Merge branch 'rs/archive'
[git/builtin-gsoc.git] / git-cvsimport.perl
blob2954fb846e3372a648124d92b7c1a1ccbd44306f
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.
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.
10 # Checking out the files is done by a single long-running CVS connection
11 # / server process.
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::Spec;
20 use File::Temp qw(tempfile tmpnam);
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;
29 $SIG{'PIPE'}="IGNORE";
30 $ENV{'TZ'}="UTC";
32 our ($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, $opt_a, $opt_r);
33 my (%conv_author_name, %conv_author_email);
35 sub usage(;$) {
36 my $msg = shift;
37 print(STDERR "Error: $msg\n") if $msg;
38 print STDERR <<END;
39 Usage: ${\basename $0} # fetch/update GIT from CVS
40 [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
41 [-p opts-for-cvsps] [-P file] [-C GIT_repository] [-z fuzz] [-i] [-k]
42 [-u] [-s subst] [-a] [-m] [-M regex] [-S regex] [-L commitlimit]
43 [-r remote] [CVS_module]
44 END
45 exit(1);
48 sub read_author_info($) {
49 my ($file) = @_;
50 my $user;
51 open my $f, '<', "$file" or die("Failed to open $file: $!\n");
53 while (<$f>) {
54 # Expected format is this:
55 # exon=Andreas Ericsson <ae@op5.se>
56 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
57 $user = $1;
58 $conv_author_name{$user} = $2;
59 $conv_author_email{$user} = $3;
61 # However, we also read from CVSROOT/users format
62 # to ease migration.
63 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
64 my $mapped;
65 ($user, $mapped) = ($1, $3);
66 if ($mapped =~ /^\s*(.*?)\s*<(.*)>\s*$/) {
67 $conv_author_name{$user} = $1;
68 $conv_author_email{$user} = $2;
70 elsif ($mapped =~ /^<?(.*)>?$/) {
71 $conv_author_name{$user} = $user;
72 $conv_author_email{$user} = $1;
75 # NEEDSWORK: Maybe warn on unrecognized lines?
77 close ($f);
80 sub write_author_info($) {
81 my ($file) = @_;
82 open my $f, '>', $file or
83 die("Failed to open $file for writing: $!");
85 foreach (keys %conv_author_name) {
86 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
88 close ($f);
91 # convert getopts specs for use by git-repo-config
92 sub read_repo_config {
93 # Split the string between characters, unless there is a ':'
94 # So "abc:de" becomes ["a", "b", "c:", "d", "e"]
95 my @opts = split(/ *(?!:)/, shift);
96 foreach my $o (@opts) {
97 my $key = $o;
98 $key =~ s/://g;
99 my $arg = 'git-repo-config';
100 $arg .= ' --bool' if ($o !~ /:$/);
102 chomp(my $tmp = `$arg --get cvsimport.$key`);
103 if ($tmp && !($arg =~ /--bool/ && $tmp eq 'false')) {
104 no strict 'refs';
105 my $opt_name = "opt_" . $key;
106 if (!$$opt_name) {
107 $$opt_name = $tmp;
111 if (@ARGV == 0) {
112 chomp(my $module = `git-repo-config --get cvsimport.module`);
113 push(@ARGV, $module);
117 my $opts = "haivmkuo:d:p:r:C:z:s:M:P:A:S:L:";
118 read_repo_config($opts);
119 getopts($opts) or usage();
120 usage if $opt_h;
122 @ARGV <= 1 or usage("You can't specify more than one CVS module");
124 if ($opt_d) {
125 $ENV{"CVSROOT"} = $opt_d;
126 } elsif (-f 'CVS/Root') {
127 open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
128 $opt_d = <$f>;
129 chomp $opt_d;
130 close $f;
131 $ENV{"CVSROOT"} = $opt_d;
132 } elsif ($ENV{"CVSROOT"}) {
133 $opt_d = $ENV{"CVSROOT"};
134 } else {
135 usage("CVSROOT needs to be set");
137 $opt_s ||= "-";
138 $opt_a ||= 0;
140 my $git_tree = $opt_C;
141 $git_tree ||= ".";
143 my $remote;
144 if (defined $opt_r) {
145 $remote = 'refs/remotes/' . $opt_r;
146 $opt_o ||= "master";
147 } else {
148 $opt_o ||= "origin";
149 $remote = 'refs/heads';
152 my $cvs_tree;
153 if ($#ARGV == 0) {
154 $cvs_tree = $ARGV[0];
155 } elsif (-f 'CVS/Repository') {
156 open my $f, '<', 'CVS/Repository' or
157 die 'Failed to open CVS/Repository';
158 $cvs_tree = <$f>;
159 chomp $cvs_tree;
160 close $f;
161 } else {
162 usage("CVS module has to be specified");
165 our @mergerx = ();
166 if ($opt_m) {
167 @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
169 if ($opt_M) {
170 push (@mergerx, qr/$opt_M/);
173 # Remember UTC of our starting time
174 # we'll want to avoid importing commits
175 # that are too recent
176 our $starttime = time();
178 select(STDERR); $|=1; select(STDOUT);
181 package CVSconn;
182 # Basic CVS dialog.
183 # We're only interested in connecting and downloading, so ...
185 use File::Spec;
186 use File::Temp qw(tempfile);
187 use POSIX qw(strftime dup2);
189 sub new {
190 my ($what,$repo,$subdir) = @_;
191 $what=ref($what) if ref($what);
193 my $self = {};
194 $self->{'buffer'} = "";
195 bless($self,$what);
197 $repo =~ s#/+$##;
198 $self->{'fullrep'} = $repo;
199 $self->conn();
201 $self->{'subdir'} = $subdir;
202 $self->{'lines'} = undef;
204 return $self;
207 sub conn {
208 my $self = shift;
209 my $repo = $self->{'fullrep'};
210 if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
211 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
213 my ($proxyhost,$proxyport);
214 if ($param && ($param =~ m/proxy=([^;]+)/)) {
215 $proxyhost = $1;
216 # Default proxyport, if not specified, is 8080.
217 $proxyport = 8080;
218 if ($ENV{"CVS_PROXY_PORT"}) {
219 $proxyport = $ENV{"CVS_PROXY_PORT"};
221 if ($param =~ m/proxyport=([^;]+)/) {
222 $proxyport = $1;
226 $user="anonymous" unless defined $user;
227 my $rr2 = "-";
228 unless ($port) {
229 $rr2 = ":pserver:$user\@$serv:$repo";
230 $port=2401;
232 my $rr = ":pserver:$user\@$serv:$port$repo";
234 unless ($pass) {
235 open(H,$ENV{'HOME'}."/.cvspass") and do {
236 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
237 while (<H>) {
238 chomp;
239 s/^\/\d+\s+//;
240 my ($w,$p) = split(/\s/,$_,2);
241 if ($w eq $rr or $w eq $rr2) {
242 $pass = $p;
243 last;
248 $pass="A" unless $pass;
250 my ($s, $rep);
251 if ($proxyhost) {
253 # Use a HTTP Proxy. Only works for HTTP proxies that
254 # don't require user authentication
256 # See: http://www.ietf.org/rfc/rfc2817.txt
258 $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
259 die "Socket to $proxyhost: $!\n" unless defined $s;
260 $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
261 or die "Write to $proxyhost: $!\n";
262 $s->flush();
264 $rep = <$s>;
266 # The answer should look like 'HTTP/1.x 2yy ....'
267 if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
268 die "Proxy connect: $rep\n";
270 # Skip up to the empty line of the proxy server output
271 # including the response headers.
272 while ($rep = <$s>) {
273 last if (!defined $rep ||
274 $rep eq "\n" ||
275 $rep eq "\r\n");
277 } else {
278 $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
279 die "Socket to $serv: $!\n" unless defined $s;
282 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
283 or die "Write to $serv: $!\n";
284 $s->flush();
286 $rep = <$s>;
288 if ($rep ne "I LOVE YOU\n") {
289 $rep="<unknown>" unless $rep;
290 die "AuthReply: $rep\n";
292 $self->{'socketo'} = $s;
293 $self->{'socketi'} = $s;
294 } else { # local or ext: Fork off our own cvs server.
295 my $pr = IO::Pipe->new();
296 my $pw = IO::Pipe->new();
297 my $pid = fork();
298 die "Fork: $!\n" unless defined $pid;
299 my $cvs = 'cvs';
300 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
301 my $rsh = 'rsh';
302 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
304 my @cvs = ($cvs, 'server');
305 my ($local, $user, $host);
306 $local = $repo =~ s/:local://;
307 if (!$local) {
308 $repo =~ s/:ext://;
309 $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
310 ($user, $host) = ($1, $2);
312 if (!$local) {
313 if ($user) {
314 unshift @cvs, $rsh, '-l', $user, $host;
315 } else {
316 unshift @cvs, $rsh, $host;
320 unless ($pid) {
321 $pr->writer();
322 $pw->reader();
323 dup2($pw->fileno(),0);
324 dup2($pr->fileno(),1);
325 $pr->close();
326 $pw->close();
327 exec(@cvs);
329 $pw->writer();
330 $pr->reader();
331 $self->{'socketo'} = $pw;
332 $self->{'socketi'} = $pr;
334 $self->{'socketo'}->write("Root $repo\n");
336 # Trial and error says that this probably is the minimum set
337 $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
339 $self->{'socketo'}->write("valid-requests\n");
340 $self->{'socketo'}->flush();
342 chomp(my $rep=$self->readline());
343 if ($rep !~ s/^Valid-requests\s*//) {
344 $rep="<unknown>" unless $rep;
345 die "Expected Valid-requests from server, but got: $rep\n";
347 chomp(my $res=$self->readline());
348 die "validReply: $res\n" if $res ne "ok";
350 $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
351 $self->{'repo'} = $repo;
354 sub readline {
355 my ($self) = @_;
356 return $self->{'socketi'}->getline();
359 sub _file {
360 # Request a file with a given revision.
361 # Trial and error says this is a good way to do it. :-/
362 my ($self,$fn,$rev) = @_;
363 $self->{'socketo'}->write("Argument -N\n") or return undef;
364 $self->{'socketo'}->write("Argument -P\n") or return undef;
365 # -kk: Linus' version doesn't use it - defaults to off
366 if ($opt_k) {
367 $self->{'socketo'}->write("Argument -kk\n") or return undef;
369 $self->{'socketo'}->write("Argument -r\n") or return undef;
370 $self->{'socketo'}->write("Argument $rev\n") or return undef;
371 $self->{'socketo'}->write("Argument --\n") or return undef;
372 $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
373 $self->{'socketo'}->write("Directory .\n") or return undef;
374 $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
375 # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
376 $self->{'socketo'}->write("co\n") or return undef;
377 $self->{'socketo'}->flush() or return undef;
378 $self->{'lines'} = 0;
379 return 1;
381 sub _line {
382 # Read a line from the server.
383 # ... except that 'line' may be an entire file. ;-)
384 my ($self, $fh) = @_;
385 die "Not in lines" unless defined $self->{'lines'};
387 my $line;
388 my $res=0;
389 while (defined($line = $self->readline())) {
390 # M U gnupg-cvs-rep/AUTHORS
391 # Updated gnupg-cvs-rep/
392 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
393 # /AUTHORS/1.1///T1.1
394 # u=rw,g=rw,o=rw
396 # ok
398 if ($line =~ s/^(?:Created|Updated) //) {
399 $line = $self->readline(); # path
400 $line = $self->readline(); # Entries line
401 my $mode = $self->readline(); chomp $mode;
402 $self->{'mode'} = $mode;
403 defined (my $cnt = $self->readline())
404 or die "EOF from server after 'Changed'\n";
405 chomp $cnt;
406 die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
407 $line="";
408 $res = $self->_fetchfile($fh, $cnt);
409 } elsif ($line =~ s/^ //) {
410 print $fh $line;
411 $res += length($line);
412 } elsif ($line =~ /^M\b/) {
413 # output, do nothing
414 } elsif ($line =~ /^Mbinary\b/) {
415 my $cnt;
416 die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
417 chomp $cnt;
418 die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
419 $line="";
420 $res += $self->_fetchfile($fh, $cnt);
421 } else {
422 chomp $line;
423 if ($line eq "ok") {
424 # print STDERR "S: ok (".length($res).")\n";
425 return $res;
426 } elsif ($line =~ s/^E //) {
427 # print STDERR "S: $line\n";
428 } elsif ($line =~ /^(Remove-entry|Removed) /i) {
429 $line = $self->readline(); # filename
430 $line = $self->readline(); # OK
431 chomp $line;
432 die "Unknown: $line" if $line ne "ok";
433 return -1;
434 } else {
435 die "Unknown: $line\n";
439 return undef;
441 sub file {
442 my ($self,$fn,$rev) = @_;
443 my $res;
445 my ($fh, $name) = tempfile('gitcvs.XXXXXX',
446 DIR => File::Spec->tmpdir(), UNLINK => 1);
448 $self->_file($fn,$rev) and $res = $self->_line($fh);
450 if (!defined $res) {
451 print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
452 truncate $fh, 0;
453 $self->conn();
454 $self->_file($fn,$rev) or die "No file command send";
455 $res = $self->_line($fh);
456 die "Retry failed" unless defined $res;
458 close ($fh);
460 return ($name, $res);
462 sub _fetchfile {
463 my ($self, $fh, $cnt) = @_;
464 my $res = 0;
465 my $bufsize = 1024 * 1024;
466 while ($cnt) {
467 if ($bufsize > $cnt) {
468 $bufsize = $cnt;
470 my $buf;
471 my $num = $self->{'socketi'}->read($buf,$bufsize);
472 die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
473 print $fh $buf;
474 $res += $num;
475 $cnt -= $num;
477 return $res;
481 package main;
483 my $cvs = CVSconn->new($opt_d, $cvs_tree);
486 sub pdate($) {
487 my ($d) = @_;
488 m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
489 or die "Unparseable date: $d\n";
490 my $y=$1; $y-=1900 if $y>1900;
491 return timegm($6||0,$5,$4,$3,$2-1,$y);
494 sub pmode($) {
495 my ($mode) = @_;
496 my $m = 0;
497 my $mm = 0;
498 my $um = 0;
499 for my $x(split(//,$mode)) {
500 if ($x eq ",") {
501 $m |= $mm&$um;
502 $mm = 0;
503 $um = 0;
504 } elsif ($x eq "u") { $um |= 0700;
505 } elsif ($x eq "g") { $um |= 0070;
506 } elsif ($x eq "o") { $um |= 0007;
507 } elsif ($x eq "r") { $mm |= 0444;
508 } elsif ($x eq "w") { $mm |= 0222;
509 } elsif ($x eq "x") { $mm |= 0111;
510 } elsif ($x eq "=") { # do nothing
511 } else { die "Unknown mode: $mode\n";
514 $m |= $mm&$um;
515 return $m;
518 sub getwd() {
519 my $pwd = `pwd`;
520 chomp $pwd;
521 return $pwd;
524 sub is_sha1 {
525 my $s = shift;
526 return $s =~ /^[a-f0-9]{40}$/;
529 sub get_headref ($$) {
530 my $name = shift;
531 my $git_dir = shift;
533 my $f = "$git_dir/$remote/$name";
534 if (open(my $fh, $f)) {
535 chomp(my $r = <$fh>);
536 is_sha1($r) or die "Cannot get head id for $name ($r): $!";
537 return $r;
539 die "unable to open $f: $!" unless $! == POSIX::ENOENT;
540 return undef;
543 -d $git_tree
544 or mkdir($git_tree,0777)
545 or die "Could not create $git_tree: $!";
546 chdir($git_tree);
548 my $last_branch = "";
549 my $orig_branch = "";
550 my %branch_date;
551 my $tip_at_start = undef;
553 my $git_dir = $ENV{"GIT_DIR"} || ".git";
554 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
555 $ENV{"GIT_DIR"} = $git_dir;
556 my $orig_git_index;
557 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
559 my %index; # holds filenames of one index per branch
561 unless (-d $git_dir) {
562 system("git-init");
563 die "Cannot init the GIT db at $git_tree: $?\n" if $?;
564 system("git-read-tree");
565 die "Cannot init an empty tree: $?\n" if $?;
567 $last_branch = $opt_o;
568 $orig_branch = "";
569 } else {
570 open(F, "git-symbolic-ref HEAD |") or
571 die "Cannot run git-symbolic-ref: $!\n";
572 chomp ($last_branch = <F>);
573 $last_branch = basename($last_branch);
574 close(F);
575 unless ($last_branch) {
576 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
577 $last_branch = "master";
579 $orig_branch = $last_branch;
580 $tip_at_start = `git-rev-parse --verify HEAD`;
582 # Get the last import timestamps
583 my $fmt = '($ref, $author) = (%(refname), %(author));';
584 open(H, "git-for-each-ref --perl --format='$fmt' $remote |") or
585 die "Cannot run git-for-each-ref: $!\n";
586 while (defined(my $entry = <H>)) {
587 my ($ref, $author);
588 eval($entry) || die "cannot eval refs list: $@";
589 my ($head) = ($ref =~ m|^$remote/(.*)|);
590 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
591 $branch_date{$head} = $1;
593 close(H);
594 if (!exists $branch_date{$opt_o}) {
595 die "Branch '$opt_o' does not exist.\n".
596 "Either use the correct '-o branch' option,\n".
597 "or import to a new repository.\n";
601 -d $git_dir
602 or die "Could not create git subdir ($git_dir).\n";
604 # now we read (and possibly save) author-info as well
605 -f "$git_dir/cvs-authors" and
606 read_author_info("$git_dir/cvs-authors");
607 if ($opt_A) {
608 read_author_info($opt_A);
609 write_author_info("$git_dir/cvs-authors");
614 # run cvsps into a file unless we are getting
615 # it passed as a file via $opt_P
617 my $cvspsfile;
618 unless ($opt_P) {
619 print "Running cvsps...\n" if $opt_v;
620 my $pid = open(CVSPS,"-|");
621 my $cvspsfh;
622 die "Cannot fork: $!\n" unless defined $pid;
623 unless ($pid) {
624 my @opt;
625 @opt = split(/,/,$opt_p) if defined $opt_p;
626 unshift @opt, '-z', $opt_z if defined $opt_z;
627 unshift @opt, '-q' unless defined $opt_v;
628 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
629 push @opt, '--cvs-direct';
631 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
632 die "Could not start cvsps: $!\n";
634 ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
635 DIR => File::Spec->tmpdir());
636 while (<CVSPS>) {
637 print $cvspsfh $_;
639 close CVSPS;
640 close $cvspsfh;
641 } else {
642 $cvspsfile = $opt_P;
645 open(CVS, "<$cvspsfile") or die $!;
647 ## cvsps output:
648 #---------------------
649 #PatchSet 314
650 #Date: 1999/09/18 13:03:59
651 #Author: wkoch
652 #Branch: STABLE-BRANCH-1-0
653 #Ancestor branch: HEAD
654 #Tag: (none)
655 #Log:
656 # See ChangeLog: Sat Sep 18 13:03:28 CEST 1999 Werner Koch
657 #Members:
658 # README:1.57->1.57.2.1
659 # VERSION:1.96->1.96.2.1
661 #---------------------
663 my $state = 0;
665 sub update_index (\@\@) {
666 my $old = shift;
667 my $new = shift;
668 open(my $fh, '|-', qw(git-update-index -z --index-info))
669 or die "unable to open git-update-index: $!";
670 print $fh
671 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
672 @$old),
673 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
674 @$new)
675 or die "unable to write to git-update-index: $!";
676 close $fh
677 or die "unable to write to git-update-index: $!";
678 $? and die "git-update-index reported error: $?";
681 sub write_tree () {
682 open(my $fh, '-|', qw(git-write-tree))
683 or die "unable to open git-write-tree: $!";
684 chomp(my $tree = <$fh>);
685 is_sha1($tree)
686 or die "Cannot get tree id ($tree): $!";
687 close($fh)
688 or die "Error running git-write-tree: $?\n";
689 print "Tree ID $tree\n" if $opt_v;
690 return $tree;
693 my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
694 my (@old,@new,@skipped,%ignorebranch);
696 # commits that cvsps cannot place anywhere...
697 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
699 sub commit {
700 if ($branch eq $opt_o && !$index{branch} && !get_headref($branch, $git_dir)) {
701 # looks like an initial commit
702 # use the index primed by git-init
703 $ENV{GIT_INDEX_FILE} = "$git_dir/index";
704 $index{$branch} = "$git_dir/index";
705 } else {
706 # use an index per branch to speed up
707 # imports of projects with many branches
708 unless ($index{$branch}) {
709 $index{$branch} = tmpnam();
710 $ENV{GIT_INDEX_FILE} = $index{$branch};
711 if ($ancestor) {
712 system("git-read-tree", "$remote/$ancestor");
713 } else {
714 system("git-read-tree", "$remote/$branch");
716 die "read-tree failed: $?\n" if $?;
719 $ENV{GIT_INDEX_FILE} = $index{$branch};
721 update_index(@old, @new);
722 @old = @new = ();
723 my $tree = write_tree();
724 my $parent = get_headref($last_branch, $git_dir);
725 print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
727 my @commit_args;
728 push @commit_args, ("-p", $parent) if $parent;
730 # loose detection of merges
731 # based on the commit msg
732 foreach my $rx (@mergerx) {
733 next unless $logmsg =~ $rx && $1;
734 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
735 if (my $sha1 = get_headref($mparent, $git_dir)) {
736 push @commit_args, '-p', $mparent;
737 print "Merge parent branch: $mparent\n" if $opt_v;
741 my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
742 $ENV{GIT_AUTHOR_NAME} = $author_name;
743 $ENV{GIT_AUTHOR_EMAIL} = $author_email;
744 $ENV{GIT_AUTHOR_DATE} = $commit_date;
745 $ENV{GIT_COMMITTER_NAME} = $author_name;
746 $ENV{GIT_COMMITTER_EMAIL} = $author_email;
747 $ENV{GIT_COMMITTER_DATE} = $commit_date;
748 my $pid = open2(my $commit_read, my $commit_write,
749 'git-commit-tree', $tree, @commit_args);
751 # compatibility with git2cvs
752 substr($logmsg,32767) = "" if length($logmsg) > 32767;
753 $logmsg =~ s/[\s\n]+\z//;
755 if (@skipped) {
756 $logmsg .= "\n\n\nSKIPPED:\n\t";
757 $logmsg .= join("\n\t", @skipped) . "\n";
758 @skipped = ();
761 print($commit_write "$logmsg\n") && close($commit_write)
762 or die "Error writing to git-commit-tree: $!\n";
764 print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
765 chomp(my $cid = <$commit_read>);
766 is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
767 print "Commit ID $cid\n" if $opt_v;
768 close($commit_read);
770 waitpid($pid,0);
771 die "Error running git-commit-tree: $?\n" if $?;
773 system("git-update-ref $remote/$branch $cid") == 0
774 or die "Cannot write branch $branch for update: $!\n";
776 if ($tag) {
777 my ($xtag) = $tag;
778 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
779 $xtag =~ tr/_/\./ if ( $opt_u );
780 $xtag =~ s/[\/]/$opt_s/g;
782 system('git-tag', '-f', $xtag, $cid) == 0
783 or die "Cannot create tag $xtag: $!\n";
785 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
789 my $commitcount = 1;
790 while (<CVS>) {
791 chomp;
792 if ($state == 0 and /^-+$/) {
793 $state = 1;
794 } elsif ($state == 0) {
795 $state = 1;
796 redo;
797 } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
798 $patchset = 0+$_;
799 $state=2;
800 } elsif ($state == 2 and s/^Date:\s+//) {
801 $date = pdate($_);
802 unless ($date) {
803 print STDERR "Could not parse date: $_\n";
804 $state=0;
805 next;
807 $state=3;
808 } elsif ($state == 3 and s/^Author:\s+//) {
809 s/\s+$//;
810 if (/^(.*?)\s+<(.*)>/) {
811 ($author_name, $author_email) = ($1, $2);
812 } elsif ($conv_author_name{$_}) {
813 $author_name = $conv_author_name{$_};
814 $author_email = $conv_author_email{$_};
815 } else {
816 $author_name = $author_email = $_;
818 $state = 4;
819 } elsif ($state == 4 and s/^Branch:\s+//) {
820 s/\s+$//;
821 s/[\/]/$opt_s/g;
822 $branch = $_;
823 $state = 5;
824 } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
825 s/\s+$//;
826 $ancestor = $_;
827 $ancestor = $opt_o if $ancestor eq "HEAD";
828 $state = 6;
829 } elsif ($state == 5) {
830 $ancestor = undef;
831 $state = 6;
832 redo;
833 } elsif ($state == 6 and s/^Tag:\s+//) {
834 s/\s+$//;
835 if ($_ eq "(none)") {
836 $tag = undef;
837 } else {
838 $tag = $_;
840 $state = 7;
841 } elsif ($state == 7 and /^Log:/) {
842 $logmsg = "";
843 $state = 8;
844 } elsif ($state == 8 and /^Members:/) {
845 $branch = $opt_o if $branch eq "HEAD";
846 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
847 # skip
848 print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
849 $state = 11;
850 next;
852 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
853 # skip if the commit is too recent
854 # that the cvsps default fuzz is 300s, we give ourselves another
855 # 300s just in case -- this also prevents skipping commits
856 # due to server clock drift
857 print "skip patchset $patchset: $date too recent\n" if $opt_v;
858 $state = 11;
859 next;
861 if (exists $ignorebranch{$branch}) {
862 print STDERR "Skipping $branch\n";
863 $state = 11;
864 next;
866 if ($ancestor) {
867 if ($ancestor eq $branch) {
868 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
869 $ancestor = $opt_o;
871 if (-f "$git_dir/$remote/$branch") {
872 print STDERR "Branch $branch already exists!\n";
873 $state=11;
874 next;
876 unless (open(H,"$git_dir/$remote/$ancestor")) {
877 print STDERR "Branch $ancestor does not exist!\n";
878 $ignorebranch{$branch} = 1;
879 $state=11;
880 next;
882 chomp(my $id = <H>);
883 close(H);
884 unless (open(H,"> $git_dir/$remote/$branch")) {
885 print STDERR "Could not create branch $branch: $!\n";
886 $ignorebranch{$branch} = 1;
887 $state=11;
888 next;
890 print H "$id\n"
891 or die "Could not write branch $branch: $!";
892 close(H)
893 or die "Could not write branch $branch: $!";
895 $last_branch = $branch if $branch ne $last_branch;
896 $state = 9;
897 } elsif ($state == 8) {
898 $logmsg .= "$_\n";
899 } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
900 # VERSION:1.96->1.96.2.1
901 my $init = ($2 eq "INITIAL");
902 my $fn = $1;
903 my $rev = $3;
904 $fn =~ s#^/+##;
905 if ($opt_S && $fn =~ m/$opt_S/) {
906 print "SKIPPING $fn v $rev\n";
907 push(@skipped, $fn);
908 next;
910 print "Fetching $fn v $rev\n" if $opt_v;
911 my ($tmpname, $size) = $cvs->file($fn,$rev);
912 if ($size == -1) {
913 push(@old,$fn);
914 print "Drop $fn\n" if $opt_v;
915 } else {
916 print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
917 my $pid = open(my $F, '-|');
918 die $! unless defined $pid;
919 if (!$pid) {
920 exec("git-hash-object", "-w", $tmpname)
921 or die "Cannot create object: $!\n";
923 my $sha = <$F>;
924 chomp $sha;
925 close $F;
926 my $mode = pmode($cvs->{'mode'});
927 push(@new,[$mode, $sha, $fn]); # may be resurrected!
929 unlink($tmpname);
930 } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
931 my $fn = $1;
932 $fn =~ s#^/+##;
933 push(@old,$fn);
934 print "Delete $fn\n" if $opt_v;
935 } elsif ($state == 9 and /^\s*$/) {
936 $state = 10;
937 } elsif (($state == 9 or $state == 10) and /^-+$/) {
938 $commitcount++;
939 if ($opt_L && $commitcount > $opt_L) {
940 last;
942 commit();
943 if (($commitcount & 1023) == 0) {
944 system("git repack -a -d");
946 $state = 1;
947 } elsif ($state == 11 and /^-+$/) {
948 $state = 1;
949 } elsif (/^-+$/) { # end of unknown-line processing
950 $state = 1;
951 } elsif ($state != 11) { # ignore stuff when skipping
952 print "* UNKNOWN LINE * $_\n";
955 commit() if $branch and $state != 11;
957 unless ($opt_P) {
958 unlink($cvspsfile);
961 # The heuristic of repacking every 1024 commits can leave a
962 # lot of unpacked data. If there is more than 1MB worth of
963 # not-packed objects, repack once more.
964 my $line = `git-count-objects`;
965 if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
966 my ($n_objects, $kb) = ($1, $2);
967 1024 < $kb
968 and system("git repack -a -d");
971 foreach my $git_index (values %index) {
972 if ($git_index ne "$git_dir/index") {
973 unlink($git_index);
977 if (defined $orig_git_index) {
978 $ENV{GIT_INDEX_FILE} = $orig_git_index;
979 } else {
980 delete $ENV{GIT_INDEX_FILE};
983 # Now switch back to the branch we were in before all of this happened
984 if ($orig_branch) {
985 print "DONE.\n" if $opt_v;
986 if ($opt_i) {
987 exit 0;
989 my $tip_at_end = `git-rev-parse --verify HEAD`;
990 if ($tip_at_start ne $tip_at_end) {
991 for ($tip_at_start, $tip_at_end) { chomp; }
992 print "Fetched into the current branch.\n" if $opt_v;
993 system(qw(git-read-tree -u -m),
994 $tip_at_start, $tip_at_end);
995 die "Fast-forward update failed: $?\n" if $?;
997 else {
998 system(qw(git-merge cvsimport HEAD), "$remote/$opt_o");
999 die "Could not merge $opt_o into the current branch.\n" if $?;
1001 } else {
1002 $orig_branch = "master";
1003 print "DONE; creating $orig_branch branch\n" if $opt_v;
1004 system("git-update-ref", "refs/heads/master", "$remote/$opt_o")
1005 unless -f "$git_dir/refs/heads/master";
1006 system("git-symbolic-ref", "$remote/HEAD", "$remote/$opt_o")
1007 if ($opt_r && $opt_o ne 'HEAD');
1008 system('git-update-ref', 'HEAD', "$orig_branch");
1009 unless ($opt_i) {
1010 system('git checkout -f');
1011 die "checkout failed: $?\n" if $?;