]>
Commit | Line | Data |
---|---|---|
eaf718f3 MU |
1 | #!/usr/bin/perl -w |
2 | ||
3 | # This tool is copyright (c) 2005, Matthias Urlichs. | |
4 | # It is released under the Gnu Public License, version 2. | |
5 | # | |
6 | # The basic idea is to pull and analyze SVN changes. | |
7 | # | |
4a91b796 | 8 | # Checking out the files is done by a single long-running SVN connection. |
eaf718f3 MU |
9 | # |
10 | # The head revision is on branch "origin" by default. | |
11 | # You can change that with the '-o' option. | |
12 | ||
eaf718f3 MU |
13 | use strict; |
14 | use warnings; | |
15 | use Getopt::Std; | |
d3cac2c9 | 16 | use File::Copy; |
eaf718f3 MU |
17 | use File::Spec; |
18 | use File::Temp qw(tempfile); | |
19 | use File::Path qw(mkpath); | |
20 | use File::Basename qw(basename dirname); | |
21 | use Time::Local; | |
22 | use IO::Pipe; | |
23 | use POSIX qw(strftime dup2); | |
24 | use IPC::Open2; | |
25 | use SVN::Core; | |
26 | use SVN::Ra; | |
27 | ||
f3ad0625 | 28 | die "Need SVN:Core 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1"; |
37dcf6de | 29 | |
eaf718f3 MU |
30 | $SIG{'PIPE'}="IGNORE"; |
31 | $ENV{'TZ'}="UTC"; | |
32 | ||
c55f3fff | 33 | our($opt_h,$opt_o,$opt_v,$opt_u,$opt_C,$opt_i,$opt_m,$opt_M,$opt_t,$opt_T, |
40006ea0 SK |
34 | $opt_b,$opt_r,$opt_I,$opt_A,$opt_s,$opt_l,$opt_d,$opt_D,$opt_S,$opt_F, |
35 | $opt_P,$opt_R); | |
eaf718f3 MU |
36 | |
37 | sub usage() { | |
38 | print STDERR <<END; | |
f3ad0625 | 39 | Usage: ${\basename $0} # fetch/update GIT from SVN |
40006ea0 | 40 | [-o branch-for-HEAD] [-h] [-v] [-l max_rev] [-R repack_each_revs] |
eaf718f3 | 41 | [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname] |
c55f3fff | 42 | [-d|-D] [-i] [-u] [-r] [-I ignorefilename] [-s start_chg] |
ec1e4689 | 43 | [-m] [-M regex] [-A author_file] [-S] [-F] [-P project_name] [SVN_URL] |
eaf718f3 MU |
44 | END |
45 | exit(1); | |
46 | } | |
47 | ||
40006ea0 | 48 | getopts("A:b:C:dDFhiI:l:mM:o:rs:t:T:SP:R:uv") or usage(); |
eaf718f3 MU |
49 | usage if $opt_h; |
50 | ||
51 | my $tag_name = $opt_t || "tags"; | |
0476786e | 52 | my $trunk_name = defined $opt_T ? $opt_T : "trunk"; |
eaf718f3 | 53 | my $branch_name = $opt_b || "branches"; |
ec1e4689 SK |
54 | my $project_name = $opt_P || ""; |
55 | $project_name = "/" . $project_name if ($project_name); | |
40006ea0 | 56 | my $repack_after = $opt_R || 1000; |
eaf718f3 | 57 | |
25f6f325 | 58 | @ARGV == 1 or @ARGV == 2 or usage(); |
eaf718f3 MU |
59 | |
60 | $opt_o ||= "origin"; | |
2fa92046 | 61 | $opt_s ||= 1; |
eaf718f3 MU |
62 | my $git_tree = $opt_C; |
63 | $git_tree ||= "."; | |
64 | ||
4a91b796 | 65 | my $svn_url = $ARGV[0]; |
25f6f325 | 66 | my $svn_dir = $ARGV[1]; |
eaf718f3 MU |
67 | |
68 | our @mergerx = (); | |
69 | if ($opt_m) { | |
65160b8b FF |
70 | my $branch_esc = quotemeta ($branch_name); |
71 | my $trunk_esc = quotemeta ($trunk_name); | |
72 | @mergerx = | |
73 | ( | |
74 | qr!\b(?:merg(?:ed?|ing))\b.*?\b((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i, | |
75 | qr!\b(?:from|of)\W+((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i, | |
76 | qr!\b(?:from|of)\W+(?:the )?([\w\.\-]+)[-\s]branch\b!i | |
77 | ); | |
eaf718f3 MU |
78 | } |
79 | if ($opt_M) { | |
65160b8b | 80 | unshift (@mergerx, qr/$opt_M/); |
eaf718f3 MU |
81 | } |
82 | ||
d3cac2c9 KW |
83 | # Absolutize filename now, since we will have chdir'ed by the time we |
84 | # get around to opening it. | |
85 | $opt_A = File::Spec->rel2abs($opt_A) if $opt_A; | |
86 | ||
36610b24 | 87 | our %users = (); |
d3cac2c9 KW |
88 | our $users_file = undef; |
89 | sub read_users($) { | |
90 | $users_file = File::Spec->rel2abs(@_); | |
91 | die "Cannot open $users_file\n" unless -f $users_file; | |
92 | open(my $authors,$users_file); | |
36610b24 KW |
93 | while(<$authors>) { |
94 | chomp; | |
80804d0a | 95 | next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/; |
36610b24 KW |
96 | (my $user,my $name,my $email) = ($1,$2,$3); |
97 | $users{$user} = [$name,$email]; | |
98 | } | |
99 | close($authors); | |
100 | } | |
101 | ||
eaf718f3 MU |
102 | select(STDERR); $|=1; select(STDOUT); |
103 | ||
104 | ||
105 | package SVNconn; | |
106 | # Basic SVN connection. | |
107 | # We're only interested in connecting and downloading, so ... | |
108 | ||
109 | use File::Spec; | |
110 | use File::Temp qw(tempfile); | |
111 | use POSIX qw(strftime dup2); | |
08ddd4f7 | 112 | use Fcntl qw(SEEK_SET); |
eaf718f3 MU |
113 | |
114 | sub new { | |
115 | my($what,$repo) = @_; | |
116 | $what=ref($what) if ref($what); | |
117 | ||
118 | my $self = {}; | |
119 | $self->{'buffer'} = ""; | |
120 | bless($self,$what); | |
121 | ||
122 | $repo =~ s#/+$##; | |
123 | $self->{'fullrep'} = $repo; | |
124 | $self->conn(); | |
125 | ||
eaf718f3 MU |
126 | return $self; |
127 | } | |
128 | ||
129 | sub conn { | |
130 | my $self = shift; | |
131 | my $repo = $self->{'fullrep'}; | |
2961e0ee EW |
132 | my $auth = SVN::Core::auth_open ([SVN::Client::get_simple_provider, |
133 | SVN::Client::get_ssl_server_trust_file_provider, | |
134 | SVN::Client::get_username_provider]); | |
135 | my $s = SVN::Ra->new(url => $repo, auth => $auth); | |
eaf718f3 MU |
136 | die "SVN connection to $repo: $!\n" unless defined $s; |
137 | $self->{'svn'} = $s; | |
138 | $self->{'repo'} = $repo; | |
139 | $self->{'maxrev'} = $s->get_latest_revnum(); | |
140 | } | |
141 | ||
142 | sub file { | |
143 | my($self,$path,$rev) = @_; | |
eaf718f3 | 144 | |
29504118 | 145 | my ($fh, $name) = tempfile('gitsvn.XXXXXX', |
eaf718f3 MU |
146 | DIR => File::Spec->tmpdir(), UNLINK => 1); |
147 | ||
2b5e63d1 | 148 | print "... $rev $path ...\n" if $opt_v; |
4802426d | 149 | my (undef, $properties); |
d598075e | 150 | my $pool = SVN::Pool->new(); |
09c3a408 | 151 | $path =~ s#^/*##; |
4802426d | 152 | eval { (undef, $properties) |
d598075e SB |
153 | = $self->{'svn'}->get_file($path,$rev,$fh,$pool); }; |
154 | $pool->clear; | |
25f6f325 MU |
155 | if($@) { |
156 | return undef if $@ =~ /Attempted to get checksum/; | |
157 | die $@; | |
158 | } | |
4802426d KW |
159 | my $mode; |
160 | if (exists $properties->{'svn:executable'}) { | |
08ddd4f7 HVR |
161 | $mode = '100755'; |
162 | } elsif (exists $properties->{'svn:special'}) { | |
163 | my ($special_content, $filesize); | |
164 | $filesize = tell $fh; | |
165 | seek $fh, 0, SEEK_SET; | |
166 | read $fh, $special_content, $filesize; | |
167 | if ($special_content =~ s/^link //) { | |
168 | $mode = '120000'; | |
169 | seek $fh, 0, SEEK_SET; | |
170 | truncate $fh, 0; | |
171 | print $fh $special_content; | |
172 | } else { | |
173 | die "unexpected svn:special file encountered"; | |
174 | } | |
4802426d | 175 | } else { |
08ddd4f7 | 176 | $mode = '100644'; |
4802426d | 177 | } |
eaf718f3 MU |
178 | close ($fh); |
179 | ||
4802426d | 180 | return ($name, $mode); |
eaf718f3 MU |
181 | } |
182 | ||
c55f3fff KW |
183 | sub ignore { |
184 | my($self,$path,$rev) = @_; | |
185 | ||
186 | print "... $rev $path ...\n" if $opt_v; | |
09c3a408 | 187 | $path =~ s#^/*##; |
c55f3fff KW |
188 | my (undef,undef,$properties) |
189 | = $self->{'svn'}->get_dir($path,$rev,undef); | |
190 | if (exists $properties->{'svn:ignore'}) { | |
191 | my ($fh, $name) = tempfile('gitsvn.XXXXXX', | |
192 | DIR => File::Spec->tmpdir(), | |
193 | UNLINK => 1); | |
194 | print $fh $properties->{'svn:ignore'}; | |
195 | close($fh); | |
196 | return $name; | |
197 | } else { | |
198 | return undef; | |
199 | } | |
200 | } | |
201 | ||
83936a29 SK |
202 | sub dir_list { |
203 | my($self,$path,$rev) = @_; | |
09c3a408 | 204 | $path =~ s#^/*##; |
83936a29 SK |
205 | my ($dirents,undef,$properties) |
206 | = $self->{'svn'}->get_dir($path,$rev,undef); | |
207 | return $dirents; | |
208 | } | |
209 | ||
eaf718f3 | 210 | package main; |
25f6f325 | 211 | use URI; |
eaf718f3 | 212 | |
03490804 | 213 | our $svn = $svn_url; |
25f6f325 | 214 | $svn .= "/$svn_dir" if defined $svn_dir; |
03490804 | 215 | my $svn2 = SVNconn->new($svn); |
25f6f325 | 216 | $svn = SVNconn->new($svn); |
eaf718f3 | 217 | |
25f6f325 MU |
218 | my $lwp_ua; |
219 | if($opt_d or $opt_D) { | |
220 | $svn_url = URI->new($svn_url)->canonical; | |
221 | if($opt_D) { | |
222 | $svn_dir =~ s#/*$#/#; | |
223 | } else { | |
224 | $svn_dir = ""; | |
225 | } | |
226 | if ($svn_url->scheme eq "http") { | |
227 | use LWP::UserAgent; | |
228 | $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []); | |
229 | } else { | |
230 | print STDERR "Warning: not HTTP; turning off direct file access\n"; | |
231 | $opt_d=0; | |
232 | } | |
233 | } | |
eaf718f3 MU |
234 | |
235 | sub pdate($) { | |
236 | my($d) = @_; | |
237 | $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)# | |
238 | or die "Unparseable date: $d\n"; | |
239 | my $y=$1; $y-=1900 if $y>1900; | |
240 | return timegm($6||0,$5,$4,$3,$2-1,$y); | |
241 | } | |
242 | ||
eaf718f3 MU |
243 | sub getwd() { |
244 | my $pwd = `pwd`; | |
245 | chomp $pwd; | |
246 | return $pwd; | |
247 | } | |
248 | ||
249 | ||
250 | sub get_headref($$) { | |
251 | my $name = shift; | |
29504118 | 252 | my $git_dir = shift; |
eaf718f3 | 253 | my $sha; |
29504118 | 254 | |
eaf718f3 MU |
255 | if (open(C,"$git_dir/refs/heads/$name")) { |
256 | chomp($sha = <C>); | |
257 | close(C); | |
258 | length($sha) == 40 | |
259 | or die "Cannot get head id for $name ($sha): $!\n"; | |
260 | } | |
261 | return $sha; | |
262 | } | |
263 | ||
264 | ||
265 | -d $git_tree | |
266 | or mkdir($git_tree,0777) | |
267 | or die "Could not create $git_tree: $!"; | |
268 | chdir($git_tree); | |
269 | ||
270 | my $orig_branch = ""; | |
271 | my $forward_master = 0; | |
272 | my %branches; | |
273 | ||
274 | my $git_dir = $ENV{"GIT_DIR"} || ".git"; | |
275 | $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#; | |
276 | $ENV{"GIT_DIR"} = $git_dir; | |
277 | my $orig_git_index; | |
278 | $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE}; | |
279 | my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx', | |
280 | DIR => File::Spec->tmpdir()); | |
281 | close ($git_ih); | |
282 | $ENV{GIT_INDEX_FILE} = $git_index; | |
283 | my $maxnum = 0; | |
284 | my $last_rev = ""; | |
285 | my $last_branch; | |
03490804 | 286 | my $current_rev = $opt_s || 1; |
eaf718f3 | 287 | unless(-d $git_dir) { |
5c94f87e | 288 | system("git-init"); |
eaf718f3 MU |
289 | die "Cannot init the GIT db at $git_tree: $?\n" if $?; |
290 | system("git-read-tree"); | |
291 | die "Cannot init an empty tree: $?\n" if $?; | |
292 | ||
293 | $last_branch = $opt_o; | |
294 | $orig_branch = ""; | |
295 | } else { | |
296 | -f "$git_dir/refs/heads/$opt_o" | |
297 | or die "Branch '$opt_o' does not exist.\n". | |
298 | "Either use the correct '-o branch' option,\n". | |
299 | "or import to a new repository.\n"; | |
300 | ||
301 | -f "$git_dir/svn2git" | |
302 | or die "'$git_dir/svn2git' does not exist.\n". | |
303 | "You need that file for incremental imports.\n"; | |
8366a10a PR |
304 | open(F, "git-symbolic-ref HEAD |") or |
305 | die "Cannot run git-symbolic-ref: $!\n"; | |
306 | chomp ($last_branch = <F>); | |
307 | $last_branch = basename($last_branch); | |
308 | close(F); | |
eaf718f3 MU |
309 | unless($last_branch) { |
310 | warn "Cannot read the last branch name: $! -- assuming 'master'\n"; | |
311 | $last_branch = "master"; | |
312 | } | |
313 | $orig_branch = $last_branch; | |
314 | $last_rev = get_headref($orig_branch, $git_dir); | |
315 | if (-f "$git_dir/SVN2GIT_HEAD") { | |
316 | die <<EOM; | |
317 | SVN2GIT_HEAD exists. | |
318 | Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD. | |
319 | You may need to run | |
320 | ||
321 | git-read-tree -m -u SVN2GIT_HEAD HEAD | |
322 | EOM | |
323 | } | |
324 | system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD"); | |
325 | ||
326 | $forward_master = | |
327 | $opt_o ne 'master' && -f "$git_dir/refs/heads/master" && | |
29504118 | 328 | system('cmp', '-s', "$git_dir/refs/heads/master", |
eaf718f3 MU |
329 | "$git_dir/refs/heads/$opt_o") == 0; |
330 | ||
331 | # populate index | |
332 | system('git-read-tree', $last_rev); | |
333 | die "read-tree failed: $?\n" if $?; | |
334 | ||
335 | # Get the last import timestamps | |
336 | open my $B,"<", "$git_dir/svn2git"; | |
337 | while(<$B>) { | |
338 | chomp; | |
339 | my($num,$branch,$ref) = split; | |
340 | $branches{$branch}{$num} = $ref; | |
341 | $branches{$branch}{"LAST"} = $ref; | |
03490804 | 342 | $current_rev = $num+1 if $current_rev <= $num; |
eaf718f3 MU |
343 | } |
344 | close($B); | |
345 | } | |
346 | -d $git_dir | |
347 | or die "Could not create git subdir ($git_dir).\n"; | |
348 | ||
d3cac2c9 KW |
349 | my $default_authors = "$git_dir/svn-authors"; |
350 | if ($opt_A) { | |
351 | read_users($opt_A); | |
352 | copy($opt_A,$default_authors) or die "Copy failed: $!"; | |
353 | } else { | |
354 | read_users($default_authors) if -f $default_authors; | |
355 | } | |
356 | ||
eaf718f3 MU |
357 | open BRANCHES,">>", "$git_dir/svn2git"; |
358 | ||
83936a29 SK |
359 | sub node_kind($$) { |
360 | my ($svnpath, $revision) = @_; | |
cbce5d89 | 361 | my $pool=SVN::Pool->new; |
09c3a408 | 362 | $svnpath =~ s#^/*##; |
83936a29 | 363 | my $kind = $svn->{'svn'}->check_path($svnpath,$revision,$pool); |
cbce5d89 YAS |
364 | $pool->clear; |
365 | return $kind; | |
366 | } | |
367 | ||
cbce5d89 | 368 | sub get_file($$$) { |
83936a29 | 369 | my($svnpath,$rev,$path) = @_; |
cbce5d89 | 370 | |
eaf718f3 | 371 | # now get it |
4802426d | 372 | my ($name,$mode); |
25f6f325 MU |
373 | if($opt_d) { |
374 | my($req,$res); | |
375 | ||
376 | # /svn/!svn/bc/2/django/trunk/django-docs/build.py | |
377 | my $url=$svn_url->clone(); | |
378 | $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath"); | |
0090a618 | 379 | print "... $path...\n" if $opt_v; |
25f6f325 MU |
380 | $req = HTTP::Request->new(GET => $url); |
381 | $res = $lwp_ua->request($req); | |
382 | if ($res->is_success) { | |
383 | my $fh; | |
29504118 JH |
384 | ($fh, $name) = tempfile('gitsvn.XXXXXX', |
385 | DIR => File::Spec->tmpdir(), UNLINK => 1); | |
25f6f325 MU |
386 | print $fh $res->content; |
387 | close($fh) or die "Could not write $name: $!\n"; | |
388 | } else { | |
389 | return undef if $res->code == 301; # directory? | |
390 | die $res->status_line." at $url\n"; | |
391 | } | |
4802426d | 392 | $mode = '0644'; # can't obtain mode via direct http request? |
25f6f325 | 393 | } else { |
4802426d | 394 | ($name,$mode) = $svn->file("$svnpath",$rev); |
25f6f325 MU |
395 | return undef unless defined $name; |
396 | } | |
eaf718f3 | 397 | |
7ae0dc01 JH |
398 | my $pid = open(my $F, '-|'); |
399 | die $! unless defined $pid; | |
400 | if (!$pid) { | |
401 | exec("git-hash-object", "-w", $name) | |
eaf718f3 | 402 | or die "Cannot create object: $!\n"; |
7ae0dc01 | 403 | } |
eaf718f3 MU |
404 | my $sha = <$F>; |
405 | chomp $sha; | |
406 | close $F; | |
22dcbb75 | 407 | unlink $name; |
eaf718f3 MU |
408 | return [$mode, $sha, $path]; |
409 | } | |
410 | ||
c55f3fff | 411 | sub get_ignore($$$$$) { |
83936a29 | 412 | my($new,$old,$rev,$path,$svnpath) = @_; |
c55f3fff KW |
413 | |
414 | return unless $opt_I; | |
c55f3fff KW |
415 | my $name = $svn->ignore("$svnpath",$rev); |
416 | if ($path eq '/') { | |
417 | $path = $opt_I; | |
418 | } else { | |
419 | $path = File::Spec->catfile($path,$opt_I); | |
420 | } | |
421 | if (defined $name) { | |
422 | my $pid = open(my $F, '-|'); | |
423 | die $! unless defined $pid; | |
424 | if (!$pid) { | |
425 | exec("git-hash-object", "-w", $name) | |
426 | or die "Cannot create object: $!\n"; | |
427 | } | |
428 | my $sha = <$F>; | |
429 | chomp $sha; | |
430 | close $F; | |
431 | unlink $name; | |
432 | push(@$new,['0644',$sha,$path]); | |
83936a29 | 433 | } elsif (defined $old) { |
c55f3fff KW |
434 | push(@$old,$path); |
435 | } | |
436 | } | |
437 | ||
ec1e4689 SK |
438 | sub project_path($$) |
439 | { | |
440 | my ($path, $project) = @_; | |
441 | ||
442 | $path = "/".$path unless ($path =~ m#^\/#) ; | |
443 | return $1 if ($path =~ m#^$project\/(.*)$#); | |
444 | ||
445 | $path =~ s#\.#\\\.#g; | |
446 | $path =~ s#\+#\\\+#g; | |
447 | return "/" if ($project =~ m#^$path.*$#); | |
448 | ||
449 | return undef; | |
450 | } | |
451 | ||
eaf718f3 MU |
452 | sub split_path($$) { |
453 | my($rev,$path) = @_; | |
454 | my $branch; | |
455 | ||
456 | if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) { | |
457 | $branch = "/$1"; | |
458 | } elsif($path =~ s#^/\Q$trunk_name\E/?##) { | |
459 | $branch = "/"; | |
460 | } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) { | |
461 | $branch = $1; | |
462 | } else { | |
c2c07a5c YAS |
463 | my %no_error = ( |
464 | "/" => 1, | |
465 | "/$tag_name" => 1, | |
466 | "/$branch_name" => 1 | |
467 | ); | |
468 | print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path}); | |
eaf718f3 MU |
469 | return () |
470 | } | |
ec1e4689 SK |
471 | if ($path eq "") { |
472 | $path = "/"; | |
473 | } elsif ($project_name) { | |
474 | $path = project_path($path, $project_name); | |
475 | } | |
eaf718f3 MU |
476 | return ($branch,$path); |
477 | } | |
478 | ||
8168373f YAS |
479 | sub branch_rev($$) { |
480 | ||
481 | my ($srcbranch,$uptorev) = @_; | |
482 | ||
483 | my $bbranches = $branches{$srcbranch}; | |
484 | my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches; | |
485 | my $therev; | |
486 | foreach my $arev(@revs) { | |
487 | next if ($arev eq 'LAST'); | |
488 | if ($arev <= $uptorev) { | |
489 | $therev = $arev; | |
490 | last; | |
491 | } | |
492 | } | |
493 | return $therev; | |
494 | } | |
495 | ||
83936a29 SK |
496 | sub expand_svndir($$$); |
497 | ||
498 | sub expand_svndir($$$) | |
499 | { | |
500 | my ($svnpath, $rev, $path) = @_; | |
501 | my @list; | |
502 | get_ignore(\@list, undef, $rev, $path, $svnpath); | |
503 | my $dirents = $svn->dir_list($svnpath, $rev); | |
504 | foreach my $p(keys %$dirents) { | |
505 | my $kind = node_kind($svnpath.'/'.$p, $rev); | |
506 | if ($kind eq $SVN::Node::file) { | |
507 | my $f = get_file($svnpath.'/'.$p, $rev, $path.'/'.$p); | |
508 | push(@list, $f) if $f; | |
509 | } elsif ($kind eq $SVN::Node::dir) { | |
510 | push(@list, | |
511 | expand_svndir($svnpath.'/'.$p, $rev, $path.'/'.$p)); | |
512 | } | |
513 | } | |
514 | return @list; | |
515 | } | |
516 | ||
109fc2b9 | 517 | sub copy_path($$$$$$$$) { |
0090a618 MU |
518 | # Somebody copied a whole subdirectory. |
519 | # We need to find the index entries from the old version which the | |
520 | # SVN log entry points to, and add them to the new place. | |
521 | ||
109fc2b9 | 522 | my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_; |
0090a618 | 523 | |
8168373f | 524 | my($srcbranch,$srcpath) = split_path($rev,$oldpath); |
83936a29 SK |
525 | unless(defined $srcbranch && defined $srcpath) { |
526 | print "Path not found when copying from $oldpath @ $rev.\n". | |
527 | "Will try to copy from original SVN location...\n" | |
528 | if $opt_v; | |
529 | push (@$new, expand_svndir($oldpath, $rev, $path)); | |
4b1ca25e MU |
530 | return; |
531 | } | |
8168373f YAS |
532 | my $therev = branch_rev($srcbranch, $rev); |
533 | my $gitrev = $branches{$srcbranch}{$therev}; | |
0090a618 MU |
534 | unless($gitrev) { |
535 | print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n"; | |
536 | return; | |
537 | } | |
109fc2b9 YAS |
538 | if ($srcbranch ne $newbranch) { |
539 | push(@$parents, $branches{$srcbranch}{'LAST'}); | |
540 | } | |
8168373f YAS |
541 | print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v; |
542 | if ($node_kind eq $SVN::Node::dir) { | |
83936a29 | 543 | $srcpath =~ s#/*$#/#; |
8168373f | 544 | } |
a6080a0a | 545 | |
7ae0dc01 JH |
546 | my $pid = open my $f,'-|'; |
547 | die $! unless defined $pid; | |
548 | if (!$pid) { | |
549 | exec("git-ls-tree","-r","-z",$gitrev,$srcpath) | |
550 | or die $!; | |
551 | } | |
0090a618 MU |
552 | local $/ = "\0"; |
553 | while(<$f>) { | |
554 | chomp; | |
555 | my($m,$p) = split(/\t/,$_,2); | |
556 | my($mode,$type,$sha1) = split(/ /,$m); | |
557 | next if $type ne "blob"; | |
8168373f YAS |
558 | if ($node_kind eq $SVN::Node::dir) { |
559 | $p = $path . substr($p,length($srcpath)-1); | |
560 | } else { | |
561 | $p = $path; | |
562 | } | |
a6080a0a | 563 | push(@$new,[$mode,$sha1,$p]); |
0090a618 | 564 | } |
29504118 | 565 | close($f) or |
0090a618 MU |
566 | print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n"; |
567 | } | |
568 | ||
eaf718f3 MU |
569 | sub commit { |
570 | my($branch, $changed_paths, $revision, $author, $date, $message) = @_; | |
ae35b304 SK |
571 | my($committer_name,$committer_email,$dest); |
572 | my($author_name,$author_email); | |
109fc2b9 | 573 | my(@old,@new,@parents); |
eaf718f3 | 574 | |
35c636ec | 575 | if (not defined $author or $author eq "") { |
ae35b304 | 576 | $committer_name = $committer_email = "unknown"; |
d3cac2c9 KW |
577 | } elsif (defined $users_file) { |
578 | die "User $author is not listed in $users_file\n" | |
36610b24 | 579 | unless exists $users{$author}; |
ae35b304 | 580 | ($committer_name,$committer_email) = @{$users{$author}}; |
2b5e63d1 | 581 | } elsif ($author =~ /^(.*?)\s+<(.*)>$/) { |
ae35b304 | 582 | ($committer_name, $committer_email) = ($1, $2); |
eaf718f3 MU |
583 | } else { |
584 | $author =~ s/^<(.*)>$/$1/; | |
ae35b304 SK |
585 | $committer_name = $committer_email = $author; |
586 | } | |
587 | ||
7b40e7d1 | 588 | if ($opt_F && $message =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) { |
ae35b304 | 589 | ($author_name, $author_email) = ($1, $2); |
7b40e7d1 AW |
590 | print "Author from From: $1 <$2>\n" if ($opt_v);; |
591 | } elsif ($opt_S && $message =~ /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) { | |
592 | ($author_name, $author_email) = ($1, $2); | |
593 | print "Author from Signed-off-by: $1 <$2>\n" if ($opt_v);; | |
ae35b304 SK |
594 | } else { |
595 | $author_name = $committer_name; | |
596 | $author_email = $committer_email; | |
eaf718f3 | 597 | } |
ae35b304 | 598 | |
eaf718f3 MU |
599 | $date = pdate($date); |
600 | ||
601 | my $tag; | |
602 | my $parent; | |
603 | if($branch eq "/") { # trunk | |
604 | $parent = $opt_o; | |
605 | } elsif($branch =~ m#^/(.+)#) { # tag | |
606 | $tag = 1; | |
607 | $parent = $1; | |
608 | } else { # "normal" branch | |
609 | # nothing to do | |
610 | $parent = $branch; | |
611 | } | |
612 | $dest = $parent; | |
613 | ||
614 | my $prev = $changed_paths->{"/"}; | |
f0daa628 | 615 | if($prev and $prev->[0] eq "A") { |
eaf718f3 | 616 | delete $changed_paths->{"/"}; |
f0daa628 | 617 | my $oldpath = $prev->[1]; |
eaf718f3 MU |
618 | my $rev; |
619 | if(defined $oldpath) { | |
620 | my $p; | |
621 | ($parent,$p) = split_path($revision,$oldpath); | |
83936a29 SK |
622 | if(defined $parent) { |
623 | if($parent eq "/") { | |
624 | $parent = $opt_o; | |
625 | } else { | |
626 | $parent =~ s#^/##; # if it's a tag | |
627 | } | |
eaf718f3 MU |
628 | } |
629 | } else { | |
630 | $parent = undef; | |
631 | } | |
632 | } | |
633 | ||
634 | my $rev; | |
7ee74a99 | 635 | if($revision > $opt_s and defined $parent) { |
bf1ee636 | 636 | open(H,'-|',"git-rev-parse","--verify",$parent); |
eaf718f3 MU |
637 | $rev = <H>; |
638 | close(H) or do { | |
639 | print STDERR "$revision: cannot find commit '$parent'!\n"; | |
640 | return; | |
641 | }; | |
642 | chop $rev; | |
643 | if(length($rev) != 40) { | |
644 | print STDERR "$revision: cannot find commit '$parent'!\n"; | |
645 | return; | |
646 | } | |
647 | $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"}; | |
7ee74a99 | 648 | if($revision != $opt_s and not $rev) { |
eaf718f3 MU |
649 | print STDERR "$revision: do not know ancestor for '$parent'!\n"; |
650 | return; | |
651 | } | |
652 | } else { | |
653 | $rev = undef; | |
654 | } | |
655 | ||
f0daa628 | 656 | # if($prev and $prev->[0] eq "A") { |
eaf718f3 MU |
657 | # if(not $tag) { |
658 | # unless(open(H,"> $git_dir/refs/heads/$branch")) { | |
659 | # print STDERR "$revision: Could not create branch $branch: $!\n"; | |
660 | # $state=11; | |
661 | # next; | |
662 | # } | |
663 | # print H "$rev\n" | |
664 | # or die "Could not write branch $branch: $!"; | |
665 | # close(H) | |
666 | # or die "Could not write branch $branch: $!"; | |
667 | # } | |
668 | # } | |
669 | if(not defined $rev) { | |
670 | unlink($git_index); | |
671 | } elsif ($rev ne $last_rev) { | |
672 | print "Switching from $last_rev to $rev ($branch)\n" if $opt_v; | |
673 | system("git-read-tree", $rev); | |
674 | die "read-tree failed for $rev: $?\n" if $?; | |
675 | $last_rev = $rev; | |
676 | } | |
677 | ||
109fc2b9 YAS |
678 | push (@parents, $rev) if defined $rev; |
679 | ||
bf267d99 MU |
680 | my $cid; |
681 | if($tag and not %$changed_paths) { | |
682 | $cid = $rev; | |
683 | } else { | |
0090a618 MU |
684 | my @paths = sort keys %$changed_paths; |
685 | foreach my $path(@paths) { | |
686 | my $action = $changed_paths->{$path}; | |
687 | ||
8168373f YAS |
688 | if ($action->[0] eq "R") { |
689 | # refer to a file/tree in an earlier commit | |
690 | push(@old,$path); # remove any old stuff | |
691 | } | |
692 | if(($action->[0] eq "A") || ($action->[0] eq "R")) { | |
83936a29 | 693 | my $node_kind = node_kind($action->[3], $revision); |
e67c6625 | 694 | if ($node_kind eq $SVN::Node::file) { |
83936a29 SK |
695 | my $f = get_file($action->[3], |
696 | $revision, $path); | |
8168373f YAS |
697 | if ($f) { |
698 | push(@new,$f) if $f; | |
699 | } else { | |
700 | my $opath = $action->[3]; | |
701 | print STDERR "$revision: $branch: could not fetch '$opath'\n"; | |
702 | } | |
c55f3fff | 703 | } elsif ($node_kind eq $SVN::Node::dir) { |
e67c6625 KW |
704 | if($action->[1]) { |
705 | copy_path($revision, $branch, | |
706 | $path, $action->[1], | |
707 | $action->[2], $node_kind, | |
708 | \@new, \@parents); | |
709 | } else { | |
710 | get_ignore(\@new, \@old, $revision, | |
83936a29 | 711 | $path, $action->[3]); |
e67c6625 | 712 | } |
0090a618 | 713 | } |
bf267d99 MU |
714 | } elsif ($action->[0] eq "D") { |
715 | push(@old,$path); | |
716 | } elsif ($action->[0] eq "M") { | |
83936a29 | 717 | my $node_kind = node_kind($action->[3], $revision); |
8168373f | 718 | if ($node_kind eq $SVN::Node::file) { |
83936a29 SK |
719 | my $f = get_file($action->[3], |
720 | $revision, $path); | |
8168373f | 721 | push(@new,$f) if $f; |
c55f3fff KW |
722 | } elsif ($node_kind eq $SVN::Node::dir) { |
723 | get_ignore(\@new, \@old, $revision, | |
83936a29 | 724 | $path, $action->[3]); |
bf267d99 MU |
725 | } |
726 | } else { | |
727 | die "$revision: unknown action '".$action->[0]."' for $path\n"; | |
728 | } | |
729 | } | |
730 | ||
d9e2e127 SK |
731 | while(@old) { |
732 | my @o1; | |
733 | if(@old > 55) { | |
734 | @o1 = splice(@old,0,50); | |
735 | } else { | |
736 | @o1 = @old; | |
737 | @old = (); | |
738 | } | |
7ae0dc01 JH |
739 | my $pid = open my $F, "-|"; |
740 | die "$!" unless defined $pid; | |
741 | if (!$pid) { | |
742 | exec("git-ls-files", "-z", @o1) or die $!; | |
743 | } | |
d9e2e127 | 744 | @o1 = (); |
0090a618 | 745 | local $/ = "\0"; |
eaf718f3 MU |
746 | while(<$F>) { |
747 | chomp; | |
d9e2e127 | 748 | push(@o1,$_); |
bf267d99 MU |
749 | } |
750 | close($F); | |
751 | ||
d9e2e127 | 752 | while(@o1) { |
bf267d99 | 753 | my @o2; |
d9e2e127 SK |
754 | if(@o1 > 55) { |
755 | @o2 = splice(@o1,0,50); | |
bf267d99 | 756 | } else { |
d9e2e127 SK |
757 | @o2 = @o1; |
758 | @o1 = (); | |
bf267d99 MU |
759 | } |
760 | system("git-update-index","--force-remove","--",@o2); | |
761 | die "Cannot remove files: $?\n" if $?; | |
eaf718f3 | 762 | } |
eaf718f3 | 763 | } |
bf267d99 MU |
764 | while(@new) { |
765 | my @n2; | |
766 | if(@new > 12) { | |
767 | @n2 = splice(@new,0,10); | |
eaf718f3 | 768 | } else { |
bf267d99 MU |
769 | @n2 = @new; |
770 | @new = (); | |
eaf718f3 | 771 | } |
bf267d99 MU |
772 | system("git-update-index","--add", |
773 | (map { ('--cacheinfo', @$_) } @n2)); | |
774 | die "Cannot add files: $?\n" if $?; | |
eaf718f3 | 775 | } |
eaf718f3 | 776 | |
bf267d99 MU |
777 | my $pid = open(C,"-|"); |
778 | die "Cannot fork: $!" unless defined $pid; | |
779 | unless($pid) { | |
780 | exec("git-write-tree"); | |
781 | die "Cannot exec git-write-tree: $!\n"; | |
eaf718f3 | 782 | } |
bf267d99 MU |
783 | chomp(my $tree = <C>); |
784 | length($tree) == 40 | |
785 | or die "Cannot get tree id ($tree): $!\n"; | |
786 | close(C) | |
787 | or die "Error running git-write-tree: $?\n"; | |
788 | print "Tree ID $tree\n" if $opt_v; | |
789 | ||
790 | my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n"; | |
791 | my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n"; | |
792 | $pid = fork(); | |
793 | die "Fork: $!\n" unless defined $pid; | |
794 | unless($pid) { | |
795 | $pr->writer(); | |
796 | $pw->reader(); | |
797 | open(OUT,">&STDOUT"); | |
798 | dup2($pw->fileno(),0); | |
799 | dup2($pr->fileno(),1); | |
800 | $pr->close(); | |
801 | $pw->close(); | |
802 | ||
803 | my @par = (); | |
bf267d99 MU |
804 | |
805 | # loose detection of merges | |
806 | # based on the commit msg | |
807 | foreach my $rx (@mergerx) { | |
808 | if ($message =~ $rx) { | |
809 | my $mparent = $1; | |
810 | if ($mparent eq 'HEAD') { $mparent = $opt_o }; | |
811 | if ( -e "$git_dir/refs/heads/$mparent") { | |
812 | $mparent = get_headref($mparent, $git_dir); | |
109fc2b9 | 813 | push (@parents, $mparent); |
bf267d99 MU |
814 | print OUT "Merge parent branch: $mparent\n" if $opt_v; |
815 | } | |
29504118 | 816 | } |
bf267d99 | 817 | } |
109fc2b9 YAS |
818 | my %seen_parents = (); |
819 | my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents; | |
820 | foreach my $bparent (@unique_parents) { | |
821 | push @par, '-p', $bparent; | |
822 | print OUT "Merge parent branch: $bparent\n" if $opt_v; | |
823 | } | |
eaf718f3 | 824 | |
bf267d99 MU |
825 | exec("env", |
826 | "GIT_AUTHOR_NAME=$author_name", | |
827 | "GIT_AUTHOR_EMAIL=$author_email", | |
828 | "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)), | |
ae35b304 SK |
829 | "GIT_COMMITTER_NAME=$committer_name", |
830 | "GIT_COMMITTER_EMAIL=$committer_email", | |
bf267d99 MU |
831 | "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)), |
832 | "git-commit-tree", $tree,@par); | |
833 | die "Cannot exec git-commit-tree: $!\n"; | |
834 | } | |
835 | $pw->writer(); | |
836 | $pr->reader(); | |
eaf718f3 | 837 | |
bf267d99 | 838 | $message =~ s/[\s\n]+\z//; |
0a48a344 | 839 | $message = "r$revision: $message" if $opt_r; |
eaf718f3 | 840 | |
bf267d99 MU |
841 | print $pw "$message\n" |
842 | or die "Error writing to git-commit-tree: $!\n"; | |
843 | $pw->close(); | |
eaf718f3 | 844 | |
bf267d99 MU |
845 | print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v; |
846 | chomp($cid = <$pr>); | |
847 | length($cid) == 40 | |
848 | or die "Cannot get commit id ($cid): $!\n"; | |
849 | print "Commit ID $cid\n" if $opt_v; | |
850 | $pr->close(); | |
eaf718f3 | 851 | |
bf267d99 MU |
852 | waitpid($pid,0); |
853 | die "Error running git-commit-tree: $?\n" if $?; | |
854 | } | |
eaf718f3 | 855 | |
a16db4f4 YAS |
856 | if (not defined $cid) { |
857 | $cid = $branches{"/"}{"LAST"}; | |
858 | } | |
859 | ||
f02b3eba MU |
860 | if(not defined $dest) { |
861 | print "... no known parent\n" if $opt_v; | |
862 | } elsif(not $tag) { | |
eaf718f3 | 863 | print "Writing to refs/heads/$dest\n" if $opt_v; |
29504118 | 864 | open(C,">$git_dir/refs/heads/$dest") and |
eaf718f3 MU |
865 | print C ("$cid\n") and |
866 | close(C) | |
867 | or die "Cannot write branch $dest for update: $!\n"; | |
eaf718f3 | 868 | } |
eaf718f3 | 869 | |
47ee8ed2 | 870 | if ($tag) { |
eaf718f3 MU |
871 | $last_rev = "-" if %$changed_paths; |
872 | # the tag was 'complex', i.e. did not refer to a "real" revision | |
29504118 | 873 | |
f02b3eba | 874 | $dest =~ tr/_/\./ if $opt_u; |
eaf718f3 | 875 | |
ee834cf0 | 876 | system('git-tag', '-f', $dest, $cid) == 0 |
47ee8ed2 | 877 | or die "Cannot create tag $dest: $!\n"; |
eaf718f3 | 878 | |
f02b3eba | 879 | print "Created tag '$dest' on '$branch'\n" if $opt_v; |
eaf718f3 | 880 | } |
e7e477df MU |
881 | $branches{$branch}{"LAST"} = $cid; |
882 | $branches{$branch}{$revision} = $cid; | |
883 | $last_rev = $cid; | |
884 | print BRANCHES "$revision $branch $cid\n"; | |
885 | print "DONE: $revision $dest $cid\n" if $opt_v; | |
eaf718f3 MU |
886 | } |
887 | ||
03490804 MU |
888 | sub commit_all { |
889 | # Recursive use of the SVN connection does not work | |
890 | local $svn = $svn2; | |
891 | ||
892 | my ($changed_paths, $revision, $author, $date, $message, $pool) = @_; | |
f0daa628 MU |
893 | my %p; |
894 | while(my($path,$action) = each %$changed_paths) { | |
0090a618 | 895 | $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ]; |
f0daa628 MU |
896 | } |
897 | $changed_paths = \%p; | |
f0daa628 | 898 | |
eaf718f3 MU |
899 | my %done; |
900 | my @col; | |
901 | my $pref; | |
902 | my $branch; | |
903 | ||
904 | while(my($path,$action) = each %$changed_paths) { | |
905 | ($branch,$path) = split_path($revision,$path); | |
906 | next if not defined $branch; | |
ec1e4689 | 907 | next if not defined $path; |
eaf718f3 MU |
908 | $done{$branch}{$path} = $action; |
909 | } | |
910 | while(($branch,$changed_paths) = each %done) { | |
911 | commit($branch, $changed_paths, $revision, $author, $date, $message); | |
912 | } | |
913 | } | |
914 | ||
03490804 | 915 | $opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'}; |
988eece4 | 916 | |
a7cfb4a4 | 917 | if ($opt_l < $current_rev) { |
988eece4 ML |
918 | print "Up to date: no new revisions to fetch!\n" if $opt_v; |
919 | unlink("$git_dir/SVN2GIT_HEAD"); | |
920 | exit; | |
921 | } | |
922 | ||
40006ea0 | 923 | print "Processing from $current_rev to $opt_l ...\n" if $opt_v; |
03490804 | 924 | |
40006ea0 | 925 | my $from_rev; |
69216777 | 926 | my $to_rev = $current_rev - 1; |
40006ea0 SK |
927 | |
928 | while ($to_rev < $opt_l) { | |
69216777 | 929 | $from_rev = $to_rev + 1; |
40006ea0 SK |
930 | $to_rev = $from_rev + $repack_after; |
931 | $to_rev = $opt_l if $opt_l < $to_rev; | |
932 | print "Fetching from $from_rev to $to_rev ...\n" if $opt_v; | |
933 | my $pool=SVN::Pool->new; | |
934 | $svn->{'svn'}->get_log("/",$from_rev,$to_rev,0,1,1,\&commit_all,$pool); | |
935 | $pool->clear; | |
936 | my $pid = fork(); | |
937 | die "Fork: $!\n" unless defined $pid; | |
938 | unless($pid) { | |
939 | exec("git-repack", "-d") | |
940 | or die "Cannot repack: $!\n"; | |
941 | } | |
942 | waitpid($pid, 0); | |
943 | } | |
eaf718f3 MU |
944 | |
945 | ||
946 | unlink($git_index); | |
947 | ||
948 | if (defined $orig_git_index) { | |
949 | $ENV{GIT_INDEX_FILE} = $orig_git_index; | |
950 | } else { | |
951 | delete $ENV{GIT_INDEX_FILE}; | |
952 | } | |
953 | ||
954 | # Now switch back to the branch we were in before all of this happened | |
955 | if($orig_branch) { | |
3ef378a6 | 956 | print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0); |
eaf718f3 MU |
957 | system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master") |
958 | if $forward_master; | |
959 | unless ($opt_i) { | |
960 | system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD'); | |
961 | die "read-tree failed: $?\n" if $?; | |
962 | } | |
963 | } else { | |
964 | $orig_branch = "master"; | |
3ef378a6 | 965 | print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0); |
eaf718f3 MU |
966 | system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master") |
967 | unless -f "$git_dir/refs/heads/master"; | |
8366a10a | 968 | system('git-update-ref', 'HEAD', "$orig_branch"); |
eaf718f3 MU |
969 | unless ($opt_i) { |
970 | system('git checkout'); | |
971 | die "checkout failed: $?\n" if $?; | |
972 | } | |
973 | } | |
974 | unlink("$git_dir/SVN2GIT_HEAD"); | |
975 | close(BRANCHES); |