Code

cvs import: Strip whitespace at the end of the log entry
[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); $|=1; select(STDOUT);
60 package CVSconn;
61 # Basic CVS dialog.
62 # We're only interested in connecting and downloading, so ...
64 use POSIX qw(strftime dup2);
66 sub new {
67         my($what,$repo,$subdir) = @_;
68         $what=ref($what) if ref($what);
70         my $self = {};
71         $self->{'buffer'} = "";
72         bless($self,$what);
74         $repo =~ s#/+$##;
75         $self->{'fullrep'} = $repo;
76         $self->conn();
78         $self->{'subdir'} = $subdir;
79         $self->{'lines'} = undef;
81         return $self;
82 }
84 sub conn {
85         my $self = shift;
86         my $repo = $self->{'fullrep'};
87         if($repo =~ s/^:pserver:(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
88                 my($user,$pass,$serv,$port) = ($1,$2,$3,$4);
89                 $user="anonymous" unless defined $user;
90                 my $rr2 = "-";
91                 unless($port) {
92                         $rr2 = ":pserver:$user\@$serv:$repo";
93                         $port=2401;
94                 }
95                 my $rr = ":pserver:$user\@$serv:$port$repo";
97                 unless($pass) {
98                         open(H,$ENV{'HOME'}."/.cvspass") and do {
99                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
100                                 while(<H>) {
101                                         chomp;
102                                         s/^\/\d+\s+//;
103                                         my ($w,$p) = split(/\s/,$_,2);
104                                         if($w eq $rr or $w eq $rr2) {
105                                                 $pass = $p;
106                                                 last;
107                                         }
108                                 }
109                         };
110                 }
111                 $pass="A" unless $pass;
113                 my $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
114                 die "Socket to $serv: $!\n" unless defined $s;
115                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
116                         or die "Write to $serv: $!\n";
117                 $s->flush();
119                 my $rep = <$s>;
121                 if($rep ne "I LOVE YOU\n") {
122                         $rep="<unknown>" unless $rep;
123                         die "AuthReply: $rep\n";
124                 }
125                 $self->{'socketo'} = $s;
126                 $self->{'socketi'} = $s;
127         } else { # local: Fork off our own cvs server.
128                 my $pr = IO::Pipe->new();
129                 my $pw = IO::Pipe->new();
130                 my $pid = fork();
131                 die "Fork: $!\n" unless defined $pid;
132                 unless($pid) {
133                         $pr->writer();
134                         $pw->reader();
135                         dup2($pw->fileno(),0);
136                         dup2($pr->fileno(),1);
137                         $pr->close();
138                         $pw->close();
139                         exec("cvs","server");
140                 }
141                 $pw->writer();
142                 $pr->reader();
143                 $self->{'socketo'} = $pw;
144                 $self->{'socketi'} = $pr;
145         }
146         $self->{'socketo'}->write("Root $repo\n");
148         # Trial and error says that this probably is the minimum set
149         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E F Checked-in Created Updated Merged Removed\n");
151         $self->{'socketo'}->write("valid-requests\n");
152         $self->{'socketo'}->flush();
154         chomp(my $rep=$self->readline());
155         if($rep !~ s/^Valid-requests\s*//) {
156                 $rep="<unknown>" unless $rep;
157                 die "Expected Valid-requests from server, but got: $rep\n";
158         }
159         chomp(my $res=$self->readline());
160         die "validReply: $res\n" if $res ne "ok";
162         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
163         $self->{'repo'} = $repo;
166 sub readline {
167         my($self) = @_;
168         return $self->{'socketi'}->getline();
171 sub _file {
172         # Request a file with a given revision.
173         # Trial and error says this is a good way to do it. :-/
174         my($self,$fn,$rev) = @_;
175         $self->{'socketo'}->write("Argument -N\n") or return undef;
176         $self->{'socketo'}->write("Argument -P\n") or return undef;
177         # $self->{'socketo'}->write("Argument -ko\n") or return undef;
178         # -ko: Linus' version doesn't use it
179         $self->{'socketo'}->write("Argument -r\n") or return undef;
180         $self->{'socketo'}->write("Argument $rev\n") or return undef;
181         $self->{'socketo'}->write("Argument --\n") or return undef;
182         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
183         $self->{'socketo'}->write("Directory .\n") or return undef;
184         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
185         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
186         $self->{'socketo'}->write("co\n") or return undef;
187         $self->{'socketo'}->flush() or return undef;
188         $self->{'lines'} = 0;
189         return 1;
191 sub _line {
192         # Read a line from the server.
193         # ... except that 'line' may be an entire file. ;-)
194         my($self) = @_;
195         die "Not in lines" unless defined $self->{'lines'};
197         my $line;
198         my $res="";
199         while(defined($line = $self->readline())) {
200                 # M U gnupg-cvs-rep/AUTHORS
201                 # Updated gnupg-cvs-rep/
202                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
203                 # /AUTHORS/1.1///T1.1
204                 # u=rw,g=rw,o=rw
205                 # 0
206                 # ok
208                 if($line =~ s/^(?:Created|Updated) //) {
209                         $line = $self->readline(); # path
210                         $line = $self->readline(); # Entries line
211                         my $mode = $self->readline(); chomp $mode;
212                         $self->{'mode'} = $mode;
213                         defined (my $cnt = $self->readline())
214                                 or die "EOF from server after 'Changed'\n";
215                         chomp $cnt;
216                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
217                         $line="";
218                         $res="";
219                         while($cnt) {
220                                 my $buf;
221                                 my $num = $self->{'socketi'}->read($buf,$cnt);
222                                 die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
223                                 $res .= $buf;
224                                 $cnt -= $num;
225                         }
226                 } elsif($line =~ s/^ //) {
227                         $res .= $line;
228                 } elsif($line =~ /^M\b/) {
229                         # output, do nothing
230                 } elsif($line =~ /^Mbinary\b/) {
231                         my $cnt;
232                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
233                         chomp $cnt;
234                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
235                         $line="";
236                         while($cnt) {
237                                 my $buf;
238                                 my $num = $self->{'socketi'}->read($buf,$cnt);
239                                 die "S: Mbinary $cnt: $num: $!\n" if not defined $num or $num<=0;
240                                 $res .= $buf;
241                                 $cnt -= $num;
242                         }
243                 } else {
244                         chomp $line;
245                         if($line eq "ok") {
246                                 # print STDERR "S: ok (".length($res).")\n";
247                                 return $res;
248                         } elsif($line =~ s/^E //) {
249                                 # print STDERR "S: $line\n";
250                         } else {
251                                 die "Unknown: $line\n";
252                         }
253                 }
254         }
256 sub file {
257         my($self,$fn,$rev) = @_;
258         my $res;
260         if ($self->_file($fn,$rev)) {
261                 $res = $self->_line();
262                 return $res if defined $res;
263         }
265         # retry
266         $self->conn();
267         $self->_file($fn,$rev)
268                 or die "No file command send\n";
269         $res = $self->_line();
270         die "No input: $fn $rev\n" unless defined $res;
271         return $res;
275 package main;
277 my $cvs = CVSconn->new($opt_d, $cvs_tree);
280 sub pdate($) {
281         my($d) = @_;
282         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
283                 or die "Unparseable date: $d\n";
284         my $y=$1; $y-=1900 if $y>1900;
285         return timegm($6||0,$5,$4,$3,$2-1,$y);
288 sub pmode($) {
289         my($mode) = @_;
290         my $m = 0;
291         my $mm = 0;
292         my $um = 0;
293         for my $x(split(//,$mode)) {
294                 if($x eq ",") {
295                         $m |= $mm&$um;
296                         $mm = 0;
297                         $um = 0;
298                 } elsif($x eq "u") { $um |= 0700;
299                 } elsif($x eq "g") { $um |= 0070;
300                 } elsif($x eq "o") { $um |= 0007;
301                 } elsif($x eq "r") { $mm |= 0444;
302                 } elsif($x eq "w") { $mm |= 0222;
303                 } elsif($x eq "x") { $mm |= 0111;
304                 } elsif($x eq "=") { # do nothing
305                 } else { die "Unknown mode: $mode\n";
306                 }
307         }
308         $m |= $mm&$um;
309         return $m;
312 my $tmpcv = "/var/cache/cvs";
314 sub getwd() {
315         my $pwd = `pwd`;
316         chomp $pwd;
317         return $pwd;
320 -d $git_tree
321         or mkdir($git_tree,0777)
322         or die "Could not create $git_tree: $!";
323 chdir($git_tree);
325 my $last_branch = "";
326 my $orig_branch = "";
327 my %branch_date;
329 my $git_dir = $ENV{"GIT_DIR"} || ".git";
330 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
331 $ENV{"GIT_DIR"} = $git_dir;
332 unless(-d $git_dir) {
333         system("git-init-db");
334         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
335         system("git-read-tree");
336         die "Cannot init an empty tree: $?\n" if $?;
338         $last_branch = $opt_o;
339         $orig_branch = "";
340 } else {
341         $last_branch = basename(readlink("$git_dir/HEAD"));
342         unless($last_branch) {
343                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
344                 $last_branch = "master";
345         }
346         $orig_branch = $last_branch;
348         # Get the last import timestamps
349         opendir(D,"$git_dir/refs/heads");
350         while(defined(my $head = readdir(D))) {
351                 next if $head =~ /^\./;
352                 open(F,"$git_dir/refs/heads/$head")
353                         or die "Bad head branch: $head: $!\n";
354                 chomp(my $ftag = <F>);
355                 close(F);
356                 open(F,"git-cat-file commit $ftag |");
357                 while(<F>) {
358                         next unless /^author\s.*\s(\d+)\s[-+]\d{4}$/;
359                         $branch_date{$head} = $1;
360                         last;
361                 }
362                 close(F);
363         }
364         closedir(D);
367 -d $git_dir
368         or die "Could not create git subdir ($git_dir).\n";
370 my $pid = open(CVS,"-|");
371 die "Cannot fork: $!\n" unless defined $pid;
372 unless($pid) {
373         exec("cvsps","-A","--cvs-direct",$cvs_tree);
374         die "Could not start cvsps: $!\n";
378 ## cvsps output:
379 #---------------------
380 #PatchSet 314
381 #Date: 1999/09/18 13:03:59
382 #Author: wkoch
383 #Branch: STABLE-BRANCH-1-0
384 #Ancestor branch: HEAD
385 #Tag: (none)
386 #Log:
387 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
388 #Members:
389 #       README:1.57->1.57.2.1
390 #       VERSION:1.96->1.96.2.1
392 #---------------------
394 my $state = 0;
396 my($patchset,$date,$author,$branch,$ancestor,$tag,$logmsg);
397 my(@old,@new);
398 my $commit = sub {
399         my $pid;
400         while(@old) {
401                 my @o2;
402                 if(@old > 55) {
403                         @o2 = splice(@old,0,50);
404                 } else {
405                         @o2 = @old;
406                         @old = ();
407                 }
408                 system("git-update-cache","--force-remove","--",@o2);
409                 die "Cannot remove files: $?\n" if $?;
410         }
411         while(@new) {
412                 my @n2;
413                 if(@new > 55) {
414                         @n2 = splice(@new,0,50);
415                 } else {
416                         @n2 = @new;
417                         @new = ();
418                 }
419                 system("git-update-cache","--add","--",@n2);
420                 die "Cannot add files: $?\n" if $?;
421         }
423         $pid = open(C,"-|");
424         die "Cannot fork: $!" unless defined $pid;
425         unless($pid) {
426                 exec("git-write-tree");
427                 die "Cannot exec git-write-tree: $!\n";
428         }
429         chomp(my $tree = <C>);
430         length($tree) == 40
431                 or die "Cannot get tree id ($tree): $!\n";
432         close(C)
433                 or die "Error running git-write-tree: $?\n";
434         print "Tree ID $tree\n" if $opt_v;
436         my $parent = "";
437         if(open(C,"$git_dir/refs/heads/$last_branch")) {
438                 chomp($parent = <C>);
439                 close(C);
440                 length($parent) == 40
441                         or die "Cannot get parent id ($parent): $!\n";
442                 print "Parent ID $parent\n" if $opt_v;
443         }
445         my $pr = IO::Pipe->new();
446         my $pw = IO::Pipe->new();
447         $pid = fork();
448         die "Fork: $!\n" unless defined $pid;
449         unless($pid) {
450                 $pr->writer();
451                 $pw->reader();
452                 dup2($pw->fileno(),0);
453                 dup2($pr->fileno(),1);
454                 $pr->close();
455                 $pw->close();
457                 my @par = ();
458                 @par = ("-p",$parent) if $parent;
459                 exec("env",
460                         "GIT_AUTHOR_NAME=$author",
461                         "GIT_AUTHOR_EMAIL=$author",
462                         "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
463                         "GIT_COMMITTER_NAME=$author",
464                         "GIT_COMMITTER_EMAIL=$author",
465                         "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
466                         "git-commit-tree", $tree,@par);
467                 die "Cannot exec git-commit-tree: $!\n";
468         }
469         $pw->writer();
470         $pr->reader();
472         # compatibility with git2cvs
473         substr($logmsg,32767) = "" if length($logmsg) > 32767;
474         $logmsg =~ s/[\s\n]+\z//;
476         print $pw "$logmsg\n"
477                 or die "Error writing to git-commit-tree: $!\n";
478         $pw->close();
480         print "Committed patch $patchset ($branch)\n" if $opt_v;
481         chomp(my $cid = <$pr>);
482         length($cid) == 40
483                 or die "Cannot get commit id ($cid): $!\n";
484         print "Commit ID $cid\n" if $opt_v;
485         $pr->close();
487         waitpid($pid,0);
488         die "Error running git-commit-tree: $?\n" if $?;
490         open(C,">$git_dir/refs/heads/$branch")
491                 or die "Cannot open branch $branch for update: $!\n";
492         print C "$cid\n"
493                 or die "Cannot write branch $branch for update: $!\n";
494         close(C)
495                 or die "Cannot write branch $branch for update: $!\n";
497         if($tag) {
498                 open(C,">$git_dir/refs/tags/$tag")
499                         or die "Cannot create tag $tag: $!\n";
500                 print C "$cid\n"
501                         or die "Cannot write tag $branch: $!\n";
502                 close(C)
503                         or die "Cannot write tag $branch: $!\n";
504                 print "Created tag '$tag' on '$branch'\n" if $opt_v;
505         }
506 };
508 while(<CVS>) {
509         chomp;
510         if($state == 0 and /^-+$/) {
511                 $state = 1;
512         } elsif($state == 0) {
513                 $state = 1;
514                 redo;
515         } elsif(($state==0 or $state==1) and s/^PatchSet\s+//) {
516                 $patchset = 0+$_;
517                 $state=2;
518         } elsif($state == 2 and s/^Date:\s+//) {
519                 $date = pdate($_);
520                 unless($date) {
521                         print STDERR "Could not parse date: $_\n";
522                         $state=0;
523                         next;
524                 }
525                 $state=3;
526         } elsif($state == 3 and s/^Author:\s+//) {
527                 s/\s+$//;
528                 $author = $_;
529                 $state = 4;
530         } elsif($state == 4 and s/^Branch:\s+//) {
531                 s/\s+$//;
532                 $branch = $_;
533                 $state = 5;
534         } elsif($state == 5 and s/^Ancestor branch:\s+//) {
535                 s/\s+$//;
536                 $ancestor = $_;
537                 $ancestor = $opt_o if $ancestor eq "HEAD";
538                 $state = 6;
539         } elsif($state == 5) {
540                 $ancestor = undef;
541                 $state = 6;
542                 redo;
543         } elsif($state == 6 and s/^Tag:\s+//) {
544                 s/\s+$//;
545                 if($_ eq "(none)") {
546                         $tag = undef;
547                 } else {
548                         $tag = $_;
549                 }
550                 $state = 7;
551         } elsif($state == 7 and /^Log:/) {
552                 $logmsg = "";
553                 $state = 8;
554         } elsif($state == 8 and /^Members:/) {
555                 $branch = $opt_o if $branch eq "HEAD";
556                 if(defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
557                         # skip
558                         print "skip patchset $patchset: $date before $branch_date{$branch}\n";
559                         $state = 11;
560                         next;
561                 }
562                 if($ancestor) {
563                         if(-f "$git_dir/refs/heads/$branch") {
564                                 print STDERR "Branch $branch already exists!\n";
565                                 $state=11;
566                                 next;
567                         }
568                         unless(open(H,"$git_dir/refs/heads/$ancestor")) {
569                                 print STDERR "Branch $ancestor does not exist!\n";
570                                 $state=11;
571                                 next;
572                         }
573                         chomp(my $id = <H>);
574                         close(H);
575                         unless(open(H,"> $git_dir/refs/heads/$branch")) {
576                                 print STDERR "Could not create branch $branch: $!\n";
577                                 $state=11;
578                                 next;
579                         }
580                         print H "$id\n"
581                                 or die "Could not write branch $branch: $!";
582                         close(H)
583                                 or die "Could not write branch $branch: $!";
584                 }
585                 if(($ancestor || $branch) ne $last_branch) {
586                         print "Switching from $last_branch to $branch\n" if $opt_v;
587                         system("git-read-tree","-m","-u","$last_branch","$branch");
588                         die "read-tree failed: $?\n" if $?;
589                 }
590                 if($branch ne $last_branch) {
591                         unlink("$git_dir/HEAD");
592                         symlink("refs/heads/$branch","$git_dir/HEAD");
593                         $last_branch = $branch;
594                 }
595                 $state = 9;
596         } elsif($state == 8) {
597                 $logmsg .= "$_\n";
598         } elsif($state == 9 and /^\s+(\S+):(INITIAL|\d(?:\.\d+)+)->(\d(?:\.\d+)+)\s*$/) {
599 #       VERSION:1.96->1.96.2.1
600                 my $init = ($2 eq "INITIAL");
601                 my $fn = $1;
602                 my $rev = $3;
603                 $fn =~ s#^/+##;
604                 my $data = $cvs->file($fn,$rev);
605                 print "".($init ? "New" : "Update")." $fn: ".length($data)." bytes.\n";
606                 mkpath(dirname($fn),$opt_v);
607                 open(F,"> ./$fn")
608                         or die "Cannot create '$fn': $!\n";
609                 print F $data
610                         or die "Cannot write to '$fn': $!\n";
611                 close(F)
612                         or die "Cannot write to '$fn': $!\n";
613                 chmod(pmode($cvs->{'mode'}), $fn);
614                 push(@new,$fn); # may be resurrected!
615         } elsif($state == 9 and /^\s+(\S+):\d(?:\.\d+)+->(\d(?:\.\d+)+)\(DEAD\)\s*$/) {
616                 my $fn = $1;
617                 $fn =~ s#^/+##;
618                 push(@old,$fn);
619         } elsif($state == 9 and /^\s*$/) {
620                 $state = 10;
621         } elsif(($state == 9 or $state == 10) and /^-+$/) {
622                 &$commit();
623                 $state = 1;
624         } elsif($state == 11 and /^-+$/) {
625                 $state = 1;
626         } elsif(/^-+$/) { # end of unknown-line processing
627                 $state = 1;
628         } elsif($state != 11) { # ignore stuff when skipping
629                 print "* UNKNOWN LINE * $_\n";
630         }
632 &$commit() if $branch and $state != 11;
634 # Now switch back to the branch we were in before all of this happened
635 if($orig_branch) {
636         print "DONE; switching back to $orig_branch\n" if $opt_v;
637 } else {
638         $orig_branch = "master";
639         print "DONE; creating $orig_branch branch\n" if $opt_v;
640         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
641                 unless -f "$git_dir/refs/heads/master";
644 system("git-read-tree","-m","-u","$last_branch","$orig_branch");
645 die "read-tree failed: $?\n" if $?;
647 unlink("$git_dir/HEAD");
648 symlink("refs/heads/$orig_branch","$git_dir/HEAD");