e480a50f4888369b1b1f7c27671ea803d4e4279d
[git.git] / git-cvsimport-script
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 aggregate CVS check-ins into related changes.
7 # Fortunately, "cvsps" does that for us; all we have to do is to parse
8 # its output.
9 #
10 # Checking out the files is done by a single long-running CVS connection
11 # / server process.
12 #
13 # The head revision is on branch "origin" by default.
14 # You can change that with the '-o' option.
15
16 use strict;
17 use warnings;
18 use Getopt::Std;
19 use File::Path qw(mkpath);
20 use File::Basename qw(basename dirname);
21 use Time::Local;
22 use POSIX qw(strftime);
23
24 $SIG{'PIPE'}="IGNORE";
25 $ENV{'TZ'}="UTC";
26
27 our($opt_h,$opt_o,$opt_v);
28
29 sub usage() {
30         print STDERR <<END;
31 Usage: ${\basename $0}  -- fetch/update GIT from CVS
32        [ -h (help) ]
33            [ -v (verbose) ] [ -o name-of-HEAD ]
34        CVS_archive GIT_dest
35 END
36         exit(1);
37 }
38
39 getopts("hqvo:") or usage();
40 usage if $opt_h;
41
42 @ARGV == 2 or usage();
43 $opt_o ||= "origin";
44
45 my($cvs_tree, $git_tree) = @ARGV;
46
47 select(STDERR); $|=1; select(STDOUT);
48
49
50 package CVSconn;
51 # Basic CVS dialog.
52 # We're only interested in downloading files.
53
54 sub new {
55         my($what,$repo,$subdir) = @_;
56         $what=ref($what) if ref($what);
57
58         my $self = {};
59         $self->{'buffer'} = "";
60         bless($self,$what);
61
62         $repo =~ s#/+$##;
63         $self->{'fullrep'} = $repo;
64         $self->conn();
65
66         $self->{'subdir'} = $subdir;
67         $self->{'lines'} = undef;
68
69         return $self;
70 }
71
72 sub conn {
73         my $self = shift;
74         my $repo = $self->{'fullrep'};
75         if($repo =~ s/^:pserver:(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
76                 my($user,$pass,$serv,$port) = ($1,$2,$3,$4);
77                 $user="anonymous" unless defined $user;
78                 my $rr2;
79                 unless($port) {
80                         $rr2 = ":pserver:$user\@$serv:$repo";
81                         $port=2401;
82                 }
83                 my $rr = ":pserver:$user\@$serv:$port$repo";
84
85                 unless($pass) {
86                         open(H,$ENV{'HOME'}."/.cvspass") and do {
87                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
88                                 while(<H>) {
89                                         chomp;
90                                         s/^\/\d+\s+//;
91                                         my ($w,$p) = split(/\s/,$_,2);
92                                         if($w eq $rr or $w eq $rr2) {
93                                                 $pass = $p;
94                                                 last;
95                                         }
96                                 }
97                         };
98                 }
99                 $pass="A" unless $pass;
100
101                 use IO::Socket;
102                 my $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
103                 die "Socket to $serv: $!\n" unless defined $s;
104                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
105                         or die "Write to $serv: $!\n";
106                 $s->flush();
107
108                 my $rep = <$s>;
109
110                 if($rep ne "I LOVE YOU\n") {
111                         $rep="<unknown>" unless $rep;
112                         die "AuthReply: $rep\n";
113                 }
114                 $self->{'socketo'} = $s;
115                 $self->{'socketi'} = $s;
116         } else { # local: Fork off our own cvs server.
117                 use IO::Pipe;
118                 my $pr = IO::Pipe->new();
119                 my $pw = IO::Pipe->new();
120                 my $pid = fork();
121                 die "Fork: $!\n" unless defined $pid;
122                 unless($pid) {
123                         $pr->writer();
124                         $pw->reader();
125                         use POSIX qw(dup2);
126                         dup2($pw->fileno(),0);
127                         dup2($pr->fileno(),1);
128                         $pr->close();
129                         $pw->close();
130                         exec("cvs","server");
131                 }
132                 $pw->writer();
133                 $pr->reader();
134                 $self->{'socketo'} = $pw;
135                 $self->{'socketi'} = $pr;
136         }
137         $self->{'socketo'}->write("Root $repo\n");
138
139         # Trial and error says that this probably is the minimum set
140         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E F Checked-in Created Updated Merged Removed\n");
141
142         $self->{'socketo'}->write("valid-requests\n");
143         $self->{'socketo'}->flush();
144
145         chomp(my $rep=$self->readline());
146         if($rep !~ s/^Valid-requests\s*//) {
147                 $rep="<unknown>" unless $rep;
148                 die "Expected Valid-requests from server, but got: $rep\n";
149         }
150         chomp(my $res=$self->readline());
151         die "validReply: $res\n" if $res ne "ok";
152
153         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
154         $self->{'repo'} = $repo;
155 }
156
157 sub readline {
158         my($self) = @_;
159         return $self->{'socketi'}->getline();
160 }
161
162 sub _file {
163         # Request a file with a given revision.
164         # Trial and error says this is a good way to do it. :-/
165         my($self,$fn,$rev) = @_;
166         $self->{'socketo'}->write("Argument -N\n") or return undef;
167         $self->{'socketo'}->write("Argument -P\n") or return undef;
168         # $self->{'socketo'}->write("Argument -ko\n") or return undef;
169         # -ko: Linus' version doesn't use it
170         $self->{'socketo'}->write("Argument -r\n") or return undef;
171         $self->{'socketo'}->write("Argument $rev\n") or return undef;
172         $self->{'socketo'}->write("Argument --\n") or return undef;
173         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
174         $self->{'socketo'}->write("Directory .\n") or return undef;
175         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
176         $self->{'socketo'}->write("Sticky T1.1\n") or return undef;
177         $self->{'socketo'}->write("co\n") or return undef;
178         $self->{'socketo'}->flush() or return undef;
179         $self->{'lines'} = 0;
180         return 1;
181 }
182 sub _line {
183         # Read a line from the server.
184         # ... except that 'line' may be an entire file. ;-)
185         my($self) = @_;
186         die "Not in lines" unless defined $self->{'lines'};
187
188         my $line;
189         my $res="";
190         while(defined($line = $self->readline())) {
191                 # M U gnupg-cvs-rep/AUTHORS
192                 # Updated gnupg-cvs-rep/
193                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
194                 # /AUTHORS/1.1///T1.1
195                 # u=rw,g=rw,o=rw
196                 # 0
197                 # ok
198
199                 if($line =~ s/^(?:Created|Updated) //) {
200                         $line = $self->readline(); # path
201                         $line = $self->readline(); # Entries line
202                         my $mode = $self->readline(); chomp $mode;
203                         $self->{'mode'} = $mode;
204                         defined (my $cnt = $self->readline())
205                                 or die "EOF from server after 'Changed'\n";
206                         chomp $cnt;
207                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
208                         $line="";
209                         $res="";
210                         while($cnt) {
211                                 my $buf;
212                                 my $num = $self->{'socketi'}->read($buf,$cnt);
213                                 die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
214                                 $res .= $buf;
215                                 $cnt -= $num;
216                         }
217                 } elsif($line =~ s/^ //) {
218                         $res .= $line;
219                 } elsif($line =~ /^M\b/) {
220                         # output, do nothing
221                 } elsif($line =~ /^Mbinary\b/) {
222                         my $cnt;
223                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
224                         chomp $cnt;
225                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
226                         $line="";
227                         while($cnt) {
228                                 my $buf;
229                                 my $num = $self->{'socketi'}->read($buf,$cnt);
230                                 die "S: Mbinary $cnt: $num: $!\n" if not defined $num or $num<=0;
231                                 $res .= $buf;
232                                 $cnt -= $num;
233                         }
234                 } else {
235                         chomp $line;
236                         if($line eq "ok") {
237                                 # print STDERR "S: ok (".length($res).")\n";
238                                 return $res;
239                         } elsif($line =~ s/^E //) {
240                                 # print STDERR "S: $line\n";
241                         } else {
242                                 die "Unknown: $line\n";
243                         }
244                 }
245         }
246 }
247 sub file {
248         my($self,$fn,$rev) = @_;
249         my $res;
250
251         if ($self->_file($fn,$rev)) {
252                 $res = $self->_line();
253                 return $res if defined $res;
254         }
255
256         # retry
257         $self->conn();
258         $self->_file($fn,$rev)
259                 or die "No file command send\n";
260         $res = $self->_line();
261         die "No input: $fn $rev\n" unless defined $res;
262         return $res;
263 }
264
265
266 package main;
267
268 my $cvs = CVSconn->new(dirname($cvs_tree), basename($cvs_tree));
269
270
271 sub pdate($) {
272         my($d) = @_;
273         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
274                 or die "Unparseable date: $d\n";
275         my $y=$1; $y-=1900 if $y>1900;
276         return timegm($6||0,$5,$4,$3,$2-1,$y);
277 }
278
279 sub pmode($) {
280         my($mode) = @_;
281         my $m = 0;
282         my $mm = 0;
283         my $um = 0;
284         for my $x(split(//,$mode)) {
285                 if($x eq ",") {
286                         $m |= $mm&$um;
287                         $mm = 0;
288                         $um = 0;
289                 } elsif($x eq "u") { $um |= 0700;
290                 } elsif($x eq "g") { $um |= 0070;
291                 } elsif($x eq "o") { $um |= 0007;
292                 } elsif($x eq "r") { $mm |= 0444;
293                 } elsif($x eq "w") { $mm |= 0222;
294                 } elsif($x eq "x") { $mm |= 0111;
295                 } elsif($x eq "=") { # do nothing
296                 } else { die "Unknown mode: $mode\n";
297                 }
298         }
299         $m |= $mm&$um;
300         return $m;
301 }
302
303 my $tmpcv = "/var/cache/cvs";
304
305 sub getwd() {
306         my $pwd = `pwd`;
307         chomp $pwd;
308         return $pwd;
309 }
310
311 -d $git_tree
312         or mkdir($git_tree,0777)
313         or die "Could not create $git_tree: $!";
314 chdir($git_tree);
315
316 my $last_branch = "";
317 my %branch_date;
318
319 my $git_dir = $ENV{"GIT_DIR"} || ".git";
320 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
321 $ENV{"GIT_DIR"} = $git_dir;
322 unless(-d $git_dir) {
323         system("git-init-db");
324         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
325         system("git-read-tree");
326         die "Cannot init an empty tree: $?\n" if $?;
327
328         $last_branch = $opt_o;
329 } else {
330         $last_branch = basename(readlink("$git_dir/HEAD"));
331         die "Cannot read the last branch name: $!\n" unless $last_branch;
332
333         # Get the last import timestamps
334         opendir(D,"$git_dir/refs/heads");
335         while(defined(my $head = readdir(D))) {
336                 next if $head =~ /^\./;
337                 open(F,"$git_dir/refs/heads/$head")
338                         or die "Bad head branch: $head: $!\n";
339                 chomp(my $ftag = <F>);
340                 close(F);
341                 open(F,"git-cat-file commit $ftag |");
342                 while(<F>) {
343                         next unless /^author\s.*\s(\d+)\s[-+]\d{4}$/;
344                         $branch_date{$head} = $1;
345                         last;
346                 }
347                 close(F);
348         }
349         closedir(D);
350 }
351
352 -d $git_dir
353         or die "Could not create git subdir ($git_dir).\n";
354
355 my $pid = open(CVS,"-|");
356 die "Cannot fork: $!\n" unless defined $pid;
357 unless($pid) {
358         exec("cvsps","-A","--cvs-direct",$cvs_tree);
359         die "Could not start cvsps: $!\n";
360 }
361
362
363 ## cvsps output:
364 #---------------------
365 #PatchSet 314
366 #Date: 1999/09/18 13:03:59
367 #Author: wkoch
368 #Branch: STABLE-BRANCH-1-0
369 #Ancestor branch: HEAD
370 #Tag: (none)
371 #Log:
372 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
373 #Members:
374 #       README:1.57->1.57.2.1
375 #       VERSION:1.96->1.96.2.1
376 #
377 #---------------------
378
379 my $state = 0;
380
381 my($patchset,$date,$author,$branch,$ancestor,$tag,$logmsg);
382 my(@old,@new);
383 my $commit = sub {
384         my $pid;
385         system("git-update-cache","--force-remove","--",@old) if @old;
386         die "Cannot remove files: $?\n" if $?;
387         system("git-update-cache","--add","--",@new) if @new;
388         die "Cannot add files: $?\n" if $?;
389
390         $pid = open(C,"-|");
391         die "Cannot fork: $!" unless defined $pid;
392         unless($pid) {
393                 exec("git-write-tree");
394                 die "Cannot exec git-write-tree: $!\n";
395         }
396         chomp(my $tree = <C>);
397         length($tree) == 40
398                 or die "Cannot get tree id ($tree): $!\n";
399         close(C)
400                 or die "Error running git-write-tree: $?\n";
401         print "Tree ID $tree\n" if $opt_v;
402
403         my $parent = "";
404         if(open(C,"$git_dir/refs/heads/$last_branch")) {
405                 chomp($parent = <C>);
406                 close(C);
407                 length($parent) == 40
408                         or die "Cannot get parent id ($parent): $!\n";
409                 print "Parent ID $parent\n" if $opt_v;
410         }
411
412         pipe(I,O)
413                 or die "No pipe: $!\n";
414         $pid = open(C,"|-");
415         die "Cannot fork: $!" unless defined $pid;
416         unless($pid) {
417                 close(I);
418                 open(STDOUT,">&O");
419                 my @par = ();
420                 @par = ("-p",$parent) if $parent;
421                 exec("env",
422                         "GIT_AUTHOR_NAME=$author",
423                         "GIT_AUTHOR_EMAIL=$author",
424                         "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
425                         "GIT_COMMITTER_NAME=$author",
426                         "GIT_COMMITTER_EMAIL=$author",
427                         "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
428                         "git-commit-tree", $tree,@par);
429                 die "Cannot exec git-commit-tree: $!\n";
430         }
431         close(O);
432         print C $logmsg
433                 or die "Error writing to git-commit-tree: $!\n";
434         close(C)
435                 or die "Error running git-commit-tree: $?\n";
436         print "Committed patch $patchset ($branch)\n" if $opt_v;
437         chomp(my $cid = <I>);
438         length($cid) == 40
439                 or die "Cannot get commit id ($cid): $!\n";
440         print "Commit ID $cid\n" if $opt_v;
441         close(I);
442
443         open(C,">$git_dir/refs/heads/$branch")
444                 or die "Cannot open branch $branch for update: $!\n";
445         print C "$cid\n"
446                 or die "Cannot write branch $branch for update: $!\n";
447         close(C)
448                 or die "Cannot write branch $branch for update: $!\n";
449
450         if($tag) {
451                 open(C,">$git_dir/refs/tags/$tag")
452                         or die "Cannot create tag $tag: $!\n";
453                 print C "$cid\n"
454                         or die "Cannot write tag $branch: $!\n";
455                 close(C)
456                         or die "Cannot write tag $branch: $!\n";
457                 print "Created tag '$tag' on '$branch'\n" if $opt_v;
458         }
459
460         @old = ();
461         @new = ();
462 };
463
464 while(<CVS>) {
465         chomp;
466         if($state == 0 and /^-+$/) {
467                 $state = 1;
468         } elsif($state == 0) {
469                 $state = 1;
470                 redo;
471         } elsif(($state==0 or $state==1) and s/^PatchSet\s+//) {
472                 $patchset = 0+$_;
473                 $state=2;
474         } elsif($state == 2 and s/^Date:\s+//) {
475                 $date = pdate($_);
476                 unless($date) {
477                         print STDERR "Could not parse date: $_\n";
478                         $state=0;
479                         next;
480                 }
481                 $state=3;
482         } elsif($state == 3 and s/^Author:\s+//) {
483                 s/\s+$//;
484                 $author = $_;
485                 $state = 4;
486         } elsif($state == 4 and s/^Branch:\s+//) {
487                 s/\s+$//;
488                 $branch = $_;
489                 $state = 5;
490         } elsif($state == 5 and s/^Ancestor branch:\s+//) {
491                 s/\s+$//;
492                 $ancestor = $_;
493                 $ancestor = $opt_o if $ancestor == "HEAD";
494                 $state = 6;
495         } elsif($state == 5) {
496                 $ancestor = undef;
497                 $state = 6;
498                 redo;
499         } elsif($state == 6 and s/^Tag:\s+//) {
500                 s/\s+$//;
501                 if($_ eq "(none)") {
502                         $tag = undef;
503                 } else {
504                         $tag = $_;
505                 }
506                 $state = 7;
507         } elsif($state == 7 and /^Log:/) {
508                 $logmsg = "";
509                 $state = 8;
510         } elsif($state == 8 and /^Members:/) {
511                 $branch = $opt_o if $branch eq "HEAD";
512                 if(defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
513                         # skip
514                         print "skip patchset $patchset: $date before $branch_date{$branch}\n";
515                         $state = 11;
516                         next;
517                 }
518                 if($ancestor) {
519                         if(-f "$git_dir/refs/heads/$branch") {
520                                 print STDERR "Branch $branch already exists!\n";
521                                 $state=11;
522                                 next;
523                         }
524                         unless(open(H,"$git_dir/refs/heads/$ancestor")) {
525                                 print STDERR "Branch $ancestor does not exist!\n";
526                                 $state=11;
527                                 next;
528                         }
529                         chomp(my $id = <H>);
530                         close(H);
531                         unless(open(H,"> $git_dir/refs/heads/$branch")) {
532                                 print STDERR "Could not create branch $branch: $!\n";
533                                 $state=11;
534                                 next;
535                         }
536                         print H "$id\n"
537                                 or die "Could not write branch $branch: $!";
538                         close(H)
539                                 or die "Could not write branch $branch: $!";
540                 }
541                 if(($ancestor || $branch) ne $last_branch) {
542                         system("git-read-tree -m -u $last_branch $branch");
543                         die "read-tree $branch failed at $branch: $?\n" if $?;
544                 }
545                 if($branch ne $last_branch) {
546                         unlink("$git_dir/HEAD");
547                         symlink("refs/heads/$branch","$git_dir/HEAD");
548                         $last_branch = $branch;
549                 }
550                 $state = 9;
551         } elsif($state == 8) {
552                 $logmsg .= "$_\n";
553         } elsif($state == 9 and /^\s+(\S+):(?:INITIAL|\d(?:\.\d+)+)->(\d(?:\.\d+)+)\s*$/) {
554 #       VERSION:1.96->1.96.2.1
555                 my $fn = $1;
556                 my $data = $cvs->file($fn,$2);
557                 print "Update $fn: ".length($data)." bytes.\n";
558                 mkpath(dirname($fn),$opt_v);
559                 open(F,"> ./$fn")
560                         or die "Cannot create '$fn': $!\n";
561                 print F $data
562                         or die "Cannot write to '$fn': $!\n";
563                 close(F)
564                         or die "Cannot write to '$fn': $!\n";
565                 chmod(pmode($cvs->{'mode'}), $fn);
566                 push(@new,$fn); # may be resurrected!
567         } elsif($state == 9 and /^\s+(\S+):\d(?:\.\d+)+->(\d(?:\.\d+)+)\(DEAD\)\s*$/) {
568                 push(@old,$1);
569         } elsif($state == 9 and /^\s*$/) {
570                 $state = 10;
571         } elsif(($state == 9 or $state == 10) and /^-+$/) {
572                 &$commit();
573                 $state = 1;
574         } elsif($state == 11 and /^-+$/) {
575                 $state = 1;
576         } elsif(/^-+$/) { # end of unknown-line processing
577                 $state = 1;
578         } elsif($state != 11) { # ignore stuff when skipping
579                 print "* UNKNOWN LINE * $_\n";
580         }
581 }
582 &$commit() if $branch and $state != 11;
583
584 print "DONE\n" if $opt_v;