261e33d02302d68c0d10fc3d3ad5e59d13975e01
1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $SVN_URL
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB
10 $_follow_parent $sha1 $sha1_short $_revision
11 $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
12 $_find_copies_harder $_l $_authors %users/;
13$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
14$VERSION = '@@GIT_VERSION@@';
15use Cwd qw/abs_path/;
17$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
18$ENV{GIT_DIR} = $GIT_DIR;
19my $LC_ALL = $ENV{LC_ALL};
21$Git::SVN::Log::TZ = $ENV{TZ};
22# make sure the svn binary gives consistent output between locales and TZs:
23$ENV{TZ} = 'UTC';
24$ENV{LC_ALL} = 'C';
25$| = 1; # unbuffer STDOUT
26sub fatal (@) { print STDERR @_; exit 1 }
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') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
33}
34push @Git::SVN::Ra::ISA, 'SVN::Ra';
35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37use Carp qw/croak/;
38use IO::File qw//;
39use File::Basename qw/dirname basename/;
40use File::Path qw/mkpath/;
41use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
42use IPC::Open3;
43use Git;
44BEGIN {
46 my $s;
47 foreach (qw/command command_oneline command_noisy command_output_pipe
48 command_input_pipe command_close_pipe/) {
49 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
50 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
51 }
52 eval $s;
53}
54my ($SVN);
56my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
58$sha1 = qr/[a-f\d]{40}/;
59$sha1_short = qr/[a-f\d]{4,40}/;
60my ($_stdin, $_help, $_edit,
61 $_repack, $_repack_nr, $_repack_flags,
62 $_message, $_file, $_no_metadata,
63 $_template, $_shared, $_no_default_regex, $_no_graft_copy,
64 $_version, $_upgrade, $_branch_all_refs, @_opt_m,
65 $_merge, $_strategy, $_dry_run,
66 $_prefix);
67my (@_branch_from, %tree_map);
68my @repo_path_split_cache;
69my %fc_opts = ( 'branch|b=s' => \@_branch_from,
71 'follow-parent|follow' => \$_follow_parent,
72 'branch-all-refs|B' => \$_branch_all_refs,
73 'authors-file|A=s' => \$_authors,
74 'repack:i' => \$_repack,
75 'no-metadata' => \$_no_metadata,
76 'quiet|q' => \$_q,
77 'username=s' => \$Git::SVN::Prompt::_username,
78 'config-dir=s' => \$Git::SVN::Ra::config_dir,
79 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
80 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
81my ($_trunk, $_tags, $_branches);
83my %multi_opts = ( 'trunk|T=s' => \$_trunk,
84 'tags|t=s' => \$_tags,
85 'branches|b=s' => \$_branches );
86my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
87my %cmt_opts = ( 'edit|e' => \$_edit,
88 'rmdir' => \$_rmdir,
89 'find-copies-harder' => \$_find_copies_harder,
90 'l=i' => \$_l,
91 'copy-similarity|C=i'=> \$_cp_similarity
92);
93my %cmd = (
95 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
96 { 'revision|r=s' => \$_revision, %fc_opts } ],
97 init => [ \&cmd_init, "Initialize a repo for tracking" .
98 " (requires URL argument)",
99 \%init_opts ],
100 dcommit => [ \&cmd_dcommit,
101 'Commit several diffs to merge with upstream',
102 { 'merge|m|M' => \$_merge,
103 'strategy|s=s' => \$_strategy,
104 'dry-run|n' => \$_dry_run,
105 %cmt_opts, %fc_opts } ],
106 'set-tree' => [ \&cmd_set_tree,
107 "Set an SVN repository to a git tree-ish",
108 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
109 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
110 { 'revision|r=i' => \$_revision } ],
111 rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
112 { 'copy-remote|remote=s' => \$_cp_remote,
113 'upgrade' => \$_upgrade } ],
114 'graft-branches' => [ \&graft_branches,
115 'Detect merges/branches from already imported history',
116 { 'merge-rx|m' => \@_opt_m,
117 'branch|b=s' => \@_branch_from,
118 'branch-all-refs|B' => \$_branch_all_refs,
119 'no-default-regex' => \$_no_default_regex,
120 'no-graft-copy' => \$_no_graft_copy } ],
121 'multi-init' => [ \&cmd_multi_init,
122 'Initialize multiple trees (like git-svnimport)',
123 { %multi_opts, %init_opts,
124 'revision|r=i' => \$_revision,
125 'username=s' => \$Git::SVN::Prompt::_username,
126 'config-dir=s' => \$Git::SVN::Ra::config_dir,
127 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
128 'prefix=s' => \$_prefix,
129 } ],
130 'multi-fetch' => [ \&cmd_multi_fetch,
131 'Fetch multiple trees (like git-svnimport)',
132 \%fc_opts ],
133 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
134 { 'limit=i' => \$Git::SVN::Log::limit,
135 'revision|r=s' => \$_revision,
136 'verbose|v' => \$Git::SVN::Log::verbose,
137 'incremental' => \$Git::SVN::Log::incremental,
138 'oneline' => \$Git::SVN::Log::oneline,
139 'show-commit' => \$Git::SVN::Log::show_commit,
140 'non-recursive' => \$Git::SVN::Log::non_recursive,
141 'authors-file|A=s' => \$_authors,
142 'color' => \$Git::SVN::Log::color,
143 'pager=s' => \$Git::SVN::Log::pager,
144 } ],
145 'commit-diff' => [ \&cmd_commit_diff,
146 'Commit a diff between two trees',
147 { 'message|m=s' => \$_message,
148 'file|F=s' => \$_file,
149 'revision|r=s' => \$_revision,
150 %cmt_opts } ],
151);
152my $cmd;
154for (my $i = 0; $i < @ARGV; $i++) {
155 if (defined $cmd{$ARGV[$i]}) {
156 $cmd = $ARGV[$i];
157 splice @ARGV, $i, 1;
158 last;
159 }
160};
161my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
163read_repo_config(\%opts);
165my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
166 'version|V' => \$_version,
167 'id|i=s' => \$GIT_SVN);
168exit 1 if (!$rv && $cmd ne 'log');
169set_default_vals();
171usage(0) if $_help;
172version() if $_version;
173usage(1) unless defined $cmd;
174init_vars();
175load_authors() if $_authors;
176load_all_refs() if $_branch_all_refs;
177migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
178$cmd{$cmd}->[0]->(@ARGV);
179exit 0;
180####################### primary functions ######################
182sub usage {
183 my $exit = shift || 0;
184 my $fd = $exit ? \*STDERR : \*STDOUT;
185 print $fd <<"";
186git-svn - bidirectional operations between a single Subversion tree and git
187Usage: $0 <command> [options] [arguments]\n
188 print $fd "Available commands:\n" unless $cmd;
190 foreach (sort keys %cmd) {
192 next if $cmd && $cmd ne $_;
193 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
194 foreach (keys %{$cmd{$_}->[2]}) {
195 # prints out arguments as they should be passed:
196 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
197 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
198 "--$_" : "-$_" }
199 split /\|/,$_)," $x\n";
200 }
201 }
202 print $fd <<"";
203\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
204arbitrary identifier if you're tracking multiple SVN branches/repositories in
205one git repository and want to keep them separate. See git-svn(1) for more
206information.
207 exit $exit;
209}
210sub version {
212 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
213 exit 0;
214}
215sub cmd_rebuild {
217 my $url = shift;
218 my $gs = $url ? Git::SVN->init(undef, $url)
219 : eval { Git::SVN->new };
220 $gs ||= Git::SVN->_new;
221 if (!verify_ref($gs->refname.'^0')) {
222 $gs->copy_remote_ref;
223 }
224 if ($_upgrade) {
225 command_noisy('update-ref',$gs->refname, $gs->{id}.'-HEAD');
226 } else {
227 $gs->check_upgrade_needed;
228 }
229 my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
231 my $latest;
232 my $svn_uuid;
233 while (<$rev_list>) {
234 chomp;
235 my $c = $_;
236 fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
237 my ($url, $rev, $uuid) = cmt_metadata($c);
238 # ignore merges (from set-tree)
240 next if (!defined $rev || !$uuid);
241 # if we merged or otherwise started elsewhere, this is
243 # how we break out of it
244 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
245 ($gs->{url} && $url && ($url ne $gs->{url}))) {
246 next;
247 }
248 unless (defined $latest) {
250 if (!$gs->{url} && !$url) {
251 fatal "SVN repository location required\n";
252 }
253 $gs = Git::SVN->init(undef, $url);
254 $latest = $rev;
255 }
256 $gs->rev_db_set($rev, $c);
257 print "r$rev = $c\n";
258 }
259 command_close_pipe($rev_list, $ctx);
260}
261sub do_git_init_db {
263 unless (-d $ENV{GIT_DIR}) {
264 my @init_db = ('init');
265 push @init_db, "--template=$_template" if defined $_template;
266 push @init_db, "--shared" if defined $_shared;
267 command_noisy(@init_db);
268 }
269}
270sub cmd_init {
272 my $url = shift or die "SVN repository location required " .
273 "as a command-line argument\n";
274 if (my $repo_path = shift) {
275 unless (-d $repo_path) {
276 mkpath([$repo_path]);
277 }
278 chdir $repo_path or croak $!;
279 $ENV{GIT_DIR} = $repo_path . "/.git";
280 }
281 do_git_init_db();
282 Git::SVN->init(undef, $url);
284}
285sub cmd_fetch {
287 my $gs = Git::SVN->new;
288 $gs->fetch(@_);
289 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
290 command_noisy(qw(update-ref refs/heads/master),
291 $gs->{last_commit});
292 }
293}
294sub cmd_set_tree {
296 my (@commits) = @_;
297 if ($_stdin || !@commits) {
298 print "Reading from stdin...\n";
299 @commits = ();
300 while (<STDIN>) {
301 if (/\b($sha1_short)\b/o) {
302 unshift @commits, $1;
303 }
304 }
305 }
306 my @revs;
307 foreach my $c (@commits) {
308 my @tmp = command('rev-parse',$c);
309 if (scalar @tmp == 1) {
310 push @revs, $tmp[0];
311 } elsif (scalar @tmp > 1) {
312 push @revs, reverse(command('rev-list',@tmp));
313 } else {
314 fatal "Failed to rev-parse $c\n";
315 }
316 }
317 my $gs = Git::SVN->new;
318 my ($r_last, $cmt_last) = $gs->last_rev_commit;
319 $gs->fetch;
320 if ($r_last != $gs->{last_rev}) {
321 fatal "There are new revisions that were fetched ",
322 "and need to be merged (or acknowledged) ",
323 "before committing.\nlast rev: $r_last\n",
324 " current: $gs->{last_rev}\n";
325 }
326 $gs->set_tree($_) foreach @revs;
327 print "Done committing ",scalar @revs," revisions to SVN\n";
328}
329sub cmd_dcommit {
331 my $head = shift;
332 my $gs = Git::SVN->new;
333 $head ||= 'HEAD';
334 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
335 my $last_rev;
336 foreach my $d (reverse @refs) {
337 if (!verify_ref("$d~1")) {
338 fatal "Commit $d\n",
339 "has no parent commit, and therefore ",
340 "nothing to diff against.\n",
341 "You should be working from a repository ",
342 "originally created by git-svn\n";
343 }
344 unless (defined $last_rev) {
345 (undef, $last_rev, undef) = cmt_metadata("$d~1");
346 unless (defined $last_rev) {
347 fatal "Unable to extract revision information ",
348 "from commit $d~1\n";
349 }
350 }
351 if ($_dry_run) {
352 print "diff-tree $d~1 $d\n";
353 } else {
354 my $ra = $gs->ra;
355 my $pool = SVN::Pool->new;
356 my %ed_opts = ( r => $last_rev,
357 ra => $ra->dup,
358 svn_path => $ra->{svn_path} );
359 my $ed = SVN::Git::Editor->new(\%ed_opts,
360 $ra->get_commit_editor($::_message,
361 sub { print "Committed r$_[0]\n";
362 $last_rev = $_[0]; }),
363 $pool);
364 my $mods = $ed->apply_diff("$d~1", $d);
365 if (@$mods == 0) {
366 print "No changes\n$d~1 == $d\n";
367 }
368 }
369 }
370 return if $_dry_run;
371 $gs->fetch;
372 # we always want to rebase against the current HEAD, not any
373 # head that was passed to us
374 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
375 my @finish;
376 if (@diff) {
377 @finish = qw/rebase/;
378 push @finish, qw/--merge/ if $_merge;
379 push @finish, "--strategy=$_strategy" if $_strategy;
380 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
381 "using @finish:\n", "@diff";
382 } else {
383 print "No changes between current HEAD and ",
384 $gs->refname, "\nResetting to the latest ",
385 $gs->refname, "\n";
386 @finish = qw/reset --mixed/;
387 }
388 command_noisy(@finish, $gs->refname);
389}
390sub cmd_show_ignore {
392 my $gs = Git::SVN->new;
393 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
394 $gs->traverse_ignore(\*STDOUT, '', $r);
395}
396sub graft_branches {
398 my $gr_file = "$GIT_DIR/info/grafts";
399 my ($grafts, $comments) = read_grafts($gr_file);
400 my $gr_sha1;
401 if (%$grafts) {
403 # temporarily disable our grafts file to make this idempotent
404 chomp($gr_sha1 = command(qw/hash-object -w/,$gr_file));
405 rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
406 }
407 my $l_map = read_url_paths();
409 my @re = map { qr/$_/is } @_opt_m if @_opt_m;
410 unless ($_no_default_regex) {
411 push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
412 qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
413 qr/\b(?:from|of)\s+([\w\.\-]+)/i );
414 }
415 foreach my $u (keys %$l_map) {
416 if (@re) {
417 foreach my $p (keys %{$l_map->{$u}}) {
418 graft_merge_msg($grafts,$l_map,$u,$p,@re);
419 }
420 }
421 unless ($_no_graft_copy) {
422 graft_file_copy_lib($grafts,$l_map,$u);
423 }
424 }
425 graft_tree_joins($grafts);
426 write_grafts($grafts, $comments, $gr_file);
428 unlink "$gr_file~$gr_sha1" if $gr_sha1;
429}
430sub cmd_multi_init {
432 my $url = shift;
433 unless (defined $_trunk || defined $_branches || defined $_tags) {
434 usage(1);
435 }
436 do_git_init_db();
437 $_prefix = '' unless defined $_prefix;
438 if (defined $_trunk) {
439 my $gs_trunk = eval { Git::SVN->new($_prefix . 'trunk') };
440 unless ($gs_trunk) {
441 my $trunk_url = complete_svn_url($url, $_trunk);
442 $gs_trunk = Git::SVN->init($_prefix . 'trunk',
443 $trunk_url);
444 command_noisy('config', 'svn.trunk', $trunk_url);
445 }
446 }
447 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
448 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
449 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
450}
451sub cmd_multi_fetch {
453 # try to do trunk first, since branches/tags
454 # may be descended from it.
455 if (-e "$ENV{GIT_DIR}/svn/trunk/info/url") {
456 my $gs = Git::SVN->new('trunk');
457 $gs->fetch(@_);
458 }
459 rec_fetch('', "$ENV{GIT_DIR}/svn", @_);
460}
461# this command is special because it requires no metadata
463sub cmd_commit_diff {
464 my ($ta, $tb, $url) = @_;
465 my $usage = "Usage: $0 commit-diff -r<revision> ".
466 "<tree-ish> <tree-ish> [<URL>]\n";
467 fatal($usage) if (!defined $ta || !defined $tb);
468 if (!defined $url) {
469 my $gs = eval { Git::SVN->new };
470 if (!$gs) {
471 fatal("Needed URL or usable git-svn --id in ",
472 "the command-line\n", $usage);
473 }
474 $url = $gs->{url};
475 }
476 unless (defined $_revision) {
477 fatal("-r|--revision is a required argument\n", $usage);
478 }
479 if (defined $_message && defined $_file) {
480 fatal("Both --message/-m and --file/-F specified ",
481 "for the commit message.\n",
482 "I have no idea what you mean\n");
483 }
484 if (defined $_file) {
485 $_message = file_to_s($_file);
486 } else {
487 $_message ||= get_commit_entry($tb)->{log};
488 }
489 my $ra ||= Git::SVN::Ra->new($url);
490 my $r = $_revision;
491 if ($r eq 'HEAD') {
492 $r = $ra->get_latest_revnum;
493 } elsif ($r !~ /^\d+$/) {
494 die "revision argument: $r not understood by git-svn\n";
495 }
496 my $pool = SVN::Pool->new;
497 my %ed_opts = ( r => $r,
498 ra => $ra->dup,
499 svn_path => $ra->{svn_path} );
500 my $ed = SVN::Git::Editor->new(\%ed_opts,
501 $ra->get_commit_editor($_message,
502 sub { print "Committed r$_[0]\n" }),
503 $pool);
504 my $mods = $ed->apply_diff($ta, $tb);
505 if (@$mods == 0) {
506 print "No changes\n$ta == $tb\n";
507 }
508 $pool->clear;
509}
510########################### utility functions #########################
512sub rec_fetch {
514 my ($pfx, $p, @args) = @_;
515 my @dir;
516 foreach (sort <$p/*>) {
517 if (-r "$_/info/url") {
518 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
519 my $id = $pfx . basename $_;
520 next if $id eq 'trunk';
521 my $gs = Git::SVN->new($id);
522 $gs->fetch(@args);
523 } elsif (-d $_) {
524 push @dir, $_;
525 }
526 }
527 foreach (@dir) {
528 my $x = $_;
529 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
530 rec_fetch($x, $_, @args);
531 }
532}
533sub complete_svn_url {
535 my ($url, $path) = @_;
536 $path =~ s#/+$##;
537 $url =~ s#/+$## if $url;
538 if ($path !~ m#^[a-z\+]+://#) {
539 $path = '/' . $path if ($path !~ m#^/#);
540 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
541 fatal("E: '$path' is not a complete URL ",
542 "and a separate URL is not specified\n");
543 }
544 $path = $url . $path;
545 }
546 return $path;
547}
548sub complete_url_ls_init {
550 my ($ra, $path, $switch, $pfx) = @_;
551 unless ($path) {
552 print STDERR "W: $switch not specified\n";
553 return;
554 }
555 $path =~ s#/+$##;
556 if ($path =~ m#^[a-z\+]+://#) {
557 $ra = Git::SVN::Ra->new($path);
558 $path = '';
559 } else {
560 $path =~ s#^/+##;
561 unless ($ra) {
562 fatal("E: '$path' is not a complete URL ",
563 "and a separate URL is not specified\n");
564 }
565 }
566 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
567 my ($dirent, undef, undef) = $ra->get_dir($path, $r);
568 my $url = $ra->{url} . (length $path ? "/$path" : '');
569 foreach my $d (sort keys %$dirent) {
570 next if ($dirent->{$d}->kind != $SVN::Node::dir);
571 my $u = "$url/$d";
572 my $id = "$pfx$d";
573 my $gs = eval { Git::SVN->new($id) };
574 # don't try to init already existing refs
575 unless ($gs) {
576 print "init $u => $id\n";
577 Git::SVN->init($id, $u);
578 }
579 }
580 my ($n) = ($switch =~ /^--(\w+)/);
581 command_noisy('config', "svn.$n", $url);
582}
583sub common_prefix {
585 my $paths = shift;
586 my %common;
587 foreach (@$paths) {
588 my @tmp = split m#/#, $_;
589 my $p = '';
590 while (my $x = shift @tmp) {
591 $p .= "/$x";
592 $common{$p} ||= 0;
593 $common{$p}++;
594 }
595 }
596 foreach (sort {length $b <=> length $a} keys %common) {
597 if ($common{$_} == @$paths) {
598 return $_;
599 }
600 }
601 return '';
602}
603# grafts set here are 'stronger' in that they're based on actual tree
605# matches, and won't be deleted from merge-base checking in write_grafts()
606sub graft_tree_joins {
607 my $grafts = shift;
608 map_tree_joins() if (@_branch_from && !%tree_map);
609 return unless %tree_map;
610 git_svn_each(sub {
612 my $i = shift;
613 my @args = (qw/rev-list --pretty=raw/, "refs/remotes/$i");
614 my ($fh, $ctx) = command_output_pipe(@args);
615 while (<$fh>) {
616 next unless /^commit ($sha1)$/o;
617 my $c = $1;
618 my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
619 next unless $tree_map{$t};
620 my $l;
622 do {
623 $l = readline $fh;
624 } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
625 my ($s, $tz) = ($1, $2);
627 if ($tz =~ s/^\+//) {
628 $s += tz_to_s_offset($tz);
629 } elsif ($tz =~ s/^\-//) {
630 $s -= tz_to_s_offset($tz);
631 }
632 my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
634 foreach my $p (@{$tree_map{$t}}) {
636 next if $p eq $c;
637 my $mb = eval { command('merge-base', $c, $p) };
638 next unless ($@ || $?);
639 if (defined $r_a) {
640 # see if SVN says it's a relative
641 my ($url_b, $r_b, $uuid_b) =
642 cmt_metadata($p);
643 next if (defined $url_b &&
644 defined $url_a &&
645 ($url_a eq $url_b) &&
646 ($uuid_a eq $uuid_b));
647 if ($uuid_a eq $uuid_b) {
648 if ($r_b < $r_a) {
649 $grafts->{$c}->{$p} = 2;
650 next;
651 } elsif ($r_b > $r_a) {
652 $grafts->{$p}->{$c} = 2;
653 next;
654 }
655 }
656 }
657 my $ct = get_commit_time($p);
658 if ($ct < $s) {
659 $grafts->{$c}->{$p} = 2;
660 } elsif ($ct > $s) {
661 $grafts->{$p}->{$c} = 2;
662 }
663 # what should we do when $ct == $s ?
664 }
665 }
666 command_close_pipe($fh, $ctx);
667 });
668}
669sub graft_file_copy_lib {
671 my ($grafts, $l_map, $u) = @_;
672 my $tree_paths = $l_map->{$u};
673 my $pfx = common_prefix([keys %$tree_paths]);
674 my ($repo, $path) = repo_path_split($u.$pfx);
675 $SVN = Git::SVN::Ra->new($repo);
676 my ($base, $head) = libsvn_parse_revision();
678 my $inc = 1000;
679 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
680 my $eh = $SVN::Error::handler;
681 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
682 while (1) {
683 $SVN->dup->get_log([$path], $min, $max, 0, 2, 1,
684 sub {
685 libsvn_graft_file_copies($grafts, $tree_paths,
686 $path, @_);
687 });
688 last if ($max >= $head);
689 $min = $max + 1;
690 $max += $inc;
691 $max = $head if ($max > $head);
692 }
693 $SVN::Error::handler = $eh;
694}
695sub process_merge_msg_matches {
697 my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
698 my (@strong, @weak);
699 foreach (@matches) {
700 # merging with ourselves is not interesting
701 next if $_ eq $p;
702 if ($l_map->{$u}->{$_}) {
703 push @strong, $_;
704 } else {
705 push @weak, $_;
706 }
707 }
708 foreach my $w (@weak) {
709 last if @strong;
710 # no exact match, use branch name as regexp.
711 my $re = qr/\Q$w\E/i;
712 foreach (keys %{$l_map->{$u}}) {
713 if (/$re/) {
714 push @strong, $l_map->{$u}->{$_};
715 last;
716 }
717 }
718 last if @strong;
719 $w = basename($w);
720 $re = qr/\Q$w\E/i;
721 foreach (keys %{$l_map->{$u}}) {
722 if (/$re/) {
723 push @strong, $l_map->{$u}->{$_};
724 last;
725 }
726 }
727 }
728 my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
729 \s(?:[a-f\d\-]+)$/xsm);
730 unless (defined $rev) {
731 ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
732 \@(?:[a-f\d\-]+)/xsm);
733 return unless defined $rev;
734 }
735 foreach my $m (@strong) {
736 my ($r0, $s0) = find_rev_before($rev, $m, 1);
737 $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
738 }
739}
740sub graft_merge_msg {
742 my ($grafts, $l_map, $u, $p, @re) = @_;
743 my $x = $l_map->{$u}->{$p};
745 my $rl = rev_list_raw("refs/remotes/$x");
746 while (my $c = next_rev_list_entry($rl)) {
747 foreach my $re (@re) {
748 my (@br) = ($c->{m} =~ /$re/g);
749 next unless @br;
750 process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
751 }
752 }
753}
754sub verify_ref {
756 my ($ref) = @_;
757 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
758 { STDERR => 0 }); };
759}
760sub repo_path_split {
762 my $full_url = shift;
763 $full_url =~ s#/+$##;
764 foreach (@repo_path_split_cache) {
766 if ($full_url =~ s#$_##) {
767 my $u = $1;
768 $full_url =~ s#^/+##;
769 return ($u, $full_url);
770 }
771 }
772 my $tmp = Git::SVN::Ra->new($full_url);
773 return ($tmp->{repos_root}, $tmp->{svn_path});
774}
775sub setup_git_svn {
777 defined $SVN_URL or croak "SVN repository location required\n";
778 unless (-d $GIT_DIR) {
779 croak "GIT_DIR=$GIT_DIR does not exist!\n";
780 }
781 mkpath([$GIT_SVN_DIR]);
782 mkpath(["$GIT_SVN_DIR/info"]);
783 open my $fh, '>>',$REVDB or croak $!;
784 close $fh;
785 s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
786}
788sub get_tree_from_treeish {
790 my ($treeish) = @_;
791 # $treeish can be a symbolic ref, too:
792 my $type = command_oneline(qw/cat-file -t/, $treeish);
793 my $expected;
794 while ($type eq 'tag') {
795 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
796 }
797 if ($type eq 'commit') {
798 $expected = (grep /^tree /, command(qw/cat-file commit/,
799 $treeish))[0];
800 ($expected) = ($expected =~ /^tree ($sha1)$/o);
801 die "Unable to get tree from $treeish\n" unless $expected;
802 } elsif ($type eq 'tree') {
803 $expected = $treeish;
804 } else {
805 die "$treeish is a $type, expected tree, tag or commit\n";
806 }
807 return $expected;
808}
809sub get_diff {
811 my ($from, $treeish) = @_;
812 print "diff-tree $from $treeish\n";
813 my @diff_tree = qw(diff-tree -z -r);
814 if ($_cp_similarity) {
815 push @diff_tree, "-C$_cp_similarity";
816 } else {
817 push @diff_tree, '-C';
818 }
819 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
820 push @diff_tree, "-l$_l" if defined $_l;
821 push @diff_tree, $from, $treeish;
822 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
823 local $/ = "\0";
824 my $state = 'meta';
825 my @mods;
826 while (<$diff_fh>) {
827 chomp $_; # this gets rid of the trailing "\0"
828 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
829 $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
830 push @mods, { mode_a => $1, mode_b => $2,
831 sha1_b => $3, chg => $4 };
832 if ($4 =~ /^(?:C|R)$/) {
833 $state = 'file_a';
834 } else {
835 $state = 'file_b';
836 }
837 } elsif ($state eq 'file_a') {
838 my $x = $mods[$#mods] or croak "Empty array\n";
839 if ($x->{chg} !~ /^(?:C|R)$/) {
840 croak "Error parsing $_, $x->{chg}\n";
841 }
842 $x->{file_a} = $_;
843 $state = 'file_b';
844 } elsif ($state eq 'file_b') {
845 my $x = $mods[$#mods] or croak "Empty array\n";
846 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
847 croak "Error parsing $_, $x->{chg}\n";
848 }
849 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
850 croak "Error parsing $_, $x->{chg}\n";
851 }
852 $x->{file_b} = $_;
853 $state = 'meta';
854 } else {
855 croak "Error parsing $_\n";
856 }
857 }
858 command_close_pipe($diff_fh, $ctx);
859 return \@mods;
860}
861sub get_commit_entry {
863 my ($treeish) = shift;
864 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
865 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
866 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
867 open my $log_fh, '>', $commit_editmsg or croak $!;
868 my $type = command_oneline(qw/cat-file -t/, $treeish);
870 if ($type eq 'commit' || $type eq 'tag') {
871 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
872 $type, $treeish);
873 my $in_msg = 0;
874 while (<$msg_fh>) {
875 if (!$in_msg) {
876 $in_msg = 1 if (/^\s*$/);
877 } elsif (/^git-svn-id: /) {
878 # skip this for now, we regenerate the
879 # correct one on re-fetch anyways
880 # TODO: set *:merge properties or like...
881 } else {
882 print $log_fh $_ or croak $!;
883 }
884 }
885 command_close_pipe($msg_fh, $ctx);
886 }
887 close $log_fh or croak $!;
888 if ($_edit || ($type eq 'tree')) {
890 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
891 # TODO: strip out spaces, comments, like git-commit.sh
892 system($editor, $commit_editmsg);
893 }
894 rename $commit_editmsg, $commit_msg or croak $!;
895 open $log_fh, '<', $commit_msg or croak $!;
896 { local $/; chomp($log_entry{log} = <$log_fh>); }
897 close $log_fh or croak $!;
898 unlink $commit_msg;
899 \%log_entry;
900}
901sub rev_list_raw {
903 my ($fh, $c) = command_output_pipe(qw/rev-list --pretty=raw/, @_);
904 return { fh => $fh, ctx => $c, t => { } };
905}
906sub next_rev_list_entry {
908 my $rl = shift;
909 my $fh = $rl->{fh};
910 my $x = $rl->{t};
911 while (<$fh>) {
912 if (/^commit ($sha1)$/o) {
913 if ($x->{c}) {
914 $rl->{t} = { c => $1 };
915 return $x;
916 } else {
917 $x->{c} = $1;
918 }
919 } elsif (/^parent ($sha1)$/o) {
920 $x->{p}->{$1} = 1;
921 } elsif (s/^ //) {
922 $x->{m} ||= '';
923 $x->{m} .= $_;
924 }
925 }
926 command_close_pipe($fh, $rl->{ctx});
927 return ($x != $rl->{t}) ? $x : undef;
928}
929sub s_to_file {
931 my ($str, $file, $mode) = @_;
932 open my $fd,'>',$file or croak $!;
933 print $fd $str,"\n" or croak $!;
934 close $fd or croak $!;
935 chmod ($mode &~ umask, $file) if (defined $mode);
936}
937sub file_to_s {
939 my $file = shift;
940 open my $fd,'<',$file or croak "$!: file: $file\n";
941 local $/;
942 my $ret = <$fd>;
943 close $fd or croak $!;
944 $ret =~ s/\s*$//s;
945 return $ret;
946}
947sub check_upgrade_needed {
949 if (!-r $REVDB) {
950 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
951 open my $fh, '>>',$REVDB or croak $!;
952 close $fh;
953 }
954 return unless eval {
955 command([qw/rev-parse --verify/,"$GIT_SVN-HEAD^0"],
956 {STDERR => 0});
957 };
958 my $head = eval { command('rev-parse',"refs/remotes/$GIT_SVN") };
959 if ($@ || !$head) {
960 print STDERR "Please run: $0 rebuild --upgrade\n";
961 exit 1;
962 }
963}
964# fills %tree_map with a reverse mapping of trees to commits. Useful
966# for finding parents to commit on.
967sub map_tree_joins {
968 my %seen;
969 foreach my $br (@_branch_from) {
970 my $pipe = command_output_pipe(qw/rev-list
971 --topo-order --pretty=raw/, $br);
972 while (<$pipe>) {
973 if (/^commit ($sha1)$/o) {
974 my $commit = $1;
975 # if we've seen a commit,
977 # we've seen its parents
978 last if $seen{$commit};
979 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
980 unless (defined $tree) {
981 die "Failed to parse commit $commit\n";
982 }
983 push @{$tree_map{$tree}}, $commit;
984 $seen{$commit} = 1;
985 }
986 }
987 close $pipe;
988 }
989}
990sub load_all_refs {
992 if (@_branch_from) {
993 print STDERR '--branch|-b parameters are ignored when ',
994 "--branch-all-refs|-B is passed\n";
995 }
996 # don't worry about rev-list on non-commit objects/tags,
998 # it shouldn't blow up if a ref is a blob or tree...
999 @_branch_from = command(qw/rev-parse --symbolic --all/);
1000}
1001# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1003sub load_authors {
1004 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1005 my $log = $cmd eq 'log';
1006 while (<$authors>) {
1007 chomp;
1008 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1009 my ($user, $name, $email) = ($1, $2, $3);
1010 if ($log) {
1011 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1012 } else {
1013 $users{$user} = [$name, $email];
1014 }
1015 }
1016 close $authors or croak $!;
1017}
1018sub git_svn_each {
1020 my $sub = shift;
1021 foreach (command(qw/rev-parse --symbolic --all/)) {
1022 next unless s#^refs/remotes/##;
1023 chomp $_;
1024 next unless -f "$GIT_DIR/svn/$_/info/url";
1025 &$sub($_);
1026 }
1027}
1028sub migrate_revdb {
1030 git_svn_each(sub {
1031 my $id = shift;
1032 defined(my $pid = fork) or croak $!;
1033 if (!$pid) {
1034 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1035 init_vars();
1036 exit 0 if -r $REVDB;
1037 print "Upgrading svn => git mapping...\n";
1038 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1039 open my $fh, '>>',$REVDB or croak $!;
1040 close $fh;
1041 rebuild();
1042 print "Done upgrading. You may now delete the ",
1043 "deprecated $GIT_SVN_DIR/revs directory\n";
1044 exit 0;
1045 }
1046 waitpid $pid, 0;
1047 croak $? if $?;
1048 });
1049}
1050sub migration_check {
1052 migrate_revdb() unless (-e $REVDB);
1053 return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
1054 print "Upgrading repository...\n";
1055 unless (-d "$GIT_DIR/svn") {
1056 mkdir "$GIT_DIR/svn" or croak $!;
1057 }
1058 print "Data from a previous version of git-svn exists, but\n\t",
1059 "$GIT_SVN_DIR\n\t(required for this version ",
1060 "($VERSION) of git-svn) does not.\n";
1061 foreach my $x (command(qw/rev-parse --symbolic --all/)) {
1063 next unless $x =~ s#^refs/remotes/##;
1064 chomp $x;
1065 next unless -f "$GIT_DIR/$x/info/url";
1066 my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
1067 next unless $u;
1068 my $dn = dirname("$GIT_DIR/svn/$x");
1069 mkpath([$dn]) unless -d $dn;
1070 rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
1071 }
1072 migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
1073 print "Done upgrading.\n";
1074}
1075sub find_rev_before {
1077 my ($r, $id, $eq_ok) = @_;
1078 my $f = "$GIT_DIR/svn/$id/.rev_db";
1079 return (undef,undef) unless -r $f;
1080 --$r unless $eq_ok;
1081 while ($r > 0) {
1082 if (my $c = revdb_get($f, $r)) {
1083 return ($r, $c);
1084 }
1085 --$r;
1086 }
1087 return (undef, undef);
1088}
1089sub init_vars {
1091 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
1092 $Git::SVN::default = $GIT_SVN;
1093 $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
1094 $REVDB = "$GIT_SVN_DIR/.rev_db";
1095 $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
1096 $SVN_URL = undef;
1097 %tree_map = ();
1098}
1099# convert GetOpt::Long specs for use by git-config
1101sub read_repo_config {
1102 return unless -d $GIT_DIR;
1103 my $opts = shift;
1104 foreach my $o (keys %$opts) {
1105 my $v = $opts->{$o};
1106 my ($key) = ($o =~ /^([a-z\-]+)/);
1107 $key =~ s/-//g;
1108 my $arg = 'git-config';
1109 $arg .= ' --int' if ($o =~ /[:=]i$/);
1110 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1111 if (ref $v eq 'ARRAY') {
1112 chomp(my @tmp = `$arg --get-all svn.$key`);
1113 @$v = @tmp if @tmp;
1114 } else {
1115 chomp(my $tmp = `$arg --get svn.$key`);
1116 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1117 $$v = $tmp;
1118 }
1119 }
1120 }
1121}
1122sub set_default_vals {
1124 if (defined $_repack) {
1125 $_repack = 1000 if ($_repack <= 0);
1126 $_repack_nr = $_repack;
1127 $_repack_flags ||= '-d';
1128 }
1129}
1130sub read_grafts {
1132 my $gr_file = shift;
1133 my ($grafts, $comments) = ({}, {});
1134 if (open my $fh, '<', $gr_file) {
1135 my @tmp;
1136 while (<$fh>) {
1137 if (/^($sha1)\s+/) {
1138 my $c = $1;
1139 if (@tmp) {
1140 @{$comments->{$c}} = @tmp;
1141 @tmp = ();
1142 }
1143 foreach my $p (split /\s+/, $_) {
1144 $grafts->{$c}->{$p} = 1;
1145 }
1146 } else {
1147 push @tmp, $_;
1148 }
1149 }
1150 close $fh or croak $!;
1151 @{$comments->{'END'}} = @tmp if @tmp;
1152 }
1153 return ($grafts, $comments);
1154}
1155sub write_grafts {
1157 my ($grafts, $comments, $gr_file) = @_;
1158 open my $fh, '>', $gr_file or croak $!;
1160 foreach my $c (sort keys %$grafts) {
1161 if ($comments->{$c}) {
1162 print $fh $_ foreach @{$comments->{$c}};
1163 }
1164 my $p = $grafts->{$c};
1165 my %x; # real parents
1166 delete $p->{$c}; # commits are not self-reproducing...
1167 my $ch = command_output_pipe(qw/cat-file commit/, $c);
1168 while (<$ch>) {
1169 if (/^parent ($sha1)/) {
1170 $x{$1} = $p->{$1} = 1;
1171 } else {
1172 last unless /^\S/;
1173 }
1174 }
1175 close $ch; # breaking the pipe
1176 # if real parents are the only ones in the grafts, drop it
1178 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1179 my (@ip, @jp, $mb);
1181 my %del = %x;
1182 @ip = @jp = keys %$p;
1183 foreach my $i (@ip) {
1184 next if $del{$i} || $p->{$i} == 2;
1185 foreach my $j (@jp) {
1186 next if $i eq $j || $del{$j} || $p->{$j} == 2;
1187 $mb = eval { command('merge-base', $i, $j) };
1188 next unless $mb;
1189 chomp $mb;
1190 next if $x{$mb};
1191 if ($mb eq $j) {
1192 delete $p->{$i};
1193 $del{$i} = 1;
1194 } elsif ($mb eq $i) {
1195 delete $p->{$j};
1196 $del{$j} = 1;
1197 }
1198 }
1199 }
1200 # if real parents are the only ones in the grafts, drop it
1202 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
1203 print $fh $c, ' ', join(' ', sort keys %$p),"\n";
1205 }
1206 if ($comments->{'END'}) {
1207 print $fh $_ foreach @{$comments->{'END'}};
1208 }
1209 close $fh or croak $!;
1210}
1211sub read_url_paths_all {
1213 my ($l_map, $pfx, $p) = @_;
1214 my @dir;
1215 foreach (<$p/*>) {
1216 if (-r "$_/info/url") {
1217 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1218 my $id = $pfx . basename $_;
1219 my $url = file_to_s("$_/info/url");
1220 my ($u, $p) = repo_path_split($url);
1221 $l_map->{$u}->{$p} = $id;
1222 } elsif (-d $_) {
1223 push @dir, $_;
1224 }
1225 }
1226 foreach (@dir) {
1227 my $x = $_;
1228 $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
1229 read_url_paths_all($l_map, $x, $_);
1230 }
1231}
1232# this one only gets ids that have been imported, not new ones
1234sub read_url_paths {
1235 my $l_map = {};
1236 git_svn_each(sub { my $x = shift;
1237 my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
1238 my ($u, $p) = repo_path_split($url);
1239 $l_map->{$u}->{$p} = $x;
1240 });
1241 return $l_map;
1242}
1243sub extract_metadata {
1245 my $id = shift or return (undef, undef, undef);
1246 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
1247 \s([a-f\d\-]+)$/x);
1248 if (!defined $rev || !$uuid || !$url) {
1249 # some of the original repositories I made had
1250 # identifiers like this:
1251 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1252 }
1253 return ($url, $rev, $uuid);
1254}
1255sub cmt_metadata {
1257 return extract_metadata((grep(/^git-svn-id: /,
1258 command(qw/cat-file commit/, shift)))[-1]);
1259}
1260sub get_commit_time {
1262 my $cmt = shift;
1263 my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
1264 while (<$fh>) {
1265 /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
1266 my ($s, $tz) = ($1, $2);
1267 if ($tz =~ s/^\+//) {
1268 $s += tz_to_s_offset($tz);
1269 } elsif ($tz =~ s/^\-//) {
1270 $s -= tz_to_s_offset($tz);
1271 }
1272 close $fh;
1273 return $s;
1274 }
1275 die "Can't get commit time for commit: $cmt\n";
1276}
1277sub tz_to_s_offset {
1279 my ($tz) = @_;
1280 $tz =~ s/(\d\d)$//;
1281 return ($1 * 60) + ($tz * 3600);
1282}
1283package Git::SVN;
1285use strict;
1286use warnings;
1287use vars qw/$default/;
1288use Carp qw/croak/;
1289use File::Path qw/mkpath/;
1290use IPC::Open3;
1291# properties that we do not log:
1293my %SKIP_PROP;
1294BEGIN {
1295 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1296 svn:special svn:executable
1297 svn:entry:committed-rev
1298 svn:entry:last-author
1299 svn:entry:uuid
1300 svn:entry:committed-date/;
1301}
1302sub init {
1304 my ($class, $id, $url) = @_;
1305 my $self = _new($class, $id);
1306 mkpath(["$self->{dir}/info"]);
1307 if (defined $url) {
1308 $url =~ s!/+$!!; # strip trailing slash
1309 ::s_to_file($url, "$self->{dir}/info/url");
1310 }
1311 $self->{url} = $url;
1312 open my $fh, '>>', $self->{db_path} or croak $!;
1313 close $fh or croak $!;
1314 $self;
1315}
1316sub new {
1318 my ($class, $id) = @_;
1319 my $self = _new($class, $id);
1320 $self->{url} = ::file_to_s("$self->{dir}/info/url");
1321 $self;
1322}
1323sub refname { "refs/remotes/$_[0]->{id}" }
1325sub ra {
1327 my ($self) = shift;
1328 $self->{ra} ||= Git::SVN::Ra->new($self->{url});
1329}
1330sub copy_remote_ref {
1332 my ($self) = @_;
1333 my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
1334 my $ref = $self->refname;
1335 if (command('ls-remote', $origin, $ref)) {
1336 command_noisy('fetch', $origin, "$ref:$ref");
1337 } elsif ($::_cp_remote && !$::_upgrade) {
1338 die "Unable to find remote reference: $ref on $origin\n";
1339 }
1340}
1341sub traverse_ignore {
1343 my ($self, $fh, $path, $r) = @_;
1344 $path =~ s#^/+##g;
1345 my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
1346 my $p = $path;
1347 $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
1348 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1349 if (my $s = $props->{'svn:ignore'}) {
1350 $s =~ s/[\r\n]+/\n/g;
1351 chomp $s;
1352 if (length $p == 0) {
1353 $s =~ s#\n#\n/$p#g;
1354 print $fh "/$s\n";
1355 } else {
1356 $s =~ s#\n#\n/$p/#g;
1357 print $fh "/$p/$s\n";
1358 }
1359 }
1360 foreach (sort keys %$dirent) {
1361 next if $dirent->{$_}->kind != $SVN::Node::dir;
1362 $self->traverse_ignore($fh, "$path/$_", $r);
1363 }
1364}
1365# returns the newest SVN revision number and newest commit SHA1
1367sub last_rev_commit {
1368 my ($self) = @_;
1369 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1370 return ($self->{last_rev}, $self->{last_commit});
1371 }
1372 my $c = ::verify_ref($self->refname.'^0');
1373 if (defined $c && length $c) {
1374 my $rev = (::cmt_metadata($c))[1];
1375 if (defined $rev) {
1376 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1377 return ($rev, $c);
1378 }
1379 }
1380 my $offset = -41; # from tail
1381 my $rl;
1382 open my $fh, '<', $self->{db_path} or
1383 croak "$self->{db_path} not readable: $!\n";
1384 seek $fh, $offset, 2;
1385 $rl = readline $fh;
1386 defined $rl or return (undef, undef);
1387 chomp $rl;
1388 while ($c ne $rl && tell $fh != 0) {
1389 $offset -= 41;
1390 seek $fh, $offset, 2;
1391 $rl = readline $fh;
1392 defined $rl or return (undef, undef);
1393 chomp $rl;
1394 }
1395 my $rev = tell $fh;
1396 croak $! if ($rev < 0);
1397 $rev = ($rev - 41) / 41;
1398 close $fh or croak $!;
1399 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1400 return ($rev, $c);
1401}
1402sub parse_revision {
1404 my ($self, $base) = @_;
1405 my $head = $self->ra->get_latest_revnum;
1406 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1407 return ($base + 1, $head) if (defined $base);
1408 return (0, $head);
1409 }
1410 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1411 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1412 if ($::_revision =~ /^BASE:(\d+)$/) {
1413 return ($base + 1, $1) if (defined $base);
1414 return (0, $head);
1415 }
1416 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1417 die "revision argument: $::_revision not understood by git-svn\n",
1418 "Try using the command-line svn client instead\n";
1419}
1420sub tmp_index_do {
1422 my ($self, $sub) = @_;
1423 my $old_index = $ENV{GIT_INDEX_FILE};
1424 $ENV{GIT_INDEX_FILE} = $self->{index};
1425 my @ret = &$sub;
1426 if ($old_index) {
1427 $ENV{GIT_INDEX_FILE} = $old_index;
1428 } else {
1429 delete $ENV{GIT_INDEX_FILE};
1430 }
1431 wantarray ? @ret : $ret[0];
1432}
1433sub assert_index_clean {
1435 my ($self, $treeish) = @_;
1436 $self->tmp_index_do(sub {
1438 command_noisy('read-tree', $treeish) unless -e $self->{index};
1439 my $x = command_oneline('write-tree');
1440 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1441 /^tree ($::sha1)/mo);
1442 if ($y ne $x) {
1443 unlink $self->{index} or croak $!;
1444 command_noisy('read-tree', $treeish);
1445 }
1446 $x = command_oneline('write-tree');
1447 if ($y ne $x) {
1448 ::fatal "trees ($treeish) $y != $x\n",
1449 "Something is seriously wrong...\n";
1450 }
1451 });
1452}
1453sub get_commit_parents {
1455 my ($self, $log_entry, @parents) = @_;
1456 my (%seen, @ret, @tmp);
1457 # commit parents can be conditionally bound to a particular
1458 # svn revision via: "svn_revno=commit_sha1", filter them out here:
1459 foreach my $p (@parents) {
1460 next unless defined $p;
1461 if ($p =~ /^(\d+)=($::sha1_short)$/o) {
1462 push @tmp, $2 if $1 == $log_entry->{revision};
1463 } else {
1464 push @tmp, $p if $p =~ /^$::sha1_short$/o;
1465 }
1466 }
1467 if (my $cur = ::verify_ref($self->refname.'^0')) {
1468 push @tmp, $cur;
1469 }
1470 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1471 while (my $p = shift @tmp) {
1472 next if $seen{$p};
1473 $seen{$p} = 1;
1474 push @ret, $p;
1475 # MAXPARENT is defined to 16 in commit-tree.c:
1476 last if @ret >= 16;
1477 }
1478 if (@tmp) {
1479 die "r$log_entry->{revision}: No room for parents:\n\t",
1480 join("\n\t", @tmp), "\n";
1481 }
1482 @ret;
1483}
1484sub check_upgrade_needed {
1486 my ($self) = @_;
1487 if (!-r $self->{db_path}) {
1488 -d $self->{dir} or mkpath([$self->{dir}]);
1489 open my $fh, '>>', $self->{db_path} or croak $!;
1490 close $fh;
1491 }
1492 return unless ::verify_ref($self->{id}.'-HEAD^0');
1493 my $head = ::verify_ref($self->refname.'^0');
1494 if ($@ || !$head) {
1495 ::fatal("Please run: $0 rebuild --upgrade\n");
1496 }
1497}
1498sub do_git_commit {
1500 my ($self, $log_entry, @parents) = @_;
1501 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1502 croak "$log_entry->{revision} = $c already exists! ",
1503 "Why are we refetching it?\n";
1504 }
1505 my $author = $log_entry->{author};
1506 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1507 : ($author, "$author\@".$self->ra->uuid));
1508 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1509 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1510 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1511 my $tree = $log_entry->{tree};
1513 if (!defined $tree) {
1514 $tree = $self->tmp_index_do(sub {
1515 command_oneline('write-tree') });
1516 }
1517 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1518 my @exec = ('git-commit-tree', $tree);
1520 foreach ($self->get_commit_parents($log_entry, @parents)) {
1521 push @exec, '-p', $_;
1522 }
1523 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1524 or croak $!;
1525 print $msg_fh $log_entry->{log} or croak $!;
1526 print $msg_fh "\ngit-svn-id: ", $self->ra->{url}, '@',
1527 $log_entry->{revision}, ' ',
1528 $self->ra->uuid, "\n" or croak $!;
1529 $msg_fh->flush == 0 or croak $!;
1530 close $msg_fh or croak $!;
1531 chomp(my $commit = do { local $/; <$out_fh> });
1532 close $out_fh or croak $!;
1533 waitpid $pid, 0;
1534 croak $? if $?;
1535 if ($commit !~ /^$::sha1$/o) {
1536 die "Failed to commit, invalid sha1: $commit\n";
1537 }
1538 command_noisy('update-ref',$self->refname, $commit);
1540 $self->rev_db_set($log_entry->{revision}, $commit);
1541 $self->{last_rev} = $log_entry->{revision};
1543 $self->{last_commit} = $commit;
1544 print "r$log_entry->{revision} = $commit\n";
1545 return $commit;
1546}
1547sub do_fetch {
1549 my ($self, $paths, $rev) = @_; #, $author, $date, $log) = @_;
1550 my $ed = SVN::Git::Fetcher->new($self);
1551 my ($last_rev, @parents);
1552 if ($self->{last_commit}) {
1553 $last_rev = $self->{last_rev};
1554 $ed->{c} = $self->{last_commit};
1555 @parents = ($self->{last_commit});
1556 } else {
1557 $last_rev = $rev;
1558 }
1559 unless ($self->ra->gs_do_update($last_rev, $rev, '', 1, $ed)) {
1560 die "SVN connection failed somewhere...\n";
1561 }
1562 $self->make_log_entry($rev, \@parents, $ed);
1563}
1564sub write_untracked {
1566 my ($self, $rev, $fh, $untracked) = @_;
1567 my $h;
1568 print $fh "r$rev\n" or croak $!;
1569 $h = $untracked->{empty};
1570 foreach (sort keys %$h) {
1571 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1572 print $fh " $act: ", uri_encode($_), "\n" or croak $!;
1573 warn "W: $act: $_\n";
1574 }
1575 foreach my $t (qw/dir_prop file_prop/) {
1576 $h = $untracked->{$t} or next;
1577 foreach my $path (sort keys %$h) {
1578 my $ppath = $path eq '' ? '.' : $path;
1579 foreach my $prop (sort keys %{$h->{$path}}) {
1580 next if $SKIP_PROP{$prop};
1581 my $v = $h->{$path}->{$prop};
1582 if (defined $v) {
1583 print $fh " +$t: ",
1584 uri_encode($ppath), ' ',
1585 uri_encode($prop), ' ',
1586 uri_encode($v), "\n"
1587 or croak $!;
1588 } else {
1589 print $fh " -$t: ",
1590 uri_encode($ppath), ' ',
1591 uri_encode($prop), "\n"
1592 or croak $!;
1593 }
1594 }
1595 }
1596 }
1597 foreach my $t (qw/absent_file absent_directory/) {
1598 $h = $untracked->{$t} or next;
1599 foreach my $parent (sort keys %$h) {
1600 foreach my $path (sort @{$h->{$parent}}) {
1601 print $fh " $t: ",
1602 uri_encode("$parent/$path"), "\n"
1603 or croak $!;
1604 warn "W: $t: $parent/$path ",
1605 "Insufficient permissions?\n";
1606 }
1607 }
1608 }
1609}
1610sub parse_svn_date {
1612 my $date = shift || return '+0000 1970-01-01 00:00:00';
1613 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1614 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1615 croak "Unable to parse date: $date\n";
1616 "+0000 $Y-$m-$d $H:$M:$S";
1617}
1618sub check_author {
1620 my ($author) = @_;
1621 if (!defined $author || length $author == 0) {
1622 $author = '(no author)';
1623 }
1624 if (defined $::_authors && ! defined $::users{$author}) {
1625 die "Author: $author not defined in $::_authors file\n";
1626 }
1627 $author;
1628}
1629sub make_log_entry {
1631 my ($self, $rev, $parents, $untracked) = @_;
1632 my $rp = $self->ra->rev_proplist($rev);
1633 my %log_entry = ( parents => $parents || [], revision => $rev,
1634 revprops => $rp, log => '');
1635 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1636 $self->write_untracked($rev, $un, $untracked);
1637 foreach (sort keys %$rp) {
1638 my $v = $rp->{$_};
1639 if (/^svn:(author|date|log)$/) {
1640 $log_entry{$1} = $v;
1641 } else {
1642 print $un " rev_prop: ", uri_encode($_), ' ',
1643 uri_encode($v), "\n";
1644 }
1645 }
1646 close $un or croak $!;
1647 $log_entry{date} = parse_svn_date($log_entry{date});
1648 $log_entry{author} = check_author($log_entry{author});
1649 $log_entry{log} .= "\n";
1650 \%log_entry;
1651}
1652sub fetch {
1654 my ($self, @parents) = @_;
1655 my ($last_rev, $last_commit) = $self->last_rev_commit;
1656 my ($base, $head) = $self->parse_revision($last_rev);
1657 return if ($base > $head);
1658 if (defined $last_commit) {
1659 $self->assert_index_clean($last_commit);
1660 }
1661 my $inc = 1000;
1662 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1663 my $err_handler = $SVN::Error::handler;
1664 $SVN::Error::handler = \&skip_unknown_revs;
1665 while (1) {
1666 my @revs;
1667 $self->ra->get_log([''], $min, $max, 0, 1, 1, sub {
1668 my ($paths, $rev, $author, $date, $log) = @_;
1669 push @revs, $rev });
1670 foreach (@revs) {
1671 my $log_entry = $self->do_fetch(undef, $_);
1672 $self->do_git_commit($log_entry, @parents);
1673 }
1674 last if $max >= $head;
1675 $min = $max + 1;
1676 $max += $inc;
1677 $max = $head if ($max > $head);
1678 }
1679 $SVN::Error::handler = $err_handler;
1680}
1681sub set_tree_cb {
1683 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1684 # TODO: enable and test optimized commits:
1685 if (0 && $rev == ($self->{last_rev} + 1)) {
1686 $log_entry->{revision} = $rev;
1687 $log_entry->{author} = $author;
1688 $self->do_git_commit($log_entry, "$rev=$tree");
1689 } else {
1690 $self->fetch("$rev=$tree");
1691 }
1692}
1693sub set_tree {
1695 my ($self, $tree) = (shift, shift);
1696 my $log_entry = ::get_commit_entry($tree);
1697 unless ($self->{last_rev}) {
1698 fatal("Must have an existing revision to commit\n");
1699 }
1700 my $pool = SVN::Pool->new;
1701 my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1702 ra => $self->ra->dup,
1703 svn_path => $self->ra->{svn_path}
1704 },
1705 $self->ra->get_commit_editor(
1706 $log_entry->{log}, sub {
1707 $self->set_tree_cb($log_entry,
1708 $tree, @_);
1709 }),
1710 $pool);
1711 my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1712 if (@$mods == 0) {
1713 print "No changes\nr$self->{last_rev} = $tree\n";
1714 }
1715 $pool->clear;
1716}
1717sub skip_unknown_revs {
1719 my ($err) = @_;
1720 my $errno = $err->apr_err();
1721 # Maybe the branch we're tracking didn't
1722 # exist when the repo started, so it's
1723 # not an error if it doesn't, just continue
1724 #
1725 # Wonderfully consistent library, eh?
1726 # 160013 - svn:// and file://
1727 # 175002 - http(s)://
1728 # 175007 - http(s):// (this repo required authorization, too...)
1729 # More codes may be discovered later...
1730 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1731 return;
1732 }
1733 croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1734}
1735# rev_db:
1737# Tie::File seems to be prone to offset errors if revisions get sparse,
1738# it's not that fast, either. Tie::File is also not in Perl 5.6. So
1739# one of my favorite modules is out :< Next up would be one of the DBM
1740# modules, but I'm not sure which is most portable... So I'll just
1741# go with something that's plain-text, but still capable of
1742# being randomly accessed. So here's my ultra-simple fixed-width
1743# database. All records are 40 characters + "\n", so it's easy to seek
1744# to a revision: (41 * rev) is the byte offset.
1745# A record of 40 0s denotes an empty revision.
1746# And yes, it's still pretty fast (faster than Tie::File).
1747sub rev_db_set {
1749 my ($self, $rev, $commit) = @_;
1750 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1751 open my $fh, '+<', $self->{db_path} or croak $!;
1752 my $offset = $rev * 41;
1753 # assume that append is the common case:
1754 seek $fh, 0, 2 or croak $!;
1755 my $pos = tell $fh;
1756 if ($pos < $offset) {
1757 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1758 or croak $!;
1759 }
1760 seek $fh, $offset, 0 or croak $!;
1761 print $fh $commit,"\n" or croak $!;
1762 close $fh or croak $!;
1763}
1764sub rev_db_get {
1766 my ($self, $rev) = @_;
1767 my $ret;
1768 my $offset = $rev * 41;
1769 open my $fh, '<', $self->{db_path} or croak $!;
1770 if (seek $fh, $offset, 0) {
1771 $ret = readline $fh;
1772 if (defined $ret) {
1773 chomp $ret;
1774 $ret = undef if ($ret =~ /^0{40}$/);
1775 }
1776 }
1777 close $fh or croak $!;
1778 $ret;
1779}
1780sub _new {
1782 my ($class, $id) = @_;
1783 $id ||= $Git::SVN::default;
1784 my $dir = "$ENV{GIT_DIR}/svn/$id";
1785 bless { id => $id, dir => $dir, index => "$dir/index",
1786 db_path => "$dir/.rev_db" }, $class;
1787}
1788sub uri_encode {
1790 my ($f) = @_;
1791 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1792 $f
1793}
1794package Git::SVN::Prompt;
1796use strict;
1797use warnings;
1798require SVN::Core;
1799use vars qw/$_no_auth_cache $_username/;
1800sub simple {
1802 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1803 $may_save = undef if $_no_auth_cache;
1804 $default_username = $_username if defined $_username;
1805 if (defined $default_username && length $default_username) {
1806 if (defined $realm && length $realm) {
1807 print STDERR "Authentication realm: $realm\n";
1808 STDERR->flush;
1809 }
1810 $cred->username($default_username);
1811 } else {
1812 username($cred, $realm, $may_save, $pool);
1813 }
1814 $cred->password(_read_password("Password for '" .
1815 $cred->username . "': ", $realm));
1816 $cred->may_save($may_save);
1817 $SVN::_Core::SVN_NO_ERROR;
1818}
1819sub ssl_server_trust {
1821 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1822 $may_save = undef if $_no_auth_cache;
1823 print STDERR "Error validating server certificate for '$realm':\n";
1824 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1825 print STDERR " - The certificate is not issued by a trusted ",
1826 "authority. Use the\n",
1827 " fingerprint to validate the certificate manually!\n";
1828 }
1829 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1830 print STDERR " - The certificate hostname does not match.\n";
1831 }
1832 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1833 print STDERR " - The certificate is not yet valid.\n";
1834 }
1835 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1836 print STDERR " - The certificate has expired.\n";
1837 }
1838 if ($failures & $SVN::Auth::SSL::OTHER) {
1839 print STDERR " - The certificate has an unknown error.\n";
1840 }
1841 printf STDERR
1842 "Certificate information:\n".
1843 " - Hostname: %s\n".
1844 " - Valid: from %s until %s\n".
1845 " - Issuer: %s\n".
1846 " - Fingerprint: %s\n",
1847 map $cert_info->$_, qw(hostname valid_from valid_until
1848 issuer_dname fingerprint);
1849 my $choice;
1850prompt:
1851 print STDERR $may_save ?
1852 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1853 "(R)eject or accept (t)emporarily? ";
1854 STDERR->flush;
1855 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1856 if ($choice =~ /^t$/i) {
1857 $cred->may_save(undef);
1858 } elsif ($choice =~ /^r$/i) {
1859 return -1;
1860 } elsif ($may_save && $choice =~ /^p$/i) {
1861 $cred->may_save($may_save);
1862 } else {
1863 goto prompt;
1864 }
1865 $cred->accepted_failures($failures);
1866 $SVN::_Core::SVN_NO_ERROR;
1867}
1868sub ssl_client_cert {
1870 my ($cred, $realm, $may_save, $pool) = @_;
1871 $may_save = undef if $_no_auth_cache;
1872 print STDERR "Client certificate filename: ";
1873 STDERR->flush;
1874 chomp(my $filename = <STDIN>);
1875 $cred->cert_file($filename);
1876 $cred->may_save($may_save);
1877 $SVN::_Core::SVN_NO_ERROR;
1878}
1879sub ssl_client_cert_pw {
1881 my ($cred, $realm, $may_save, $pool) = @_;
1882 $may_save = undef if $_no_auth_cache;
1883 $cred->password(_read_password("Password: ", $realm));
1884 $cred->may_save($may_save);
1885 $SVN::_Core::SVN_NO_ERROR;
1886}
1887sub username {
1889 my ($cred, $realm, $may_save, $pool) = @_;
1890 $may_save = undef if $_no_auth_cache;
1891 if (defined $realm && length $realm) {
1892 print STDERR "Authentication realm: $realm\n";
1893 }
1894 my $username;
1895 if (defined $_username) {
1896 $username = $_username;
1897 } else {
1898 print STDERR "Username: ";
1899 STDERR->flush;
1900 chomp($username = <STDIN>);
1901 }
1902 $cred->username($username);
1903 $cred->may_save($may_save);
1904 $SVN::_Core::SVN_NO_ERROR;
1905}
1906sub _read_password {
1908 my ($prompt, $realm) = @_;
1909 print STDERR $prompt;
1910 STDERR->flush;
1911 require Term::ReadKey;
1912 Term::ReadKey::ReadMode('noecho');
1913 my $password = '';
1914 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1915 last if $key =~ /[\012\015]/; # \n\r
1916 $password .= $key;
1917 }
1918 Term::ReadKey::ReadMode('restore');
1919 print STDERR "\n";
1920 STDERR->flush;
1921 $password;
1922}
1923package main;
1925sub uri_encode {
1927 my ($f) = @_;
1928 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1929 $f
1930}
1931sub uri_decode {
1933 my ($f) = @_;
1934 $f =~ tr/+/ /;
1935 $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1936 $f
1937}
1938sub libsvn_parse_revision {
1940 my $base = shift;
1941 my $head = $SVN->get_latest_revnum();
1942 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
1943 return ($base + 1, $head) if (defined $base);
1944 return (0, $head);
1945 }
1946 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
1947 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
1948 if ($_revision =~ /^BASE:(\d+)$/) {
1949 return ($base + 1, $1) if (defined $base);
1950 return (0, $head);
1951 }
1952 return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
1953 die "revision argument: $_revision not understood by git-svn\n",
1954 "Try using the command-line svn client instead\n";
1955}
1956sub libsvn_traverse_ignore {
1958 my ($fh, $path, $r) = @_;
1959 $path =~ s#^/+##g;
1960 my ($dirent, undef, $props) = $SVN->get_dir($path, $r);
1961 my $p = $path;
1962 $p =~ s#^\Q$SVN->{svn_path}\E/##;
1963 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1964 if (my $s = $props->{'svn:ignore'}) {
1965 $s =~ s/[\r\n]+/\n/g;
1966 chomp $s;
1967 if (length $p == 0) {
1968 $s =~ s#\n#\n/$p#g;
1969 print $fh "/$s\n";
1970 } else {
1971 $s =~ s#\n#\n/$p/#g;
1972 print $fh "/$p/$s\n";
1973 }
1974 }
1975 foreach (sort keys %$dirent) {
1976 next if $dirent->{$_}->kind != $SVN::Node::dir;
1977 libsvn_traverse_ignore($fh, "$path/$_", $r);
1978 }
1979}
1980sub revisions_eq {
1982 my ($path, $r0, $r1) = @_;
1983 return 1 if $r0 == $r1;
1984 my $nr = 0;
1985 # should be OK to use Pool here (r1 - r0) should be small
1986 $SVN->get_log([$path], $r0, $r1, 0, 0, 1, sub {$nr++});
1987 return 0 if ($nr > 1);
1988 return 1;
1989}
1990sub libsvn_find_parent_branch {
1992 my ($paths, $rev, $author, $date, $log) = @_;
1993 my $svn_path = '/'.$SVN->{svn_path};
1994 # look for a parent from another branch:
1996 my $i = $paths->{$svn_path} or return;
1997 my $branch_from = $i->copyfrom_path or return;
1998 my $r = $i->copyfrom_rev;
1999 print STDERR "Found possible branch point: ",
2000 "$branch_from => $svn_path, $r\n";
2001 $branch_from =~ s#^/##;
2002 my $l_map = {};
2003 read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2004 my $url = $SVN->{repos_root};
2005 defined $l_map->{$url} or return;
2006 my $id = $l_map->{$url}->{$branch_from};
2007 if (!defined $id && $_follow_parent) {
2008 print STDERR "Following parent: $branch_from\@$r\n";
2009 # auto create a new branch and follow it
2010 $id = basename($branch_from);
2011 $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2012 while (-r "$GIT_DIR/svn/$id") {
2013 # just grow a tail if we're not unique enough :x
2014 $id .= '-';
2015 }
2016 }
2017 return unless defined $id;
2018 my ($r0, $parent) = find_rev_before($r,$id,1);
2020 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2021 defined(my $pid = fork) or croak $!;
2022 if (!$pid) {
2023 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2024 init_vars();
2025 $SVN_URL = "$url/$branch_from";
2026 $SVN = undef;
2027 setup_git_svn();
2028 # we can't assume SVN_URL exists at r+1:
2029 $_revision = "0:$r";
2030 fetch_lib();
2031 exit 0;
2032 }
2033 waitpid $pid, 0;
2034 croak $? if $?;
2035 ($r0, $parent) = find_rev_before($r,$id,1);
2036 }
2037 return unless (defined $r0 && defined $parent);
2038 if (revisions_eq($branch_from, $r0, $r)) {
2039 unlink $GIT_SVN_INDEX;
2040 print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2041 command_noisy('read-tree', $parent);
2042 unless ($SVN->can_do_switch) {
2043 return _libsvn_new_tree($paths, $rev, $author, $date,
2044 $log, [$parent]);
2045 }
2046 # do_switch works with svn/trunk >= r22312, but that is not
2047 # included with SVN 1.4.2 (the latest version at the moment),
2048 # so we can't rely on it.
2049 my $ra = Git::SVN::Ra->new("$url/$branch_from");
2050 my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q });
2051 $ra->gs_do_switch($r0, $rev, '', 1, $SVN->{url}, $ed) or
2052 die "SVN connection failed somewhere...\n";
2053 return libsvn_log_entry($rev, $author, $date, $log, [$parent]);
2054 }
2055 print STDERR "Nope, branch point not imported or unknown\n";
2056 return undef;
2057}
2058sub _libsvn_new_tree {
2060 my ($paths, $rev, $author, $date, $log, $parents) = @_;
2061 my $ed = SVN::Git::Fetcher->new({q => $_q});
2062 unless ($SVN->gs_do_update($rev, $rev, '', 1, $ed)) {
2063 die "SVN connection failed somewhere...\n";
2064 }
2065 libsvn_log_entry($rev, $author, $date, $log, $parents, $ed);
2066}
2067sub find_graft_path_commit {
2069 my ($tree_paths, $p1, $r1) = @_;
2070 foreach my $x (keys %$tree_paths) {
2071 next unless ($p1 =~ /^\Q$x\E/);
2072 my $i = $tree_paths->{$x};
2073 my ($r0, $parent) = find_rev_before($r1,$i,1);
2074 return $parent if (defined $r0 && $r0 == $r1);
2075 print STDERR "r$r1 of $i not imported\n";
2076 next;
2077 }
2078 return undef;
2079}
2080sub find_graft_path_parents {
2082 my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2083 foreach my $x (keys %$tree_paths) {
2084 next unless ($p0 =~ /^\Q$x\E/);
2085 my $i = $tree_paths->{$x};
2086 my ($r, $parent) = find_rev_before($r0, $i, 1);
2087 if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2088 my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2089 my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2090 next if ($url_a && $url_b && $url_a eq $url_b &&
2091 $uuid_b eq $uuid_a);
2092 $grafts->{$c}->{$parent} = 1;
2093 }
2094 }
2095}
2096sub libsvn_graft_file_copies {
2098 my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2099 foreach (keys %$paths) {
2100 my $i = $paths->{$_};
2101 my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2102 $i->copyfrom_rev);
2103 next unless (defined $p0 && defined $r0);
2104 my $p1 = $_;
2106 $p1 =~ s#^/##;
2107 $p0 =~ s#^/##;
2108 my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2109 next unless $c;
2110 find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2111 }
2112}
2113sub revdb_get {
2115 my ($file, $rev) = @_;
2116 my $ret;
2117 my $offset = $rev * 41;
2118 open my $fh, '<', $file or croak $!;
2119 seek $fh, $offset, 0;
2120 if (tell $fh == $offset) {
2121 $ret = readline $fh;
2122 if (defined $ret) {
2123 chomp $ret;
2124 $ret = undef if ($ret =~ /^0{40}$/);
2125 }
2126 }
2127 close $fh or croak $!;
2128 return $ret;
2129}
2130{
2132 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2133 $SVN::Node::dir.$SVN::Node::unknown.
2134 $SVN::Node::none.$SVN::Node::file.
2135 $SVN::Node::dir.$SVN::Node::unknown.
2136 $SVN::Auth::SSL::CNMISMATCH.
2137 $SVN::Auth::SSL::NOTYETVALID.
2138 $SVN::Auth::SSL::EXPIRED.
2139 $SVN::Auth::SSL::UNKNOWNCA.
2140 $SVN::Auth::SSL::OTHER;
2141}
2142package SVN::Git::Fetcher;
2144use vars qw/@ISA/;
2145use strict;
2146use warnings;
2147use Carp qw/croak/;
2148use IO::File qw//;
2149# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2151sub new {
2152 my ($class, $git_svn) = @_;
2153 my $self = SVN::Delta::Editor->new;
2154 bless $self, $class;
2155 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2156 $self->{empty} = {};
2157 $self->{dir_prop} = {};
2158 $self->{file_prop} = {};
2159 $self->{absent_dir} = {};
2160 $self->{absent_file} = {};
2161 ($self->{gui}, $self->{ctx}) = $git_svn->tmp_index_do(
2162 sub { command_input_pipe(qw/update-index -z --index-info/) } );
2163 require Digest::MD5;
2164 $self;
2165}
2166sub open_root {
2168 { path => '' };
2169}
2170sub open_directory {
2172 my ($self, $path, $pb, $rev) = @_;
2173 { path => $path };
2174}
2175sub delete_entry {
2177 my ($self, $path, $rev, $pb) = @_;
2178 my $gui = $self->{gui};
2179 # remove entire directories.
2181 if (command('ls-tree', $self->{c}, '--', $path) =~ /^040000 tree/) {
2182 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2183 -r --name-only -z/,
2184 $self->{c}, '--', $path);
2185 local $/ = "\0";
2186 while (<$ls>) {
2187 print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2188 print "\tD\t$_\n" unless $self->{q};
2189 }
2190 print "\tD\t$path/\n" unless $self->{q};
2191 command_close_pipe($ls, $ctx);
2192 $self->{empty}->{$path} = 0
2193 } else {
2194 print $gui '0 ',0 x 40,"\t",$path,"\0" or croak $!;
2195 print "\tD\t$path\n" unless $self->{q};
2196 }
2197 undef;
2198}
2199sub open_file {
2201 my ($self, $path, $pb, $rev) = @_;
2202 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--',$path)
2203 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2204 unless (defined $mode && defined $blob) {
2205 die "$path was not found in commit $self->{c} (r$rev)\n";
2206 }
2207 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2208 pool => SVN::Pool->new, action => 'M' };
2209}
2210sub add_file {
2212 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2213 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2214 delete $self->{empty}->{$dir};
2215 { path => $path, mode_a => 100644, mode_b => 100644,
2216 pool => SVN::Pool->new, action => 'A' };
2217}
2218sub add_directory {
2220 my ($self, $path, $cp_path, $cp_rev) = @_;
2221 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2222 delete $self->{empty}->{$dir};
2223 $self->{empty}->{$path} = 1;
2224 { path => $path };
2225}
2226sub change_dir_prop {
2228 my ($self, $db, $prop, $value) = @_;
2229 $self->{dir_prop}->{$db->{path}} ||= {};
2230 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2231 undef;
2232}
2233sub absent_directory {
2235 my ($self, $path, $pb) = @_;
2236 $self->{absent_dir}->{$pb->{path}} ||= [];
2237 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2238 undef;
2239}
2240sub absent_file {
2242 my ($self, $path, $pb) = @_;
2243 $self->{absent_file}->{$pb->{path}} ||= [];
2244 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2245 undef;
2246}
2247sub change_file_prop {
2249 my ($self, $fb, $prop, $value) = @_;
2250 if ($prop eq 'svn:executable') {
2251 if ($fb->{mode_b} != 120000) {
2252 $fb->{mode_b} = defined $value ? 100755 : 100644;
2253 }
2254 } elsif ($prop eq 'svn:special') {
2255 $fb->{mode_b} = defined $value ? 120000 : 100644;
2256 } else {
2257 $self->{file_prop}->{$fb->{path}} ||= {};
2258 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2259 }
2260 undef;
2261}
2262sub apply_textdelta {
2264 my ($self, $fb, $exp) = @_;
2265 my $fh = IO::File->new_tmpfile;
2266 $fh->autoflush(1);
2267 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2268 # (but $base does not,) so dup() it for reading in close_file
2269 open my $dup, '<&', $fh or croak $!;
2270 my $base = IO::File->new_tmpfile;
2271 $base->autoflush(1);
2272 if ($fb->{blob}) {
2273 defined (my $pid = fork) or croak $!;
2274 if (!$pid) {
2275 open STDOUT, '>&', $base or croak $!;
2276 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2277 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2278 }
2279 waitpid $pid, 0;
2280 croak $? if $?;
2281 if (defined $exp) {
2283 seek $base, 0, 0 or croak $!;
2284 my $md5 = Digest::MD5->new;
2285 $md5->addfile($base);
2286 my $got = $md5->hexdigest;
2287 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2288 "expected: $exp\n",
2289 " got: $got\n" if ($got ne $exp);
2290 }
2291 }
2292 seek $base, 0, 0 or croak $!;
2293 $fb->{fh} = $dup;
2294 $fb->{base} = $base;
2295 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2296}
2297sub close_file {
2299 my ($self, $fb, $exp) = @_;
2300 my $hash;
2301 my $path = $fb->{path};
2302 if (my $fh = $fb->{fh}) {
2303 seek($fh, 0, 0) or croak $!;
2304 my $md5 = Digest::MD5->new;
2305 $md5->addfile($fh);
2306 my $got = $md5->hexdigest;
2307 die "Checksum mismatch: $path\n",
2308 "expected: $exp\n got: $got\n" if ($got ne $exp);
2309 seek($fh, 0, 0) or croak $!;
2310 if ($fb->{mode_b} == 120000) {
2311 read($fh, my $buf, 5) == 5 or croak $!;
2312 $buf eq 'link ' or die "$path has mode 120000",
2313 "but is not a link\n";
2314 }
2315 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2316 if (!$pid) {
2317 open STDIN, '<&', $fh or croak $!;
2318 exec qw/git-hash-object -w --stdin/ or croak $!;
2319 }
2320 chomp($hash = do { local $/; <$out> });
2321 close $out or croak $!;
2322 close $fh or croak $!;
2323 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2324 close $fb->{base} or croak $!;
2325 } else {
2326 $hash = $fb->{blob} or die "no blob information\n";
2327 }
2328 $fb->{pool}->clear;
2329 my $gui = $self->{gui};
2330 print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
2331 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
2332 undef;
2333}
2334sub abort_edit {
2336 my $self = shift;
2337 eval { command_close_pipe($self->{gui}, $self->{ctx}) };
2338 $self->SUPER::abort_edit(@_);
2339}
2340sub close_edit {
2342 my $self = shift;
2343 command_close_pipe($self->{gui}, $self->{ctx});
2344 $self->{git_commit_ok} = 1;
2345 $self->SUPER::close_edit(@_);
2346}
2347package SVN::Git::Editor;
2349use vars qw/@ISA/;
2350use strict;
2351use warnings;
2352use Carp qw/croak/;
2353use IO::File;
2354sub new {
2356 my $class = shift;
2357 my $git_svn = shift;
2358 my $self = SVN::Delta::Editor->new(@_);
2359 bless $self, $class;
2360 foreach (qw/svn_path r ra/) {
2361 die "$_ required!\n" unless (defined $git_svn->{$_});
2362 $self->{$_} = $git_svn->{$_};
2363 }
2364 $self->{pool} = SVN::Pool->new;
2365 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2366 $self->{rm} = { };
2367 require Digest::MD5;
2368 return $self;
2369}
2370sub split_path {
2372 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2373}
2374sub repo_path {
2376 (defined $_[1] && length $_[1]) ? $_[1] : ''
2377}
2378sub url_path {
2380 my ($self, $path) = @_;
2381 $self->{ra}->{url} . '/' . $self->repo_path($path);
2382}
2383sub rmdirs {
2385 my ($self, $tree_b) = @_;
2386 my $rm = $self->{rm};
2387 delete $rm->{''}; # we never delete the url we're tracking
2388 return unless %$rm;
2389 foreach (keys %$rm) {
2391 my @d = split m#/#, $_;
2392 my $c = shift @d;
2393 $rm->{$c} = 1;
2394 while (@d) {
2395 $c .= '/' . shift @d;
2396 $rm->{$c} = 1;
2397 }
2398 }
2399 delete $rm->{$self->{svn_path}};
2400 delete $rm->{''}; # we never delete the url we're tracking
2401 return unless %$rm;
2402 my ($fh, $ctx) = command_output_pipe(
2404 qw/ls-tree --name-only -r -z/, $tree_b);
2405 local $/ = "\0";
2406 while (<$fh>) {
2407 chomp;
2408 my @dn = split m#/#, $_;
2409 while (pop @dn) {
2410 delete $rm->{join '/', @dn};
2411 }
2412 unless (%$rm) {
2413 close $fh;
2414 return;
2415 }
2416 }
2417 command_close_pipe($fh, $ctx);
2418 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2420 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2421 $self->close_directory($bat->{$d}, $p);
2422 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2423 print "\tD+\t$d/\n" unless $::_q;
2424 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2425 delete $bat->{$d};
2426 }
2427}
2428sub open_or_add_dir {
2430 my ($self, $full_path, $baton) = @_;
2431 my $t = $self->{ra}->check_path($full_path, $self->{r});
2432 if ($t == $SVN::Node::none) {
2433 return $self->add_directory($full_path, $baton,
2434 undef, -1, $self->{pool});
2435 } elsif ($t == $SVN::Node::dir) {
2436 return $self->open_directory($full_path, $baton,
2437 $self->{r}, $self->{pool});
2438 }
2439 print STDERR "$full_path already exists in repository at ",
2440 "r$self->{r} and it is not a directory (",
2441 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2442 exit 1;
2443}
2444sub ensure_path {
2446 my ($self, $path) = @_;
2447 my $bat = $self->{bat};
2448 $path = $self->repo_path($path);
2449 return $bat->{''} unless (length $path);
2450 my @p = split m#/+#, $path;
2451 my $c = shift @p;
2452 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2453 while (@p) {
2454 my $c0 = $c;
2455 $c .= '/' . shift @p;
2456 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2457 }
2458 return $bat->{$c};
2459}
2460sub A {
2462 my ($self, $m) = @_;
2463 my ($dir, $file) = split_path($m->{file_b});
2464 my $pbat = $self->ensure_path($dir);
2465 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2466 undef, -1);
2467 print "\tA\t$m->{file_b}\n" unless $::_q;
2468 $self->chg_file($fbat, $m);
2469 $self->close_file($fbat,undef,$self->{pool});
2470}
2471sub C {
2473 my ($self, $m) = @_;
2474 my ($dir, $file) = split_path($m->{file_b});
2475 my $pbat = $self->ensure_path($dir);
2476 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2477 $self->url_path($m->{file_a}), $self->{r});
2478 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2479 $self->chg_file($fbat, $m);
2480 $self->close_file($fbat,undef,$self->{pool});
2481}
2482sub delete_entry {
2484 my ($self, $path, $pbat) = @_;
2485 my $rpath = $self->repo_path($path);
2486 my ($dir, $file) = split_path($rpath);
2487 $self->{rm}->{$dir} = 1;
2488 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2489}
2490sub R {
2492 my ($self, $m) = @_;
2493 my ($dir, $file) = split_path($m->{file_b});
2494 my $pbat = $self->ensure_path($dir);
2495 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2496 $self->url_path($m->{file_a}), $self->{r});
2497 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2498 $self->chg_file($fbat, $m);
2499 $self->close_file($fbat,undef,$self->{pool});
2500 ($dir, $file) = split_path($m->{file_a});
2502 $pbat = $self->ensure_path($dir);
2503 $self->delete_entry($m->{file_a}, $pbat);
2504}
2505sub M {
2507 my ($self, $m) = @_;
2508 my ($dir, $file) = split_path($m->{file_b});
2509 my $pbat = $self->ensure_path($dir);
2510 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2511 $pbat,$self->{r},$self->{pool});
2512 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2513 $self->chg_file($fbat, $m);
2514 $self->close_file($fbat,undef,$self->{pool});
2515}
2516sub T { shift->M(@_) }
2518sub change_file_prop {
2520 my ($self, $fbat, $pname, $pval) = @_;
2521 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2522}
2523sub chg_file {
2525 my ($self, $fbat, $m) = @_;
2526 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2527 $self->change_file_prop($fbat,'svn:executable','*');
2528 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2529 $self->change_file_prop($fbat,'svn:executable',undef);
2530 }
2531 my $fh = IO::File->new_tmpfile or croak $!;
2532 if ($m->{mode_b} =~ /^120/) {
2533 print $fh 'link ' or croak $!;
2534 $self->change_file_prop($fbat,'svn:special','*');
2535 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2536 $self->change_file_prop($fbat,'svn:special',undef);
2537 }
2538 defined(my $pid = fork) or croak $!;
2539 if (!$pid) {
2540 open STDOUT, '>&', $fh or croak $!;
2541 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2542 }
2543 waitpid $pid, 0;
2544 croak $? if $?;
2545 $fh->flush == 0 or croak $!;
2546 seek $fh, 0, 0 or croak $!;
2547 my $md5 = Digest::MD5->new;
2549 $md5->addfile($fh) or croak $!;
2550 seek $fh, 0, 0 or croak $!;
2551 my $exp = $md5->hexdigest;
2553 my $pool = SVN::Pool->new;
2554 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2555 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2556 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2557 $pool->clear;
2558 close $fh or croak $!;
2560}
2561sub D {
2563 my ($self, $m) = @_;
2564 my ($dir, $file) = split_path($m->{file_b});
2565 my $pbat = $self->ensure_path($dir);
2566 print "\tD\t$m->{file_b}\n" unless $::_q;
2567 $self->delete_entry($m->{file_b}, $pbat);
2568}
2569sub close_edit {
2571 my ($self) = @_;
2572 my ($p,$bat) = ($self->{pool}, $self->{bat});
2573 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2574 $self->close_directory($bat->{$_}, $p);
2575 }
2576 $self->SUPER::close_edit($p);
2577 $p->clear;
2578}
2579sub abort_edit {
2581 my ($self) = @_;
2582 $self->SUPER::abort_edit($self->{pool});
2583 $self->{pool}->clear;
2584}
2585# this drives the editor
2587sub apply_diff {
2588 my ($self, $tree_a, $tree_b) = @_;
2589 my @diff_tree = qw(diff-tree -z -r);
2590 if ($::_cp_similarity) {
2591 push @diff_tree, "-C$::_cp_similarity";
2592 } else {
2593 push @diff_tree, '-C';
2594 }
2595 push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
2596 push @diff_tree, "-l$::_l" if defined $::_l;
2597 push @diff_tree, $tree_a, $tree_b;
2598 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2599 my $nl = $/;
2600 local $/ = "\0";
2601 my $state = 'meta';
2602 my @mods;
2603 while (<$diff_fh>) {
2604 chomp $_; # this gets rid of the trailing "\0"
2605 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2606 $::sha1\s($::sha1)\s
2607 ([MTCRAD])\d*$/xo) {
2608 push @mods, { mode_a => $1, mode_b => $2,
2609 sha1_b => $3, chg => $4 };
2610 if ($4 =~ /^(?:C|R)$/) {
2611 $state = 'file_a';
2612 } else {
2613 $state = 'file_b';
2614 }
2615 } elsif ($state eq 'file_a') {
2616 my $x = $mods[$#mods] or croak "Empty array\n";
2617 if ($x->{chg} !~ /^(?:C|R)$/) {
2618 croak "Error parsing $_, $x->{chg}\n";
2619 }
2620 $x->{file_a} = $_;
2621 $state = 'file_b';
2622 } elsif ($state eq 'file_b') {
2623 my $x = $mods[$#mods] or croak "Empty array\n";
2624 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2625 croak "Error parsing $_, $x->{chg}\n";
2626 }
2627 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2628 croak "Error parsing $_, $x->{chg}\n";
2629 }
2630 $x->{file_b} = $_;
2631 $state = 'meta';
2632 } else {
2633 croak "Error parsing $_\n";
2634 }
2635 }
2636 command_close_pipe($diff_fh, $ctx);
2637 $/ = $nl;
2638 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2640 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2641 my $f = $m->{chg};
2642 if (defined $o{$f}) {
2643 $self->$f($m);
2644 } else {
2645 fatal("Invalid change type: $f\n");
2646 }
2647 }
2648 $self->rmdirs($tree_b) if $::_rmdir;
2649 if (@mods == 0) {
2650 $self->abort_edit;
2651 } else {
2652 $self->close_edit;
2653 }
2654 \@mods;
2655}
2656package Git::SVN::Ra;
2658use vars qw/@ISA $config_dir/;
2659use strict;
2660use warnings;
2661my ($can_do_switch);
2662BEGIN {
2664 # enforce temporary pool usage for some simple functions
2665 my $e;
2666 foreach (qw/get_latest_revnum rev_proplist get_file
2667 check_path get_dir get_uuid get_repos_root/) {
2668 $e .= "sub $_ {
2669 my \$self = shift;
2670 my \$pool = SVN::Pool->new;
2671 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2672 \$pool->clear;
2673 wantarray ? \@ret : \$ret[0]; }\n";
2674 }
2675 eval $e;
2676}
2677sub new {
2679 my ($class, $url) = @_;
2680 SVN::_Core::svn_config_ensure($config_dir, undef);
2681 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2682 SVN::Client::get_simple_provider(),
2683 SVN::Client::get_ssl_server_trust_file_provider(),
2684 SVN::Client::get_simple_prompt_provider(
2685 \&Git::SVN::Prompt::simple, 2),
2686 SVN::Client::get_ssl_client_cert_prompt_provider(
2687 \&Git::SVN::Prompt::ssl_client_cert, 2),
2688 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2689 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2690 SVN::Client::get_username_provider(),
2691 SVN::Client::get_ssl_server_trust_prompt_provider(
2692 \&Git::SVN::Prompt::ssl_server_trust),
2693 SVN::Client::get_username_prompt_provider(
2694 \&Git::SVN::Prompt::username, 2),
2695 ]);
2696 my $config = SVN::Core::config_get_config($config_dir);
2697 my $self = SVN::Ra->new(url => $url, auth => $baton,
2698 config => $config,
2699 pool => SVN::Pool->new,
2700 auth_provider_callbacks => $callbacks);
2701 $self->{svn_path} = $url;
2702 $self->{repos_root} = $self->get_repos_root;
2703 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2704 bless $self, $class;
2705}
2706sub DESTROY {
2708 my $self = shift;
2709 $self->{pool}->clear if $self->{pool};
2710 $self->SUPER::DESTROY(@_);
2711}
2712sub dup {
2714 my ($self) = @_;
2715 my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2716 map { $_ => $self->{$_} } qw/config url
2717 auth auth_provider_callbacks repos_root svn_path/);
2718 bless $dup, ref $self;
2719}
2720sub get_log {
2722 my ($self, @args) = @_;
2723 my $pool = SVN::Pool->new;
2724 $args[4]-- if $args[4] && ! $::_follow_parent;
2725 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2726 my $ret = $self->SUPER::get_log(@args, $pool);
2727 $pool->clear;
2728 $ret;
2729}
2730sub get_commit_editor {
2732 my ($self, $log, $cb, $pool) = @_;
2733 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2734 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2735}
2736sub uuid {
2738 my ($self) = @_;
2739 $self->{uuid} ||= $self->get_uuid;
2740}
2741sub gs_do_update {
2743 my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2744 my $pool = SVN::Pool->new;
2745 my $reporter = $self->do_update($rev_b, $path, $recurse,
2746 $editor, $pool);
2747 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2748 my $new = ($rev_a == $rev_b);
2749 $reporter->set_path($path, $rev_a, $new, @lock, $pool);
2750 $reporter->finish_report($pool);
2751 $pool->clear;
2752 $editor->{git_commit_ok};
2753}
2754sub gs_do_switch {
2756 my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2757 my $pool = SVN::Pool->new;
2758 my $reporter = $self->do_switch($rev_b, $path, $recurse,
2759 $url_b, $editor, $pool);
2760 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2761 $reporter->set_path($path, $rev_a, 0, @lock, $pool);
2762 $reporter->finish_report($pool);
2763 $pool->clear;
2764 $editor->{git_commit_ok};
2765}
2766sub can_do_switch {
2768 my $self = shift;
2769 unless (defined $can_do_switch) {
2770 my $pool = SVN::Pool->new;
2771 my $rep = eval {
2772 $self->do_switch(1, '', 0, $self->{url},
2773 SVN::Delta::Editor->new, $pool);
2774 };
2775 if ($@) {
2776 $can_do_switch = 0;
2777 } else {
2778 $rep->abort_report($pool);
2779 $can_do_switch = 1;
2780 }
2781 $pool->clear;
2782 }
2783 $can_do_switch;
2784}
2785package Git::SVN::Log;
2787use strict;
2788use warnings;
2789use POSIX qw/strftime/;
2790use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2791 %rusers $show_commit $incremental/;
2792my $l_fmt;
2793sub cmt_showable {
2795 my ($c) = @_;
2796 return 1 if defined $c->{r};
2797 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2798 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2799 my @log = command(qw/cat-file commit/, $c->{c});
2800 shift @log while ($log[0] ne "\n");
2801 shift @log;
2802 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2803 (undef, $c->{r}, undef) = ::extract_metadata(
2805 (grep(/^git-svn-id: /, @log))[-1]);
2806 }
2807 return defined $c->{r};
2808}
2809sub log_use_color {
2811 return 1 if $color;
2812 my ($dc, $dcvar);
2813 $dcvar = 'color.diff';
2814 $dc = `git-config --get $dcvar`;
2815 if ($dc eq '') {
2816 # nothing at all; fallback to "diff.color"
2817 $dcvar = 'diff.color';
2818 $dc = `git-config --get $dcvar`;
2819 }
2820 chomp($dc);
2821 if ($dc eq 'auto') {
2822 my $pc;
2823 $pc = `git-config --get color.pager`;
2824 if ($pc eq '') {
2825 # does not have it -- fallback to pager.color
2826 $pc = `git-config --bool --get pager.color`;
2827 }
2828 else {
2829 $pc = `git-config --bool --get color.pager`;
2830 if ($?) {
2831 $pc = 'false';
2832 }
2833 }
2834 chomp($pc);
2835 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2836 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2837 }
2838 return 0;
2839 }
2840 return 0 if $dc eq 'never';
2841 return 1 if $dc eq 'always';
2842 chomp($dc = `git-config --bool --get $dcvar`);
2843 return ($dc eq 'true');
2844}
2845sub git_svn_log_cmd {
2847 my ($r_min, $r_max) = @_;
2848 my $gs = Git::SVN->_new;
2849 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2850 $gs->refname);
2851 push @cmd, '-r' unless $non_recursive;
2852 push @cmd, qw/--raw --name-status/ if $verbose;
2853 push @cmd, '--color' if log_use_color();
2854 return @cmd unless defined $r_max;
2855 if ($r_max == $r_min) {
2856 push @cmd, '--max-count=1';
2857 if (my $c = $gs->rev_db_get($r_max)) {
2858 push @cmd, $c;
2859 }
2860 } else {
2861 my ($c_min, $c_max);
2862 $c_max = $gs->rev_db_get($r_max);
2863 $c_min = $gs->rev_db_get($r_min);
2864 if (defined $c_min && defined $c_max) {
2865 if ($r_max > $r_max) {
2866 push @cmd, "$c_min..$c_max";
2867 } else {
2868 push @cmd, "$c_max..$c_min";
2869 }
2870 } elsif ($r_max > $r_min) {
2871 push @cmd, $c_max;
2872 } else {
2873 push @cmd, $c_min;
2874 }
2875 }
2876 return @cmd;
2877}
2878# adapted from pager.c
2880sub config_pager {
2881 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2882 if (!defined $pager) {
2883 $pager = 'less';
2884 } elsif (length $pager == 0 || $pager eq 'cat') {
2885 $pager = undef;
2886 }
2887}
2888sub run_pager {
2890 return unless -t *STDOUT;
2891 pipe my $rfd, my $wfd or return;
2892 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2893 if (!$pid) {
2894 open STDOUT, '>&', $wfd or
2895 ::fatal "Can't redirect to stdout: $!\n";
2896 return;
2897 }
2898 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2899 $ENV{LESS} ||= 'FRSX';
2900 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2901}
2902sub get_author_info {
2904 my ($dest, $author, $t, $tz) = @_;
2905 $author =~ s/(?:^\s*|\s*$)//g;
2906 $dest->{a_raw} = $author;
2907 my $au;
2908 if ($::_authors) {
2909 $au = $rusers{$author} || undef;
2910 }
2911 if (!$au) {
2912 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2913 }
2914 $dest->{t} = $t;
2915 $dest->{tz} = $tz;
2916 $dest->{a} = $au;
2917 # Date::Parse isn't in the standard Perl distro :(
2918 if ($tz =~ s/^\+//) {
2919 $t += ::tz_to_s_offset($tz);
2920 } elsif ($tz =~ s/^\-//) {
2921 $t -= ::tz_to_s_offset($tz);
2922 }
2923 $dest->{t_utc} = $t;
2924}
2925sub process_commit {
2927 my ($c, $r_min, $r_max, $defer) = @_;
2928 if (defined $r_min && defined $r_max) {
2929 if ($r_min == $c->{r} && $r_min == $r_max) {
2930 show_commit($c);
2931 return 0;
2932 }
2933 return 1 if $r_min == $r_max;
2934 if ($r_min < $r_max) {
2935 # we need to reverse the print order
2936 return 0 if (defined $limit && --$limit < 0);
2937 push @$defer, $c;
2938 return 1;
2939 }
2940 if ($r_min != $r_max) {
2941 return 1 if ($r_min < $c->{r});
2942 return 1 if ($r_max > $c->{r});
2943 }
2944 }
2945 return 0 if (defined $limit && --$limit < 0);
2946 show_commit($c);
2947 return 1;
2948}
2949sub show_commit {
2951 my $c = shift;
2952 if ($oneline) {
2953 my $x = "\n";
2954 if (my $l = $c->{l}) {
2955 while ($l->[0] =~ /^\s*$/) { shift @$l }
2956 $x = $l->[0];
2957 }
2958 $l_fmt ||= 'A' . length($c->{r});
2959 print 'r',pack($l_fmt, $c->{r}),' | ';
2960 print "$c->{c} | " if $show_commit;
2961 print $x;
2962 } else {
2963 show_commit_normal($c);
2964 }
2965}
2966sub show_commit_changed_paths {
2968 my ($c) = @_;
2969 return unless $c->{changed};
2970 print "Changed paths:\n", @{$c->{changed}};
2971}
2972sub show_commit_normal {
2974 my ($c) = @_;
2975 print '-' x72, "\nr$c->{r} | ";
2976 print "$c->{c} | " if $show_commit;
2977 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2978 localtime($c->{t_utc})), ' | ';
2979 my $nr_line = 0;
2980 if (my $l = $c->{l}) {
2982 while ($l->[$#$l] eq "\n" && $#$l > 0
2983 && $l->[($#$l - 1)] eq "\n") {
2984 pop @$l;
2985 }
2986 $nr_line = scalar @$l;
2987 if (!$nr_line) {
2988 print "1 line\n\n\n";
2989 } else {
2990 if ($nr_line == 1) {
2991 $nr_line = '1 line';
2992 } else {
2993 $nr_line .= ' lines';
2994 }
2995 print $nr_line, "\n";
2996 show_commit_changed_paths($c);
2997 print "\n";
2998 print $_ foreach @$l;
2999 }
3000 } else {
3001 print "1 line\n";
3002 show_commit_changed_paths($c);
3003 print "\n";
3004 }
3006 foreach my $x (qw/raw diff/) {
3007 if ($c->{$x}) {
3008 print "\n";
3009 print $_ foreach @{$c->{$x}}
3010 }
3011 }
3012}
3013sub cmd_show_log {
3015 my (@args) = @_;
3016 my ($r_min, $r_max);
3017 my $r_last = -1; # prevent dupes
3018 if (defined $TZ) {
3019 $ENV{TZ} = $TZ;
3020 } else {
3021 delete $ENV{TZ};
3022 }
3023 if (defined $::_revision) {
3024 if ($::_revision =~ /^(\d+):(\d+)$/) {
3025 ($r_min, $r_max) = ($1, $2);
3026 } elsif ($::_revision =~ /^\d+$/) {
3027 $r_min = $r_max = $::_revision;
3028 } else {
3029 ::fatal "-r$::_revision is not supported, use ",
3030 "standard \'git log\' arguments instead\n";
3031 }
3032 }
3033 config_pager();
3035 @args = (git_svn_log_cmd($r_min, $r_max), @args);
3036 my $log = command_output_pipe(@args);
3037 run_pager();
3038 my (@k, $c, $d);
3039 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3040 while (<$log>) {
3041 if (/^${esc_color}commit ($::sha1_short)/o) {
3042 my $cmt = $1;
3043 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3044 $r_last = $c->{r};
3045 process_commit($c, $r_min, $r_max, \@k) or
3046 goto out;
3047 }
3048 $d = undef;
3049 $c = { c => $cmt };
3050 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3051 get_author_info($c, $1, $2, $3);
3052 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3053 # ignore
3054 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3055 push @{$c->{raw}}, $_;
3056 } elsif (/^${esc_color}[ACRMDT]\t/) {
3057 # we could add $SVN->{svn_path} here, but that requires
3058 # remote access at the moment (repo_path_split)...
3059 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3060 push @{$c->{changed}}, $_;
3061 } elsif (/^${esc_color}diff /o) {
3062 $d = 1;
3063 push @{$c->{diff}}, $_;
3064 } elsif ($d) {
3065 push @{$c->{diff}}, $_;
3066 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3067 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3068 } elsif (s/^${esc_color} //o) {
3069 push @{$c->{l}}, $_;
3070 }
3071 }
3072 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3073 $r_last = $c->{r};
3074 process_commit($c, $r_min, $r_max, \@k);
3075 }
3076 if (@k) {
3077 my $swap = $r_max;
3078 $r_max = $r_min;
3079 $r_min = $swap;
3080 process_commit($_, $r_min, $r_max) foreach reverse @k;
3081 }
3082out:
3083 close $log;
3084 print '-' x72,"\n" unless $incremental || $oneline;
3085}
3086__END__
3088Data structures:
3090$log_entry hashref as returned by libsvn_log_entry()
3092{
3093 log => 'whitespace-formatted log entry
3094', # trailing newline is preserved
3095 revision => '8', # integer
3096 date => '2004-02-24T17:01:44.108345Z', # commit date
3097 author => 'committer name'
3098};
3099@mods = array of diff-index line hashes, each element represents one line
3101 of diff-index output
3102diff-index line ($m hash)
3104{
3105 mode_a => first column of diff-index output, no leading ':',
3106 mode_b => second column of diff-index output,
3107 sha1_b => sha1sum of the final blob,
3108 chg => change type [MCRADT],
3109 file_a => original file name of a file (iff chg is 'C' or 'R')
3110 file_b => new/current file name of a file (any chg)
3111}
3112;
3113# retval of read_url_paths{,_all}();
3115$l_map = {
3116 # repository root url
3117 'https://svn.musicpd.org' => {
3118 # repository path # GIT_SVN_ID
3119 'mpd/trunk' => 'trunk',
3120 'mpd/tags/0.11.5' => 'tags/0.11.5',
3121 },
3122}
3123Notes:
3125 I don't trust the each() function on unless I created %hash myself
3126 because the internal iterator may not have started at base.