Code

Fix remote CVS import.
[git.git] / git-cvsimport-script
1 #!/usr/bin/perl -w
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.
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 IO::Socket;
23 use IO::Pipe;
24 use POSIX qw(strftime dup2);
26 $SIG{'PIPE'}="IGNORE";
27 $ENV{'TZ'}="UTC";
29 our($opt_h,$opt_o,$opt_v,$opt_d);
31 sub usage() {
32         print STDERR <<END;
33 Usage: ${\basename $0}     # fetch/update GIT from CVS
34            [ -o branch-for-HEAD ] [ -h ] [ -v ] [ -d CVSROOT ]
35        CVS_module [ GIT_repository ]
36 END
37         exit(1);
38 }
40 getopts("hqvo:d:") or usage();
41 usage if $opt_h;
43 @ARGV == 1 or @ARGV == 2 or usage();
45 my($cvs_tree, $git_tree) = @ARGV;
47 if($opt_d) {
48         $ENV{"CVSROOT"} = $opt_d;
49 } elsif($ENV{"CVSROOT"}) {
50         $opt_d = $ENV{"CVSROOT"};
51 } else {
52         die "CVSROOT needs to be set";
53 }
54 $opt_o ||= "origin";
55 $git_tree ||= ".";
57 select(STDERR); $|=1; select(STDOUT);
60 package CVSconn;
61 # Basic CVS dialog.
62 # We're only interested in connecting and downloading, so ...
64 sub new {
65         my($what,$repo,$subdir) = @_;
66         $what=ref($what) if ref($what);
68         my $self = {};
69         $self->{'buffer'} = "";
70         bless($self,$what);
72         $repo =~ s#/+$##;
73         $self->{'fullrep'} = $repo;
74         $self->conn();
76         $self->{'subdir'} = $subdir;
77         $self->{'lines'} = undef;
79         return $self;
80 }
82 sub conn {
83         my $self = shift;
84         my $repo = $self->{'fullrep'};
85         if($repo =~ s/^:pserver:(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
86                 my($user,$pass,$serv,$port) = ($1,$2,$3,$4);
87                 $user="anonymous" unless defined $user;
88                 my $rr2 = "-";
89                 unless($port) {
90                         $rr2 = ":pserver:$user\@$serv:$repo";
91                         $port=2401;
92                 }
93                 my $rr = ":pserver:$user\@$serv:$port$repo";
95                 unless($pass) {
96                         open(H,$ENV{'HOME'}."/.cvspass") and do {
97                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
98                                 while(<H>) {
99                                         chomp;
100                                         s/^\/\d+\s+//;
101                                         my ($w,$p) = split(/\s/,$_,2);
102                                         if($w eq $rr or $w eq $rr2) {
103                                                 $pass = $p;
104                                                 last;
105                                         }
106                                 }
107                         };
108                 }
109                 $pass="A" unless $pass;
111                 my $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
112                 die "Socket to $serv: $!\n" unless defined $s;
113                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
114                         or die "Write to $serv: $!\n";
115                 $s->flush();
117                 my $rep = <$s>;
119                 if($rep ne "I LOVE YOU\n") {
120                         $rep="<unknown>" unless $rep;
121                         die "AuthReply: $rep\n";
122                 }
123                 $self->{'socketo'} = $s;
124                 $self->{'socketi'} = $s;
125         } else { # local: Fork off our own cvs server.
126                 my $pr = IO::Pipe->new();
127                 my $pw = IO::Pipe->new();
128                 my $pid = fork();
129                 die "Fork: $!\n" unless defined $pid;
130                 unless($pid) {
131                         $pr->writer();
132                         $pw->reader();
133                         dup2($pw->fileno(),0);
134                         dup2($pr->fileno(),1);
135                         $pr->close();
136                         $pw->close();
137                         exec("cvs","server");
138                 }
139                 $pw->writer();
140                 $pr->reader();
141                 $self->{'socketo'} = $pw;
142                 $self->{'socketi'} = $pr;
143         }
144         $self->{'socketo'}->write("Root $repo\n");
146         # Trial and error says that this probably is the minimum set
147         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E F Checked-in Created Updated Merged Removed\n");
149         $self->{'socketo'}->write("valid-requests\n");
150         $self->{'socketo'}->flush();
152         chomp(my $rep=$self->readline());
153         if($rep !~ s/^Valid-requests\s*//) {
154                 $rep="<unknown>" unless $rep;
155                 die "Expected Valid-requests from server, but got: $rep\n";
156         }
157         chomp(my $res=$self->readline());
158         die "validReply: $res\n" if $res ne "ok";
160         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
161         $self->{'repo'} = $repo;
164 sub readline {
165         my($self) = @_;
166         return $self->{'socketi'}->getline();
169 sub _file {
170         # Request a file with a given revision.
171         # Trial and error says this is a good way to do it. :-/
172         my($self,$fn,$rev) = @_;
173         $self->{'socketo'}->write("Argument -N\n") or return undef;
174         $self->{'socketo'}->write("Argument -P\n") or return undef;
175         # $self->{'socketo'}->write("Argument -ko\n") or return undef;
176         # -ko: Linus' version doesn't use it
177         $self->{'socketo'}->write("Argument -r\n") or return undef;
178         $self->{'socketo'}->write("Argument $rev\n") or return undef;
179         $self->{'socketo'}->write("Argument --\n") or return undef;
180         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
181         $self->{'socketo'}->write("Directory .\n") or return undef;
182         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
183         $self->{'socketo'}->write("Sticky T1.1\n") or return undef;
184         $self->{'socketo'}->write("co\n") or return undef;
185         $self->{'socketo'}->flush() or return undef;
186         $self->{'lines'} = 0;
187         return 1;
189 sub _line {
190         # Read a line from the server.
191         # ... except that 'line' may be an entire file. ;-)
192         my($self) = @_;
193         die "Not in lines" unless defined $self->{'lines'};
195         my $line;
196         my $res="";
197         while(defined($line = $self->readline())) {
198                 # M U gnupg-cvs-rep/AUTHORS
199                 # Updated gnupg-cvs-rep/
200                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
201                 # /AUTHORS/1.1///T1.1
202                 # u=rw,g=rw,o=rw
203                 # 0
204                 # ok
206                 if($line =~ s/^(?:Created|Updated) //) {
207                         $line = $self->readline(); # path
208                         $line = $self->readline(); # Entries line
209                         my $mode = $self->readline(); chomp $mode;
210                         $self->{'mode'} = $mode;
211                         defined (my $cnt = $self->readline())
212                                 or die "EOF from server after 'Changed'\n";
213                         chomp $cnt;
214                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
215                         $line="";
216                         $res="";
217                         while($cnt) {
218                                 my $buf;
219                                 my $num = $self->{'socketi'}->read($buf,$cnt);
220                                 die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
221                                 $res .= $buf;
222                                 $cnt -= $num;
223                         }
224                 } elsif($line =~ s/^ //) {
225                         $res .= $line;
226                 } elsif($line =~ /^M\b/) {
227                         # output, do nothing
228                 } elsif($line =~ /^Mbinary\b/) {
229                         my $cnt;
230                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
231                         chomp $cnt;
232                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
233                         $line="";
234                         while($cnt) {
235                                 my $buf;
236                                 my $num = $self->{'socketi'}->read($buf,$cnt);
237                                 die "S: Mbinary $cnt: $num: $!\n" if not defined $num or $num<=0;
238                                 $res .= $buf;
239                                 $cnt -= $num;
240                         }
241                 } else {
242                         chomp $line;
243                         if($line eq "ok") {
244                                 # print STDERR "S: ok (".length($res).")\n";
245                                 return $res;
246                         } elsif($line =~ s/^E //) {
247                                 # print STDERR "S: $line\n";
248                         } else {
249                                 die "Unknown: $line\n";
250                         }
251                 }
252         }
254 sub file {
255         my($self,$fn,$rev) = @_;
256         my $res;
258         if ($self->_file($fn,$rev)) {
259                 $res = $self->_line();
260                 return $res if defined $res;
261         }
263         # retry
264         $self->conn();
265         $self->_file($fn,$rev)
266                 or die "No file command send\n";
267         $res = $self->_line();
268         die "No input: $fn $rev\n" unless defined $res;
269         return $res;
273 package main;
275 my $cvs = CVSconn->new($opt_d, $cvs_tree);
278 sub pdate($) {
279         my($d) = @_;
280         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
281                 or die "Unparseable date: $d\n";
282         my $y=$1; $y-=1900 if $y>1900;
283         return timegm($6||0,$5,$4,$3,$2-1,$y);
286 sub pmode($) {
287         my($mode) = @_;
288         my $m = 0;
289         my $mm = 0;
290         my $um = 0;
291         for my $x(split(//,$mode)) {
292                 if($x eq ",") {
293                         $m |= $mm&$um;
294                         $mm = 0;
295                         $um = 0;
296                 } elsif($x eq "u") { $um |= 0700;
297                 } elsif($x eq "g") { $um |= 0070;
298                 } elsif($x eq "o") { $um |= 0007;
299                 } elsif($x eq "r") { $mm |= 0444;
300                 } elsif($x eq "w") { $mm |= 0222;
301                 } elsif($x eq "x") { $mm |= 0111;
302                 } elsif($x eq "=") { # do nothing
303                 } else { die "Unknown mode: $mode\n";
304                 }
305         }
306         $m |= $mm&$um;
307         return $m;
310 my $tmpcv = "/var/cache/cvs";
312 sub getwd() {
313         my $pwd = `pwd`;
314         chomp $pwd;
315         return $pwd;
318 -d $git_tree
319         or mkdir($git_tree,0777)
320         or die "Could not create $git_tree: $!";
321 chdir($git_tree);
323 my $last_branch = "";
324 my %branch_date;
326 my $git_dir = $ENV{"GIT_DIR"} || ".git";
327 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
328 $ENV{"GIT_DIR"} = $git_dir;
329 unless(-d $git_dir) {
330         system("git-init-db");
331         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
332         system("git-read-tree");
333         die "Cannot init an empty tree: $?\n" if $?;
335         $last_branch = $opt_o;
336 } else {
337         $last_branch = basename(readlink("$git_dir/HEAD"));
338         die "Cannot read the last branch name: $!\n" unless $last_branch;
340         # Get the last import timestamps
341         opendir(D,"$git_dir/refs/heads");
342         while(defined(my $head = readdir(D))) {
343                 next if $head =~ /^\./;
344                 open(F,"$git_dir/refs/heads/$head")
345                         or die "Bad head branch: $head: $!\n";
346                 chomp(my $ftag = <F>);
347                 close(F);
348                 open(F,"git-cat-file commit $ftag |");
349                 while(<F>) {
350                         next unless /^author\s.*\s(\d+)\s[-+]\d{4}$/;
351                         $branch_date{$head} = $1;
352                         last;
353                 }
354                 close(F);
355         }
356         closedir(D);
359 -d $git_dir
360         or die "Could not create git subdir ($git_dir).\n";
362 my $pid = open(CVS,"-|");
363 die "Cannot fork: $!\n" unless defined $pid;
364 unless($pid) {
365         exec("cvsps","-A","--cvs-direct",$cvs_tree);
366         die "Could not start cvsps: $!\n";
370 ## cvsps output:
371 #---------------------
372 #PatchSet 314
373 #Date: 1999/09/18 13:03:59
374 #Author: wkoch
375 #Branch: STABLE-BRANCH-1-0
376 #Ancestor branch: HEAD
377 #Tag: (none)
378 #Log:
379 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
380 #Members:
381 #       README:1.57->1.57.2.1
382 #       VERSION:1.96->1.96.2.1
384 #---------------------
386 my $state = 0;
388 my($patchset,$date,$author,$branch,$ancestor,$tag,$logmsg);
389 my(@old,@new);
390 my $commit = sub {
391         my $pid;
392         system("git-update-cache","--force-remove","--",@old) if @old;
393         die "Cannot remove files: $?\n" if $?;
394         system("git-update-cache","--add","--",@new) if @new;
395         die "Cannot add files: $?\n" if $?;
397         $pid = open(C,"-|");
398         die "Cannot fork: $!" unless defined $pid;
399         unless($pid) {
400                 exec("git-write-tree");
401                 die "Cannot exec git-write-tree: $!\n";
402         }
403         chomp(my $tree = <C>);
404         length($tree) == 40
405                 or die "Cannot get tree id ($tree): $!\n";
406         close(C)
407                 or die "Error running git-write-tree: $?\n";
408         print "Tree ID $tree\n" if $opt_v;
410         my $parent = "";
411         if(open(C,"$git_dir/refs/heads/$last_branch")) {
412                 chomp($parent = <C>);
413                 close(C);
414                 length($parent) == 40
415                         or die "Cannot get parent id ($parent): $!\n";
416                 print "Parent ID $parent\n" if $opt_v;
417         }
419         my $pr = IO::Pipe->new();
420         my $pw = IO::Pipe->new();
421         $pid = fork();
422         die "Fork: $!\n" unless defined $pid;
423         unless($pid) {
424                 $pr->writer();
425                 $pw->reader();
426                 dup2($pw->fileno(),0);
427                 dup2($pr->fileno(),1);
428                 $pr->close();
429                 $pw->close();
431                 my @par = ();
432                 @par = ("-p",$parent) if $parent;
433                 exec("env",
434                         "GIT_AUTHOR_NAME=$author",
435                         "GIT_AUTHOR_EMAIL=$author",
436                         "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
437                         "GIT_COMMITTER_NAME=$author",
438                         "GIT_COMMITTER_EMAIL=$author",
439                         "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
440                         "git-commit-tree", $tree,@par);
441                 die "Cannot exec git-commit-tree: $!\n";
442         }
443         $pw->writer();
444         $pr->reader();
445         print $pw $logmsg
446                 or die "Error writing to git-commit-tree: $!\n";
447         $pw->close();
449         print "Committed patch $patchset ($branch)\n" if $opt_v;
450         chomp(my $cid = <$pr>);
451         length($cid) == 40
452                 or die "Cannot get commit id ($cid): $!\n";
453         print "Commit ID $cid\n" if $opt_v;
454         $pr->close();
456         waitpid($pid,0);
457         die "Error running git-commit-tree: $?\n" if $?;
459         open(C,">$git_dir/refs/heads/$branch")
460                 or die "Cannot open branch $branch for update: $!\n";
461         print C "$cid\n"
462                 or die "Cannot write branch $branch for update: $!\n";
463         close(C)
464                 or die "Cannot write branch $branch for update: $!\n";
466         if($tag) {
467                 open(C,">$git_dir/refs/tags/$tag")
468                         or die "Cannot create tag $tag: $!\n";
469                 print C "$cid\n"
470                         or die "Cannot write tag $branch: $!\n";
471                 close(C)
472                         or die "Cannot write tag $branch: $!\n";
473                 print "Created tag '$tag' on '$branch'\n" if $opt_v;
474         }
476         @old = ();
477         @new = ();
478 };
480 while(<CVS>) {
481         chomp;
482         if($state == 0 and /^-+$/) {
483                 $state = 1;
484         } elsif($state == 0) {
485                 $state = 1;
486                 redo;
487         } elsif(($state==0 or $state==1) and s/^PatchSet\s+//) {
488                 $patchset = 0+$_;
489                 $state=2;
490         } elsif($state == 2 and s/^Date:\s+//) {
491                 $date = pdate($_);
492                 unless($date) {
493                         print STDERR "Could not parse date: $_\n";
494                         $state=0;
495                         next;
496                 }
497                 $state=3;
498         } elsif($state == 3 and s/^Author:\s+//) {
499                 s/\s+$//;
500                 $author = $_;
501                 $state = 4;
502         } elsif($state == 4 and s/^Branch:\s+//) {
503                 s/\s+$//;
504                 $branch = $_;
505                 $state = 5;
506         } elsif($state == 5 and s/^Ancestor branch:\s+//) {
507                 s/\s+$//;
508                 $ancestor = $_;
509                 $ancestor = $opt_o if $ancestor == "HEAD";
510                 $state = 6;
511         } elsif($state == 5) {
512                 $ancestor = undef;
513                 $state = 6;
514                 redo;
515         } elsif($state == 6 and s/^Tag:\s+//) {
516                 s/\s+$//;
517                 if($_ eq "(none)") {
518                         $tag = undef;
519                 } else {
520                         $tag = $_;
521                 }
522                 $state = 7;
523         } elsif($state == 7 and /^Log:/) {
524                 $logmsg = "";
525                 $state = 8;
526         } elsif($state == 8 and /^Members:/) {
527                 $branch = $opt_o if $branch eq "HEAD";
528                 if(defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
529                         # skip
530                         print "skip patchset $patchset: $date before $branch_date{$branch}\n";
531                         $state = 11;
532                         next;
533                 }
534                 if($ancestor) {
535                         if(-f "$git_dir/refs/heads/$branch") {
536                                 print STDERR "Branch $branch already exists!\n";
537                                 $state=11;
538                                 next;
539                         }
540                         unless(open(H,"$git_dir/refs/heads/$ancestor")) {
541                                 print STDERR "Branch $ancestor does not exist!\n";
542                                 $state=11;
543                                 next;
544                         }
545                         chomp(my $id = <H>);
546                         close(H);
547                         unless(open(H,"> $git_dir/refs/heads/$branch")) {
548                                 print STDERR "Could not create branch $branch: $!\n";
549                                 $state=11;
550                                 next;
551                         }
552                         print H "$id\n"
553                                 or die "Could not write branch $branch: $!";
554                         close(H)
555                                 or die "Could not write branch $branch: $!";
556                 }
557                 if(($ancestor || $branch) ne $last_branch) {
558                         print "Switching from $last_branch to $branch\n" if $opt_v;
559                         system("git-read-tree -m -u $last_branch $branch");
560                         die "read-tree $branch failed at $branch: $?\n" if $?;
561                 }
562                 if($branch ne $last_branch) {
563                         unlink("$git_dir/HEAD");
564                         symlink("refs/heads/$branch","$git_dir/HEAD");
565                         $last_branch = $branch;
566                 }
567                 $state = 9;
568         } elsif($state == 8) {
569                 $logmsg .= "$_\n";
570         } elsif($state == 9 and /^\s+(\S+):(INITIAL|\d(?:\.\d+)+)->(\d(?:\.\d+)+)\s*$/) {
571 #       VERSION:1.96->1.96.2.1
572                 my $init = ($2 eq "INITIAL");
573                 my $fn = $1;
574                 my $data = $cvs->file($fn,$3);
575                 print "".($init ? "New" : "Update")." $fn: ".length($data)." bytes.\n";
576                 mkpath(dirname($fn),$opt_v);
577                 open(F,"> ./$fn")
578                         or die "Cannot create '$fn': $!\n";
579                 print F $data
580                         or die "Cannot write to '$fn': $!\n";
581                 close(F)
582                         or die "Cannot write to '$fn': $!\n";
583                 chmod(pmode($cvs->{'mode'}), $fn);
584                 push(@new,$fn); # may be resurrected!
585         } elsif($state == 9 and /^\s+(\S+):\d(?:\.\d+)+->(\d(?:\.\d+)+)\(DEAD\)\s*$/) {
586                 push(@old,$1);
587         } elsif($state == 9 and /^\s*$/) {
588                 $state = 10;
589         } elsif(($state == 9 or $state == 10) and /^-+$/) {
590                 &$commit();
591                 $state = 1;
592         } elsif($state == 11 and /^-+$/) {
593                 $state = 1;
594         } elsif(/^-+$/) { # end of unknown-line processing
595                 $state = 1;
596         } elsif($state != 11) { # ignore stuff when skipping
597                 print "* UNKNOWN LINE * $_\n";
598         }
600 &$commit() if $branch and $state != 11;
602 print "DONE\n" if $opt_v;