?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Strongly discourage --update-head-ok in fetch-options documentation.
[git/jnareb-git.git] / git-svn.perl
blobb28c5bbc7238120d1d98445f1bf0f116dd7a0b8a
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB/;
10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11 $VERSION = '@@GIT_VERSION@@';
12
13 use Cwd qw/abs_path/;
14 $GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
15 $ENV{GIT_DIR} = $GIT_DIR;
16
17 my $LC_ALL = $ENV{LC_ALL};
18 my $TZ = $ENV{TZ};
19 # make sure the svn binary gives consistent output between locales and TZs:
20 $ENV{TZ} = 'UTC';
21 $ENV{LC_ALL} = 'C';
22 $| = 1; # unbuffer STDOUT
23
24 # properties that we do not log:
25 my %SKIP = ( 'svn:wc:ra_dav:version-url' => 1,
26 'svn:special' => 1,
27 'svn:executable' => 1,
28 'svn:entry:committed-rev' => 1,
29 'svn:entry:last-author' => 1,
30 'svn:entry:uuid' => 1,
31 'svn:entry:committed-date' => 1,
32 );
33
34 sub fatal (@) { print STDERR @_; exit 1 }
35 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
36 require SVN::Ra;
37 require SVN::Delta;
38 if ($SVN::Core::VERSION lt '1.1.0') {
39 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
40 }
41 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
42 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
43 *SVN::Git::Fetcher::process_rm = *process_rm;
44 use Carp qw/croak/;
45 use IO::File qw//;
46 use File::Basename qw/dirname basename/;
47 use File::Path qw/mkpath/;
48 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
49 use POSIX qw/strftime/;
50 use IPC::Open3;
51 use Memoize;
52 use Git qw/command command_oneline command_noisy
53 command_output_pipe command_input_pipe command_close_pipe/;
54 memoize('revisions_eq');
55 memoize('cmt_metadata');
56 memoize('get_commit_time');
57
58 my ($SVN);
59
60 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
61 my $sha1 = qr/[a-f\d]{40}/;
62 my $sha1_short = qr/[a-f\d]{4,40}/;
63 my $_esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
64 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
65 $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
66 $_repack, $_repack_nr, $_repack_flags, $_q,
67 $_message, $_file, $_follow_parent, $_no_metadata,
68 $_template, $_shared, $_no_default_regex, $_no_graft_copy,
69 $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
70 $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
71 $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive,
72 $_username, $_config_dir, $_no_auth_cache,
73 $_pager, $_color);
74 my (@_branch_from, %tree_map, %users, %rusers, %equiv);
75 my ($_svn_can_do_switch);
76 my @repo_path_split_cache;
77
78 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
79 'branch|b=s' => \@_branch_from,
80 'follow-parent|follow' => \$_follow_parent,
81 'branch-all-refs|B' => \$_branch_all_refs,
82 'authors-file|A=s' => \$_authors,
83 'repack:i' => \$_repack,
84 'no-metadata' => \$_no_metadata,
85 'quiet|q' => \$_q,
86 'username=s' => \$_username,
87 'config-dir=s' => \$_config_dir,
88 'no-auth-cache' => \$_no_auth_cache,
89 'ignore-nodate' => \$_ignore_nodate,
90 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
91
92 my ($_trunk, $_tags, $_branches);
93 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
94 'tags|t=s' => \$_tags,
95 'branches|b=s' => \$_branches );
96 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
97 my %cmt_opts = ( 'edit|e' => \$_edit,
98 'rmdir' => \$_rmdir,
99 'find-copies-harder' => \$_find_copies_harder,
100 'l=i' => \$_l,
101 'copy-similarity|C=i'=> \$_cp_similarity
102 );
103
104 my %cmd = (
105 fetch => [ \&fetch, "Download new revisions from SVN",
106 { 'revision|r=s' => \$_revision, %fc_opts } ],
107 init => [ \&init, "Initialize a repo for tracking" .
108 " (requires URL argument)",
109 \%init_opts ],
110 dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
111 { 'merge|m|M' => \$_merge,
112 'strategy|s=s' => \$_strategy,
113 'dry-run|n' => \$_dry_run,
114 %cmt_opts, %fc_opts } ],
115 'set-tree' => [ \&commit, "Set an SVN repository to a git tree-ish",
116 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
117 'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
118 { 'revision|r=i' => \$_revision } ],
119 rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
120 { 'no-ignore-externals' => \$_no_ignore_ext,
121 'copy-remote|remote=s' => \$_cp_remote,
122 'upgrade' => \$_upgrade } ],
123 'graft-branches' => [ \&graft_branches,
124 'Detect merges/branches from already imported history',
125 { 'merge-rx|m' => \@_opt_m,
126 'branch|b=s' => \@_branch_from,
127 'branch-all-refs|B' => \$_branch_all_refs,
128 'no-default-regex' => \$_no_default_regex,
129 'no-graft-copy' => \$_no_graft_copy } ],
130 'multi-init' => [ \&multi_init,
131 'Initialize multiple trees (like git-svnimport)',
132 { %multi_opts, %init_opts,
133 'revision|r=i' => \$_revision,
134 'username=s' => \$_username,
135 'config-dir=s' => \$_config_dir,
136 'no-auth-cache' => \$_no_auth_cache,
137 } ],
138 'multi-fetch' => [ \&multi_fetch,
139 'Fetch multiple trees (like git-svnimport)',
140 \%fc_opts ],
141 'log' => [ \&show_log, 'Show commit logs',
142 { 'limit=i' => \$_limit,
143 'revision|r=s' => \$_revision,
144 'verbose|v' => \$_verbose,
145 'incremental' => \$_incremental,
146 'oneline' => \$_oneline,
147 'show-commit' => \$_show_commit,
148 'non-recursive' => \$_non_recursive,
149 'authors-file|A=s' => \$_authors,
150 'color' => \$_color,
151 'pager=s' => \$_pager,
152 } ],
153 'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
154 { 'message|m=s' => \$_message,
155 'file|F=s' => \$_file,
156 'revision|r=s' => \$_revision,
157 %cmt_opts } ],
158 );
159
160 my $cmd;
161 for (my $i = 0; $i < @ARGV; $i++) {
162 if (defined $cmd{$ARGV[$i]}) {
163 $cmd = $ARGV[$i];
164 splice @ARGV, $i, 1;
165 last;
166 }
167 };
168
169 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
170
171 read_repo_config(\%opts);
172 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
173 'version|V' => \$_version,
174 'id|i=s' => \$GIT_SVN);
175 exit 1 if (!$rv && $cmd ne 'log');
176
177 set_default_vals();
178 usage(0) if $_help;
179 version() if $_version;
180 usage(1) unless defined $cmd;
181 init_vars();
182 load_authors() if $_authors;
183 load_all_refs() if $_branch_all_refs;
184 migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
185 $cmd{$cmd}->[0]->(@ARGV);
186 exit 0;
187
188 ####################### primary functions ######################
189 sub usage {
190 my $exit = shift || 0;
191 my $fd = $exit ? \*STDERR : \*STDOUT;
192 print $fd <<"";
193 git-svn - bidirectional operations between a single Subversion tree and git
194 Usage: $0 <command> [options] [arguments]\n
195
196 print $fd "Available commands:\n" unless $cmd;
197
198 foreach (sort keys %cmd) {
199 next if $cmd && $cmd ne $_;
200 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
201 foreach (keys %{$cmd{$_}->[2]}) {
202 # prints out arguments as they should be passed:
203 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
204 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
205 "--$_" : "-$_" }
206 split /\|/,$_)," $x\n";
207 }
208 }
209 print $fd <<"";
210 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
211 arbitrary identifier if you're tracking multiple SVN branches/repositories in
212 one git repository and want to keep them separate. See git-svn(1) for more
213 information.
214
215 exit $exit;
216 }
217
218 sub version {
219 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
220 exit 0;
221 }
222
223 sub rebuild {
224 if (!verify_ref("refs/remotes/$GIT_SVN^0")) {
225 copy_remote_ref();
226 }
227 $SVN_URL = shift or undef;
228 my $newest_rev = 0;
229 if ($_upgrade) {
230 command_noisy('update-ref',"refs/remotes/$GIT_SVN","
231 $GIT_SVN-HEAD");
232 } else {
233 check_upgrade_needed();
234 }
235
236 my ($rev_list, $ctx) = command_output_pipe("rev-list",
237 "refs/remotes/$GIT_SVN");
238 my $latest;
239 while (<$rev_list>) {
240 chomp;
241 my $c = $_;
242 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
243 my @commit = grep(/^git-svn-id: /,
244 command(qw/cat-file commit/, $c));
245 next if (!@commit); # skip merges
246 my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
247 if (!defined $rev || !$uuid) {
248 croak "Unable to extract revision or UUID from ",
249 "$c, $commit[$#commit]\n";
250 }
251
252 # if we merged or otherwise started elsewhere, this is
253 # how we break out of it
254 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
255 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
256
257 unless (defined $latest) {
258 if (!$SVN_URL && !$url) {
259 croak "SVN repository location required: $url\n";
260 }
261 $SVN_URL ||= $url;
262 $SVN_UUID ||= $uuid;
263 setup_git_svn();
264 $latest = $rev;
265 }
266 revdb_set($REVDB, $rev, $c);
267 print "r$rev = $c\n";
268 $newest_rev = $rev if ($rev > $newest_rev);
269 }
270 command_close_pipe($rev_list, $ctx);
271 }
272
273 sub init {
274 my $url = shift or die "SVN repository location required " .
275 "as a command-line argument\n";
276 $url =~ s!/+$!!; # strip trailing slash
277
278 if (my $repo_path = shift) {
279 unless (-d $repo_path) {
280 mkpath([$repo_path]);
281 }
282 $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
283 init_vars();
284 }
285
286 $SVN_URL = $url;
287 unless (-d $GIT_DIR) {
288 my @init_db = ('init-db');
289 push @init_db, "--template=$_template" if defined $_template;
290 push @init_db, "--shared" if defined $_shared;
291 command_noisy(@init_db);
292 }
293 setup_git_svn();
294 }
295
296 sub fetch {
297 check_upgrade_needed();
298 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
299 my $ret = fetch_lib(@_);
300 if ($ret->{commit} && !verify_ref('refs/heads/master^0')) {
301 command_noisy(qw(update-ref refs/heads/master),$ret->{commit});
302 }
303 return $ret;
304 }
305
306 sub fetch_lib {
307 my (@parents) = @_;
308 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
309 $SVN ||= libsvn_connect($SVN_URL);
310 my ($last_rev, $last_commit) = svn_grab_base_rev();
311 my ($base, $head) = libsvn_parse_revision($last_rev);
312 if ($base > $head) {
313 return { revision => $last_rev, commit => $last_commit }
314 }
315 my $index = set_index($GIT_SVN_INDEX);
316
317 # limit ourselves and also fork() since get_log won't release memory
318 # after processing a revision and SVN stuff seems to leak
319 my $inc = 1000;
320 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
321 read_uuid();
322 if (defined $last_commit) {
323 unless (-e $GIT_SVN_INDEX) {
324 command_noisy('read-tree', $last_commit);
325 }
326 my $x = command_oneline('write-tree');
327 my ($y) = (command(qw/cat-file commit/, $last_commit)
328 =~ /^tree ($sha1)/m);
329 if ($y ne $x) {
330 unlink $GIT_SVN_INDEX or croak $!;
331 command_noisy('read-tree', $last_commit);
332 }
333 $x = command_oneline('write-tree');
334 if ($y ne $x) {
335 print STDERR "trees ($last_commit) $y != $x\n",
336 "Something is seriously wrong...\n";
337 }
338 }
339 while (1) {
340 # fork, because using SVN::Pool with get_log() still doesn't
341 # seem to help enough to keep memory usage down.
342 defined(my $pid = fork) or croak $!;
343 if (!$pid) {
344 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
345
346 # Yes I'm perfectly aware that the fourth argument
347 # below is the limit revisions number. Unfortunately
348 # performance sucks with it enabled, so it's much
349 # faster to fetch revision ranges instead of relying
350 # on the limiter.
351 libsvn_get_log(libsvn_dup_ra($SVN), [''],
352 $min, $max, 0, 1, 1,
353 sub {
354 my $log_msg;
355 if ($last_commit) {
356 $log_msg = libsvn_fetch(
357 $last_commit, @_);
358 $last_commit = git_commit(
359 $log_msg,
360 $last_commit,
361 @parents);
362 } else {
363 $log_msg = libsvn_new_tree(@_);
364 $last_commit = git_commit(
365 $log_msg, @parents);
366 }
367 });
368 exit 0;
369 }
370 waitpid $pid, 0;
371 croak $? if $?;
372 ($last_rev, $last_commit) = svn_grab_base_rev();
373 last if ($max >= $head);
374 $min = $max + 1;
375 $max += $inc;
376 $max = $head if ($max > $head);
377 $SVN = libsvn_connect($SVN_URL);
378 }
379 restore_index($index);
380 return { revision => $last_rev, commit => $last_commit };
381 }
382
383 sub commit {
384 my (@commits) = @_;
385 check_upgrade_needed();
386 if ($_stdin || !@commits) {
387 print "Reading from stdin...\n";
388 @commits = ();
389 while (<STDIN>) {
390 if (/\b($sha1_short)\b/o) {
391 unshift @commits, $1;
392 }
393 }
394 }
395 my @revs;
396 foreach my $c (@commits) {
397 my @tmp = command('rev-parse',$c);
398 if (scalar @tmp == 1) {
399 push @revs, $tmp[0];
400 } elsif (scalar @tmp > 1) {
401 push @revs, reverse(command('rev-list',@tmp));
402 } else {
403 die "Failed to rev-parse $c\n";
404 }
405 }
406 commit_lib(@revs);
407 print "Done committing ",scalar @revs," revisions to SVN\n";
408 }
409
410 sub commit_lib {
411 my (@revs) = @_;
412 my ($r_last, $cmt_last) = svn_grab_base_rev();
413 defined $r_last or die "Must have an existing revision to commit\n";
414 my $fetched = fetch();
415 if ($r_last != $fetched->{revision}) {
416 print STDERR "There are new revisions that were fetched ",
417 "and need to be merged (or acknowledged) ",
418 "before committing.\n",
419 "last rev: $r_last\n",
420 " current: $fetched->{revision}\n";
421 exit 1;
422 }
423 read_uuid();
424 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
425 my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
426
427 my $repo;
428 set_svn_commit_env();
429 foreach my $c (@revs) {
430 my $log_msg = get_commit_message($c, $commit_msg);
431
432 # fork for each commit because there's a memory leak I
433 # can't track down... (it's probably in the SVN code)
434 defined(my $pid = open my $fh, '-|') or croak $!;
435 if (!$pid) {
436 my $ed = SVN::Git::Editor->new(
437 { r => $r_last,
438 ra => libsvn_dup_ra($SVN),
439 c => $c,
440 svn_path => $SVN->{svn_path},
441 },
442 $SVN->get_commit_editor(
443 $log_msg->{msg},
444 sub {
445 libsvn_commit_cb(
446 @_, $c,
447 $log_msg->{msg},
448 $r_last,
449 $cmt_last)
450 },
451 @lock)
452 );
453 my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
454 if (@$mods == 0) {
455 print "No changes\nr$r_last = $cmt_last\n";
456 $ed->abort_edit;
457 } else {
458 $ed->close_edit;
459 }
460 exit 0;
461 }
462 my ($r_new, $cmt_new, $no);
463 while (<$fh>) {
464 print $_;
465 chomp;
466 if (/^r(\d+) = ($sha1)$/o) {
467 ($r_new, $cmt_new) = ($1, $2);
468 } elsif ($_ eq 'No changes') {
469 $no = 1;
470 }
471 }
472 close $fh or exit 1;
473 if (! defined $r_new && ! defined $cmt_new) {
474 unless ($no) {
475 die "Failed to parse revision information\n";
476 }
477 } else {
478 ($r_last, $cmt_last) = ($r_new, $cmt_new);
479 }
480 }
481 $ENV{LC_ALL} = 'C';
482 unlink $commit_msg;
483 }
484
485 sub dcommit {
486 my $head = shift || 'HEAD';
487 my $gs = "refs/remotes/$GIT_SVN";
488 my @refs = command(qw/rev-list --no-merges/, "$gs..$head");
489 my $last_rev;
490 foreach my $d (reverse @refs) {
491 if (!verify_ref("$d~1")) {
492 die "Commit $d\n",
493 "has no parent commit, and therefore ",
494 "nothing to diff against.\n",
495 "You should be working from a repository ",
496 "originally created by git-svn\n";
497 }
498 unless (defined $last_rev) {
499 (undef, $last_rev, undef) = cmt_metadata("$d~1");
500 unless (defined $last_rev) {
501 die "Unable to extract revision information ",
502 "from commit $d~1\n";
503 }
504 }
505 if ($_dry_run) {
506 print "diff-tree $d~1 $d\n";
507 } else {
508 if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
509 $last_rev = $r;
510 } # else: no changes, same $last_rev
511 }
512 }
513 return if $_dry_run;
514 fetch();
515 my @diff = command('diff-tree', 'HEAD', $gs, '--');
516 my @finish;
517 if (@diff) {
518 @finish = qw/rebase/;
519 push @finish, qw/--merge/ if $_merge;
520 push @finish, "--strategy=$_strategy" if $_strategy;
521 print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
522 } else {
523 print "No changes between current HEAD and $gs\n",
524 "Resetting to the latest $gs\n";
525 @finish = qw/reset --mixed/;
526 }
527 command_noisy(@finish, $gs);
528 }
529
530 sub show_ignore {
531 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
532 my $repo;
533 $SVN ||= libsvn_connect($SVN_URL);
534 my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
535 libsvn_traverse_ignore(\*STDOUT, $SVN->{svn_path}, $r);
536 }
537
538 sub graft_branches {
539 my $gr_file = "$GIT_DIR/info/grafts";
540 my ($grafts, $comments) = read_grafts($gr_file);
541 my $gr_sha1;
542
543 if (%$grafts) {
544 # temporarily disable our grafts file to make this idempotent
545 chomp($gr_sha1 = command(qw/hash-object -w/,$gr_file));
546 rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
547 }
548
549 my $l_map = read_url_paths();
550 my @re = map { qr/$_/is } @_opt_m if @_opt_m;
551 unless ($_no_default_regex) {
552 push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
553 qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
554 qr/\b(?:from|of)\s+([\w\.\-]+)/i );
555 }
556 foreach my $u (keys %$l_map) {
557 if (@re) {
558 foreach my $p (keys %{$l_map->{$u}}) {
559 graft_merge_msg($grafts,$l_map,$u,$p,@re);
560 }
561 }
562 unless ($_no_graft_copy) {
563 graft_file_copy_lib($grafts,$l_map,$u);
564 }
565 }
566 graft_tree_joins($grafts);
567
568 write_grafts($grafts, $comments, $gr_file);
569 unlink "$gr_file~$gr_sha1" if $gr_sha1;
570 }
571
572 sub multi_init {
573 my $url = shift;
574 $_trunk ||= 'trunk';
575 $_trunk =~ s#/+$##;
576 $url =~ s#/+$## if $url;
577 if ($_trunk !~ m#^[a-z\+]+://#) {
578 $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
579 unless ($url) {
580 print STDERR "E: '$_trunk' is not a complete URL ",
581 "and a separate URL is not specified\n";
582 exit 1;
583 }
584 $_trunk = $url . $_trunk;
585 }
586 my $ch_id;
587 if ($GIT_SVN eq 'git-svn') {
588 $ch_id = 1;
589 $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
590 }
591 init_vars();
592 unless (-d $GIT_SVN_DIR) {
593 print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
594 init($_trunk);
595 command_noisy('repo-config', 'svn.trunk', $_trunk);
596 }
597 complete_url_ls_init($url, $_branches, '--branches/-b', '');
598 complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
599 }
600
601 sub multi_fetch {
602 # try to do trunk first, since branches/tags
603 # may be descended from it.
604 if (-e "$GIT_DIR/svn/trunk/info/url") {
605 fetch_child_id('trunk', @_);
606 }
607 rec_fetch('', "$GIT_DIR/svn", @_);
608 }
609
610 sub show_log {
611 my (@args) = @_;
612 my ($r_min, $r_max);
613 my $r_last = -1; # prevent dupes
614 rload_authors() if $_authors;
615 if (defined $TZ) {
616 $ENV{TZ} = $TZ;
617 } else {
618 delete $ENV{TZ};
619 }
620 if (defined $_revision) {
621 if ($_revision =~ /^(\d+):(\d+)$/) {
622 ($r_min, $r_max) = ($1, $2);
623 } elsif ($_revision =~ /^\d+$/) {
624 $r_min = $r_max = $_revision;
625 } else {
626 print STDERR "-r$_revision is not supported, use ",
627 "standard \'git log\' arguments instead\n";
628 exit 1;
629 }
630 }
631
632 config_pager();
633 @args = (git_svn_log_cmd($r_min, $r_max), @args);
634 my $log = command_output_pipe(@args);
635 run_pager();
636 my (@k, $c, $d);
637
638 while (<$log>) {
639 if (/^${_esc_color}commit ($sha1_short)/o) {
640 my $cmt = $1;
641 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
642 $r_last = $c->{r};
643 process_commit($c, $r_min, $r_max, \@k) or
644 goto out;
645 }
646 $d = undef;
647 $c = { c => $cmt };
648 } elsif (/^${_esc_color}author (.+) (\d+) ([\-\+]?\d+)$/) {
649 get_author_info($c, $1, $2, $3);
650 } elsif (/^${_esc_color}(?:tree|parent|committer) /) {
651 # ignore
652 } elsif (/^${_esc_color}:\d{6} \d{6} $sha1_short/o) {
653 push @{$c->{raw}}, $_;
654 } elsif (/^${_esc_color}[ACRMDT]\t/) {
655 # we could add $SVN->{svn_path} here, but that requires
656 # remote access at the moment (repo_path_split)...
657 s#^(${_esc_color})([ACRMDT])\t#$1 $2 #;
658 push @{$c->{changed}}, $_;
659 } elsif (/^${_esc_color}diff /) {
660 $d = 1;
661 push @{$c->{diff}}, $_;
662 } elsif ($d) {
663 push @{$c->{diff}}, $_;
664 } elsif (/^${_esc_color} (git-svn-id:.+)$/) {
665 ($c->{url}, $c->{r}, undef) = extract_metadata($1);
666 } elsif (s/^${_esc_color} //) {
667 push @{$c->{l}}, $_;
668 }
669 }
670 if ($c && defined $c->{r} && $c->{r} != $r_last) {
671 $r_last = $c->{r};
672 process_commit($c, $r_min, $r_max, \@k);
673 }
674 if (@k) {
675 my $swap = $r_max;
676 $r_max = $r_min;
677 $r_min = $swap;
678 process_commit($_, $r_min, $r_max) foreach reverse @k;
679 }
680 out:
681 eval { command_close_pipe($log) };
682 print '-' x72,"\n" unless $_incremental || $_oneline;
683 }
684
685 sub commit_diff_usage {
686 print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
687 exit 1
688 }
689
690 sub commit_diff {
691 my $ta = shift or commit_diff_usage();
692 my $tb = shift or commit_diff_usage();
693 if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
694 print STDERR "Needed URL or usable git-svn id command-line\n";
695 commit_diff_usage();
696 }
697 my $r = shift;
698 unless (defined $r) {
699 if (defined $_revision) {
700 $r = $_revision
701 } else {
702 die "-r|--revision is a required argument\n";
703 }
704 }
705 if (defined $_message && defined $_file) {
706 print STDERR "Both --message/-m and --file/-F specified ",
707 "for the commit message.\n",
708 "I have no idea what you mean\n";
709 exit 1;
710 }
711 if (defined $_file) {
712 $_message = file_to_s($_file);
713 } else {
714 $_message ||= get_commit_message($tb,
715 "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
716 }
717 $SVN ||= libsvn_connect($SVN_URL);
718 if ($r eq 'HEAD') {
719 $r = $SVN->get_latest_revnum;
720 } elsif ($r !~ /^\d+$/) {
721 die "revision argument: $r not understood by git-svn\n";
722 }
723 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
724 my $rev_committed;
725 my $ed = SVN::Git::Editor->new({ r => $r,
726 ra => libsvn_dup_ra($SVN),
727 c => $tb,
728 svn_path => $SVN->{svn_path}
729 },
730 $SVN->get_commit_editor($_message,
731 sub {
732 $rev_committed = $_[0];
733 print "Committed $_[0]\n";
734 }, @lock)
735 );
736 eval {
737 my $mods = libsvn_checkout_tree($ta, $tb, $ed);
738 if (@$mods == 0) {
739 print "No changes\n$ta == $tb\n";
740 $ed->abort_edit;
741 } else {
742 $ed->close_edit;
743 }
744 };
745 fatal "$@\n" if $@;
746 $_message = $_file = undef;
747 return $rev_committed;
748 }
749
750 ########################### utility functions #########################
751
752 sub cmt_showable {
753 my ($c) = @_;
754 return 1 if defined $c->{r};
755 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
756 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
757 my @msg = command(qw/cat-file commit/, $c->{c});
758 shift @msg while ($msg[0] ne "\n");
759 shift @msg;
760 @{$c->{l}} = grep !/^git-svn-id: /, @msg;
761
762 (undef, $c->{r}, undef) = extract_metadata(
763 (grep(/^git-svn-id: /, @msg))[-1]);
764 }
765 return defined $c->{r};
766 }
767
768 sub log_use_color {
769 return 1 if $_color;
770 my ($dc, $dcvar);
771 $dcvar = 'color.diff';
772 $dc = `git-repo-config --get $dcvar`;
773 if ($dc eq '') {
774 # nothing at all; fallback to "diff.color"
775 $dcvar = 'diff.color';
776 $dc = `git-repo-config --get $dcvar`;
777 }
778 chomp($dc);
779 if ($dc eq 'auto') {
780 my $pc;
781 $pc = `git-repo-config --get color.pager`;
782 if ($pc eq '') {
783 # does not have it -- fallback to pager.color
784 $pc = `git-repo-config --bool --get pager.color`;
785 }
786 else {
787 $pc = `git-repo-config --bool --get color.pager`;
788 if ($?) {
789 $pc = 'false';
790 }
791 }
792 chomp($pc);
793 if (-t *STDOUT || (defined $_pager && $pc eq 'true')) {
794 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
795 }
796 return 0;
797 }
798 return 0 if $dc eq 'never';
799 return 1 if $dc eq 'always';
800 chomp($dc = `git-repo-config --bool --get $dcvar`);
801 return ($dc eq 'true');
802 }
803
804 sub git_svn_log_cmd {
805 my ($r_min, $r_max) = @_;
806 my @cmd = (qw/log --abbrev-commit --pretty=raw
807 --default/, "refs/remotes/$GIT_SVN");
808 push @cmd, '-r' unless $_non_recursive;
809 push @cmd, qw/--raw --name-status/ if $_verbose;
810 push @cmd, '--color' if log_use_color();
811 return @cmd unless defined $r_max;
812 if ($r_max == $r_min) {
813 push @cmd, '--max-count=1';
814 if (my $c = revdb_get($REVDB, $r_max)) {
815 push @cmd, $c;
816 }
817 } else {
818 my ($c_min, $c_max);
819 $c_max = revdb_get($REVDB, $r_max);
820 $c_min = revdb_get($REVDB, $r_min);
821 if (defined $c_min && defined $c_max) {
822 if ($r_max > $r_max) {
823 push @cmd, "$c_min..$c_max";
824 } else {
825 push @cmd, "$c_max..$c_min";
826 }
827 } elsif ($r_max > $r_min) {
828 push @cmd, $c_max;
829 } else {
830 push @cmd, $c_min;
831 }
832 }
833 return @cmd;
834 }
835
836 sub fetch_child_id {
837 my $id = shift;
838 print "Fetching $id\n";
839 my $ref = "$GIT_DIR/refs/remotes/$id";
840 defined(my $pid = open my $fh, '-|') or croak $!;
841 if (!$pid) {
842 $_repack = undef;
843 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
844 init_vars();
845 fetch(@_);
846 exit 0;
847 }
848 while (<$fh>) {
849 print $_;
850 check_repack() if (/^r\d+ = $sha1/);
851 }
852 close $fh or croak $?;
853 }
854
855 sub rec_fetch {
856 my ($pfx, $p, @args) = @_;
857 my @dir;
858 foreach (sort <$p/*>) {
859 if (-r "$_/info/url") {
860 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
861 my $id = $pfx . basename $_;
862 next if $id eq 'trunk';
863 fetch_child_id($id, @args);
864 } elsif (-d $_) {
865 push @dir, $_;
866 }
867 }
868 foreach (@dir) {
869 my $x = $_;
870 $x =~ s!^\Q$GIT_DIR\E/svn/!!;
871 rec_fetch($x, $_);
872 }
873 }
874
875 sub complete_url_ls_init {
876 my ($url, $var, $switch, $pfx) = @_;
877 unless ($var) {
878 print STDERR "W: $switch not specified\n";
879 return;
880 }
881 $var =~ s#/+$##;
882 if ($var !~ m#^[a-z\+]+://#) {
883 $var = '/' . $var if ($var !~ m#^/#);
884 unless ($url) {
885 print STDERR "E: '$var' is not a complete URL ",
886 "and a separate URL is not specified\n";
887 exit 1;
888 }
889 $var = $url . $var;
890 }
891 my @ls = libsvn_ls_fullurl($var);
892 my $old = $GIT_SVN;
893 defined(my $pid = fork) or croak $!;
894 if (!$pid) {
895 foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
896 $u =~ s#/+$##;
897 if ($u !~ m!\Q$var\E/(.+)$!) {
898 print STDERR "W: Unrecognized URL: $u\n";
899 die "This should never happen\n";
900 }
901 # don't try to init already existing refs
902 my $id = $pfx.$1;
903 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
904 init_vars();
905 unless (-d $GIT_SVN_DIR) {
906 print "init $u => $id\n";
907 init($u);
908 }
909 }
910 exit 0;
911 }
912 waitpid $pid, 0;
913 croak $? if $?;
914 my ($n) = ($switch =~ /^--(\w+)/);
915 command_noisy('repo-config', "svn.$n", $var);
916 }
917
918 sub common_prefix {
919 my $paths = shift;
920 my %common;
921 foreach (@$paths) {
922 my @tmp = split m#/#, $_;
923 my $p = '';
924 while (my $x = shift @tmp) {
925 $p .= "/$x";
926 $common{$p} ||= 0;
927 $common{$p}++;
928 }
929 }
930 foreach (sort {length $b <=> length $a} keys %common) {
931 if ($common{$_} == @$paths) {
932 return $_;
933 }
934 }
935 return '';
936 }
937
938 # grafts set here are 'stronger' in that they're based on actual tree
939 # matches, and won't be deleted from merge-base checking in write_grafts()
940 sub graft_tree_joins {
941 my $grafts = shift;
942 map_tree_joins() if (@_branch_from && !%tree_map);
943 return unless %tree_map;
944
945 git_svn_each(sub {
946 my $i = shift;
947 my @args = (qw/rev-list --pretty=raw/, "refs/remotes/$i");
948 my ($fh, $ctx) = command_output_pipe(@args);
949 while (<$fh>) {
950 next unless /^commit ($sha1)$/o;
951 my $c = $1;
952 my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
953 next unless $tree_map{$t};
954
955 my $l;
956 do {
957 $l = readline $fh;
958 } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
959
960 my ($s, $tz) = ($1, $2);
961 if ($tz =~ s/^\+//) {
962 $s += tz_to_s_offset($tz);
963 } elsif ($tz =~ s/^\-//) {
964 $s -= tz_to_s_offset($tz);
965 }
966
967 my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
968
969 foreach my $p (@{$tree_map{$t}}) {
970 next if $p eq $c;
971 my $mb = eval { command('merge-base', $c, $p) };
972 next unless ($@ || $?);
973 if (defined $r_a) {
974 # see if SVN says it's a relative
975 my ($url_b, $r_b, $uuid_b) =
976 cmt_metadata($p);
977 next if (defined $url_b &&
978 defined $url_a &&
979 ($url_a eq $url_b) &&
980 ($uuid_a eq $uuid_b));
981 if ($uuid_a eq $uuid_b) {
982 if ($r_b < $r_a) {
983 $grafts->{$c}->{$p} = 2;
984 next;
985 } elsif ($r_b > $r_a) {
986 $grafts->{$p}->{$c} = 2;
987 next;
988 }
989 }
990 }
991 my $ct = get_commit_time($p);
992 if ($ct < $s) {
993 $grafts->{$c}->{$p} = 2;
994 } elsif ($ct > $s) {
995 $grafts->{$p}->{$c} = 2;
996 }
997 # what should we do when $ct == $s ?
998 }
999 }
1000 command_close_pipe($fh, $ctx);
1001 });
1002 }
1003
1004 sub graft_file_copy_lib {
1005 my ($grafts, $l_map, $u) = @_;
1006 my $tree_paths = $l_map->{$u};
1007 my $pfx = common_prefix([keys %$tree_paths]);
1008 my ($repo, $path) = repo_path_split($u.$pfx);
1009 $SVN = libsvn_connect($repo);
1010
1011 my ($base, $head) = libsvn_parse_revision();
1012 my $inc = 1000;
1013 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1014 my $eh = $SVN::Error::handler;
1015 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1016 while (1) {
1017 my $pool = SVN::Pool->new;
1018 libsvn_get_log(libsvn_dup_ra($SVN), [$path],
1019 $min, $max, 0, 2, 1,
1020 sub {
1021 libsvn_graft_file_copies($grafts, $tree_paths,
1022 $path, @_);
1023 }, $pool);
1024 $pool->clear;
1025 last if ($max >= $head);
1026 $min = $max + 1;
1027 $max += $inc;
1028 $max = $head if ($max > $head);
1029 }
1030 $SVN::Error::handler = $eh;
1031 }
1032
1033 sub process_merge_msg_matches {
1034 my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1035 my (@strong, @weak);
1036 foreach (@matches) {
1037 # merging with ourselves is not interesting
1038 next if $_ eq $p;
1039 if ($l_map->{$u}->{$_}) {
1040 push @strong, $_;
1041 } else {
1042 push @weak, $_;
1043 }
1044 }
1045 foreach my $w (@weak) {
1046 last if @strong;
1047 # no exact match, use branch name as regexp.
1048 my $re = qr/\Q$w\E/i;
1049 foreach (keys %{$l_map->{$u}}) {
1050 if (/$re/) {
1051 push @strong, $l_map->{$u}->{$_};
1052 last;
1053 }
1054 }
1055 last if @strong;
1056 $w = basename($w);
1057 $re = qr/\Q$w\E/i;
1058 foreach (keys %{$l_map->{$u}}) {
1059 if (/$re/) {
1060 push @strong, $l_map->{$u}->{$_};
1061 last;
1062 }
1063 }
1064 }
1065 my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1066 \s(?:[a-f\d\-]+)$/xsm);
1067 unless (defined $rev) {
1068 ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1069 \@(?:[a-f\d\-]+)/xsm);
1070 return unless defined $rev;
1071 }
1072 foreach my $m (@strong) {
1073 my ($r0, $s0) = find_rev_before($rev, $m, 1);
1074 $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1075 }
1076 }
1077
1078 sub graft_merge_msg {
1079 my ($grafts, $l_map, $u, $p, @re) = @_;
1080
1081 my $x = $l_map->{$u}->{$p};
1082 my $rl = rev_list_raw($x);
1083 while (my $c = next_rev_list_entry($rl)) {
1084 foreach my $re (@re) {
1085 my (@br) = ($c->{m} =~ /$re/g);
1086 next unless @br;
1087 process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1088 }
1089 }
1090 }
1091
1092 sub read_uuid {
1093 return if $SVN_UUID;
1094 my $pool = SVN::Pool->new;
1095 $SVN_UUID = $SVN->get_uuid($pool);
1096 $pool->clear;
1097 }
1098
1099 sub verify_ref {
1100 my ($ref) = @_;
1101 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1102 { STDERR => 0 }); };
1103 }
1104
1105 sub repo_path_split {
1106 my $full_url = shift;
1107 $full_url =~ s#/+$##;
1108
1109 foreach (@repo_path_split_cache) {
1110 if ($full_url =~ s#$_##) {
1111 my $u = $1;
1112 $full_url =~ s#^/+##;
1113 return ($u, $full_url);
1114 }
1115 }
1116 my $tmp = libsvn_connect($full_url);
1117 return ($tmp->{repos_root}, $tmp->{svn_path});
1118 }
1119
1120 sub setup_git_svn {
1121 defined $SVN_URL or croak "SVN repository location required\n";
1122 unless (-d $GIT_DIR) {
1123 croak "GIT_DIR=$GIT_DIR does not exist!\n";
1124 }
1125 mkpath([$GIT_SVN_DIR]);
1126 mkpath(["$GIT_SVN_DIR/info"]);
1127 open my $fh, '>>',$REVDB or croak $!;
1128 close $fh;
1129 s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1130
1131 }
1132
1133 sub get_tree_from_treeish {
1134 my ($treeish) = @_;
1135 croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1136 my $type = command_oneline(qw/cat-file -t/, $treeish);
1137 my $expected;
1138 while ($type eq 'tag') {
1139 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1140 }
1141 if ($type eq 'commit') {
1142 $expected = (grep /^tree /, command(qw/cat-file commit/,
1143 $treeish))[0];
1144 ($expected) = ($expected =~ /^tree ($sha1)$/);
1145 die "Unable to get tree from $treeish\n" unless $expected;
1146 } elsif ($type eq 'tree') {
1147 $expected = $treeish;
1148 } else {
1149 die "$treeish is a $type, expected tree, tag or commit\n";
1150 }
1151 return $expected;
1152 }
1153
1154 sub get_diff {
1155 my ($from, $treeish) = @_;
1156 print "diff-tree $from $treeish\n";
1157 my @diff_tree = qw(diff-tree -z -r);
1158 if ($_cp_similarity) {
1159 push @diff_tree, "-C$_cp_similarity";
1160 } else {
1161 push @diff_tree, '-C';
1162 }
1163 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1164 push @diff_tree, "-l$_l" if defined $_l;
1165 push @diff_tree, $from, $treeish;
1166 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1167 local $/ = "\0";
1168 my $state = 'meta';
1169 my @mods;
1170 while (<$diff_fh>) {
1171 chomp $_; # this gets rid of the trailing "\0"
1172 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1173 $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1174 push @mods, { mode_a => $1, mode_b => $2,
1175 sha1_b => $3, chg => $4 };
1176 if ($4 =~ /^(?:C|R)$/) {
1177 $state = 'file_a';
1178 } else {
1179 $state = 'file_b';
1180 }
1181 } elsif ($state eq 'file_a') {
1182 my $x = $mods[$#mods] or croak "Empty array\n";
1183 if ($x->{chg} !~ /^(?:C|R)$/) {
1184 croak "Error parsing $_, $x->{chg}\n";
1185 }
1186 $x->{file_a} = $_;
1187 $state = 'file_b';
1188 } elsif ($state eq 'file_b') {
1189 my $x = $mods[$#mods] or croak "Empty array\n";
1190 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1191 croak "Error parsing $_, $x->{chg}\n";
1192 }
1193 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1194 croak "Error parsing $_, $x->{chg}\n";
1195 }
1196 $x->{file_b} = $_;
1197 $state = 'meta';
1198 } else {
1199 croak "Error parsing $_\n";
1200 }
1201 }
1202 command_close_pipe($diff_fh, $ctx);
1203 return \@mods;
1204 }
1205
1206 sub libsvn_checkout_tree {
1207 my ($from, $treeish, $ed) = @_;
1208 my $mods = get_diff($from, $treeish);
1209 return $mods unless (scalar @$mods);
1210 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1211 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1212 my $f = $m->{chg};
1213 if (defined $o{$f}) {
1214 $ed->$f($m, $_q);
1215 } else {
1216 croak "Invalid change type: $f\n";
1217 }
1218 }
1219 $ed->rmdirs($_q) if $_rmdir;
1220 return $mods;
1221 }
1222
1223 sub get_commit_message {
1224 my ($commit, $commit_msg) = (@_);
1225 my %log_msg = ( msg => '' );
1226 open my $msg, '>', $commit_msg or croak $!;
1227
1228 my $type = command_oneline(qw/cat-file -t/, $commit);
1229 if ($type eq 'commit' || $type eq 'tag') {
1230 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1231 $type, $commit);
1232 my $in_msg = 0;
1233 while (<$msg_fh>) {
1234 if (!$in_msg) {
1235 $in_msg = 1 if (/^\s*$/);
1236 } elsif (/^git-svn-id: /) {
1237 # skip this, we regenerate the correct one
1238 # on re-fetch anyways
1239 } else {
1240 print $msg $_ or croak $!;
1241 }
1242 }
1243 command_close_pipe($msg_fh, $ctx);
1244 }
1245 close $msg or croak $!;
1246
1247 if ($_edit || ($type eq 'tree')) {
1248 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1249 system($editor, $commit_msg);
1250 }
1251
1252 # file_to_s removes all trailing newlines, so just use chomp() here:
1253 open $msg, '<', $commit_msg or croak $!;
1254 { local $/; chomp($log_msg{msg} = <$msg>); }
1255 close $msg or croak $!;
1256
1257 return \%log_msg;
1258 }
1259
1260 sub set_svn_commit_env {
1261 if (defined $LC_ALL) {
1262 $ENV{LC_ALL} = $LC_ALL;
1263 } else {
1264 delete $ENV{LC_ALL};
1265 }
1266 }
1267
1268 sub rev_list_raw {
1269 my ($fh, $c) = command_output_pipe(qw/rev-list --pretty=raw/, @_);
1270 return { fh => $fh, ctx => $c, t => { } };
1271 }
1272
1273 sub next_rev_list_entry {
1274 my $rl = shift;
1275 my $fh = $rl->{fh};
1276 my $x = $rl->{t};
1277 while (<$fh>) {
1278 if (/^commit ($sha1)$/o) {
1279 if ($x->{c}) {
1280 $rl->{t} = { c => $1 };
1281 return $x;
1282 } else {
1283 $x->{c} = $1;
1284 }
1285 } elsif (/^parent ($sha1)$/o) {
1286 $x->{p}->{$1} = 1;
1287 } elsif (s/^ //) {
1288 $x->{m} ||= '';
1289 $x->{m} .= $_;
1290 }
1291 }
1292 command_close_pipe($fh, $rl->{ctx});
1293 return ($x != $rl->{t}) ? $x : undef;
1294 }
1295
1296 sub s_to_file {
1297 my ($str, $file, $mode) = @_;
1298 open my $fd,'>',$file or croak $!;
1299 print $fd $str,"\n" or croak $!;
1300 close $fd or croak $!;
1301 chmod ($mode &~ umask, $file) if (defined $mode);
1302 }
1303
1304 sub file_to_s {
1305 my $file = shift;
1306 open my $fd,'<',$file or croak "$!: file: $file\n";
1307 local $/;
1308 my $ret = <$fd>;
1309 close $fd or croak $!;
1310 $ret =~ s/\s*$//s;
1311 return $ret;
1312 }
1313
1314 sub assert_revision_unknown {
1315 my $r = shift;
1316 if (my $c = revdb_get($REVDB, $r)) {
1317 croak "$r = $c already exists! Why are we refetching it?";
1318 }
1319 }
1320
1321 sub git_commit {
1322 my ($log_msg, @parents) = @_;
1323 assert_revision_unknown($log_msg->{revision});
1324 map_tree_joins() if (@_branch_from && !%tree_map);
1325
1326 my (@tmp_parents, @exec_parents, %seen_parent);
1327 if (my $lparents = $log_msg->{parents}) {
1328 @tmp_parents = @$lparents
1329 }
1330 # commit parents can be conditionally bound to a particular
1331 # svn revision via: "svn_revno=commit_sha1", filter them out here:
1332 foreach my $p (@parents) {
1333 next unless defined $p;
1334 if ($p =~ /^(\d+)=($sha1_short)$/o) {
1335 if ($1 == $log_msg->{revision}) {
1336 push @tmp_parents, $2;
1337 }
1338 } else {
1339 push @tmp_parents, $p if $p =~ /$sha1_short/o;
1340 }
1341 }
1342 my $tree = $log_msg->{tree};
1343 if (!defined $tree) {
1344 my $index = set_index($GIT_SVN_INDEX);
1345 $tree = command_oneline('write-tree');
1346 croak $? if $?;
1347 restore_index($index);
1348 }
1349 # just in case we clobber the existing ref, we still want that ref
1350 # as our parent:
1351 if (my $cur = verify_ref("refs/remotes/$GIT_SVN^0")) {
1352 chomp $cur;
1353 push @tmp_parents, $cur;
1354 }
1355
1356 if (exists $tree_map{$tree}) {
1357 foreach my $p (@{$tree_map{$tree}}) {
1358 my $skip;
1359 foreach (@tmp_parents) {
1360 # see if a common parent is found
1361 my $mb = eval { command('merge-base', $_, $p) };
1362 next if ($@ || $?);
1363 $skip = 1;
1364 last;
1365 }
1366 next if $skip;
1367 my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1368 next if (($SVN_UUID eq $uuid_p) &&
1369 ($log_msg->{revision} > $r_p));
1370 next if (defined $url_p && defined $SVN_URL &&
1371 ($SVN_UUID eq $uuid_p) &&
1372 ($url_p eq $SVN_URL));
1373 push @tmp_parents, $p;
1374 }
1375 }
1376 foreach (@tmp_parents) {
1377 next if $seen_parent{$_};
1378 $seen_parent{$_} = 1;
1379 push @exec_parents, $_;
1380 # MAXPARENT is defined to 16 in commit-tree.c:
1381 last if @exec_parents > 16;
1382 }
1383
1384 set_commit_env($log_msg);
1385 my @exec = ('git-commit-tree', $tree);
1386 push @exec, '-p', $_ foreach @exec_parents;
1387 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1388 or croak $!;
1389 print $msg_fh $log_msg->{msg} or croak $!;
1390 unless ($_no_metadata) {
1391 print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1392 " $SVN_UUID\n" or croak $!;
1393 }
1394 $msg_fh->flush == 0 or croak $!;
1395 close $msg_fh or croak $!;
1396 chomp(my $commit = do { local $/; <$out_fh> });
1397 close $out_fh or croak $!;
1398 waitpid $pid, 0;
1399 croak $? if $?;
1400 if ($commit !~ /^$sha1$/o) {
1401 die "Failed to commit, invalid sha1: $commit\n";
1402 }
1403 command_noisy('update-ref',"refs/remotes/$GIT_SVN",$commit);
1404 revdb_set($REVDB, $log_msg->{revision}, $commit);
1405
1406 # this output is read via pipe, do not change:
1407 print "r$log_msg->{revision} = $commit\n";
1408 check_repack();
1409 return $commit;
1410 }
1411
1412 sub check_repack {
1413 if ($_repack && (--$_repack_nr == 0)) {
1414 $_repack_nr = $_repack;
1415 # repack doesn't use any arguments with spaces in them, does it?
1416 command_noisy('repack', split(/\s+/, $_repack_flags));
1417 }
1418 }
1419
1420 sub set_commit_env {
1421 my ($log_msg) = @_;
1422 my $author = $log_msg->{author};
1423 if (!defined $author || length $author == 0) {
1424 $author = '(no author)';
1425 }
1426 my ($name,$email) = defined $users{$author} ? @{$users{$author}}
1427 : ($author,"$author\@$SVN_UUID");
1428 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1429 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1430 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1431 }
1432
1433 sub check_upgrade_needed {
1434 if (!-r $REVDB) {
1435 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1436 open my $fh, '>>',$REVDB or croak $!;
1437 close $fh;
1438 }
1439 return unless eval {
1440 command([qw/rev-parse --verify/,"$GIT_SVN-HEAD^0"],
1441 {STDERR => 0});
1442 };
1443 my $head = eval { command('rev-parse',"refs/remotes/$GIT_SVN") };
1444 if ($@ || !$head) {
1445 print STDERR "Please run: $0 rebuild --upgrade\n";
1446 exit 1;
1447 }
1448 }
1449
1450 # fills %tree_map with a reverse mapping of trees to commits. Useful
1451 # for finding parents to commit on.
1452 sub map_tree_joins {
1453 my %seen;
1454 foreach my $br (@_branch_from) {
1455 my $pipe = command_output_pipe(qw/rev-list
1456 --topo-order --pretty=raw/, $br);
1457 while (<$pipe>) {
1458 if (/^commit ($sha1)$/o) {
1459 my $commit = $1;
1460
1461 # if we've seen a commit,
1462 # we've seen its parents
1463 last if $seen{$commit};
1464 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1465 unless (defined $tree) {
1466 die "Failed to parse commit $commit\n";
1467 }
1468 push @{$tree_map{$tree}}, $commit;
1469 $seen{$commit} = 1;
1470 }
1471 }
1472 eval { command_close_pipe($pipe) };
1473 }
1474 }
1475
1476 sub load_all_refs {
1477 if (@_branch_from) {
1478 print STDERR '--branch|-b parameters are ignored when ',
1479 "--branch-all-refs|-B is passed\n";
1480 }
1481
1482 # don't worry about rev-list on non-commit objects/tags,
1483 # it shouldn't blow up if a ref is a blob or tree...
1484 @_branch_from = command(qw/rev-parse --symbolic --all/);
1485 }
1486
1487 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1488 sub load_authors {
1489 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1490 while (<$authors>) {
1491 chomp;
1492 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1493 my ($user, $name, $email) = ($1, $2, $3);
1494 $users{$user} = [$name, $email];
1495 }
1496 close $authors or croak $!;
1497 }
1498
1499 sub rload_authors {
1500 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1501 while (<$authors>) {
1502 chomp;
1503 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1504 my ($user, $name, $email) = ($1, $2, $3);
1505 $rusers{"$name <$email>"} = $user;
1506 }
1507 close $authors or croak $!;
1508 }
1509
1510 sub git_svn_each {
1511 my $sub = shift;
1512 foreach (command(qw/rev-parse --symbolic --all/)) {
1513 next unless s#^refs/remotes/##;
1514 chomp $_;
1515 next unless -f "$GIT_DIR/svn/$_/info/url";
1516 &$sub($_);
1517 }
1518 }
1519
1520 sub migrate_revdb {
1521 git_svn_each(sub {
1522 my $id = shift;
1523 defined(my $pid = fork) or croak $!;
1524 if (!$pid) {
1525 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1526 init_vars();
1527 exit 0 if -r $REVDB;
1528 print "Upgrading svn => git mapping...\n";
1529 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1530 open my $fh, '>>',$REVDB or croak $!;
1531 close $fh;
1532 rebuild();
1533 print "Done upgrading. You may now delete the ",
1534 "deprecated $GIT_SVN_DIR/revs directory\n";
1535 exit 0;
1536 }
1537 waitpid $pid, 0;
1538 croak $? if $?;
1539 });
1540 }
1541
1542 sub migration_check {
1543 migrate_revdb() unless (-e $REVDB);
1544 return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
1545 print "Upgrading repository...\n";
1546 unless (-d "$GIT_DIR/svn") {
1547 mkdir "$GIT_DIR/svn" or croak $!;
1548 }
1549 print "Data from a previous version of git-svn exists, but\n\t",
1550 "$GIT_SVN_DIR\n\t(required for this version ",
1551 "($VERSION) of git-svn) does not.\n";
1552
1553 foreach my $x (command(qw/rev-parse --symbolic --all/)) {
1554 next unless $x =~ s#^refs/remotes/##;
1555 chomp $x;
1556 next unless -f "$GIT_DIR/$x/info/url";
1557 my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
1558 next unless $u;
1559 my $dn = dirname("$GIT_DIR/svn/$x");
1560 mkpath([$dn]) unless -d $dn;
1561 rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
1562 }
1563 migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
1564 print "Done upgrading.\n";
1565 }
1566
1567 sub find_rev_before {
1568 my ($r, $id, $eq_ok) = @_;
1569 my $f = "$GIT_DIR/svn/$id/.rev_db";
1570 return (undef,undef) unless -r $f;
1571 --$r unless $eq_ok;
1572 while ($r > 0) {
1573 if (my $c = revdb_get($f, $r)) {
1574 return ($r, $c);
1575 }
1576 --$r;
1577 }
1578 return (undef, undef);
1579 }
1580
1581 sub init_vars {
1582 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
1583 $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
1584 $REVDB = "$GIT_SVN_DIR/.rev_db";
1585 $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
1586 $SVN_URL = undef;
1587 $SVN_WC = "$GIT_SVN_DIR/tree";
1588 %tree_map = ();
1589 }
1590
1591 # convert GetOpt::Long specs for use by git-repo-config
1592 sub read_repo_config {
1593 return unless -d $GIT_DIR;
1594 my $opts = shift;
1595 foreach my $o (keys %$opts) {
1596 my $v = $opts->{$o};
1597 my ($key) = ($o =~ /^([a-z\-]+)/);
1598 $key =~ s/-//g;
1599 my $arg = 'git-repo-config';
1600 $arg .= ' --int' if ($o =~ /[:=]i$/);
1601 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1602 if (ref $v eq 'ARRAY') {
1603 chomp(my @tmp = `$arg --get-all svn.$key`);
1604 @$v = @tmp if @tmp;
1605 } else {
1606 chomp(my $tmp = `$arg --get svn.$key`);
1607 if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
1608 $$v = $tmp;
1609 }
1610 }
1611 }
1612 }
1613
1614 sub set_default_vals {
1615 if (defined $_repack) {
1616 $_repack = 1000 if ($_repack <= 0);
1617 $_repack_nr = $_repack;
1618 $_repack_flags ||= '-d';
1619 }
1620 }
1621
1622 sub read_grafts {
1623 my $gr_file = shift;
1624 my ($grafts, $comments) = ({}, {});
1625 if (open my $fh, '<', $gr_file) {
1626 my @tmp;
1627 while (<$fh>) {
1628 if (/^($sha1)\s+/) {
1629 my $c = $1;
1630 if (@tmp) {
1631 @{$comments->{$c}} = @tmp;
1632 @tmp = ();
1633 }
1634 foreach my $p (split /\s+/, $_) {
1635 $grafts->{$c}->{$p} = 1;
1636 }
1637 } else {
1638 push @tmp, $_;
1639 }
1640 }
1641 close $fh or croak $!;
1642 @{$comments->{'END'}} = @tmp if @tmp;
1643 }
1644 return ($grafts, $comments);
1645 }
1646
1647 sub write_grafts {
1648 my ($grafts, $comments, $gr_file) = @_;
1649
1650 open my $fh, '>', $gr_file or croak $!;
1651 foreach my $c (sort keys %$grafts) {
1652 if ($comments->{$c}) {
1653 print $fh $_ foreach @{$comments->{$c}};
1654 }
1655 my $p = $grafts->{$c};
1656 my %x; # real parents
1657 delete $p->{$c}; # commits are not self-reproducing...
1658 my $ch = command_output_pipe(qw/cat-file commit/, $c);
1659 while (<$ch>) {
1660 if (/^parent ($sha1)/) {
1661 $x{$1} = $p->{$1} = 1;
1662 } else {
1663 last unless /^\S/;
1664 }
1665 }
1666 eval { command_close_pipe($ch) }; # breaking the pipe
1667
1668 # if real parents are the only ones in the grafts, drop it
1669 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1670
1671 my (@ip, @jp, $mb);
1672 my %del = %x;
1673 @ip = @jp = keys %$p;
1674 foreach my $i (@ip) {
1675 next if $del{$i} || $p->{$i} == 2;
1676 foreach my $j (@jp) {
1677 next if $i eq $j || $del{$j} || $p->{$j} == 2;
1678 $mb = eval { command('merge-base', $i, $j) };
1679 next unless $mb;
1680 chomp $mb;
1681 next if $x{$mb};
1682 if ($mb eq $j) {
1683 delete $p->{$i};
1684 $del{$i} = 1;
1685 } elsif ($mb eq $i) {
1686 delete $p->{$j};
1687 $del{$j} = 1;
1688 }
1689 }
1690 }
1691
1692 # if real parents are the only ones in the grafts, drop it
1693 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1694
1695 print $fh $c, ' ', join(' ', sort keys %$p),"\n";
1696 }
1697 if ($comments->{'END'}) {
1698 print $fh $_ foreach @{$comments->{'END'}};
1699 }
1700 close $fh or croak $!;
1701 }
1702
1703 sub read_url_paths_all {
1704 my ($l_map, $pfx, $p) = @_;
1705 my @dir;
1706 foreach (<$p/*>) {
1707 if (-r "$_/info/url") {
1708 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1709 my $id = $pfx . basename $_;
1710 my $url = file_to_s("$_/info/url");
1711 my ($u, $p) = repo_path_split($url);
1712 $l_map->{$u}->{$p} = $id;
1713 } elsif (-d $_) {
1714 push @dir, $_;
1715 }
1716 }
1717 foreach (@dir) {
1718 my $x = $_;
1719 $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
1720 read_url_paths_all($l_map, $x, $_);
1721 }
1722 }
1723
1724 # this one only gets ids that have been imported, not new ones
1725 sub read_url_paths {
1726 my $l_map = {};
1727 git_svn_each(sub { my $x = shift;
1728 my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
1729 my ($u, $p) = repo_path_split($url);
1730 $l_map->{$u}->{$p} = $x;
1731 });
1732 return $l_map;
1733 }
1734
1735 sub extract_metadata {
1736 my $id = shift or return (undef, undef, undef);
1737 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
1738 \s([a-f\d\-]+)$/x);
1739 if (!defined $rev || !$uuid || !$url) {
1740 # some of the original repositories I made had
1741 # identifiers like this:
1742 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1743 }
1744 return ($url, $rev, $uuid);
1745 }
1746
1747 sub cmt_metadata {
1748 return extract_metadata((grep(/^git-svn-id: /,
1749 command(qw/cat-file commit/, shift)))[-1]);
1750 }
1751
1752 sub get_commit_time {
1753 my $cmt = shift;
1754 my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
1755 while (<$fh>) {
1756 /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
1757 my ($s, $tz) = ($1, $2);
1758 if ($tz =~ s/^\+//) {
1759 $s += tz_to_s_offset($tz);
1760 } elsif ($tz =~ s/^\-//) {
1761 $s -= tz_to_s_offset($tz);
1762 }
1763 eval { command_close_pipe($fh) };
1764 return $s;
1765 }
1766 die "Can't get commit time for commit: $cmt\n";
1767 }
1768
1769 sub tz_to_s_offset {
1770 my ($tz) = @_;
1771 $tz =~ s/(\d\d)$//;
1772 return ($1 * 60) + ($tz * 3600);
1773 }
1774
1775 # adapted from pager.c
1776 sub config_pager {
1777 $_pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
1778 if (!defined $_pager) {
1779 $_pager = 'less';
1780 } elsif (length $_pager == 0 || $_pager eq 'cat') {
1781 $_pager = undef;
1782 }
1783 }
1784
1785 sub run_pager {
1786 return unless -t *STDOUT;
1787 pipe my $rfd, my $wfd or return;
1788 defined(my $pid = fork) or croak $!;
1789 if (!$pid) {
1790 open STDOUT, '>&', $wfd or croak $!;
1791 return;
1792 }
1793 open STDIN, '<&', $rfd or croak $!;
1794 $ENV{LESS} ||= 'FRSX';
1795 exec $_pager or croak "Can't run pager: $! ($_pager)\n";
1796 }
1797
1798 sub get_author_info {
1799 my ($dest, $author, $t, $tz) = @_;
1800 $author =~ s/(?:^\s*|\s*$)//g;
1801 $dest->{a_raw} = $author;
1802 my $_a;
1803 if ($_authors) {
1804 $_a = $rusers{$author} || undef;
1805 }
1806 if (!$_a) {
1807 ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
1808 }
1809 $dest->{t} = $t;
1810 $dest->{tz} = $tz;
1811 $dest->{a} = $_a;
1812 # Date::Parse isn't in the standard Perl distro :(
1813 if ($tz =~ s/^\+//) {
1814 $t += tz_to_s_offset($tz);
1815 } elsif ($tz =~ s/^\-//) {
1816 $t -= tz_to_s_offset($tz);
1817 }
1818 $dest->{t_utc} = $t;
1819 }
1820
1821 sub process_commit {
1822 my ($c, $r_min, $r_max, $defer) = @_;
1823 if (defined $r_min && defined $r_max) {
1824 if ($r_min == $c->{r} && $r_min == $r_max) {
1825 show_commit($c);
1826 return 0;
1827 }
1828 return 1 if $r_min == $r_max;
1829 if ($r_min < $r_max) {
1830 # we need to reverse the print order
1831 return 0 if (defined $_limit && --$_limit < 0);
1832 push @$defer, $c;
1833 return 1;
1834 }
1835 if ($r_min != $r_max) {
1836 return 1 if ($r_min < $c->{r});
1837 return 1 if ($r_max > $c->{r});
1838 }
1839 }
1840 return 0 if (defined $_limit && --$_limit < 0);
1841 show_commit($c);
1842 return 1;
1843 }
1844
1845 sub show_commit {
1846 my $c = shift;
1847 if ($_oneline) {
1848 my $x = "\n";
1849 if (my $l = $c->{l}) {
1850 while ($l->[0] =~ /^\s*$/) { shift @$l }
1851 $x = $l->[0];
1852 }
1853 $_l_fmt ||= 'A' . length($c->{r});
1854 print 'r',pack($_l_fmt, $c->{r}),' | ';
1855 print "$c->{c} | " if $_show_commit;
1856 print $x;
1857 } else {
1858 show_commit_normal($c);
1859 }
1860 }
1861
1862 sub show_commit_changed_paths {
1863 my ($c) = @_;
1864 return unless $c->{changed};
1865 print "Changed paths:\n", @{$c->{changed}};
1866 }
1867
1868 sub show_commit_normal {
1869 my ($c) = @_;
1870 print '-' x72, "\nr$c->{r} | ";
1871 print "$c->{c} | " if $_show_commit;
1872 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
1873 localtime($c->{t_utc})), ' | ';
1874 my $nr_line = 0;
1875
1876 if (my $l = $c->{l}) {
1877 while ($l->[$#$l] eq "\n" && $#$l > 0
1878 && $l->[($#$l - 1)] eq "\n") {
1879 pop @$l;
1880 }
1881 $nr_line = scalar @$l;
1882 if (!$nr_line) {
1883 print "1 line\n\n\n";
1884 } else {
1885 if ($nr_line == 1) {
1886 $nr_line = '1 line';
1887 } else {
1888 $nr_line .= ' lines';
1889 }
1890 print $nr_line, "\n";
1891 show_commit_changed_paths($c);
1892 print "\n";
1893 print $_ foreach @$l;
1894 }
1895 } else {
1896 print "1 line\n";
1897 show_commit_changed_paths($c);
1898 print "\n";
1899
1900 }
1901 foreach my $x (qw/raw diff/) {
1902 if ($c->{$x}) {
1903 print "\n";
1904 print $_ foreach @{$c->{$x}}
1905 }
1906 }
1907 }
1908
1909 sub _simple_prompt {
1910 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1911 $may_save = undef if $_no_auth_cache;
1912 $default_username = $_username if defined $_username;
1913 if (defined $default_username && length $default_username) {
1914 if (defined $realm && length $realm) {
1915 print "Authentication realm: $realm\n";
1916 }
1917 $cred->username($default_username);
1918 } else {
1919 _username_prompt($cred, $realm, $may_save, $pool);
1920 }
1921 $cred->password(_read_password("Password for '" .
1922 $cred->username . "': ", $realm));
1923 $cred->may_save($may_save);
1924 $SVN::_Core::SVN_NO_ERROR;
1925 }
1926
1927 sub _ssl_server_trust_prompt {
1928 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1929 $may_save = undef if $_no_auth_cache;
1930 print "Error validating server certificate for '$realm':\n";
1931 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1932 print " - The certificate is not issued by a trusted ",
1933 "authority. Use the\n",
1934 " fingerprint to validate the certificate manually!\n";
1935 }
1936 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1937 print " - The certificate hostname does not match.\n";
1938 }
1939 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1940 print " - The certificate is not yet valid.\n";
1941 }
1942 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1943 print " - The certificate has expired.\n";
1944 }
1945 if ($failures & $SVN::Auth::SSL::OTHER) {
1946 print " - The certificate has an unknown error.\n";
1947 }
1948 printf( "Certificate information:\n".
1949 " - Hostname: %s\n".
1950 " - Valid: from %s until %s\n".
1951 " - Issuer: %s\n".
1952 " - Fingerprint: %s\n",
1953 map $cert_info->$_, qw(hostname valid_from valid_until
1954 issuer_dname fingerprint) );
1955 my $choice;
1956 prompt:
1957 print $may_save ?
1958 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1959 "(R)eject or accept (t)emporarily? ";
1960 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1961 if ($choice =~ /^t$/i) {
1962 $cred->may_save(undef);
1963 } elsif ($choice =~ /^r$/i) {
1964 return -1;
1965 } elsif ($may_save && $choice =~ /^p$/i) {
1966 $cred->may_save($may_save);
1967 } else {
1968 goto prompt;
1969 }
1970 $cred->accepted_failures($failures);
1971 $SVN::_Core::SVN_NO_ERROR;
1972 }
1973
1974 sub _ssl_client_cert_prompt {
1975 my ($cred, $realm, $may_save, $pool) = @_;
1976 $may_save = undef if $_no_auth_cache;
1977 print "Client certificate filename: ";
1978 chomp(my $filename = <STDIN>);
1979 $cred->cert_file($filename);
1980 $cred->may_save($may_save);
1981 $SVN::_Core::SVN_NO_ERROR;
1982 }
1983
1984 sub _ssl_client_cert_pw_prompt {
1985 my ($cred, $realm, $may_save, $pool) = @_;
1986 $may_save = undef if $_no_auth_cache;
1987 $cred->password(_read_password("Password: ", $realm));
1988 $cred->may_save($may_save);
1989 $SVN::_Core::SVN_NO_ERROR;
1990 }
1991
1992 sub _username_prompt {
1993 my ($cred, $realm, $may_save, $pool) = @_;
1994 $may_save = undef if $_no_auth_cache;
1995 if (defined $realm && length $realm) {
1996 print "Authentication realm: $realm\n";
1997 }
1998 my $username;
1999 if (defined $_username) {
2000 $username = $_username;
2001 } else {
2002 print "Username: ";
2003 chomp($username = <STDIN>);
2004 }
2005 $cred->username($username);
2006 $cred->may_save($may_save);
2007 $SVN::_Core::SVN_NO_ERROR;
2008 }
2009
2010 sub _read_password {
2011 my ($prompt, $realm) = @_;
2012 print $prompt;
2013 require Term::ReadKey;
2014 Term::ReadKey::ReadMode('noecho');
2015 my $password = '';
2016 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2017 last if $key =~ /[\012\015]/; # \n\r
2018 $password .= $key;
2019 }
2020 Term::ReadKey::ReadMode('restore');
2021 print "\n";
2022 $password;
2023 }
2024
2025 sub libsvn_connect {
2026 my ($url) = @_;
2027 SVN::_Core::svn_config_ensure($_config_dir, undef);
2028 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2029 SVN::Client::get_simple_provider(),
2030 SVN::Client::get_ssl_server_trust_file_provider(),
2031 SVN::Client::get_simple_prompt_provider(
2032 \&_simple_prompt, 2),
2033 SVN::Client::get_ssl_client_cert_prompt_provider(
2034 \&_ssl_client_cert_prompt, 2),
2035 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2036 \&_ssl_client_cert_pw_prompt, 2),
2037 SVN::Client::get_username_provider(),
2038 SVN::Client::get_ssl_server_trust_prompt_provider(
2039 \&_ssl_server_trust_prompt),
2040 SVN::Client::get_username_prompt_provider(
2041 \&_username_prompt, 2),
2042 ]);
2043 my $config = SVN::Core::config_get_config($_config_dir);
2044 my $ra = SVN::Ra->new(url => $url, auth => $baton,
2045 config => $config,
2046 pool => SVN::Pool->new,
2047 auth_provider_callbacks => $callbacks);
2048 $ra->{svn_path} = $url;
2049 $ra->{repos_root} = $ra->get_repos_root;
2050 $ra->{svn_path} =~ s#^\Q$ra->{repos_root}\E/*##;
2051 push @repo_path_split_cache, qr/^(\Q$ra->{repos_root}\E)/;
2052 return $ra;
2053 }
2054
2055 sub libsvn_can_do_switch {
2056 unless (defined $_svn_can_do_switch) {
2057 my $pool = SVN::Pool->new;
2058 my $rep = eval {
2059 $SVN->do_switch(1, '', 0, $SVN->{url},
2060 SVN::Delta::Editor->new, $pool);
2061 };
2062 if ($@) {
2063 $_svn_can_do_switch = 0;
2064 } else {
2065 $rep->abort_report($pool);
2066 $_svn_can_do_switch = 1;
2067 }
2068 $pool->clear;
2069 }
2070 $_svn_can_do_switch;
2071 }
2072
2073 sub libsvn_dup_ra {
2074 my ($ra) = @_;
2075 SVN::Ra->new(map { $_ => $ra->{$_} } qw/config url
2076 auth auth_provider_callbacks repos_root svn_path/);
2077 }
2078
2079 sub uri_encode {
2080 my ($f) = @_;
2081 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2082 $f
2083 }
2084
2085 sub uri_decode {
2086 my ($f) = @_;
2087 $f =~ tr/+/ /;
2088 $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
2089 $f
2090 }
2091
2092 sub libsvn_log_entry {
2093 my ($rev, $author, $date, $msg, $parents, $untracked) = @_;
2094 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2095 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2096 or die "Unable to parse date: $date\n";
2097 if (defined $author && length $author > 0 &&
2098 defined $_authors && ! defined $users{$author}) {
2099 die "Author: $author not defined in $_authors file\n";
2100 }
2101 $msg = '' if ($rev == 0 && !defined $msg);
2102
2103 open my $un, '>>', "$GIT_SVN_DIR/unhandled.log" or croak $!;
2104 my $h;
2105 print $un "r$rev\n" or croak $!;
2106 $h = $untracked->{empty};
2107 foreach (sort keys %$h) {
2108 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2109 print $un " $act: ", uri_encode($_), "\n" or croak $!;
2110 warn "W: $act: $_\n";
2111 }
2112 foreach my $t (qw/dir_prop file_prop/) {
2113 $h = $untracked->{$t} or next;
2114 foreach my $path (sort keys %$h) {
2115 my $ppath = $path eq '' ? '.' : $path;
2116 foreach my $prop (sort keys %{$h->{$path}}) {
2117 next if $SKIP{$prop};
2118 my $v = $h->{$path}->{$prop};
2119 if (defined $v) {
2120 print $un " +$t: ",
2121 uri_encode($ppath), ' ',
2122 uri_encode($prop), ' ',
2123 uri_encode($v), "\n"
2124 or croak $!;
2125 } else {
2126 print $un " -$t: ",
2127 uri_encode($ppath), ' ',
2128 uri_encode($prop), "\n"
2129 or croak $!;
2130 }
2131 }
2132 }
2133 }
2134 foreach my $t (qw/absent_file absent_directory/) {
2135 $h = $untracked->{$t} or next;
2136 foreach my $parent (sort keys %$h) {
2137 foreach my $path (sort @{$h->{$parent}}) {
2138 print $un " $t: ",
2139 uri_encode("$parent/$path"), "\n"
2140 or croak $!;
2141 warn "W: $t: $parent/$path ",
2142 "Insufficient permissions?\n";
2143 }
2144 }
2145 }
2146
2147 # revprops (make this optional? it's an extra network trip...)
2148 my $pool = SVN::Pool->new;
2149 my $rp = $SVN->rev_proplist($rev, $pool);
2150 foreach (sort keys %$rp) {
2151 next if /^svn:(?:author|date|log)$/;
2152 print $un " rev_prop: ", uri_encode($_), ' ',
2153 uri_encode($rp->{$_}), "\n";
2154 }
2155 $pool->clear;
2156 close $un or croak $!;
2157
2158 { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2159 author => $author, msg => $msg."\n", parents => $parents || [],
2160 revprops => $rp }
2161 }
2162
2163 sub process_rm {
2164 my ($gui, $last_commit, $f, $q) = @_;
2165 # remove entire directories.
2166 if (command('ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2167 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2168 -r --name-only -z/,
2169 $last_commit,'--',$f);
2170 local $/ = "\0";
2171 while (<$ls>) {
2172 print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2173 print "\tD\t$_\n" unless $q;
2174 }
2175 print "\tD\t$f/\n" unless $q;
2176 command_close_pipe($ls, $ctx);
2177 return $SVN::Node::dir;
2178 } else {
2179 print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2180 print "\tD\t$f\n" unless $q;
2181 return $SVN::Node::file;
2182 }
2183 }
2184
2185 sub libsvn_fetch {
2186 my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2187 my $pool = SVN::Pool->new;
2188 my $ed = SVN::Git::Fetcher->new({ c => $last_commit, q => $_q });
2189 my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
2190 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2191 my (undef, $last_rev, undef) = cmt_metadata($last_commit);
2192 $reporter->set_path('', $last_rev, 0, @lock, $pool);
2193 $reporter->finish_report($pool);
2194 $pool->clear;
2195 unless ($ed->{git_commit_ok}) {
2196 die "SVN connection failed somewhere...\n";
2197 }
2198 libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ed);
2199 }
2200
2201 sub svn_grab_base_rev {
2202 my $c = eval { command_oneline([qw/rev-parse --verify/,
2203 "refs/remotes/$GIT_SVN^0"],
2204 { STDERR => 0 }) };
2205 if (defined $c && length $c) {
2206 my ($url, $rev, $uuid) = cmt_metadata($c);
2207 return ($rev, $c) if defined $rev;
2208 }
2209 if ($_no_metadata) {
2210 my $offset = -41; # from tail
2211 my $rl;
2212 open my $fh, '<', $REVDB or
2213 die "--no-metadata specified and $REVDB not readable\n";
2214 seek $fh, $offset, 2;
2215 $rl = readline $fh;
2216 defined $rl or return (undef, undef);
2217 chomp $rl;
2218 while ($c ne $rl && tell $fh != 0) {
2219 $offset -= 41;
2220 seek $fh, $offset, 2;
2221 $rl = readline $fh;
2222 defined $rl or return (undef, undef);
2223 chomp $rl;
2224 }
2225 my $rev = tell $fh;
2226 croak $! if ($rev < -1);
2227 $rev = ($rev - 41) / 41;
2228 close $fh or croak $!;
2229 return ($rev, $c);
2230 }
2231 return (undef, undef);
2232 }
2233
2234 sub libsvn_parse_revision {
2235 my $base = shift;
2236 my $head = $SVN->get_latest_revnum();
2237 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2238 return ($base + 1, $head) if (defined $base);
2239 return (0, $head);
2240 }
2241 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2242 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2243 if ($_revision =~ /^BASE:(\d+)$/) {
2244 return ($base + 1, $1) if (defined $base);
2245 return (0, $head);
2246 }
2247 return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2248 die "revision argument: $_revision not understood by git-svn\n",
2249 "Try using the command-line svn client instead\n";
2250 }
2251
2252 sub libsvn_traverse_ignore {
2253 my ($fh, $path, $r) = @_;
2254 $path =~ s#^/+##g;
2255 my $pool = SVN::Pool->new;
2256 my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2257 my $p = $path;
2258 $p =~ s#^\Q$SVN->{svn_path}\E/##;
2259 print $fh length $p ? "\n# $p\n" : "\n# /\n";
2260 if (my $s = $props->{'svn:ignore'}) {
2261 $s =~ s/[\r\n]+/\n/g;
2262 chomp $s;
2263 if (length $p == 0) {
2264 $s =~ s#\n#\n/$p#g;
2265 print $fh "/$s\n";
2266 } else {
2267 $s =~ s#\n#\n/$p/#g;
2268 print $fh "/$p/$s\n";
2269 }
2270 }
2271 foreach (sort keys %$dirent) {
2272 next if $dirent->{$_}->kind != $SVN::Node::dir;
2273 libsvn_traverse_ignore($fh, "$path/$_", $r);
2274 }
2275 $pool->clear;
2276 }
2277
2278 sub revisions_eq {
2279 my ($path, $r0, $r1) = @_;
2280 return 1 if $r0 == $r1;
2281 my $nr = 0;
2282 # should be OK to use Pool here (r1 - r0) should be small
2283 my $pool = SVN::Pool->new;
2284 libsvn_get_log($SVN, [$path], $r0, $r1,
2285 0, 0, 1, sub {$nr++}, $pool);
2286 $pool->clear;
2287 return 0 if ($nr > 1);
2288 return 1;
2289 }
2290
2291 sub libsvn_find_parent_branch {
2292 my ($paths, $rev, $author, $date, $msg) = @_;
2293 my $svn_path = '/'.$SVN->{svn_path};
2294
2295 # look for a parent from another branch:
2296 my $i = $paths->{$svn_path} or return;
2297 my $branch_from = $i->copyfrom_path or return;
2298 my $r = $i->copyfrom_rev;
2299 print STDERR "Found possible branch point: ",
2300 "$branch_from => $svn_path, $r\n";
2301 $branch_from =~ s#^/##;
2302 my $l_map = {};
2303 read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2304 my $url = $SVN->{repos_root};
2305 defined $l_map->{$url} or return;
2306 my $id = $l_map->{$url}->{$branch_from};
2307 if (!defined $id && $_follow_parent) {
2308 print STDERR "Following parent: $branch_from\@$r\n";
2309 # auto create a new branch and follow it
2310 $id = basename($branch_from);
2311 $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2312 while (-r "$GIT_DIR/svn/$id") {
2313 # just grow a tail if we're not unique enough :x
2314 $id .= '-';
2315 }
2316 }
2317 return unless defined $id;
2318
2319 my ($r0, $parent) = find_rev_before($r,$id,1);
2320 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2321 defined(my $pid = fork) or croak $!;
2322 if (!$pid) {
2323 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2324 init_vars();
2325 $SVN_URL = "$url/$branch_from";
2326 $SVN = undef;
2327 setup_git_svn();
2328 # we can't assume SVN_URL exists at r+1:
2329 $_revision = "0:$r";
2330 fetch_lib();
2331 exit 0;
2332 }
2333 waitpid $pid, 0;
2334 croak $? if $?;
2335 ($r0, $parent) = find_rev_before($r,$id,1);
2336 }
2337 return unless (defined $r0 && defined $parent);
2338 if (revisions_eq($branch_from, $r0, $r)) {
2339 unlink $GIT_SVN_INDEX;
2340 print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2341 command_noisy('read-tree', $parent);
2342 unless (libsvn_can_do_switch()) {
2343 return _libsvn_new_tree($paths, $rev, $author, $date,
2344 $msg, [$parent]);
2345 }
2346 # do_switch works with svn/trunk >= r22312, but that is not
2347 # included with SVN 1.4.2 (the latest version at the moment),
2348 # so we can't rely on it.
2349 my $ra = libsvn_connect("$url/$branch_from");
2350 my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q });
2351 my $pool = SVN::Pool->new;
2352 my $reporter = $ra->do_switch($rev, '', 1, $SVN->{url},
2353 $ed, $pool);
2354 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2355 $reporter->set_path('', $r0, 0, @lock, $pool);
2356 $reporter->finish_report($pool);
2357 $pool->clear;
2358 unless ($ed->{git_commit_ok}) {
2359 die "SVN connection failed somewhere...\n";
2360 }
2361 return libsvn_log_entry($rev, $author, $date, $msg, [$parent]);
2362 }
2363 print STDERR "Nope, branch point not imported or unknown\n";
2364 return undef;
2365 }
2366
2367 sub libsvn_get_log {
2368 my ($ra, @args) = @_;
2369 $args[4]-- if $args[4] && ! $_follow_parent;
2370 if ($SVN::Core::VERSION le '1.2.0') {
2371 splice(@args, 3, 1);
2372 }
2373 $ra->get_log(@args);
2374 }
2375
2376 sub libsvn_new_tree {
2377 if (my $log_entry = libsvn_find_parent_branch(@_)) {
2378 return $log_entry;
2379 }
2380 my ($paths, $rev, $author, $date, $msg) = @_; # $pool is last
2381 _libsvn_new_tree($paths, $rev, $author, $date, $msg, []);
2382 }
2383
2384 sub _libsvn_new_tree {
2385 my ($paths, $rev, $author, $date, $msg, $parents) = @_;
2386 my $pool = SVN::Pool->new;
2387 my $ed = SVN::Git::Fetcher->new({q => $_q});
2388 my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
2389 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2390 $reporter->set_path('', $rev, 1, @lock, $pool);
2391 $reporter->finish_report($pool);
2392 $pool->clear;
2393 unless ($ed->{git_commit_ok}) {
2394 die "SVN connection failed somewhere...\n";
2395 }
2396 libsvn_log_entry($rev, $author, $date, $msg, $parents, $ed);
2397 }
2398
2399 sub find_graft_path_commit {
2400 my ($tree_paths, $p1, $r1) = @_;
2401 foreach my $x (keys %$tree_paths) {
2402 next unless ($p1 =~ /^\Q$x\E/);
2403 my $i = $tree_paths->{$x};
2404 my ($r0, $parent) = find_rev_before($r1,$i,1);
2405 return $parent if (defined $r0 && $r0 == $r1);
2406 print STDERR "r$r1 of $i not imported\n";
2407 next;
2408 }
2409 return undef;
2410 }
2411
2412 sub find_graft_path_parents {
2413 my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2414 foreach my $x (keys %$tree_paths) {
2415 next unless ($p0 =~ /^\Q$x\E/);
2416 my $i = $tree_paths->{$x};
2417 my ($r, $parent) = find_rev_before($r0, $i, 1);
2418 if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2419 my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2420 my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2421 next if ($url_a && $url_b && $url_a eq $url_b &&
2422 $uuid_b eq $uuid_a);
2423 $grafts->{$c}->{$parent} = 1;
2424 }
2425 }
2426 }
2427
2428 sub libsvn_graft_file_copies {
2429 my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2430 foreach (keys %$paths) {
2431 my $i = $paths->{$_};
2432 my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2433 $i->copyfrom_rev);
2434 next unless (defined $p0 && defined $r0);
2435
2436 my $p1 = $_;
2437 $p1 =~ s#^/##;
2438 $p0 =~ s#^/##;
2439 my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2440 next unless $c;
2441 find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2442 }
2443 }
2444
2445 sub set_index {
2446 my $old = $ENV{GIT_INDEX_FILE};
2447 $ENV{GIT_INDEX_FILE} = shift;
2448 return $old;
2449 }
2450
2451 sub restore_index {
2452 my ($old) = @_;
2453 if (defined $old) {
2454 $ENV{GIT_INDEX_FILE} = $old;
2455 } else {
2456 delete $ENV{GIT_INDEX_FILE};
2457 }
2458 }
2459
2460 sub libsvn_commit_cb {
2461 my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2462 if ($_optimize_commits && $rev == ($r_last + 1)) {
2463 my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2464 $log->{tree} = get_tree_from_treeish($c);
2465 my $cmt = git_commit($log, $cmt_last, $c);
2466 my @diff = command('diff-tree', $cmt, $c);
2467 if (@diff) {
2468 print STDERR "Trees differ: $cmt $c\n",
2469 join('',@diff),"\n";
2470 exit 1;
2471 }
2472 } else {
2473 fetch("$rev=$c");
2474 }
2475 }
2476
2477 sub libsvn_ls_fullurl {
2478 my $fullurl = shift;
2479 my $ra = libsvn_connect($fullurl);
2480 my @ret;
2481 my $pool = SVN::Pool->new;
2482 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
2483 my ($dirent, undef, undef) = $ra->get_dir('', $r, $pool);
2484 foreach my $d (sort keys %$dirent) {
2485 if ($dirent->{$d}->kind == $SVN::Node::dir) {
2486 push @ret, "$d/"; # add '/' for compat with cli svn
2487 }
2488 }
2489 $pool->clear;
2490 return @ret;
2491 }
2492
2493
2494 sub libsvn_skip_unknown_revs {
2495 my $err = shift;
2496 my $errno = $err->apr_err();
2497 # Maybe the branch we're tracking didn't
2498 # exist when the repo started, so it's
2499 # not an error if it doesn't, just continue
2500 #
2501 # Wonderfully consistent library, eh?
2502 # 160013 - svn:// and file://
2503 # 175002 - http(s)://
2504 # 175007 - http(s):// (this repo required authorization, too...)
2505 # More codes may be discovered later...
2506 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2507 return;
2508 }
2509 croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2510 };
2511
2512 # Tie::File seems to be prone to offset errors if revisions get sparse,
2513 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2514 # one of my favorite modules is out :< Next up would be one of the DBM
2515 # modules, but I'm not sure which is most portable... So I'll just
2516 # go with something that's plain-text, but still capable of
2517 # being randomly accessed. So here's my ultra-simple fixed-width
2518 # database. All records are 40 characters + "\n", so it's easy to seek
2519 # to a revision: (41 * rev) is the byte offset.
2520 # A record of 40 0s denotes an empty revision.
2521 # And yes, it's still pretty fast (faster than Tie::File).
2522 sub revdb_set {
2523 my ($file, $rev, $commit) = @_;
2524 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
2525 open my $fh, '+<', $file or croak $!;
2526 my $offset = $rev * 41;
2527 # assume that append is the common case:
2528 seek $fh, 0, 2 or croak $!;
2529 my $pos = tell $fh;
2530 if ($pos < $offset) {
2531 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
2532 }
2533 seek $fh, $offset, 0 or croak $!;
2534 print $fh $commit,"\n";
2535 close $fh or croak $!;
2536 }
2537
2538 sub revdb_get {
2539 my ($file, $rev) = @_;
2540 my $ret;
2541 my $offset = $rev * 41;
2542 open my $fh, '<', $file or croak $!;
2543 seek $fh, $offset, 0;
2544 if (tell $fh == $offset) {
2545 $ret = readline $fh;
2546 if (defined $ret) {
2547 chomp $ret;
2548 $ret = undef if ($ret =~ /^0{40}$/);
2549 }
2550 }
2551 close $fh or croak $!;
2552 return $ret;
2553 }
2554
2555 sub copy_remote_ref {
2556 my $origin = $_cp_remote ? $_cp_remote : 'origin';
2557 my $ref = "refs/remotes/$GIT_SVN";
2558 if (command('ls-remote', $origin, $ref)) {
2559 command_noisy('fetch', $origin, "$ref:$ref");
2560 } elsif ($_cp_remote && !$_upgrade) {
2561 die "Unable to find remote reference: ",
2562 "refs/remotes/$GIT_SVN on $origin\n";
2563 }
2564 }
2565
2566 {
2567 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2568 $SVN::Node::dir.$SVN::Node::unknown.
2569 $SVN::Node::none.$SVN::Node::file.
2570 $SVN::Node::dir.$SVN::Node::unknown.
2571 $SVN::Auth::SSL::CNMISMATCH.
2572 $SVN::Auth::SSL::NOTYETVALID.
2573 $SVN::Auth::SSL::EXPIRED.
2574 $SVN::Auth::SSL::UNKNOWNCA.
2575 $SVN::Auth::SSL::OTHER;
2576 }
2577
2578 package SVN::Git::Fetcher;
2579 use vars qw/@ISA/;
2580 use strict;
2581 use warnings;
2582 use Carp qw/croak/;
2583 use IO::File qw//;
2584 use Git qw/command command_oneline command_noisy
2585 command_output_pipe command_input_pipe command_close_pipe/;
2586
2587 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2588 sub new {
2589 my ($class, $git_svn) = @_;
2590 my $self = SVN::Delta::Editor->new;
2591 bless $self, $class;
2592 $self->{c} = $git_svn->{c} if exists $git_svn->{c};
2593 $self->{q} = $git_svn->{q};
2594 $self->{empty} = {};
2595 $self->{dir_prop} = {};
2596 $self->{file_prop} = {};
2597 $self->{absent_dir} = {};
2598 $self->{absent_file} = {};
2599 ($self->{gui}, $self->{ctx}) = command_input_pipe(
2600 qw/update-index -z --index-info/);
2601 require Digest::MD5;
2602 $self;
2603 }
2604
2605 sub open_root {
2606 { path => '' };
2607 }
2608
2609 sub open_directory {
2610 my ($self, $path, $pb, $rev) = @_;
2611 { path => $path };
2612 }
2613
2614 sub delete_entry {
2615 my ($self, $path, $rev, $pb) = @_;
2616 my $t = process_rm($self->{gui}, $self->{c}, $path, $self->{q});
2617 $self->{empty}->{$path} = 0 if $t == $SVN::Node::dir;
2618 undef;
2619 }
2620
2621 sub open_file {
2622 my ($self, $path, $pb, $rev) = @_;
2623 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--',$path)
2624 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2625 unless (defined $mode && defined $blob) {
2626 die "$path was not found in commit $self->{c} (r$rev)\n";
2627 }
2628 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2629 pool => SVN::Pool->new, action => 'M' };
2630 }
2631
2632 sub add_file {
2633 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2634 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2635 delete $self->{empty}->{$dir};
2636 { path => $path, mode_a => 100644, mode_b => 100644,
2637 pool => SVN::Pool->new, action => 'A' };
2638 }
2639
2640 sub add_directory {
2641 my ($self, $path, $cp_path, $cp_rev) = @_;
2642 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2643 delete $self->{empty}->{$dir};
2644 $self->{empty}->{$path} = 1;
2645 { path => $path };
2646 }
2647
2648 sub change_dir_prop {
2649 my ($self, $db, $prop, $value) = @_;
2650 $self->{dir_prop}->{$db->{path}} ||= {};
2651 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2652 undef;
2653 }
2654
2655 sub absent_directory {
2656 my ($self, $path, $pb) = @_;
2657 $self->{absent_dir}->{$pb->{path}} ||= [];
2658 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2659 undef;
2660 }
2661
2662 sub absent_file {
2663 my ($self, $path, $pb) = @_;
2664 $self->{absent_file}->{$pb->{path}} ||= [];
2665 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2666 undef;
2667 }
2668
2669 sub change_file_prop {
2670 my ($self, $fb, $prop, $value) = @_;
2671 if ($prop eq 'svn:executable') {
2672 if ($fb->{mode_b} != 120000) {
2673 $fb->{mode_b} = defined $value ? 100755 : 100644;
2674 }
2675 } elsif ($prop eq 'svn:special') {
2676 $fb->{mode_b} = defined $value ? 120000 : 100644;
2677 } else {
2678 $self->{file_prop}->{$fb->{path}} ||= {};
2679 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2680 }
2681 undef;
2682 }
2683
2684 sub apply_textdelta {
2685 my ($self, $fb, $exp) = @_;
2686 my $fh = IO::File->new_tmpfile;
2687 $fh->autoflush(1);
2688 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2689 # (but $base does not,) so dup() it for reading in close_file
2690 open my $dup, '<&', $fh or croak $!;
2691 my $base = IO::File->new_tmpfile;
2692 $base->autoflush(1);
2693 if ($fb->{blob}) {
2694 defined (my $pid = fork) or croak $!;
2695 if (!$pid) {
2696 open STDOUT, '>&', $base or croak $!;
2697 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2698 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2699 }
2700 waitpid $pid, 0;
2701 croak $? if $?;
2702
2703 if (defined $exp) {
2704 seek $base, 0, 0 or croak $!;
2705 my $md5 = Digest::MD5->new;
2706 $md5->addfile($base);
2707 my $got = $md5->hexdigest;
2708 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2709 "expected: $exp\n",
2710 " got: $got\n" if ($got ne $exp);
2711 }
2712 }
2713 seek $base, 0, 0 or croak $!;
2714 $fb->{fh} = $dup;
2715 $fb->{base} = $base;
2716 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2717 }
2718
2719 sub close_file {
2720 my ($self, $fb, $exp) = @_;
2721 my $hash;
2722 my $path = $fb->{path};
2723 if (my $fh = $fb->{fh}) {
2724 seek($fh, 0, 0) or croak $!;
2725 my $md5 = Digest::MD5->new;
2726 $md5->addfile($fh);
2727 my $got = $md5->hexdigest;
2728 die "Checksum mismatch: $path\n",
2729 "expected: $exp\n got: $got\n" if ($got ne $exp);
2730 seek($fh, 0, 0) or croak $!;
2731 if ($fb->{mode_b} == 120000) {
2732 read($fh, my $buf, 5) == 5 or croak $!;
2733 $buf eq 'link ' or die "$path has mode 120000",
2734 "but is not a link\n";
2735 }
2736 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2737 if (!$pid) {
2738 open STDIN, '<&', $fh or croak $!;
2739 exec qw/git-hash-object -w --stdin/ or croak $!;
2740 }
2741 chomp($hash = do { local $/; <$out> });
2742 close $out or croak $!;
2743 close $fh or croak $!;
2744 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2745 close $fb->{base} or croak $!;
2746 } else {
2747 $hash = $fb->{blob} or die "no blob information\n";
2748 }
2749 $fb->{pool}->clear;
2750 my $gui = $self->{gui};
2751 print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
2752 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
2753 undef;
2754 }
2755
2756 sub abort_edit {
2757 my $self = shift;
2758 eval { command_close_pipe($self->{gui}, $self->{ctx}) };
2759 $self->SUPER::abort_edit(@_);
2760 }
2761
2762 sub close_edit {
2763 my $self = shift;
2764 command_close_pipe($self->{gui}, $self->{ctx});
2765 $self->{git_commit_ok} = 1;
2766 $self->SUPER::close_edit(@_);
2767 }
2768
2769 package SVN::Git::Editor;
2770 use vars qw/@ISA/;
2771 use strict;
2772 use warnings;
2773 use Carp qw/croak/;
2774 use IO::File;
2775 use Git qw/command command_oneline command_noisy
2776 command_output_pipe command_input_pipe command_close_pipe/;
2777
2778 sub new {
2779 my $class = shift;
2780 my $git_svn = shift;
2781 my $self = SVN::Delta::Editor->new(@_);
2782 bless $self, $class;
2783 foreach (qw/svn_path c r ra /) {
2784 die "$_ required!\n" unless (defined $git_svn->{$_});
2785 $self->{$_} = $git_svn->{$_};
2786 }
2787 $self->{pool} = SVN::Pool->new;
2788 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2789 $self->{rm} = { };
2790 require Digest::MD5;
2791 return $self;
2792 }
2793
2794 sub split_path {
2795 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2796 }
2797
2798 sub repo_path {
2799 (defined $_[1] && length $_[1]) ? $_[1] : ''
2800 }
2801
2802 sub url_path {
2803 my ($self, $path) = @_;
2804 $self->{ra}->{url} . '/' . $self->repo_path($path);
2805 }
2806
2807 sub rmdirs {
2808 my ($self, $q) = @_;
2809 my $rm = $self->{rm};
2810 delete $rm->{''}; # we never delete the url we're tracking
2811 return unless %$rm;
2812
2813 foreach (keys %$rm) {
2814 my @d = split m#/#, $_;
2815 my $c = shift @d;
2816 $rm->{$c} = 1;
2817 while (@d) {
2818 $c .= '/' . shift @d;
2819 $rm->{$c} = 1;
2820 }
2821 }
2822 delete $rm->{$self->{svn_path}};
2823 delete $rm->{''}; # we never delete the url we're tracking
2824 return unless %$rm;
2825
2826 my ($fh, $ctx) = command_output_pipe(
2827 qw/ls-tree --name-only -r -z/, $self->{c});
2828 local $/ = "\0";
2829 while (<$fh>) {
2830 chomp;
2831 my @dn = split m#/#, $_;
2832 while (pop @dn) {
2833 delete $rm->{join '/', @dn};
2834 }
2835 unless (%$rm) {
2836 eval { command_close_pipe($fh) };
2837 return;
2838 }
2839 }
2840 command_close_pipe($fh, $ctx);
2841
2842 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2843 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2844 $self->close_directory($bat->{$d}, $p);
2845 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2846 print "\tD+\t/$d/\n" unless $q;
2847 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2848 delete $bat->{$d};
2849 }
2850 }
2851
2852 sub open_or_add_dir {
2853 my ($self, $full_path, $baton) = @_;
2854 my $p = SVN::Pool->new;
2855 my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
2856 $p->clear;
2857 if ($t == $SVN::Node::none) {
2858 return $self->add_directory($full_path, $baton,
2859 undef, -1, $self->{pool});
2860 } elsif ($t == $SVN::Node::dir) {
2861 return $self->open_directory($full_path, $baton,
2862 $self->{r}, $self->{pool});
2863 }
2864 print STDERR "$full_path already exists in repository at ",
2865 "r$self->{r} and it is not a directory (",
2866 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2867 exit 1;
2868 }
2869
2870 sub ensure_path {
2871 my ($self, $path) = @_;
2872 my $bat = $self->{bat};
2873 $path = $self->repo_path($path);
2874 return $bat->{''} unless (length $path);
2875 my @p = split m#/+#, $path;
2876 my $c = shift @p;
2877 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2878 while (@p) {
2879 my $c0 = $c;
2880 $c .= '/' . shift @p;
2881 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2882 }
2883 return $bat->{$c};
2884 }
2885
2886 sub A {
2887 my ($self, $m, $q) = @_;
2888 my ($dir, $file) = split_path($m->{file_b});
2889 my $pbat = $self->ensure_path($dir);
2890 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2891 undef, -1);
2892 print "\tA\t$m->{file_b}\n" unless $q;
2893 $self->chg_file($fbat, $m);
2894 $self->close_file($fbat,undef,$self->{pool});
2895 }
2896
2897 sub C {
2898 my ($self, $m, $q) = @_;
2899 my ($dir, $file) = split_path($m->{file_b});
2900 my $pbat = $self->ensure_path($dir);
2901 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2902 $self->url_path($m->{file_a}), $self->{r});
2903 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
2904 $self->chg_file($fbat, $m);
2905 $self->close_file($fbat,undef,$self->{pool});
2906 }
2907
2908 sub delete_entry {
2909 my ($self, $path, $pbat) = @_;
2910 my $rpath = $self->repo_path($path);
2911 my ($dir, $file) = split_path($rpath);
2912 $self->{rm}->{$dir} = 1;
2913 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2914 }
2915
2916 sub R {
2917 my ($self, $m, $q) = @_;
2918 my ($dir, $file) = split_path($m->{file_b});
2919 my $pbat = $self->ensure_path($dir);
2920 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2921 $self->url_path($m->{file_a}), $self->{r});
2922 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
2923 $self->chg_file($fbat, $m);
2924 $self->close_file($fbat,undef,$self->{pool});
2925
2926 ($dir, $file) = split_path($m->{file_a});
2927 $pbat = $self->ensure_path($dir);
2928 $self->delete_entry($m->{file_a}, $pbat);
2929 }
2930
2931 sub M {
2932 my ($self, $m, $q) = @_;
2933 my ($dir, $file) = split_path($m->{file_b});
2934 my $pbat = $self->ensure_path($dir);
2935 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2936 $pbat,$self->{r},$self->{pool});
2937 print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
2938 $self->chg_file($fbat, $m);
2939 $self->close_file($fbat,undef,$self->{pool});
2940 }
2941
2942 sub T { shift->M(@_) }
2943
2944 sub change_file_prop {
2945 my ($self, $fbat, $pname, $pval) = @_;
2946 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2947 }
2948
2949 sub chg_file {
2950 my ($self, $fbat, $m) = @_;
2951 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2952 $self->change_file_prop($fbat,'svn:executable','*');
2953 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2954 $self->change_file_prop($fbat,'svn:executable',undef);
2955 }
2956 my $fh = IO::File->new_tmpfile or croak $!;
2957 if ($m->{mode_b} =~ /^120/) {
2958 print $fh 'link ' or croak $!;
2959 $self->change_file_prop($fbat,'svn:special','*');
2960 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2961 $self->change_file_prop($fbat,'svn:special',undef);
2962 }
2963 defined(my $pid = fork) or croak $!;
2964 if (!$pid) {
2965 open STDOUT, '>&', $fh or croak $!;
2966 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2967 }
2968 waitpid $pid, 0;
2969 croak $? if $?;
2970 $fh->flush == 0 or croak $!;
2971 seek $fh, 0, 0 or croak $!;
2972
2973 my $md5 = Digest::MD5->new;
2974 $md5->addfile($fh) or croak $!;
2975 seek $fh, 0, 0 or croak $!;
2976
2977 my $exp = $md5->hexdigest;
2978 my $pool = SVN::Pool->new;
2979 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2980 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2981 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2982 $pool->clear;
2983
2984 close $fh or croak $!;
2985 }
2986
2987 sub D {
2988 my ($self, $m, $q) = @_;
2989 my ($dir, $file) = split_path($m->{file_b});
2990 my $pbat = $self->ensure_path($dir);
2991 print "\tD\t$m->{file_b}\n" unless $q;
2992 $self->delete_entry($m->{file_b}, $pbat);
2993 }
2994
2995 sub close_edit {
2996 my ($self) = @_;
2997 my ($p,$bat) = ($self->{pool}, $self->{bat});
2998 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2999 $self->close_directory($bat->{$_}, $p);
3000 }
3001 $self->SUPER::close_edit($p);
3002 $p->clear;
3003 }
3004
3005 sub abort_edit {
3006 my ($self) = @_;
3007 $self->SUPER::abort_edit($self->{pool});
3008 $self->{pool}->clear;
3009 }
3010
3011 __END__
3012
3013 Data structures:
3014
3015 $log_msg hashref as returned by libsvn_log_entry()
3016 {
3017 msg => 'whitespace-formatted log entry
3018 ', # trailing newline is preserved
3019 revision => '8', # integer
3020 date => '2004-02-24T17:01:44.108345Z', # commit date
3021 author => 'committer name'
3022 };
3023
3024 @mods = array of diff-index line hashes, each element represents one line
3025 of diff-index output
3026
3027 diff-index line ($m hash)
3028 {
3029 mode_a => first column of diff-index output, no leading ':',
3030 mode_b => second column of diff-index output,
3031 sha1_b => sha1sum of the final blob,
3032 chg => change type [MCRADT],
3033 file_a => original file name of a file (iff chg is 'C' or 'R')
3034 file_b => new/current file name of a file (any chg)
3035 }
3036 ;
3037
3038 # retval of read_url_paths{,_all}();
3039 $l_map = {
3040 # repository root url
3041 'https://svn.musicpd.org' => {
3042 # repository path # GIT_SVN_ID
3043 'mpd/trunk' => 'trunk',
3044 'mpd/tags/0.11.5' => 'tags/0.11.5',
3045 },
3046 }
3047
3048 Notes:
3049 I don't trust the each() function on unless I created %hash myself
3050 because the internal iterator may not have started at base.
git with Jakub Narębski modifications