Merge branch 'jc/waitpid'
[git] / git-cvsimport.perl
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::Spec;
20 use File::Temp qw(tempfile);
21 use File::Path qw(mkpath);
22 use File::Basename qw(basename dirname);
23 use Time::Local;
24 use IO::Socket;
25 use IO::Pipe;
26 use POSIX qw(strftime dup2 ENOENT);
27 use IPC::Open2;
28
29 $SIG{'PIPE'}="IGNORE";
30 $ENV{'TZ'}="UTC";
31
32 our($opt_h,$opt_o,$opt_v,$opt_k,$opt_u,$opt_d,$opt_p,$opt_C,$opt_z,$opt_i,$opt_P, $opt_s,$opt_m,$opt_M,$opt_A,$opt_S,$opt_L);
33 my (%conv_author_name, %conv_author_email);
34
35 sub usage() {
36         print STDERR <<END;
37 Usage: ${\basename $0}     # fetch/update GIT from CVS
38        [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
39        [-p opts-for-cvsps] [-C GIT_repository] [-z fuzz] [-i] [-k] [-u]
40        [-s subst] [-m] [-M regex] [-S regex] [CVS_module]
41 END
42         exit(1);
43 }
44
45 sub read_author_info($) {
46         my ($file) = @_;
47         my $user;
48         open my $f, '<', "$file" or die("Failed to open $file: $!\n");
49
50         while (<$f>) {
51                 # Expected format is this:
52                 #   exon=Andreas Ericsson <ae@op5.se>
53                 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
54                         $user = $1;
55                         $conv_author_name{$user} = $2;
56                         $conv_author_email{$user} = $3;
57                 }
58                 # However, we also read from CVSROOT/users format
59                 # to ease migration.
60                 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
61                         my $mapped;
62                         ($user, $mapped) = ($1, $3);
63                         if ($mapped =~ /^\s*(.*?)\s*<(.*)>\s*$/) {
64                                 $conv_author_name{$user} = $1;
65                                 $conv_author_email{$user} = $2;
66                         }
67                         elsif ($mapped =~ /^<?(.*)>?$/) {
68                                 $conv_author_name{$user} = $user;
69                                 $conv_author_email{$user} = $1;
70                         }
71                 }
72                 # NEEDSWORK: Maybe warn on unrecognized lines?
73         }
74         close ($f);
75 }
76
77 sub write_author_info($) {
78         my ($file) = @_;
79         open my $f, '>', $file or
80           die("Failed to open $file for writing: $!");
81
82         foreach (keys %conv_author_name) {
83                 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
84         }
85         close ($f);
86 }
87
88 getopts("hivmkuo:d:p:C:z:s:M:P:A:S:L:") or usage();
89 usage if $opt_h;
90
91 @ARGV <= 1 or usage();
92
93 if($opt_d) {
94         $ENV{"CVSROOT"} = $opt_d;
95 } elsif(-f 'CVS/Root') {
96         open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
97         $opt_d = <$f>;
98         chomp $opt_d;
99         close $f;
100         $ENV{"CVSROOT"} = $opt_d;
101 } elsif($ENV{"CVSROOT"}) {
102         $opt_d = $ENV{"CVSROOT"};
103 } else {
104         die "CVSROOT needs to be set";
105 }
106 $opt_o ||= "origin";
107 $opt_s ||= "-";
108 my $git_tree = $opt_C;
109 $git_tree ||= ".";
110
111 my $cvs_tree;
112 if ($#ARGV == 0) {
113         $cvs_tree = $ARGV[0];
114 } elsif (-f 'CVS/Repository') {
115         open my $f, '<', 'CVS/Repository' or 
116             die 'Failed to open CVS/Repository';
117         $cvs_tree = <$f>;
118         chomp $cvs_tree;
119         close $f;
120 } else {
121         usage();
122 }
123
124 our @mergerx = ();
125 if ($opt_m) {
126         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
127 }
128 if ($opt_M) {
129         push (@mergerx, qr/$opt_M/);
130 }
131
132 select(STDERR); $|=1; select(STDOUT);
133
134
135 package CVSconn;
136 # Basic CVS dialog.
137 # We're only interested in connecting and downloading, so ...
138
139 use File::Spec;
140 use File::Temp qw(tempfile);
141 use POSIX qw(strftime dup2);
142
143 sub new {
144         my($what,$repo,$subdir) = @_;
145         $what=ref($what) if ref($what);
146
147         my $self = {};
148         $self->{'buffer'} = "";
149         bless($self,$what);
150
151         $repo =~ s#/+$##;
152         $self->{'fullrep'} = $repo;
153         $self->conn();
154
155         $self->{'subdir'} = $subdir;
156         $self->{'lines'} = undef;
157
158         return $self;
159 }
160
161 sub conn {
162         my $self = shift;
163         my $repo = $self->{'fullrep'};
164         if($repo =~ s/^:pserver:(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
165                 my($user,$pass,$serv,$port) = ($1,$2,$3,$4);
166                 $user="anonymous" unless defined $user;
167                 my $rr2 = "-";
168                 unless($port) {
169                         $rr2 = ":pserver:$user\@$serv:$repo";
170                         $port=2401;
171                 }
172                 my $rr = ":pserver:$user\@$serv:$port$repo";
173
174                 unless($pass) {
175                         open(H,$ENV{'HOME'}."/.cvspass") and do {
176                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
177                                 while(<H>) {
178                                         chomp;
179                                         s/^\/\d+\s+//;
180                                         my ($w,$p) = split(/\s/,$_,2);
181                                         if($w eq $rr or $w eq $rr2) {
182                                                 $pass = $p;
183                                                 last;
184                                         }
185                                 }
186                         };
187                 }
188                 $pass="A" unless $pass;
189
190                 my $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
191                 die "Socket to $serv: $!\n" unless defined $s;
192                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
193                         or die "Write to $serv: $!\n";
194                 $s->flush();
195
196                 my $rep = <$s>;
197
198                 if($rep ne "I LOVE YOU\n") {
199                         $rep="<unknown>" unless $rep;
200                         die "AuthReply: $rep\n";
201                 }
202                 $self->{'socketo'} = $s;
203                 $self->{'socketi'} = $s;
204         } else { # local or ext: Fork off our own cvs server.
205                 my $pr = IO::Pipe->new();
206                 my $pw = IO::Pipe->new();
207                 my $pid = fork();
208                 die "Fork: $!\n" unless defined $pid;
209                 my $cvs = 'cvs';
210                 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
211                 my $rsh = 'rsh';
212                 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
213
214                 my @cvs = ($cvs, 'server');
215                 my ($local, $user, $host);
216                 $local = $repo =~ s/:local://;
217                 if (!$local) {
218                     $repo =~ s/:ext://;
219                     $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
220                     ($user, $host) = ($1, $2);
221                 }
222                 if (!$local) {
223                     if ($user) {
224                         unshift @cvs, $rsh, '-l', $user, $host;
225                     } else {
226                         unshift @cvs, $rsh, $host;
227                     }
228                 }
229
230                 unless($pid) {
231                         $pr->writer();
232                         $pw->reader();
233                         dup2($pw->fileno(),0);
234                         dup2($pr->fileno(),1);
235                         $pr->close();
236                         $pw->close();
237                         exec(@cvs);
238                 }
239                 $pw->writer();
240                 $pr->reader();
241                 $self->{'socketo'} = $pw;
242                 $self->{'socketi'} = $pr;
243         }
244         $self->{'socketo'}->write("Root $repo\n");
245
246         # Trial and error says that this probably is the minimum set
247         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
248
249         $self->{'socketo'}->write("valid-requests\n");
250         $self->{'socketo'}->flush();
251
252         chomp(my $rep=$self->readline());
253         if($rep !~ s/^Valid-requests\s*//) {
254                 $rep="<unknown>" unless $rep;
255                 die "Expected Valid-requests from server, but got: $rep\n";
256         }
257         chomp(my $res=$self->readline());
258         die "validReply: $res\n" if $res ne "ok";
259
260         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
261         $self->{'repo'} = $repo;
262 }
263
264 sub readline {
265         my($self) = @_;
266         return $self->{'socketi'}->getline();
267 }
268
269 sub _file {
270         # Request a file with a given revision.
271         # Trial and error says this is a good way to do it. :-/
272         my($self,$fn,$rev) = @_;
273         $self->{'socketo'}->write("Argument -N\n") or return undef;
274         $self->{'socketo'}->write("Argument -P\n") or return undef;
275         # -kk: Linus' version doesn't use it - defaults to off
276         if ($opt_k) {
277             $self->{'socketo'}->write("Argument -kk\n") or return undef;
278         }
279         $self->{'socketo'}->write("Argument -r\n") or return undef;
280         $self->{'socketo'}->write("Argument $rev\n") or return undef;
281         $self->{'socketo'}->write("Argument --\n") or return undef;
282         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
283         $self->{'socketo'}->write("Directory .\n") or return undef;
284         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
285         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
286         $self->{'socketo'}->write("co\n") or return undef;
287         $self->{'socketo'}->flush() or return undef;
288         $self->{'lines'} = 0;
289         return 1;
290 }
291 sub _line {
292         # Read a line from the server.
293         # ... except that 'line' may be an entire file. ;-)
294         my($self, $fh) = @_;
295         die "Not in lines" unless defined $self->{'lines'};
296
297         my $line;
298         my $res=0;
299         while(defined($line = $self->readline())) {
300                 # M U gnupg-cvs-rep/AUTHORS
301                 # Updated gnupg-cvs-rep/
302                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
303                 # /AUTHORS/1.1///T1.1
304                 # u=rw,g=rw,o=rw
305                 # 0
306                 # ok
307
308                 if($line =~ s/^(?:Created|Updated) //) {
309                         $line = $self->readline(); # path
310                         $line = $self->readline(); # Entries line
311                         my $mode = $self->readline(); chomp $mode;
312                         $self->{'mode'} = $mode;
313                         defined (my $cnt = $self->readline())
314                                 or die "EOF from server after 'Changed'\n";
315                         chomp $cnt;
316                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
317                         $line="";
318                         $res = $self->_fetchfile($fh, $cnt);
319                 } elsif($line =~ s/^ //) {
320                         print $fh $line;
321                         $res += length($line);
322                 } elsif($line =~ /^M\b/) {
323                         # output, do nothing
324                 } elsif($line =~ /^Mbinary\b/) {
325                         my $cnt;
326                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
327                         chomp $cnt;
328                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
329                         $line="";
330                         $res += $self->_fetchfile($fh, $cnt);
331                 } else {
332                         chomp $line;
333                         if($line eq "ok") {
334                                 # print STDERR "S: ok (".length($res).")\n";
335                                 return $res;
336                         } elsif($line =~ s/^E //) {
337                                 # print STDERR "S: $line\n";
338                         } elsif($line =~ /^(Remove-entry|Removed) /i) {
339                                 $line = $self->readline(); # filename
340                                 $line = $self->readline(); # OK
341                                 chomp $line;
342                                 die "Unknown: $line" if $line ne "ok";
343                                 return -1;
344                         } else {
345                                 die "Unknown: $line\n";
346                         }
347                 }
348         }
349         return undef;
350 }
351 sub file {
352         my($self,$fn,$rev) = @_;
353         my $res;
354
355         my ($fh, $name) = tempfile('gitcvs.XXXXXX', 
356                     DIR => File::Spec->tmpdir(), UNLINK => 1);
357
358         $self->_file($fn,$rev) and $res = $self->_line($fh);
359
360         if (!defined $res) {
361             print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
362             truncate $fh, 0;
363             $self->conn();
364             $self->_file($fn,$rev) or die "No file command send";
365             $res = $self->_line($fh);
366             die "Retry failed" unless defined $res;
367         }
368         close ($fh);
369
370         return ($name, $res);
371 }
372 sub _fetchfile {
373         my ($self, $fh, $cnt) = @_;
374         my $res = 0;
375         my $bufsize = 1024 * 1024;
376         while($cnt) {
377             if ($bufsize > $cnt) {
378                 $bufsize = $cnt;
379             }
380             my $buf;
381             my $num = $self->{'socketi'}->read($buf,$bufsize);
382             die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
383             print $fh $buf;
384             $res += $num;
385             $cnt -= $num;
386         }
387         return $res;
388 }
389
390
391 package main;
392
393 my $cvs = CVSconn->new($opt_d, $cvs_tree);
394
395
396 sub pdate($) {
397         my($d) = @_;
398         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
399                 or die "Unparseable date: $d\n";
400         my $y=$1; $y-=1900 if $y>1900;
401         return timegm($6||0,$5,$4,$3,$2-1,$y);
402 }
403
404 sub pmode($) {
405         my($mode) = @_;
406         my $m = 0;
407         my $mm = 0;
408         my $um = 0;
409         for my $x(split(//,$mode)) {
410                 if($x eq ",") {
411                         $m |= $mm&$um;
412                         $mm = 0;
413                         $um = 0;
414                 } elsif($x eq "u") { $um |= 0700;
415                 } elsif($x eq "g") { $um |= 0070;
416                 } elsif($x eq "o") { $um |= 0007;
417                 } elsif($x eq "r") { $mm |= 0444;
418                 } elsif($x eq "w") { $mm |= 0222;
419                 } elsif($x eq "x") { $mm |= 0111;
420                 } elsif($x eq "=") { # do nothing
421                 } else { die "Unknown mode: $mode\n";
422                 }
423         }
424         $m |= $mm&$um;
425         return $m;
426 }
427
428 sub getwd() {
429         my $pwd = `pwd`;
430         chomp $pwd;
431         return $pwd;
432 }
433
434 sub is_sha1 {
435         my $s = shift;
436         return $s =~ /^[a-f0-9]{40}$/;
437 }
438
439 sub get_headref ($$) {
440     my $name    = shift;
441     my $git_dir = shift; 
442     
443     my $f = "$git_dir/refs/heads/$name";
444     if(open(my $fh, $f)) {
445             chomp(my $r = <$fh>);
446             is_sha1($r) or die "Cannot get head id for $name ($r): $!";
447             return $r;
448     }
449     die "unable to open $f: $!" unless $! == POSIX::ENOENT;
450     return undef;
451 }
452
453 -d $git_tree
454         or mkdir($git_tree,0777)
455         or die "Could not create $git_tree: $!";
456 chdir($git_tree);
457
458 my $last_branch = "";
459 my $orig_branch = "";
460 my %branch_date;
461 my $tip_at_start = undef;
462
463 my $git_dir = $ENV{"GIT_DIR"} || ".git";
464 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
465 $ENV{"GIT_DIR"} = $git_dir;
466 my $orig_git_index;
467 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
468
469 my %index; # holds filenames of one index per branch
470 {   # init with an index for origin
471     my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
472                              DIR => File::Spec->tmpdir());
473     close ($fh);
474     $index{$opt_o} = $fn;
475 }
476 $ENV{GIT_INDEX_FILE} = $index{$opt_o};
477 unless(-d $git_dir) {
478         system("git-init-db");
479         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
480         system("git-read-tree");
481         die "Cannot init an empty tree: $?\n" if $?;
482
483         $last_branch = $opt_o;
484         $orig_branch = "";
485 } else {
486         -f "$git_dir/refs/heads/$opt_o"
487                 or die "Branch '$opt_o' does not exist.\n".
488                        "Either use the correct '-o branch' option,\n".
489                        "or import to a new repository.\n";
490
491         open(F, "git-symbolic-ref HEAD |") or
492                 die "Cannot run git-symbolic-ref: $!\n";
493         chomp ($last_branch = <F>);
494         $last_branch = basename($last_branch);
495         close(F);
496         unless($last_branch) {
497                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
498                 $last_branch = "master";
499         }
500         $orig_branch = $last_branch;
501         $tip_at_start = `git-rev-parse --verify HEAD`;
502
503         # populate index
504         unless ($index{$last_branch}) {
505             my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
506                                      DIR => File::Spec->tmpdir());
507             close ($fh);
508             $index{$last_branch} = $fn;
509         }
510         $ENV{GIT_INDEX_FILE} = $index{$last_branch};
511         system('git-read-tree', $last_branch);
512         die "read-tree failed: $?\n" if $?;
513
514         # Get the last import timestamps
515         opendir(D,"$git_dir/refs/heads");
516         while(defined(my $head = readdir(D))) {
517                 next if $head =~ /^\./;
518                 open(F,"$git_dir/refs/heads/$head")
519                         or die "Bad head branch: $head: $!\n";
520                 chomp(my $ftag = <F>);
521                 close(F);
522                 open(F,"git-cat-file commit $ftag |");
523                 while(<F>) {
524                         next unless /^author\s.*\s(\d+)\s[-+]\d{4}$/;
525                         $branch_date{$head} = $1;
526                         last;
527                 }
528                 close(F);
529         }
530         closedir(D);
531 }
532
533 -d $git_dir
534         or die "Could not create git subdir ($git_dir).\n";
535
536 # now we read (and possibly save) author-info as well
537 -f "$git_dir/cvs-authors" and
538   read_author_info("$git_dir/cvs-authors");
539 if ($opt_A) {
540         read_author_info($opt_A);
541         write_author_info("$git_dir/cvs-authors");
542 }
543
544
545 #
546 # run cvsps into a file unless we are getting
547 # it passed as a file via $opt_P
548 #
549 unless ($opt_P) {
550         print "Running cvsps...\n" if $opt_v;
551         my $pid = open(CVSPS,"-|");
552         die "Cannot fork: $!\n" unless defined $pid;
553         unless($pid) {
554                 my @opt;
555                 @opt = split(/,/,$opt_p) if defined $opt_p;
556                 unshift @opt, '-z', $opt_z if defined $opt_z;
557                 unshift @opt, '-q'         unless defined $opt_v;
558                 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
559                         push @opt, '--cvs-direct';
560                 }
561                 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
562                 die "Could not start cvsps: $!\n";
563         }
564         my ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
565                                              DIR => File::Spec->tmpdir());
566         while (<CVSPS>) {
567             print $cvspsfh $_;
568         }
569         close CVSPS;
570         close $cvspsfh;
571         $opt_P = $cvspsfile;
572 }
573
574
575 open(CVS, "<$opt_P") or die $!;
576
577 ## cvsps output:
578 #---------------------
579 #PatchSet 314
580 #Date: 1999/09/18 13:03:59
581 #Author: wkoch
582 #Branch: STABLE-BRANCH-1-0
583 #Ancestor branch: HEAD
584 #Tag: (none)
585 #Log:
586 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
587 #Members:
588 #       README:1.57->1.57.2.1
589 #       VERSION:1.96->1.96.2.1
590 #
591 #---------------------
592
593 my $state = 0;
594
595 sub update_index (\@\@) {
596         my $old = shift;
597         my $new = shift;
598         open(my $fh, '|-', qw(git-update-index -z --index-info))
599                 or die "unable to open git-update-index: $!";
600         print $fh
601                 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
602                         @$old),
603                 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
604                         @$new)
605                 or die "unable to write to git-update-index: $!";
606         close $fh
607                 or die "unable to write to git-update-index: $!";
608         $? and die "git-update-index reported error: $?";
609 }
610
611 sub write_tree () {
612         open(my $fh, '-|', qw(git-write-tree))
613                 or die "unable to open git-write-tree: $!";
614         chomp(my $tree = <$fh>);
615         is_sha1($tree)
616                 or die "Cannot get tree id ($tree): $!";
617         close($fh)
618                 or die "Error running git-write-tree: $?\n";
619         print "Tree ID $tree\n" if $opt_v;
620         return $tree;
621 }
622
623 my($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
624 my(@old,@new,@skipped,%ignorebranch);
625
626 # commits that cvsps cannot place anywhere...
627 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
628
629 sub commit {
630         update_index(@old, @new);
631         @old = @new = ();
632         my $tree = write_tree();
633         my $parent = get_headref($last_branch, $git_dir);
634         print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
635
636         my @commit_args;
637         push @commit_args, ("-p", $parent) if $parent;
638
639         # loose detection of merges
640         # based on the commit msg
641         foreach my $rx (@mergerx) {
642                 next unless $logmsg =~ $rx && $1;
643                 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
644                 if(my $sha1 = get_headref($mparent, $git_dir)) {
645                         push @commit_args, '-p', $mparent;
646                         print "Merge parent branch: $mparent\n" if $opt_v;
647                 }
648         }
649
650         my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
651         $ENV{GIT_AUTHOR_NAME} = $author_name;
652         $ENV{GIT_AUTHOR_EMAIL} = $author_email;
653         $ENV{GIT_AUTHOR_DATE} = $commit_date;
654         $ENV{GIT_COMMITTER_NAME} = $author_name;
655         $ENV{GIT_COMMITTER_EMAIL} = $author_email;
656         $ENV{GIT_COMMITTER_DATE} = $commit_date;
657         my $pid = open2(my $commit_read, my $commit_write,
658                 'git-commit-tree', $tree, @commit_args);
659
660         # compatibility with git2cvs
661         substr($logmsg,32767) = "" if length($logmsg) > 32767;
662         $logmsg =~ s/[\s\n]+\z//;
663
664         if (@skipped) {
665             $logmsg .= "\n\n\nSKIPPED:\n\t";
666             $logmsg .= join("\n\t", @skipped) . "\n";
667             @skipped = ();
668         }
669
670         print($commit_write "$logmsg\n") && close($commit_write)
671                 or die "Error writing to git-commit-tree: $!\n";
672
673         print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
674         chomp(my $cid = <$commit_read>);
675         is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
676         print "Commit ID $cid\n" if $opt_v;
677         close($commit_read);
678
679         waitpid($pid,0);
680         die "Error running git-commit-tree: $?\n" if $?;
681
682         system("git-update-ref refs/heads/$branch $cid") == 0
683                 or die "Cannot write branch $branch for update: $!\n";
684
685         if($tag) {
686                 my($in, $out) = ('','');
687                 my($xtag) = $tag;
688                 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
689                 $xtag =~ tr/_/\./ if ( $opt_u );
690                 $xtag =~ s/[\/]/$opt_s/g;
691                 
692                 my $pid = open2($in, $out, 'git-mktag');
693                 print $out "object $cid\n".
694                     "type commit\n".
695                     "tag $xtag\n".
696                     "tagger $author_name <$author_email>\n"
697                     or die "Cannot create tag object $xtag: $!\n";
698                 close($out)
699                     or die "Cannot create tag object $xtag: $!\n";
700
701                 my $tagobj = <$in>;
702                 chomp $tagobj;
703
704                 if ( !close($in) or waitpid($pid, 0) != $pid or
705                      $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
706                     die "Cannot create tag object $xtag: $!\n";
707                 }
708                 
709
710                 open(C,">$git_dir/refs/tags/$xtag")
711                         or die "Cannot create tag $xtag: $!\n";
712                 print C "$tagobj\n"
713                         or die "Cannot write tag $xtag: $!\n";
714                 close(C)
715                         or die "Cannot write tag $xtag: $!\n";
716
717                 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
718         }
719 };
720
721 my $commitcount = 1;
722 while(<CVS>) {
723         chomp;
724         if($state == 0 and /^-+$/) {
725                 $state = 1;
726         } elsif($state == 0) {
727                 $state = 1;
728                 redo;
729         } elsif(($state==0 or $state==1) and s/^PatchSet\s+//) {
730                 $patchset = 0+$_;
731                 $state=2;
732         } elsif($state == 2 and s/^Date:\s+//) {
733                 $date = pdate($_);
734                 unless($date) {
735                         print STDERR "Could not parse date: $_\n";
736                         $state=0;
737                         next;
738                 }
739                 $state=3;
740         } elsif($state == 3 and s/^Author:\s+//) {
741                 s/\s+$//;
742                 if (/^(.*?)\s+<(.*)>/) {
743                     ($author_name, $author_email) = ($1, $2);
744                 } elsif ($conv_author_name{$_}) {
745                         $author_name = $conv_author_name{$_};
746                         $author_email = $conv_author_email{$_};
747                 } else {
748                     $author_name = $author_email = $_;
749                 }
750                 $state = 4;
751         } elsif($state == 4 and s/^Branch:\s+//) {
752                 s/\s+$//;
753                 s/[\/]/$opt_s/g;
754                 $branch = $_;
755                 $state = 5;
756         } elsif($state == 5 and s/^Ancestor branch:\s+//) {
757                 s/\s+$//;
758                 $ancestor = $_;
759                 $ancestor = $opt_o if $ancestor eq "HEAD";
760                 $state = 6;
761         } elsif($state == 5) {
762                 $ancestor = undef;
763                 $state = 6;
764                 redo;
765         } elsif($state == 6 and s/^Tag:\s+//) {
766                 s/\s+$//;
767                 if($_ eq "(none)") {
768                         $tag = undef;
769                 } else {
770                         $tag = $_;
771                 }
772                 $state = 7;
773         } elsif($state == 7 and /^Log:/) {
774                 $logmsg = "";
775                 $state = 8;
776         } elsif($state == 8 and /^Members:/) {
777                 $branch = $opt_o if $branch eq "HEAD";
778                 if(defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
779                         # skip
780                         print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
781                         $state = 11;
782                         next;
783                 }
784                 if (exists $ignorebranch{$branch}) {
785                         print STDERR "Skipping $branch\n";
786                         $state = 11;
787                         next;
788                 }
789                 if($ancestor) {
790                         if($ancestor eq $branch) {
791                                 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
792                                 $ancestor = $opt_o;
793                         }
794                         if(-f "$git_dir/refs/heads/$branch") {
795                                 print STDERR "Branch $branch already exists!\n";
796                                 $state=11;
797                                 next;
798                         }
799                         unless(open(H,"$git_dir/refs/heads/$ancestor")) {
800                                 print STDERR "Branch $ancestor does not exist!\n";
801                                 $ignorebranch{$branch} = 1;
802                                 $state=11;
803                                 next;
804                         }
805                         chomp(my $id = <H>);
806                         close(H);
807                         unless(open(H,"> $git_dir/refs/heads/$branch")) {
808                                 print STDERR "Could not create branch $branch: $!\n";
809                                 $ignorebranch{$branch} = 1;
810                                 $state=11;
811                                 next;
812                         }
813                         print H "$id\n"
814                                 or die "Could not write branch $branch: $!";
815                         close(H)
816                                 or die "Could not write branch $branch: $!";
817                 }
818                 if(($ancestor || $branch) ne $last_branch) {
819                         print "Switching from $last_branch to $branch\n" if $opt_v;
820                         unless ($index{$branch}) {
821                             my ($fh, $fn) = tempfile('gitXXXXXX', SUFFIX => '.idx',
822                                                      DIR => File::Spec->tmpdir());
823                             close ($fh);
824                             $index{$branch} = $fn;
825                             $ENV{GIT_INDEX_FILE} = $index{$branch};
826                             system("git-read-tree", $branch);
827                             die "read-tree failed: $?\n" if $?;
828                         } else {
829                             $ENV{GIT_INDEX_FILE} = $index{$branch};
830                         }
831                 }
832                 $last_branch = $branch if $branch ne $last_branch;
833                 $state = 9;
834         } elsif($state == 8) {
835                 $logmsg .= "$_\n";
836         } elsif($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
837 #       VERSION:1.96->1.96.2.1
838                 my $init = ($2 eq "INITIAL");
839                 my $fn = $1;
840                 my $rev = $3;
841                 $fn =~ s#^/+##;
842                 if ($opt_S && $fn =~ m/$opt_S/) {
843                     print "SKIPPING $fn v $rev\n";
844                     push(@skipped, $fn);
845                     next;
846                 }
847                 print "Fetching $fn   v $rev\n" if $opt_v;
848                 my ($tmpname, $size) = $cvs->file($fn,$rev);
849                 if($size == -1) {
850                         push(@old,$fn);
851                         print "Drop $fn\n" if $opt_v;
852                 } else {
853                         print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
854                         my $pid = open(my $F, '-|');
855                         die $! unless defined $pid;
856                         if (!$pid) {
857                             exec("git-hash-object", "-w", $tmpname)
858                                 or die "Cannot create object: $!\n";
859                         }
860                         my $sha = <$F>;
861                         chomp $sha;
862                         close $F;
863                         my $mode = pmode($cvs->{'mode'});
864                         push(@new,[$mode, $sha, $fn]); # may be resurrected!
865                 }
866                 unlink($tmpname);
867         } elsif($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
868                 my $fn = $1;
869                 $fn =~ s#^/+##;
870                 push(@old,$fn);
871                 print "Delete $fn\n" if $opt_v;
872         } elsif($state == 9 and /^\s*$/) {
873                 $state = 10;
874         } elsif(($state == 9 or $state == 10) and /^-+$/) {
875                 $commitcount++;
876                 if ($opt_L && $commitcount > $opt_L) {
877                         last;
878                 }
879                 commit();
880                 if (($commitcount & 1023) == 0) {
881                         system("git repack -a -d");
882                 }
883                 $state = 1;
884         } elsif($state == 11 and /^-+$/) {
885                 $state = 1;
886         } elsif(/^-+$/) { # end of unknown-line processing
887                 $state = 1;
888         } elsif($state != 11) { # ignore stuff when skipping
889                 print "* UNKNOWN LINE * $_\n";
890         }
891 }
892 commit() if $branch and $state != 11;
893
894 foreach my $git_index (values %index) {
895     unlink($git_index);
896 }
897
898 if (defined $orig_git_index) {
899         $ENV{GIT_INDEX_FILE} = $orig_git_index;
900 } else {
901         delete $ENV{GIT_INDEX_FILE};
902 }
903
904 # Now switch back to the branch we were in before all of this happened
905 if($orig_branch) {
906         print "DONE.\n" if $opt_v;
907         if ($opt_i) {
908                 exit 0;
909         }
910         my $tip_at_end = `git-rev-parse --verify HEAD`;
911         if ($tip_at_start ne $tip_at_end) {
912                 for ($tip_at_start, $tip_at_end) { chomp; }
913                 print "Fetched into the current branch.\n" if $opt_v;
914                 system(qw(git-read-tree -u -m),
915                        $tip_at_start, $tip_at_end);
916                 die "Fast-forward update failed: $?\n" if $?;
917         }
918         else {
919                 system(qw(git-merge cvsimport HEAD), "refs/heads/$opt_o");
920                 die "Could not merge $opt_o into the current branch.\n" if $?;
921         }
922 } else {
923         $orig_branch = "master";
924         print "DONE; creating $orig_branch branch\n" if $opt_v;
925         system("git-update-ref", "refs/heads/master", "refs/heads/$opt_o")
926                 unless -f "$git_dir/refs/heads/master";
927         system('git-update-ref', 'HEAD', "$orig_branch");
928         unless ($opt_i) {
929                 system('git checkout');
930                 die "checkout failed: $?\n" if $?;
931         }
932 }