm not sure which is most portable...
2597#
2598# This is the replacement for the rev_db format, which was too big
2599# and inefficient for large repositories with a lot of sparse history
2600# (mainly tags)
2601#
2602# The format is this:
2603# - 24 bytes for every record,
2604# * 4 bytes for the integer representing an SVN revision number
2605# * 20 bytes representing the sha1 of a git commit
2606# - No empty padding records like the old format
66ab84b9 2607# (except the last record, which can be overwritten)
060610c5
EW
2608# - new records are written append-only since SVN revision numbers
2609# increase monotonically
2610# - lookups on SVN revision number are done via a binary search
66ab84b9
EW
2611# - Piping the file to xxd -c24 is a good way of dumping it for
2612# viewing or editing (piped back through xxd -r), should the need
2613# ever arise.
2614# - The last record can be padding revision with an all-zero sha1
2615# This is used to optimize fetch performance when using multiple
2616# "fetch" directives in .git/config
060610c5 2617#
97ae0911 2618# These files are disposable unless noMetadata or useSvmProps is set
9b981fc6 2619
060610c5 2620sub _rev_map_set {
26a62d57 2621 my ($fh, $rev, $commit) = @_;
060610c5 2622
4f7ec797 2623 binmode $fh or croak "binmode: $!";
060610c5
EW
2624 my $size = (stat($fh))[7];
2625 ($size % 24) == 0 or croak "inconsistent size: $size";
2626
66ab84b9 2627 my $wr_offset = 0;
060610c5
EW
2628 if ($size > 0) {
2629 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2630 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2631 $read == 24 or croak "read only $read bytes (!= 24)";
2632 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
66ab84b9
EW
2633 if ($last_commit eq ('0' x40)) {
2634 if ($size >= 48) {
2635 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2636 $read = sysread($fh, $buf, 24) or
2637 croak "read: $!";
2638 $read == 24 or
2639 croak "read only $read bytes (!= 24)";
2640 ($last_rev, $last_commit) =
2641 unpack(rev_map_fmt, $buf);
2642 if ($last_commit eq ('0' x40)) {
2643 croak "inconsistent .rev_map\n";
2644 }
2645 }
2646 if ($last_rev >= $rev) {
2647 croak "last_rev is higher!: $last_rev >= $rev";
2648 }
2649 $wr_offset = -24;
26a62d57
EW
2650 }
2651 }
66ab84b9 2652 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
060610c5
EW
2653 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2654 croak "write: $!";
26a62d57
EW
2655}
2656
2657sub mkfile {
2658 my ($path) = @_;
2659 unless (-e $path) {
2660 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2661 mkpath([$dir]) unless -d $dir;
2662 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2663 close $fh or die "Couldn't close (create) $path: $!\n";
2664 }
2665}
2666
060610c5 2667sub rev_map_set {
26a62d57
EW
2668 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2669 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
060610c5 2670 my $db = $self->map_path($uuid);
26a62d57 2671 my $db_lock = "$db.lock";
373274f9
EW
2672 my $sig;
2673 if ($update_ref) {
2674 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2675 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2676 }
26a62d57
EW
2677 mkfile($db);
2678
373274f9 2679 $LOCKFILES{$db_lock} = 1;
97ae0911 2680 my $sync;
97ae0911
EW
2681 # both of these options make our .rev_db file very, very important
2682 # and we can't afford to lose it because rebuild() won't work
2683 if ($self->use_svm_props || $self->no_metadata) {
2684 $sync = 1;
060610c5 2685 copy($db, $db_lock) or die "rev_map_set(@_): ",
26a62d57 2686 "Failed to copy: ",
373274f9
EW
2687 "$db => $db_lock ($!)\n";
2688 } else {
060610c5 2689 rename $db, $db_lock or die "rev_map_set(@_): ",
26a62d57 2690 "Failed to rename: ",
373274f9
EW
2691 "$db => $db_lock ($!)\n";
2692 }
060610c5 2693
66ab84b9 2694 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
060610c5
EW
2695 or croak "Couldn't open $db_lock: $!\n";
2696 _rev_map_set($fh, $rev, $commit);
97ae0911
EW
2697 if ($sync) {
2698 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2699 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2700 }
9b981fc6 2701 close $fh or croak $!;
373274f9 2702 if ($update_ref) {
1e889ef3 2703 $_head = $self;
373274f9
EW
2704 command_noisy('update-ref', '-m', "r$rev",
2705 $self->refname, $commit);
2706 }
060610c5 2707 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
373274f9
EW
2708 "$db_lock => $db ($!)\n";
2709 delete $LOCKFILES{$db_lock};
2710 if ($update_ref) {
2711 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2712 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2713 kill $sig, $$ if defined $sig;
2714 }
9b981fc6
EW
2715}
2716
66ab84b9
EW
2717# If want_commit, this will return an array of (rev, commit) where
2718# commit _must_ be a valid commit in the archive.
2719# Otherwise, it'll return the max revision (whether or not the
2720# commit is valid or just a 0x40 placeholder).
060610c5 2721sub rev_map_max {
66ab84b9 2722 my ($self, $want_commit) = @_;
d6d3346b 2723 $self->rebuild;
060610c5 2724 my $map_path = $self->map_path;
66ab84b9 2725 stat $map_path or return $want_commit ? (0, undef) : 0;
060610c5 2726 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4f7ec797 2727 binmode $fh or croak "binmode: $!";
060610c5
EW
2728 my $size = (stat($fh))[7];
2729 ($size % 24) == 0 or croak "inconsistent size: $size";
2730
2731 if ($size == 0) {
2732 close $fh or croak "close: $!";
66ab84b9 2733 return $want_commit ? (0, undef) : 0;
060610c5
EW
2734 }
2735
66ab84b9 2736 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
060610c5 2737 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
060610c5 2738 my ($r, $c) = unpack(rev_map_fmt, $buf);
66ab84b9
EW
2739 if ($want_commit && $c eq ('0' x40)) {
2740 if ($size < 48) {
2741 return $want_commit ? (0, undef) : 0;
2742 }
2743 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2744 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2745 ($r, $c) = unpack(rev_map_fmt, $buf);
2746 if ($c eq ('0'x40)) {
2747 croak "Penultimate record is all-zeroes in $map_path";
2748 }
2749 }
2750 close $fh or croak "close: $!";
2751 $want_commit ? ($r, $c) : $r;
9c93fee5
EW
2752}
2753
060610c5 2754sub rev_map_get {
26a62d57 2755 my ($self, $rev, $uuid) = @_;
060610c5
EW
2756 my $map_path = $self->map_path($uuid);
2757 return undef unless -e $map_path;
2758
2759 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4f7ec797 2760 binmode $fh or croak "binmode: $!";
060610c5
EW
2761 my $size = (stat($fh))[7];
2762 ($size % 24) == 0 or croak "inconsistent size: $size";
2763
2764 if ($size == 0) {
2765 close $fh or croak "close: $fh";
2766 return undef;
2767 }
2768
2769 my ($l, $u) = (0, $size - 24);
2770 my ($r, $c, $buf);
2771
2772 while ($l <= $u) {
2773 my $i = int(($l/24 + $u/24) / 2) * 24;
2774 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2775 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2776 my ($r, $c) = unpack('NH40', $buf);
2777
2778 if ($r < $rev) {
2779 $l = $i + 24;
2780 } elsif ($r > $rev) {
2781 $u = $i - 24;
2782 } else { # $r == $rev
2783 close($fh) or croak "close: $!";
66ab84b9 2784 return $c eq ('0' x 40) ? undef : $c;
060610c5 2785 }
9b981fc6 2786 }
060610c5
EW
2787 close($fh) or croak "close: $!";
2788 undef;
9b981fc6
EW
2789}
2790
111947ef
DK
2791# Finds the first svn revision that exists on (if $eq_ok is true) or
2792# before $rev for the current branch. It will not search any lower
2793# than $min_rev. Returns the git commit hash and svn revision number
2794# if found, else (undef, undef).
15710b6f 2795sub find_rev_before {
111947ef 2796 my ($self, $rev, $eq_ok, $min_rev) = @_;
15710b6f 2797 --$rev unless $eq_ok;
111947ef
DK
2798 $min_rev ||= 1;
2799 while ($rev >= $min_rev) {
060610c5 2800 if (my $c = $self->rev_map_get($rev)) {
15710b6f
EW
2801 return ($rev, $c);
2802 }
2803 --$rev;
2804 }
2805 return (undef, undef);
2806}
2807
111947ef
DK
2808# Finds the first svn revision that exists on (if $eq_ok is true) or
2809# after $rev for the current branch. It will not search any higher
2810# than $max_rev. Returns the git commit hash and svn revision number
2811# if found, else (undef, undef).
2812sub find_rev_after {
2813 my ($self, $rev, $eq_ok, $max_rev) = @_;
2814 ++$rev unless $eq_ok;
060610c5 2815 $max_rev ||= $self->rev_map_max;
111947ef 2816 while ($rev <= $max_rev) {
060610c5 2817 if (my $c = $self->rev_map_get($rev)) {
111947ef
DK
2818 return ($rev, $c);
2819 }
2820 ++$rev;
2821 }
2822 return (undef, undef);
2823}
2824
9b981fc6 2825sub _new {
706587fc
EW
2826 my ($class, $repo_id, $ref_id, $path) = @_;
2827 unless (defined $repo_id && length $repo_id) {
2828 $repo_id = $Git::SVN::default_repo_id;
2829 }
2830 unless (defined $ref_id && length $ref_id) {
8b8fc068 2831 $_[2] = $ref_id = $Git::SVN::default_ref_id;
706587fc
EW
2832 }
2833 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2834 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2835 $_[3] = $path = '' unless (defined $path);
b4d57e5e 2836 mkpath(["$ENV{GIT_DIR}/svn"]);
26a62d57
EW
2837 bless {
2838 ref_id => $ref_id, dir => $dir, index => "$dir/index",
8a49ee97 2839 path => $path, config => "$ENV{GIT_DIR}/svn/config",
060610c5 2840 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
26a62d57
EW
2841}
2842
060610c5
EW
2843# for read-only access of old .rev_db formats
2844sub unlink_rev_db_symlink {
2845 my ($self) = @_;
2846 my $link = $self->rev_db_path;
2847 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2848 if (-l $link) {
2849 unlink $link or croak "unlink: $link failed!";
2850 }
2851}
2852
2853sub rev_db_path {
2854 my ($self, $uuid) = @_;
2855 my $db_path = $self->map_path($uuid);
2856 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2857 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2858 $db_path;
2859}
2860
2861# the new replacement for .rev_db
2862sub map_path {
26a62d57
EW
2863 my ($self, $uuid) = @_;
2864 $uuid ||= $self->ra_uuid;
060610c5 2865 "$self->{map_root}.$uuid";
9b981fc6
EW
2866}
2867
1c8443b0
EW
2868sub uri_encode {
2869 my ($f) = @_;
2870 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2871 $f
2872}
9b981fc6 2873
18ea92bd
SV
2874sub remove_username {
2875 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2876}
2877
d976acfd
EW
2878package Git::SVN::Prompt;
2879use strict;
2880use warnings;
2881require SVN::Core;
2882use vars qw/$_no_auth_cache $_username/;
2883
2884sub simple {
30d055aa
EW
2885 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2886 $may_save = undef if $_no_auth_cache;
2887 $default_username = $_username if defined $_username;
2888 if (defined $default_username && length $default_username) {
2889 if (defined $realm && length $realm) {
6f729591
EW
2890 print STDERR "Authentication realm: $realm\n";
2891 STDERR->flush;
30d055aa
EW
2892 }
2893 $cred->username($default_username);
2894 } else {
d976acfd 2895 username($cred, $realm, $may_save, $pool);
30d055aa
EW
2896 }
2897 $cred->password(_read_password("Password for '" .
2898 $cred->username . "': ", $realm));
2899 $cred->may_save($may_save);
2900 $SVN::_Core::SVN_NO_ERROR;
2901}
2902
d976acfd 2903sub ssl_server_trust {
30d055aa
EW
2904 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2905 $may_save = undef if $_no_auth_cache;
6f729591 2906 print STDERR "Error validating server certificate for '$realm':\n";
fd499bcc
ER
2907 {
2908 no warnings 'once';
2909 # All variables SVN::Auth::SSL::* are used only once,
2910 # so we're shutting up Perl warnings about this.
2911 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2912 print STDERR " - The certificate is not issued ",
2913 "by a trusted authority. Use the\n",
2914 " fingerprint to validate ",
2915 "the certificate manually!\n";
2916 }
2917 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2918 print STDERR " - The certificate hostname ",
2919 "does not match.\n";
2920 }
2921 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2922 print STDERR " - The certificate is not yet valid.\n";
2923 }
2924 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2925 print STDERR " - The certificate has expired.\n";
2926 }
2927 if ($failures & $SVN::Auth::SSL::OTHER) {
2928 print STDERR " - The certificate has ",
2929 "an unknown error.\n";
2930 }
2931 } # no warnings 'once'
6f729591
EW
2932 printf STDERR
2933 "Certificate information:\n".
30d055aa
EW
2934 " - Hostname: %s\n".
2935 " - Valid: from %s until %s\n".
2936 " - Issuer: %s\n".
2937 " - Fingerprint: %s\n",
2938 map $cert_info->$_, qw(hostname valid_from valid_until
6f729591 2939 issuer_dname fingerprint);
30d055aa
EW
2940 my $choice;
2941prompt:
6f729591 2942 print STDERR $may_save ?
30d055aa
EW
2943 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2944 "(R)eject or accept (t)emporarily? ";
6f729591 2945 STDERR->flush;
30d055aa
EW
2946 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2947 if ($choice =~ /^t$/i) {
2948 $cred->may_save(undef);
2949 } elsif ($choice =~ /^r$/i) {
2950 return -1;
2951 } elsif ($may_save && $choice =~ /^p$/i) {
2952 $cred->may_save($may_save);
2953 } else {
2954 goto prompt;
2955 }
2956 $cred->accepted_failures($failures);
2957 $SVN::_Core::SVN_NO_ERROR;
2958}
2959
d976acfd 2960sub ssl_client_cert {
30d055aa
EW
2961 my ($cred, $realm, $may_save, $pool) = @_;
2962 $may_save = undef if $_no_auth_cache;
6f729591
EW
2963 print STDERR "Client certificate filename: ";
2964 STDERR->flush;
30d055aa
EW
2965 chomp(my $filename = <STDIN>);
2966 $cred->cert_file($filename);
2967 $cred->may_save($may_save);
2968 $SVN::_Core::SVN_NO_ERROR;
2969}
2970
d976acfd 2971sub ssl_client_cert_pw {
30d055aa
EW
2972 my ($cred, $realm, $may_save, $pool) = @_;
2973 $may_save = undef if $_no_auth_cache;
2974 $cred->password(_read_password("Password: ", $realm));
2975 $cred->may_save($may_save);
2976 $SVN::_Core::SVN_NO_ERROR;
2977}
2978
d976acfd 2979sub username {
30d055aa
EW
2980 my ($cred, $realm, $may_save, $pool) = @_;
2981 $may_save = undef if $_no_auth_cache;
2982 if (defined $realm && length $realm) {
6f729591 2983 print STDERR "Authentication realm: $realm\n";
30d055aa
EW
2984 }
2985 my $username;
2986 if (defined $_username) {
2987 $username = $_username;
2988 } else {
6f729591
EW
2989 print STDERR "Username: ";
2990 STDERR->flush;
30d055aa
EW
2991 chomp($username = <STDIN>);
2992 }
2993 $cred->username($username);
2994 $cred->may_save($may_save);
2995 $SVN::_Core::SVN_NO_ERROR;
2996}
2997
2998sub _read_password {
2999 my ($prompt, $realm) = @_;
6f729591
EW
3000 print STDERR $prompt;
3001 STDERR->flush;
30d055aa
EW
3002 require Term::ReadKey;
3003 Term::ReadKey::ReadMode('noecho');
3004 my $password = '';
3005 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3006 last if $key =~ /[\012\015]/; # \n\r
3007 $password .= $key;
3008 }
3009 Term::ReadKey::ReadMode('restore');
6f729591
EW
3010 print STDERR "\n";
3011 STDERR->flush;
30d055aa
EW
3012 $password;
3013}
3014
27a1a801
EW
3015package SVN::Git::Fetcher;
3016use vars qw/@ISA/;
3017use strict;
3018use warnings;
3019use Carp qw/croak/;
3020use IO::File qw//;
3021
3022# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3023sub new {
3024 my ($class, $git_svn) = @_;
3025 my $self = SVN::Delta::Editor->new;
3026 bless $self, $class;
1c8443b0 3027 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
d2a9a87b
EW
3028 $self->{empty} = {};
3029 $self->{dir_prop} = {};
3030 $self->{file_prop} = {};
3031 $self->{absent_dir} = {};
3032 $self->{absent_file} = {};
ef3cfaad 3033 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
27a1a801
EW
3034 $self;
3035}
3036
8b8fc068
EW
3037sub set_path_strip {
3038 my ($self, $path) = @_;
4e9f6cc7 3039 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
8b8fc068
EW
3040}
3041
d2a9a87b
EW
3042sub open_root {
3043 { path => '' };
3044}
3045
3046sub open_directory {
3047 my ($self, $path, $pb, $rev) = @_;
3048 { path => $path };
3049}
3050
706587fc
EW
3051sub git_path {
3052 my ($self, $path) = @_;
2b27f6c8
EW
3053 if ($self->{path_strip}) {
3054 $path =~ s!$self->{path_strip}!! or
3055 die "Failed to strip path '$path' ($self->{path_strip})\n";
3056 }
706587fc
EW
3057 $path;
3058}
3059
27a1a801
EW
3060sub delete_entry {
3061 my ($self, $path, $rev, $pb) = @_;
4a87db0e 3062
706587fc 3063 my $gpath = $self->git_path($path);
8a603774
EW
3064 return undef if ($gpath eq '');
3065
4a87db0e 3066 # remove entire directories.
706587fc 3067 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
4a87db0e
EW
3068 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3069 -r --name-only -z/,
706587fc 3070 $self->{c}, '--', $gpath);
4a87db0e
EW
3071 local $/ = "\0";
3072 while (<$ls>) {
ef3cfaad
EW
3073 chomp;
3074 $self->{gii}->remove($_);
9e3cdbd4 3075 print "\tD\t$_\n" unless $::_q;
4a87db0e 3076 }
9e3cdbd4 3077 print "\tD\t$gpath/\n" unless $::_q;
4a87db0e
EW
3078 command_close_pipe($ls, $ctx);
3079 $self->{empty}->{$path} = 0
3080 } else {
ef3cfaad 3081 $self->{gii}->remove($gpath);
9e3cdbd4 3082 print "\tD\t$gpath\n" unless $::_q;
4a87db0e 3083 }
27a1a801
EW
3084 undef;
3085}
3086
3087sub open_file {
3088 my ($self, $path, $pb, $rev) = @_;
706587fc
EW
3089 my $gpath = $self->git_path($path);
3090 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
27a1a801 3091 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
006ede5e
EW
3092 unless (defined $mode && defined $blob) {
3093 die "$path was not found in commit $self->{c} (r$rev)\n";
3094 }
27a1a801 3095 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
0864e3ba 3096 pool => SVN::Pool->new, action => 'M' };
27a1a801
EW
3097}
3098
3099sub add_file {
3100 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
d2a9a87b
EW
3101 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3102 delete $self->{empty}->{$dir};
27a1a801 3103 { path => $path, mode_a => 100644, mode_b => 100644,
0864e3ba 3104 pool => SVN::Pool->new, action => 'A' };
27a1a801
EW
3105}
3106
d2a9a87b
EW
3107sub add_directory {
3108 my ($self, $path, $cp_path, $cp_rev) = @_;
12a6d752
EW
3109 my $gpath = $self->git_path($path);
3110 if ($gpath eq '') {
3111 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3112 -r --name-only -z/,
3113 $self->{c});
3114 local $/ = "\0";
3115 while (<$ls>) {
3116 chomp;
3117 $self->{gii}->remove($_);
3118 print "\tD\t$_\n" unless $::_q;
3119 }
3120 command_close_pipe($ls, $ctx);
3121 $self->{empty}->{$path} = 0;
3122 }
d2a9a87b
EW
3123 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3124 delete $self->{empty}->{$dir};
3125 $self->{empty}->{$path} = 1;
3126 { path => $path };
3127}
3128
3129sub change_dir_prop {
3130 my ($self, $db, $prop, $value) = @_;
3131 $self->{dir_prop}->{$db->{path}} ||= {};
3132 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3133 undef;
3134}
3135
3136sub absent_directory {
3137 my ($self, $path, $pb) = @_;
3138 $self->{absent_dir}->{$pb->{path}} ||= [];
3139 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3140 undef;
3141}
3142
3143sub absent_file {
3144 my ($self, $path, $pb) = @_;
3145 $self->{absent_file}->{$pb->{path}} ||= [];
3146 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3147 undef;
3148}
3149
27a1a801
EW
3150sub change_file_prop {
3151 my ($self, $fb, $prop, $value) = @_;
3152 if ($prop eq 'svn:executable') {
3153 if ($fb->{mode_b} != 120000) {
3154 $fb->{mode_b} = defined $value ? 100755 : 100644;
3155 }
3156 } elsif ($prop eq 'svn:special') {
3157 $fb->{mode_b} = defined $value ? 120000 : 100644;
d2a9a87b
EW
3158 } else {
3159 $self->{file_prop}->{$fb->{path}} ||= {};
3160 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
27a1a801
EW
3161 }
3162 undef;
3163}
3164
3165sub apply_textdelta {
3166 my ($self, $fb, $exp) = @_;
3167 my $fh = IO::File->new_tmpfile;
3168 $fh->autoflush(1);
3169 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3170 # (but $base does not,) so dup() it for reading in close_file
3171 open my $dup, '<&', $fh or croak $!;
3172 my $base = IO::File->new_tmpfile;
3173 $base->autoflush(1);
3174 if ($fb->{blob}) {
3175 defined (my $pid = fork) or croak $!;
3176 if (!$pid) {
3177 open STDOUT, '>&', $base or croak $!;
3178 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3179 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3180 }
3181 waitpid $pid, 0;
3182 croak $? if $?;
3183
3184 if (defined $exp) {
3185 seek $base, 0, 0 or croak $!;
8d7c4fad 3186 my $got = ::md5sum($base);
27a1a801
EW
3187 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3188 "expected: $exp\n",
3189 " got: $got\n" if ($got ne $exp);
3190 }
3191 }
3192 seek $base, 0, 0 or croak $!;
3193 $fb->{fh} = $dup;
3194 $fb->{base} = $base;
3195 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3196}
3197
3198sub close_file {
3199 my ($self, $fb, $exp) = @_;
3200 my $hash;
706587fc 3201 my $path = $self->git_path($fb->{path});
27a1a801 3202 if (my $fh = $fb->{fh}) {
7faf0686
EW
3203 if (defined $exp) {
3204 seek($fh, 0, 0) or croak $!;
8d7c4fad 3205 my $got = ::md5sum($fh);
7faf0686
EW
3206 if ($got ne $exp) {
3207 die "Checksum mismatch: $path\n",
3208 "expected: $exp\n got: $got\n";
3209 }
3210 }
bcd8ee5b 3211 sysseek($fh, 0, 0) or croak $!;
27a1a801 3212 if ($fb->{mode_b} == 120000) {
7fc35e0e
EW
3213 eval {
3214 sysread($fh, my $buf, 5) == 5 or croak $!;
3215 $buf eq 'link ' or die "$path has mode 120000",
3216 " but is not a link";
3217 };
3218 if ($@) {
3219 warn "$@\n";
3220 sysseek($fh, 0, 0) or croak $!;
3221 }
27a1a801
EW
3222 }
3223 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3224 if (!$pid) {
3225 open STDIN, '<&', $fh or croak $!;
3226 exec qw/git-hash-object -w --stdin/ or croak $!;
3227 }
3228 chomp($hash = do { local $/; <$out> });
3229 close $out or croak $!;
3230 close $fh or croak $!;
3231 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3232 close $fb->{base} or croak $!;
3233 } else {
3234 $hash = $fb->{blob} or die "no blob information\n";
3235 }
3236 $fb->{pool}->clear;
ef3cfaad 3237 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
9e3cdbd4 3238 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
27a1a801
EW
3239 undef;
3240}
3241
3242sub abort_edit {
3243 my $self = shift;
ef3cfaad
EW
3244 $self->{nr} = $self->{gii}->{nr};
3245 delete $self->{gii};
27a1a801
EW
3246 $self->SUPER::abort_edit(@_);
3247}
3248
3249sub close_edit {
3250 my $self = shift;
dad73c0b 3251 $self->{git_commit_ok} = 1;
ef3cfaad
EW
3252 $self->{nr} = $self->{gii}->{nr};
3253 delete $self->{gii};
27a1a801
EW
3254 $self->SUPER::close_edit(@_);
3255}
1a82e793 3256
a5e0cedc 3257package SVN::Git::Editor;
24e22aa8 3258use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
a5e0cedc
EW
3259use strict;
3260use warnings;
3261use Carp qw/croak/;
3262use IO::File;
3263
3264sub new {
61395354
EW
3265 my ($class, $opts) = @_;
3266 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3267 die "$_ required!\n" unless (defined $opts->{$_});
3268 }
3269
3270 my $pool = SVN::Pool->new;
3271 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3272 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3273 $opts->{r}, $mods);
3274
3275 # $opts->{ra} functions should not be used after this:
3276 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3277 $opts->{editor_cb}, $pool);
3278 my $self = SVN::Delta::Editor->new(@ce, $pool);
a5e0cedc 3279 bless $self, $class;
61395354
EW
3280 foreach (qw/svn_path r tree_a tree_b/) {
3281 $self->{$_} = $opts->{$_};
a5e0cedc 3282 }
61395354
EW
3283 $self->{url} = $opts->{ra}->{url};
3284 $self->{mods} = $mods;
3285 $self->{types} = $types;
3286 $self->{pool} = $pool;
a5e0cedc
EW
3287 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3288 $self->{rm} = { };
d3a840dc
EW
3289 $self->{path_prefix} = length $self->{svn_path} ?
3290 "$self->{svn_path}/" : '';
a5e0cedc
EW
3291 return $self;
3292}
3293
61395354
EW
3294sub generate_diff {
3295 my ($tree_a, $tree_b) = @_;
3296 my @diff_tree = qw(diff-tree -z -r);
24e22aa8
EW
3297 if ($_cp_similarity) {
3298 push @diff_tree, "-C$_cp_similarity";
61395354
EW
3299 } else {
3300 push @diff_tree, '-C';
3301 }
24e22aa8
EW
3302 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3303 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
61395354
EW
3304 push @diff_tree, $tree_a, $tree_b;
3305 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3306 local $/ = "\0";
3307 my $state = 'meta';
3308 my @mods;
3309 while (<$diff_fh>) {
3310 chomp $_; # this gets rid of the trailing "\0"
3311 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3312 $::sha1\s($::sha1)\s
3313 ([MTCRAD])\d*$/xo) {
3314 push @mods, { mode_a => $1, mode_b => $2,
3315 sha1_b => $3, chg => $4 };
3316 if ($4 =~ /^(?:C|R)$/) {
3317 $state = 'file_a';
3318 } else {
3319 $state = 'file_b';
3320 }
3321 } elsif ($state eq 'file_a') {
3322 my $x = $mods[$#mods] or croak "Empty array\n";
3323 if ($x->{chg} !~ /^(?:C|R)$/) {
3324 croak "Error parsing $_, $x->{chg}\n";
3325 }
3326 $x->{file_a} = $_;
3327 $state = 'file_b';
3328 } elsif ($state eq 'file_b') {
3329 my $x = $mods[$#mods] or croak "Empty array\n";
3330 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3331 croak "Error parsing $_, $x->{chg}\n";
3332 }
3333 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3334 croak "Error parsing $_, $x->{chg}\n";
3335 }
3336 $x->{file_b} = $_;
3337 $state = 'meta';
3338 } else {
3339 croak "Error parsing $_\n";
3340 }
3341 }
3342 command_close_pipe($diff_fh, $ctx);
3343 \@mods;
3344}
3345
3346sub check_diff_paths {
3347 my ($ra, $pfx, $rev, $mods) = @_;
3348 my %types;
3349 $pfx .= '/' if length $pfx;
3350
3351 sub type_diff_paths {
3352 my ($ra, $types, $path, $rev) = @_;
3353 my @p = split m#/+#, $path;
3354 my $c = shift @p;
3355 unless (defined $types->{$c}) {
3356 $types->{$c} = $ra->check_path($c, $rev);
3357 }
3358 while (@p) {
3359 $c .= '/' . shift @p;
3360 next if defined $types->{$c};
3361 $types->{$c} = $ra->check_path($c, $rev);
3362 }
3363 }
3364
3365 foreach my $m (@$mods) {
3366 foreach my $f (qw/file_a file_b/) {
3367 next unless defined $m->{$f};
3368 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3369 if (length $pfx.$dir && ! defined $types{$dir}) {
3370 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3371 }
3372 }
3373 }
3374 \%types;
3375}
3376
a5e0cedc
EW
3377sub split_path {
3378 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3379}
3380
3381sub repo_path {
d3a840dc
EW
3382 my ($self, $path) = @_;
3383 $self->{path_prefix}.(defined $path ? $path : '');
a5e0cedc
EW
3384}
3385
3386sub url_path {
3387 my ($self, $path) = @_;
29633bb9
EW
3388 if ($self->{url} =~ m#^https?://#) {
3389 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3390 }
6e8548cc 3391 $self->{url} . '/' . $self->repo_path($path);
a5e0cedc
EW
3392}
3393
3394sub rmdirs {
61395354 3395 my ($self) = @_;
a5e0cedc
EW
3396 my $rm = $self->{rm};
3397 delete $rm->{''}; # we never delete the url we're tracking
3398 return unless %$rm;
3399
3400 foreach (keys %$rm) {
3401 my @d = split m#/#, $_;
3402 my $c = shift @d;
3403 $rm->{$c} = 1;
3404 while (@d) {
3405 $c .= '/' . shift @d;
3406 $rm->{$c} = 1;
3407 }
3408 }
3409 delete $rm->{$self->{svn_path}};
3410 delete $rm->{''}; # we never delete the url we're tracking
3411 return unless %$rm;
3412
61395354
EW
3413 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3414 $self->{tree_b});
a5e0cedc
EW
3415 local $/ = "\0";
3416 while (<$fh>) {
3417 chomp;
747fa12c 3418 my @dn = split m#/#, $_;
c07eee1f
EW
3419 while (pop @dn) {
3420 delete $rm->{join '/', @dn};
3421 }
3422 unless (%$rm) {
22600a25 3423 close $fh;
c07eee1f
EW
3424 return;
3425 }
a5e0cedc 3426 }
aef4e921 3427 command_close_pipe($fh, $ctx);
c07eee1f 3428
a5e0cedc
EW
3429 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3430 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3431 $self->close_directory($bat->{$d}, $p);
3432 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
44320b9e 3433 print "\tD+\t$d/\n" unless $::_q;
a5e0cedc
EW
3434 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3435 delete $bat->{$d};
3436 }
3437}
3438
3439sub open_or_add_dir {
3440 my ($self, $full_path, $baton) = @_;
6e8548cc
EW
3441 my $t = $self->{types}->{$full_path};
3442 if (!defined $t) {
3443 die "$full_path not known in r$self->{r} or we have a bug!\n";
3444 }
fd499bcc
ER
3445 {
3446 no warnings 'once';
3447 # SVN::Node::none and SVN::Node::file are used only once,
3448 # so we're shutting up Perl's warnings about them.
3449 if ($t == $SVN::Node::none) {
3450 return $self->add_directory($full_path, $baton,
3451 undef, -1, $self->{pool});
3452 } elsif ($t == $SVN::Node::dir) {
3453 return $self->open_directory($full_path, $baton,
3454 $self->{r}, $self->{pool});
3455 } # no warnings 'once'
3456 print STDERR "$full_path already exists in repository at ",
3457 "r$self->{r} and it is not a directory (",
3458 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3459 } # no warnings 'once'
a5e0cedc
EW
3460 exit 1;
3461}
3462
3463sub ensure_path {
3464 my ($self, $path) = @_;
3465 my $bat = $self->{bat};
6e8548cc
EW
3466 my $repo_path = $self->repo_path($path);
3467 return $bat->{''} unless (length $repo_path);
3468 my @p = split m#/+#, $repo_path;
a5e0cedc
EW
3469 my $c = shift @p;
3470 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3471 while (@p) {
3472 my $c0 = $c;
3473 $c .= '/' . shift @p;
3474 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3475 }
3476 return $bat->{$c};
3477}
3478
3479sub A {
44320b9e 3480 my ($self, $m) = @_;
a5e0cedc
EW
3481 my ($dir, $file) = split_path($m->{file_b});
3482 my $pbat = $self->ensure_path($dir);
3483 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3484 undef, -1);
44320b9e 3485 print "\tA\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3486 $self->chg_file($fbat, $m);
3487 $self->close_file($fbat,undef,$self->{pool});
3488}
3489
3490sub C {
44320b9e 3491 my ($self, $m) = @_;
a5e0cedc
EW
3492 my ($dir, $file) = split_path($m->{file_b});
3493 my $pbat = $self->ensure_path($dir);
3494 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3495 $self->url_path($m->{file_a}), $self->{r});
44320b9e 3496 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3497 $self->chg_file($fbat, $m);
3498 $self->close_file($fbat,undef,$self->{pool});
3499}
3500
3501sub delete_entry {
3502 my ($self, $path, $pbat) = @_;
3503 my $rpath = $self->repo_path($path);
3504 my ($dir, $file) = split_path($rpath);
3505 $self->{rm}->{$dir} = 1;
3506 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3507}
3508
3509sub R {
44320b9e 3510 my ($self, $m) = @_;
a5e0cedc
EW
3511 my ($dir, $file) = split_path($m->{file_b});
3512 my $pbat = $self->ensure_path($dir);
3513 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3514 $self->url_path($m->{file_a}), $self->{r});
44320b9e 3515 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3516 $self->chg_file($fbat, $m);
3517 $self->close_file($fbat,undef,$self->{pool});
3518
3519 ($dir, $file) = split_path($m->{file_a});
3520 $pbat = $self->ensure_path($dir);
3521 $self->delete_entry($m->{file_a}, $pbat);
3522}
3523
3524sub M {
44320b9e 3525 my ($self, $m) = @_;
a5e0cedc
EW
3526 my ($dir, $file) = split_path($m->{file_b});
3527 my $pbat = $self->ensure_path($dir);
3528 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3529 $pbat,$self->{r},$self->{pool});
44320b9e 3530 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
a5e0cedc
EW
3531 $self->chg_file($fbat, $m);
3532 $self->close_file($fbat,undef,$self->{pool});
3533}
3534
3535sub T { shift->M(@_) }
3536
3537sub change_file_prop {
3538 my ($self, $fbat, $pname, $pval) = @_;