cvsimport: use rev-parse to support packed refs
[git/git.git] / git-cvsimport.perl
CommitLineData
a57a9493 1#!/usr/bin/perl -w
9718a00b 2
a57a9493
MU
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
16use strict;
17use warnings;
18use Getopt::Std;
79ee456c 19use File::Spec;
7ccd9009 20use File::Temp qw(tempfile tmpnam);
a57a9493
MU
21use File::Path qw(mkpath);
22use File::Basename qw(basename dirname);
23use Time::Local;
2a3e1a85
MU
24use IO::Socket;
25use IO::Pipe;
e49289df 26use POSIX qw(strftime dup2 ENOENT);
0d821d4d 27use IPC::Open2;
a57a9493
MU
28
29$SIG{'PIPE'}="IGNORE";
30$ENV{'TZ'}="UTC";
31
8b7f5fc1 32our ($opt_h,$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, $opt_r);
ffd97f3a 33my (%conv_author_name, %conv_author_email);
a57a9493 34
7bf77644
FL
35sub usage(;$) {
36 my $msg = shift;
37 print(STDERR "Error: $msg\n") if $msg;
a57a9493 38 print STDERR <<END;
2a3e1a85 39Usage: ${\basename $0} # fetch/update GIT from CVS
ffd97f3a 40 [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
edbe4466
FL
41 [-p opts-for-cvsps] [-P file] [-C GIT_repository] [-z fuzz] [-i] [-k]
42 [-u] [-s subst] [-a] [-m] [-M regex] [-S regex] [-L commitlimit]
cbc9be5c 43 [-r remote] [CVS_module]
a57a9493
MU
44END
45 exit(1);
46}
47
ffd97f3a
AE
48sub read_author_info($) {
49 my ($file) = @_;
50 my $user;
51 open my $f, '<', "$file" or die("Failed to open $file: $!\n");
52
53 while (<$f>) {
8cd16211 54 # Expected format is this:
ffd97f3a 55 # exon=Andreas Ericsson <ae@op5.se>
8cd16211 56 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
ffd97f3a 57 $user = $1;
8cd16211
JH
58 $conv_author_name{$user} = $2;
59 $conv_author_email{$user} = $3;
ffd97f3a 60 }
8cd16211
JH
61 # However, we also read from CVSROOT/users format
62 # to ease migration.
63 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
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} = $2;
69 }
70 elsif ($mapped =~ /^<?(.*)>?$/) {
71 $conv_author_name{$user} = $user;
72 $conv_author_email{$user} = $1;
73 }
74 }
75 # NEEDSWORK: Maybe warn on unrecognized lines?
ffd97f3a
AE
76 }
77 close ($f);
78}
79
80sub write_author_info($) {
81 my ($file) = @_;
82 open my $f, '>', $file or
83 die("Failed to open $file for writing: $!");
84
85 foreach (keys %conv_author_name) {
8cd16211 86 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
ffd97f3a
AE
87 }
88 close ($f);
89}
90
ed35dece
JB
91# convert getopts specs for use by git-repo-config
92sub 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 !~ /:$/);
101
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;
108 }
109 }
110 }
111 if (@ARGV == 0) {
112 chomp(my $module = `git-repo-config --get cvsimport.module`);
113 push(@ARGV, $module);
114 }
115}
116
8b7f5fc1 117my $opts = "haivmkuo:d:p:r:C:z:s:M:P:A:S:L:";
ed35dece
JB
118read_repo_config($opts);
119getopts($opts) or usage();
a57a9493
MU
120usage if $opt_h;
121
7bf77644 122@ARGV <= 1 or usage("You can't specify more than one CVS module");
a57a9493 123
86d11cf2 124if ($opt_d) {
2a3e1a85 125 $ENV{"CVSROOT"} = $opt_d;
86d11cf2 126} elsif (-f 'CVS/Root') {
f9714a4a
SV
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"} = $opt_d;
86d11cf2 132} elsif ($ENV{"CVSROOT"}) {
2a3e1a85
MU
133 $opt_d = $ENV{"CVSROOT"};
134} else {
7bf77644 135 usage("CVSROOT needs to be set");
2a3e1a85 136}
fbfd60d6 137$opt_s ||= "-";
ded9f400
ML
138$opt_a ||= 0;
139
f9714a4a 140my $git_tree = $opt_C;
2a3e1a85
MU
141$git_tree ||= ".";
142
8b7f5fc1
AW
143my $remote;
144if (defined $opt_r) {
145 $remote = 'refs/remotes/' . $opt_r;
146 $opt_o ||= "master";
147} else {
148 $opt_o ||= "origin";
149 $remote = 'refs/heads';
150}
151
f9714a4a
SV
152my $cvs_tree;
153if ($#ARGV == 0) {
154 $cvs_tree = $ARGV[0];
155} elsif (-f 'CVS/Repository') {
a6080a0a 156 open my $f, '<', 'CVS/Repository' or
f9714a4a
SV
157 die 'Failed to open CVS/Repository';
158 $cvs_tree = <$f>;
159 chomp $cvs_tree;
db4b6582 160 close $f;
f9714a4a 161} else {
7bf77644 162 usage("CVS module has to be specified");
f9714a4a
SV
163}
164
db4b6582
ML
165our @mergerx = ();
166if ($opt_m) {
167 @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
168}
169if ($opt_M) {
170 push (@mergerx, qr/$opt_M/);
171}
172
6211988f
ML
173# Remember UTC of our starting time
174# we'll want to avoid importing commits
175# that are too recent
176our $starttime = time();
177
a57a9493
MU
178select(STDERR); $|=1; select(STDOUT);
179
180
181package CVSconn;
182# Basic CVS dialog.
2a3e1a85 183# We're only interested in connecting and downloading, so ...
a57a9493 184
2eb6d82e
SV
185use File::Spec;
186use File::Temp qw(tempfile);
f65ae603
MU
187use POSIX qw(strftime dup2);
188
a57a9493 189sub new {
86d11cf2 190 my ($what,$repo,$subdir) = @_;
a57a9493
MU
191 $what=ref($what) if ref($what);
192
193 my $self = {};
194 $self->{'buffer'} = "";
195 bless($self,$what);
196
197 $repo =~ s#/+$##;
198 $self->{'fullrep'} = $repo;
199 $self->conn();
200
201 $self->{'subdir'} = $subdir;
202 $self->{'lines'} = undef;
203
204 return $self;
205}
206
207sub conn {
208 my $self = shift;
209 my $repo = $self->{'fullrep'};
86d11cf2
JH
210 if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
211 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
73bcf533 212
86d11cf2
JH
213 my ($proxyhost,$proxyport);
214 if ($param && ($param =~ m/proxy=([^;]+)/)) {
73bcf533
IA
215 $proxyhost = $1;
216 # Default proxyport, if not specified, is 8080.
217 $proxyport = 8080;
86d11cf2 218 if ($ENV{"CVS_PROXY_PORT"}) {
73bcf533
IA
219 $proxyport = $ENV{"CVS_PROXY_PORT"};
220 }
86d11cf2 221 if ($param =~ m/proxyport=([^;]+)/) {
73bcf533
IA
222 $proxyport = $1;
223 }
224 }
225
a57a9493 226 $user="anonymous" unless defined $user;
2a3e1a85 227 my $rr2 = "-";
86d11cf2 228 unless ($port) {
a57a9493
MU
229 $rr2 = ":pserver:$user\@$serv:$repo";
230 $port=2401;
231 }
232 my $rr = ":pserver:$user\@$serv:$port$repo";
233
86d11cf2 234 unless ($pass) {
a57a9493
MU
235 open(H,$ENV{'HOME'}."/.cvspass") and do {
236 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
86d11cf2 237 while (<H>) {
a57a9493
MU
238 chomp;
239 s/^\/\d+\s+//;
240 my ($w,$p) = split(/\s/,$_,2);
86d11cf2 241 if ($w eq $rr or $w eq $rr2) {
a57a9493
MU
242 $pass = $p;
243 last;
244 }
245 }
246 };
247 }
248 $pass="A" unless $pass;
249
73bcf533 250 my ($s, $rep);
86d11cf2 251 if ($proxyhost) {
73bcf533
IA
252
253 # Use a HTTP Proxy. Only works for HTTP proxies that
254 # don't require user authentication
255 #
256 # See: http://www.ietf.org/rfc/rfc2817.txt
257
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();
263
264 $rep = <$s>;
265
266 # The answer should look like 'HTTP/1.x 2yy ....'
86d11cf2 267 if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
73bcf533
IA
268 die "Proxy connect: $rep\n";
269 }
270 # Skip up to the empty line of the proxy server output
271 # including the response headers.
272 while ($rep = <$s>) {
273 last if (!defined $rep ||
274 $rep eq "\n" ||
275 $rep eq "\r\n");
276 }
277 } else {
278 $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
279 die "Socket to $serv: $!\n" unless defined $s;
280 }
281
a57a9493
MU
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();
285
73bcf533 286 $rep = <$s>;
a57a9493 287
86d11cf2 288 if ($rep ne "I LOVE YOU\n") {
a57a9493
MU
289 $rep="<unknown>" unless $rep;
290 die "AuthReply: $rep\n";
291 }
292 $self->{'socketo'} = $s;
293 $self->{'socketi'} = $s;
34155390 294 } else { # local or ext: Fork off our own cvs server.
a57a9493
MU
295 my $pr = IO::Pipe->new();
296 my $pw = IO::Pipe->new();
297 my $pid = fork();
298 die "Fork: $!\n" unless defined $pid;
8d0ea311
SV
299 my $cvs = 'cvs';
300 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
34155390
SV
301 my $rsh = 'rsh';
302 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
303
304 my @cvs = ($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, $2);
311 }
312 if (!$local) {
313 if ($user) {
314 unshift @cvs, $rsh, '-l', $user, $host;
315 } else {
316 unshift @cvs, $rsh, $host;
317 }
318 }
319
86d11cf2 320 unless ($pid) {
a57a9493
MU
321 $pr->writer();
322 $pw->reader();
a57a9493
MU
323 dup2($pw->fileno(),0);
324 dup2($pr->fileno(),1);
325 $pr->close();
326 $pw->close();
34155390 327 exec(@cvs);
a57a9493
MU
328 }
329 $pw->writer();
330 $pr->reader();
331 $self->{'socketo'} = $pw;
332 $self->{'socketi'} = $pr;
333 }
334 $self->{'socketo'}->write("Root $repo\n");
335
336 # Trial and error says that this probably is the minimum set
b0921331 337 $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
a57a9493
MU
338
339 $self->{'socketo'}->write("valid-requests\n");
340 $self->{'socketo'}->flush();
341
342 chomp(my $rep=$self->readline());
86d11cf2 343 if ($rep !~ s/^Valid-requests\s*//) {
a57a9493
MU
344 $rep="<unknown>" unless $rep;
345 die "Expected Valid-requests from server, but got: $rep\n";
346 }
347 chomp(my $res=$self->readline());
348 die "validReply: $res\n" if $res ne "ok";
349
350 $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
351 $self->{'repo'} = $repo;
352}
353
354sub readline {
86d11cf2 355 my ($self) = @_;
a57a9493
MU
356 return $self->{'socketi'}->getline();
357}
358
359sub _file {
360 # Request a file with a given revision.
361 # Trial and error says this is a good way to do it. :-/
86d11cf2 362 my ($self,$fn,$rev) = @_;
a57a9493
MU
363 $self->{'socketo'}->write("Argument -N\n") or return undef;
364 $self->{'socketo'}->write("Argument -P\n") or return undef;
abe05822
ML
365 # -kk: Linus' version doesn't use it - defaults to off
366 if ($opt_k) {
367 $self->{'socketo'}->write("Argument -kk\n") or return undef;
368 }
a57a9493
MU
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;
4f7c0caa 375 # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
a57a9493
MU
376 $self->{'socketo'}->write("co\n") or return undef;
377 $self->{'socketo'}->flush() or return undef;
378 $self->{'lines'} = 0;
379 return 1;
380}
381sub _line {
382 # Read a line from the server.
383 # ... except that 'line' may be an entire file. ;-)
86d11cf2 384 my ($self, $fh) = @_;
a57a9493
MU
385 die "Not in lines" unless defined $self->{'lines'};
386
387 my $line;
2eb6d82e 388 my $res=0;
86d11cf2 389 while (defined($line = $self->readline())) {
a57a9493
MU
390 # M U gnupg-cvs-rep/AUTHORS
391 # Updated gnupg-cvs-rep/
392 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
393 # /AUTHORS/1.1///T1.1
394 # u=rw,g=rw,o=rw
395 # 0
396 # ok
397
86d11cf2 398 if ($line =~ s/^(?:Created|Updated) //) {
a57a9493
MU
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="";
55cad842 408 $res = $self->_fetchfile($fh, $cnt);
86d11cf2 409 } elsif ($line =~ s/^ //) {
2eb6d82e
SV
410 print $fh $line;
411 $res += length($line);
86d11cf2 412 } elsif ($line =~ /^M\b/) {
a57a9493 413 # output, do nothing
86d11cf2 414 } elsif ($line =~ /^Mbinary\b/) {
a57a9493
MU
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="";
55cad842 420 $res += $self->_fetchfile($fh, $cnt);
a57a9493
MU
421 } else {
422 chomp $line;
86d11cf2 423 if ($line eq "ok") {
a57a9493
MU
424 # print STDERR "S: ok (".length($res).")\n";
425 return $res;
86d11cf2 426 } elsif ($line =~ s/^E //) {
a57a9493 427 # print STDERR "S: $line\n";
86d11cf2 428 } elsif ($line =~ /^(Remove-entry|Removed) /i) {
8b8840e0
MU
429 $line = $self->readline(); # filename
430 $line = $self->readline(); # OK
431 chomp $line;
432 die "Unknown: $line" if $line ne "ok";
433 return -1;
a57a9493
MU
434 } else {
435 die "Unknown: $line\n";
436 }
437 }
438 }
39ba7d54 439 return undef;
a57a9493
MU
440}
441sub file {
86d11cf2 442 my ($self,$fn,$rev) = @_;
a57a9493
MU
443 my $res;
444
a6080a0a 445 my ($fh, $name) = tempfile('gitcvs.XXXXXX',
2eb6d82e
SV
446 DIR => File::Spec->tmpdir(), UNLINK => 1);
447
448 $self->_file($fn,$rev) and $res = $self->_line($fh);
449
450 if (!defined $res) {
39ba7d54
MM
451 print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
452 truncate $fh, 0;
2eb6d82e 453 $self->conn();
39ba7d54 454 $self->_file($fn,$rev) or die "No file command send";
2eb6d82e 455 $res = $self->_line($fh);
39ba7d54 456 die "Retry failed" unless defined $res;
a57a9493 457 }
c619ad51 458 close ($fh);
a57a9493 459
2eb6d82e 460 return ($name, $res);
a57a9493 461}
55cad842
ML
462sub _fetchfile {
463 my ($self, $fh, $cnt) = @_;
61efa5e3 464 my $res = 0;
55cad842 465 my $bufsize = 1024 * 1024;
86d11cf2 466 while ($cnt) {
55cad842
ML
467 if ($bufsize > $cnt) {
468 $bufsize = $cnt;
469 }
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;
476 }
477 return $res;
478}
a57a9493
MU
479
480
481package main;
482
2a3e1a85 483my $cvs = CVSconn->new($opt_d, $cvs_tree);
a57a9493
MU
484
485
486sub pdate($) {
86d11cf2 487 my ($d) = @_;
a57a9493
MU
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);
9718a00b
TM
492}
493
a57a9493 494sub pmode($) {
86d11cf2 495 my ($mode) = @_;
a57a9493
MU
496 my $m = 0;
497 my $mm = 0;
498 my $um = 0;
499 for my $x(split(//,$mode)) {
86d11cf2 500 if ($x eq ",") {
a57a9493
MU
501 $m |= $mm&$um;
502 $mm = 0;
503 $um = 0;
86d11cf2
JH
504 } elsif ($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
a57a9493
MU
511 } else { die "Unknown mode: $mode\n";
512 }
513 }
514 $m |= $mm&$um;
515 return $m;
516}
d4f8b390 517
a57a9493
MU
518sub getwd() {
519 my $pwd = `pwd`;
520 chomp $pwd;
521 return $pwd;
d4f8b390
LT
522}
523
e73aefe4
JK
524sub is_sha1 {
525 my $s = shift;
526 return $s =~ /^[a-f0-9]{40}$/;
527}
db4b6582 528
9da0dabc
JK
529sub get_headref ($) {
530 my $name = shift;
531 my $r = `git rev-parse --verify '$name' 2>/dev/null`;
532 return undef unless $? == 0;
533 chomp $r;
534 return $r;
db4b6582
ML
535}
536
a57a9493
MU
537-d $git_tree
538 or mkdir($git_tree,0777)
539 or die "Could not create $git_tree: $!";
540chdir($git_tree);
d4f8b390 541
a57a9493 542my $last_branch = "";
46541669 543my $orig_branch = "";
a57a9493 544my %branch_date;
8a5f2eac 545my $tip_at_start = undef;
a57a9493
MU
546
547my $git_dir = $ENV{"GIT_DIR"} || ".git";
548$git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
549$ENV{"GIT_DIR"} = $git_dir;
79ee456c
SV
550my $orig_git_index;
551$orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
8f732649
ML
552
553my %index; # holds filenames of one index per branch
061303f0 554
86d11cf2 555unless (-d $git_dir) {
5c94f87e 556 system("git-init");
a57a9493
MU
557 die "Cannot init the GIT db at $git_tree: $?\n" if $?;
558 system("git-read-tree");
559 die "Cannot init an empty tree: $?\n" if $?;
560
561 $last_branch = $opt_o;
46541669 562 $orig_branch = "";
a57a9493 563} else {
8366a10a
PR
564 open(F, "git-symbolic-ref HEAD |") or
565 die "Cannot run git-symbolic-ref: $!\n";
566 chomp ($last_branch = <F>);
567 $last_branch = basename($last_branch);
568 close(F);
86d11cf2 569 unless ($last_branch) {
46541669
MU
570 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
571 $last_branch = "master";
572 }
573 $orig_branch = $last_branch;
8a5f2eac 574 $tip_at_start = `git-rev-parse --verify HEAD`;
a57a9493
MU
575
576 # Get the last import timestamps
1f24c587 577 my $fmt = '($ref, $author) = (%(refname), %(author));';
8b7f5fc1 578 open(H, "git-for-each-ref --perl --format='$fmt' $remote |") or
1f24c587 579 die "Cannot run git-for-each-ref: $!\n";
86d11cf2 580 while (defined(my $entry = <H>)) {
1f24c587
AW
581 my ($ref, $author);
582 eval($entry) || die "cannot eval refs list: $@";
8b7f5fc1 583 my ($head) = ($ref =~ m|^$remote/(.*)|);
1f24c587
AW
584 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
585 $branch_date{$head} = $1;
a57a9493 586 }
1f24c587 587 close(H);
7ca055f7
SS
588 if (!exists $branch_date{$opt_o}) {
589 die "Branch '$opt_o' does not exist.\n".
590 "Either use the correct '-o branch' option,\n".
591 "or import to a new repository.\n";
592 }
a57a9493
MU
593}
594
595-d $git_dir
596 or die "Could not create git subdir ($git_dir).\n";
597
ffd97f3a
AE
598# now we read (and possibly save) author-info as well
599-f "$git_dir/cvs-authors" and
600 read_author_info("$git_dir/cvs-authors");
601if ($opt_A) {
602 read_author_info($opt_A);
603 write_author_info("$git_dir/cvs-authors");
604}
605
2f57c697
ML
606
607#
608# run cvsps into a file unless we are getting
609# it passed as a file via $opt_P
610#
4083c2fc 611my $cvspsfile;
2f57c697
ML
612unless ($opt_P) {
613 print "Running cvsps...\n" if $opt_v;
614 my $pid = open(CVSPS,"-|");
4083c2fc 615 my $cvspsfh;
2f57c697 616 die "Cannot fork: $!\n" unless defined $pid;
86d11cf2 617 unless ($pid) {
2f57c697
ML
618 my @opt;
619 @opt = split(/,/,$opt_p) if defined $opt_p;
620 unshift @opt, '-z', $opt_z if defined $opt_z;
621 unshift @opt, '-q' unless defined $opt_v;
622 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
623 push @opt, '--cvs-direct';
624 }
625 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
626 die "Could not start cvsps: $!\n";
df73e9c6 627 }
4083c2fc
ML
628 ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
629 DIR => File::Spec->tmpdir());
2f57c697
ML
630 while (<CVSPS>) {
631 print $cvspsfh $_;
211dcac6 632 }
2f57c697
ML
633 close CVSPS;
634 close $cvspsfh;
4083c2fc
ML
635} else {
636 $cvspsfile = $opt_P;
a57a9493
MU
637}
638
4083c2fc 639open(CVS, "<$cvspsfile") or die $!;
2f57c697 640
a57a9493
MU
641## cvsps output:
642#---------------------
643#PatchSet 314
644#Date: 1999/09/18 13:03:59
645#Author: wkoch
646#Branch: STABLE-BRANCH-1-0
647#Ancestor branch: HEAD
648#Tag: (none)
649#Log:
650# See ChangeLog: Sat Sep 18 13:03:28 CEST 1999 Werner Koch
651#Members:
652# README:1.57->1.57.2.1
653# VERSION:1.96->1.96.2.1
654#
655#---------------------
656
657my $state = 0;
658
e73aefe4
JK
659sub update_index (\@\@) {
660 my $old = shift;
661 my $new = shift;
6a1871e1
JK
662 open(my $fh, '|-', qw(git-update-index -z --index-info))
663 or die "unable to open git-update-index: $!";
664 print $fh
665 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
e73aefe4 666 @$old),
6a1871e1 667 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
e73aefe4 668 @$new)
6a1871e1
JK
669 or die "unable to write to git-update-index: $!";
670 close $fh
671 or die "unable to write to git-update-index: $!";
672 $? and die "git-update-index reported error: $?";
e73aefe4 673}
a57a9493 674
e73aefe4
JK
675sub write_tree () {
676 open(my $fh, '-|', qw(git-write-tree))
677 or die "unable to open git-write-tree: $!";
678 chomp(my $tree = <$fh>);
679 is_sha1($tree)
680 or die "Cannot get tree id ($tree): $!";
681 close($fh)
a57a9493
MU
682 or die "Error running git-write-tree: $?\n";
683 print "Tree ID $tree\n" if $opt_v;
e73aefe4
JK
684 return $tree;
685}
a57a9493 686
86d11cf2
JH
687my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
688my (@old,@new,@skipped,%ignorebranch);
71b08148
ML
689
690# commits that cvsps cannot place anywhere...
691$ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
692
e73aefe4 693sub commit {
9da0dabc
JK
694 if ($branch eq $opt_o && !$index{branch} &&
695 !get_headref("$remote/$branch")) {
c5f448b0 696 # looks like an initial commit
5c94f87e 697 # use the index primed by git-init
23fcdc79
MM
698 $ENV{GIT_INDEX_FILE} = "$git_dir/index";
699 $index{$branch} = "$git_dir/index";
c5f448b0
ML
700 } else {
701 # use an index per branch to speed up
702 # imports of projects with many branches
703 unless ($index{$branch}) {
704 $index{$branch} = tmpnam();
705 $ENV{GIT_INDEX_FILE} = $index{$branch};
706 if ($ancestor) {
8b7f5fc1 707 system("git-read-tree", "$remote/$ancestor");
c5f448b0 708 } else {
8b7f5fc1 709 system("git-read-tree", "$remote/$branch");
c5f448b0
ML
710 }
711 die "read-tree failed: $?\n" if $?;
712 }
713 }
714 $ENV{GIT_INDEX_FILE} = $index{$branch};
715
e73aefe4
JK
716 update_index(@old, @new);
717 @old = @new = ();
718 my $tree = write_tree();
9da0dabc 719 my $parent = get_headref("$remote/$last_branch");
e73aefe4
JK
720 print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
721
722 my @commit_args;
723 push @commit_args, ("-p", $parent) if $parent;
724
725 # loose detection of merges
726 # based on the commit msg
727 foreach my $rx (@mergerx) {
728 next unless $logmsg =~ $rx && $1;
729 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
9da0dabc 730 if (my $sha1 = get_headref("$remote/$mparent")) {
e73aefe4
JK
731 push @commit_args, '-p', $mparent;
732 print "Merge parent branch: $mparent\n" if $opt_v;
db4b6582 733 }
a57a9493 734 }
e73aefe4
JK
735
736 my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
62bf0d96
JK
737 $ENV{GIT_AUTHOR_NAME} = $author_name;
738 $ENV{GIT_AUTHOR_EMAIL} = $author_email;
739 $ENV{GIT_AUTHOR_DATE} = $commit_date;
740 $ENV{GIT_COMMITTER_NAME} = $author_name;
741 $ENV{GIT_COMMITTER_EMAIL} = $author_email;
742 $ENV{GIT_COMMITTER_DATE} = $commit_date;
e73aefe4 743 my $pid = open2(my $commit_read, my $commit_write,
e73aefe4 744 'git-commit-tree', $tree, @commit_args);
e371046b
MU
745
746 # compatibility with git2cvs
747 substr($logmsg,32767) = "" if length($logmsg) > 32767;
748 $logmsg =~ s/[\s\n]+\z//;
749
5179c8a5
ML
750 if (@skipped) {
751 $logmsg .= "\n\n\nSKIPPED:\n\t";
752 $logmsg .= join("\n\t", @skipped) . "\n";
f396f01f 753 @skipped = ();
5179c8a5
ML
754 }
755
e73aefe4 756 print($commit_write "$logmsg\n") && close($commit_write)
a57a9493 757 or die "Error writing to git-commit-tree: $!\n";
2a3e1a85 758
e73aefe4
JK
759 print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
760 chomp(my $cid = <$commit_read>);
761 is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
a57a9493 762 print "Commit ID $cid\n" if $opt_v;
e73aefe4 763 close($commit_read);
2a3e1a85
MU
764
765 waitpid($pid,0);
766 die "Error running git-commit-tree: $?\n" if $?;
a57a9493 767
8b7f5fc1 768 system("git-update-ref $remote/$branch $cid") == 0
a57a9493
MU
769 or die "Cannot write branch $branch for update: $!\n";
770
86d11cf2 771 if ($tag) {
86d11cf2 772 my ($xtag) = $tag;
0d821d4d
PA
773 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
774 $xtag =~ tr/_/\./ if ( $opt_u );
34c99da2 775 $xtag =~ s/[\/]/$opt_s/g;
a6080a0a 776
ee834cf0 777 system('git-tag', '-f', $xtag, $cid) == 0
0d821d4d 778 or die "Cannot create tag $xtag: $!\n";
0d821d4d
PA
779
780 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
a57a9493 781 }
a57a9493
MU
782};
783
06918348 784my $commitcount = 1;
86d11cf2 785while (<CVS>) {
a57a9493 786 chomp;
86d11cf2 787 if ($state == 0 and /^-+$/) {
a57a9493 788 $state = 1;
86d11cf2 789 } elsif ($state == 0) {
a57a9493
MU
790 $state = 1;
791 redo;
86d11cf2 792 } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
a57a9493
MU
793 $patchset = 0+$_;
794 $state=2;
86d11cf2 795 } elsif ($state == 2 and s/^Date:\s+//) {
a57a9493 796 $date = pdate($_);
86d11cf2 797 unless ($date) {
a57a9493
MU
798 print STDERR "Could not parse date: $_\n";
799 $state=0;
800 next;
801 }
802 $state=3;
86d11cf2 803 } elsif ($state == 3 and s/^Author:\s+//) {
a57a9493 804 s/\s+$//;
94c23343
JH
805 if (/^(.*?)\s+<(.*)>/) {
806 ($author_name, $author_email) = ($1, $2);
ffd97f3a
AE
807 } elsif ($conv_author_name{$_}) {
808 $author_name = $conv_author_name{$_};
809 $author_email = $conv_author_email{$_};
94c23343
JH
810 } else {
811 $author_name = $author_email = $_;
812 }
a57a9493 813 $state = 4;
86d11cf2 814 } elsif ($state == 4 and s/^Branch:\s+//) {
a57a9493 815 s/\s+$//;
a0554224 816 tr/_/\./ if ( $opt_u );
fbfd60d6 817 s/[\/]/$opt_s/g;
a57a9493
MU
818 $branch = $_;
819 $state = 5;
86d11cf2 820 } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
a57a9493
MU
821 s/\s+$//;
822 $ancestor = $_;
0fa2824f 823 $ancestor = $opt_o if $ancestor eq "HEAD";
a57a9493 824 $state = 6;
86d11cf2 825 } elsif ($state == 5) {
a57a9493
MU
826 $ancestor = undef;
827 $state = 6;
828 redo;
86d11cf2 829 } elsif ($state == 6 and s/^Tag:\s+//) {
a57a9493 830 s/\s+$//;
86d11cf2 831 if ($_ eq "(none)") {
a57a9493
MU
832 $tag = undef;
833 } else {
834 $tag = $_;
835 }
836 $state = 7;
86d11cf2 837 } elsif ($state == 7 and /^Log:/) {
a57a9493
MU
838 $logmsg = "";
839 $state = 8;
86d11cf2 840 } elsif ($state == 8 and /^Members:/) {
a57a9493 841 $branch = $opt_o if $branch eq "HEAD";
86d11cf2 842 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
a57a9493 843 # skip
9da07f34 844 print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
a57a9493
MU
845 $state = 11;
846 next;
847 }
ded9f400 848 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
6211988f
ML
849 # skip if the commit is too recent
850 # that the cvsps default fuzz is 300s, we give ourselves another
851 # 300s just in case -- this also prevents skipping commits
852 # due to server clock drift
853 print "skip patchset $patchset: $date too recent\n" if $opt_v;
854 $state = 11;
855 next;
856 }
71b08148
ML
857 if (exists $ignorebranch{$branch}) {
858 print STDERR "Skipping $branch\n";
859 $state = 11;
860 next;
861 }
86d11cf2
JH
862 if ($ancestor) {
863 if ($ancestor eq $branch) {
71b08148
ML
864 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
865 $ancestor = $opt_o;
866 }
8b7f5fc1 867 if (-f "$git_dir/$remote/$branch") {
a57a9493
MU
868 print STDERR "Branch $branch already exists!\n";
869 $state=11;
870 next;
871 }
8b7f5fc1 872 unless (open(H,"$git_dir/$remote/$ancestor")) {
a57a9493 873 print STDERR "Branch $ancestor does not exist!\n";
71b08148 874 $ignorebranch{$branch} = 1;
a57a9493
MU
875 $state=11;
876 next;
877 }
878 chomp(my $id = <H>);
879 close(H);
8b7f5fc1 880 unless (open(H,"> $git_dir/$remote/$branch")) {
a57a9493 881 print STDERR "Could not create branch $branch: $!\n";
71b08148 882 $ignorebranch{$branch} = 1;
a57a9493
MU
883 $state=11;
884 next;
885 }
886 print H "$id\n"
887 or die "Could not write branch $branch: $!";
888 close(H)
889 or die "Could not write branch $branch: $!";
890 }
46e63efc 891 $last_branch = $branch if $branch ne $last_branch;
a57a9493 892 $state = 9;
86d11cf2 893 } elsif ($state == 8) {
a57a9493 894 $logmsg .= "$_\n";
86d11cf2 895 } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
a57a9493 896# VERSION:1.96->1.96.2.1
2a3e1a85 897 my $init = ($2 eq "INITIAL");
a57a9493 898 my $fn = $1;
f65ae603
MU
899 my $rev = $3;
900 $fn =~ s#^/+##;
5179c8a5
ML
901 if ($opt_S && $fn =~ m/$opt_S/) {
902 print "SKIPPING $fn v $rev\n";
903 push(@skipped, $fn);
904 next;
905 }
906 print "Fetching $fn v $rev\n" if $opt_v;
2eb6d82e 907 my ($tmpname, $size) = $cvs->file($fn,$rev);
86d11cf2 908 if ($size == -1) {
8b8840e0
MU
909 push(@old,$fn);
910 print "Drop $fn\n" if $opt_v;
911 } else {
912 print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
dd27478f
JH
913 my $pid = open(my $F, '-|');
914 die $! unless defined $pid;
915 if (!$pid) {
916 exec("git-hash-object", "-w", $tmpname)
8b8840e0 917 or die "Cannot create object: $!\n";
dd27478f 918 }
8b8840e0
MU
919 my $sha = <$F>;
920 chomp $sha;
921 close $F;
922 my $mode = pmode($cvs->{'mode'});
923 push(@new,[$mode, $sha, $fn]); # may be resurrected!
924 }
2eb6d82e 925 unlink($tmpname);
86d11cf2 926 } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
f65ae603
MU
927 my $fn = $1;
928 $fn =~ s#^/+##;
929 push(@old,$fn);
8b8840e0 930 print "Delete $fn\n" if $opt_v;
86d11cf2 931 } elsif ($state == 9 and /^\s*$/) {
a57a9493 932 $state = 10;
86d11cf2 933 } elsif (($state == 9 or $state == 10) and /^-+$/) {
4adcea99
LT
934 $commitcount++;
935 if ($opt_L && $commitcount > $opt_L) {
06918348
ML
936 last;
937 }
c4b16f8d 938 commit();
4adcea99
LT
939 if (($commitcount & 1023) == 0) {
940 system("git repack -a -d");
941 }
a57a9493 942 $state = 1;
86d11cf2 943 } elsif ($state == 11 and /^-+$/) {
a57a9493 944 $state = 1;
86d11cf2 945 } elsif (/^-+$/) { # end of unknown-line processing
a57a9493 946 $state = 1;
86d11cf2 947 } elsif ($state != 11) { # ignore stuff when skipping
a57a9493
MU
948 print "* UNKNOWN LINE * $_\n";
949 }
950}
c4b16f8d 951commit() if $branch and $state != 11;
d4f8b390 952
4083c2fc
ML
953unless ($opt_P) {
954 unlink($cvspsfile);
955}
956
efe4abd1
JM
957# The heuristic of repacking every 1024 commits can leave a
958# lot of unpacked data. If there is more than 1MB worth of
959# not-packed objects, repack once more.
960my $line = `git-count-objects`;
961if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
962 my ($n_objects, $kb) = ($1, $2);
963 1024 < $kb
964 and system("git repack -a -d");
965}
966
8f732649 967foreach my $git_index (values %index) {
23fcdc79 968 if ($git_index ne "$git_dir/index") {
c5f448b0
ML
969 unlink($git_index);
970 }
8f732649 971}
79ee456c 972
210569f9
SV
973if (defined $orig_git_index) {
974 $ENV{GIT_INDEX_FILE} = $orig_git_index;
975} else {
976 delete $ENV{GIT_INDEX_FILE};
977}
978
46541669 979# Now switch back to the branch we were in before all of this happened
86d11cf2 980if ($orig_branch) {
8a5f2eac
JH
981 print "DONE.\n" if $opt_v;
982 if ($opt_i) {
983 exit 0;
984 }
985 my $tip_at_end = `git-rev-parse --verify HEAD`;
986 if ($tip_at_start ne $tip_at_end) {
cb9594e2 987 for ($tip_at_start, $tip_at_end) { chomp; }
8a5f2eac
JH
988 print "Fetched into the current branch.\n" if $opt_v;
989 system(qw(git-read-tree -u -m),
990 $tip_at_start, $tip_at_end);
991 die "Fast-forward update failed: $?\n" if $?;
992 }
993 else {
8b7f5fc1 994 system(qw(git-merge cvsimport HEAD), "$remote/$opt_o");
8a5f2eac
JH
995 die "Could not merge $opt_o into the current branch.\n" if $?;
996 }
46541669
MU
997} else {
998 $orig_branch = "master";
999 print "DONE; creating $orig_branch branch\n" if $opt_v;
8b7f5fc1 1000 system("git-update-ref", "refs/heads/master", "$remote/$opt_o")
46541669 1001 unless -f "$git_dir/refs/heads/master";
06baffd3
AW
1002 system("git-symbolic-ref", "$remote/HEAD", "$remote/$opt_o")
1003 if ($opt_r && $opt_o ne 'HEAD');
8366a10a 1004 system('git-update-ref', 'HEAD', "$orig_branch");
c1c774e7 1005 unless ($opt_i) {
7051c3b1 1006 system('git checkout -f');
c1c774e7
SV
1007 die "checkout failed: $?\n" if $?;
1008 }
46541669 1009}