cvsimport: have default merge regex allow for dashes in the branch name
[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
cfc44a12 91# convert getopts specs for use by git config
ed35dece
JB
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;
cfc44a12 99 my $arg = 'git config';
ed35dece
JB
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 }
ed35dece
JB
111}
112
8b7f5fc1 113my $opts = "haivmkuo:d:p:r:C:z:s:M:P:A:S:L:";
ed35dece
JB
114read_repo_config($opts);
115getopts($opts) or usage();
a57a9493
MU
116usage if $opt_h;
117
67d23242 118if (@ARGV == 0) {
cfc44a12 119 chomp(my $module = `git config --get cvsimport.module`);
67d23242
JK
120 push(@ARGV, $module) if $? == 0;
121}
7bf77644 122@ARGV <= 1 or usage("You can't specify more than one CVS module");
a57a9493 123
86d11cf2 124if ($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) {
fbbbc362 167 @mergerx = ( qr/\b(?:from|of|merge|merging|merged) ([-\w]+)/i );
db4b6582
ML
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
2e458e05
GH
226 # if username is not explicit in CVSROOT, then use current user, as cvs would
227 $user=(getlogin() || $ENV{'LOGNAME'} || $ENV{'USER'} || "anonymous") unless $user;
2a3e1a85 228 my $rr2 = "-";
86d11cf2 229 unless ($port) {
a57a9493
MU
230 $rr2 = ":pserver:$user\@$serv:$repo";
231 $port=2401;
232 }
233 my $rr = ":pserver:$user\@$serv:$port$repo";
234
86d11cf2 235 unless ($pass) {
a57a9493
MU
236 open(H,$ENV{'HOME'}."/.cvspass") and do {
237 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
86d11cf2 238 while (<H>) {
a57a9493
MU
239 chomp;
240 s/^\/\d+\s+//;
241 my ($w,$p) = split(/\s/,$_,2);
86d11cf2 242 if ($w eq $rr or $w eq $rr2) {
a57a9493
MU
243 $pass = $p;
244 last;
245 }
246 }
247 };
248 }
249 $pass="A" unless $pass;
250
73bcf533 251 my ($s, $rep);
86d11cf2 252 if ($proxyhost) {
73bcf533
IA
253
254 # Use a HTTP Proxy. Only works for HTTP proxies that
255 # don't require user authentication
256 #
257 # See: http://www.ietf.org/rfc/rfc2817.txt
258
259 $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
260 die "Socket to $proxyhost: $!\n" unless defined $s;
261 $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
262 or die "Write to $proxyhost: $!\n";
263 $s->flush();
264
265 $rep = <$s>;
266
267 # The answer should look like 'HTTP/1.x 2yy ....'
86d11cf2 268 if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
73bcf533
IA
269 die "Proxy connect: $rep\n";
270 }
271 # Skip up to the empty line of the proxy server output
272 # including the response headers.
273 while ($rep = <$s>) {
274 last if (!defined $rep ||
275 $rep eq "\n" ||
276 $rep eq "\r\n");
277 }
278 } else {
279 $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
280 die "Socket to $serv: $!\n" unless defined $s;
281 }
282
a57a9493
MU
283 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
284 or die "Write to $serv: $!\n";
285 $s->flush();
286
73bcf533 287 $rep = <$s>;
a57a9493 288
86d11cf2 289 if ($rep ne "I LOVE YOU\n") {
a57a9493
MU
290 $rep="<unknown>" unless $rep;
291 die "AuthReply: $rep\n";
292 }
293 $self->{'socketo'} = $s;
294 $self->{'socketi'} = $s;
34155390 295 } else { # local or ext: Fork off our own cvs server.
a57a9493
MU
296 my $pr = IO::Pipe->new();
297 my $pw = IO::Pipe->new();
298 my $pid = fork();
299 die "Fork: $!\n" unless defined $pid;
8d0ea311
SV
300 my $cvs = 'cvs';
301 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
34155390
SV
302 my $rsh = 'rsh';
303 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
304
305 my @cvs = ($cvs, 'server');
306 my ($local, $user, $host);
307 $local = $repo =~ s/:local://;
308 if (!$local) {
309 $repo =~ s/:ext://;
310 $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
311 ($user, $host) = ($1, $2);
312 }
313 if (!$local) {
314 if ($user) {
315 unshift @cvs, $rsh, '-l', $user, $host;
316 } else {
317 unshift @cvs, $rsh, $host;
318 }
319 }
320
86d11cf2 321 unless ($pid) {
a57a9493
MU
322 $pr->writer();
323 $pw->reader();
a57a9493
MU
324 dup2($pw->fileno(),0);
325 dup2($pr->fileno(),1);
326 $pr->close();
327 $pw->close();
34155390 328 exec(@cvs);
a57a9493
MU
329 }
330 $pw->writer();
331 $pr->reader();
332 $self->{'socketo'} = $pw;
333 $self->{'socketi'} = $pr;
334 }
335 $self->{'socketo'}->write("Root $repo\n");
336
337 # Trial and error says that this probably is the minimum set
b0921331 338 $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
a57a9493
MU
339
340 $self->{'socketo'}->write("valid-requests\n");
341 $self->{'socketo'}->flush();
342
343 chomp(my $rep=$self->readline());
86d11cf2 344 if ($rep !~ s/^Valid-requests\s*//) {
a57a9493
MU
345 $rep="<unknown>" unless $rep;
346 die "Expected Valid-requests from server, but got: $rep\n";
347 }
348 chomp(my $res=$self->readline());
349 die "validReply: $res\n" if $res ne "ok";
350
351 $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
352 $self->{'repo'} = $repo;
353}
354
355sub readline {
86d11cf2 356 my ($self) = @_;
a57a9493
MU
357 return $self->{'socketi'}->getline();
358}
359
360sub _file {
361 # Request a file with a given revision.
362 # Trial and error says this is a good way to do it. :-/
86d11cf2 363 my ($self,$fn,$rev) = @_;
a57a9493
MU
364 $self->{'socketo'}->write("Argument -N\n") or return undef;
365 $self->{'socketo'}->write("Argument -P\n") or return undef;
abe05822
ML
366 # -kk: Linus' version doesn't use it - defaults to off
367 if ($opt_k) {
368 $self->{'socketo'}->write("Argument -kk\n") or return undef;
369 }
a57a9493
MU
370 $self->{'socketo'}->write("Argument -r\n") or return undef;
371 $self->{'socketo'}->write("Argument $rev\n") or return undef;
372 $self->{'socketo'}->write("Argument --\n") or return undef;
373 $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
374 $self->{'socketo'}->write("Directory .\n") or return undef;
375 $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
4f7c0caa 376 # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
a57a9493
MU
377 $self->{'socketo'}->write("co\n") or return undef;
378 $self->{'socketo'}->flush() or return undef;
379 $self->{'lines'} = 0;
380 return 1;
381}
382sub _line {
383 # Read a line from the server.
384 # ... except that 'line' may be an entire file. ;-)
86d11cf2 385 my ($self, $fh) = @_;
a57a9493
MU
386 die "Not in lines" unless defined $self->{'lines'};
387
388 my $line;
2eb6d82e 389 my $res=0;
86d11cf2 390 while (defined($line = $self->readline())) {
a57a9493
MU
391 # M U gnupg-cvs-rep/AUTHORS
392 # Updated gnupg-cvs-rep/
393 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
394 # /AUTHORS/1.1///T1.1
395 # u=rw,g=rw,o=rw
396 # 0
397 # ok
398
86d11cf2 399 if ($line =~ s/^(?:Created|Updated) //) {
a57a9493
MU
400 $line = $self->readline(); # path
401 $line = $self->readline(); # Entries line
402 my $mode = $self->readline(); chomp $mode;
403 $self->{'mode'} = $mode;
404 defined (my $cnt = $self->readline())
405 or die "EOF from server after 'Changed'\n";
406 chomp $cnt;
407 die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
408 $line="";
55cad842 409 $res = $self->_fetchfile($fh, $cnt);
86d11cf2 410 } elsif ($line =~ s/^ //) {
2eb6d82e
SV
411 print $fh $line;
412 $res += length($line);
86d11cf2 413 } elsif ($line =~ /^M\b/) {
a57a9493 414 # output, do nothing
86d11cf2 415 } elsif ($line =~ /^Mbinary\b/) {
a57a9493
MU
416 my $cnt;
417 die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
418 chomp $cnt;
419 die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
420 $line="";
55cad842 421 $res += $self->_fetchfile($fh, $cnt);
a57a9493
MU
422 } else {
423 chomp $line;
86d11cf2 424 if ($line eq "ok") {
a57a9493
MU
425 # print STDERR "S: ok (".length($res).")\n";
426 return $res;
86d11cf2 427 } elsif ($line =~ s/^E //) {
a57a9493 428 # print STDERR "S: $line\n";
86d11cf2 429 } elsif ($line =~ /^(Remove-entry|Removed) /i) {
8b8840e0
MU
430 $line = $self->readline(); # filename
431 $line = $self->readline(); # OK
432 chomp $line;
433 die "Unknown: $line" if $line ne "ok";
434 return -1;
a57a9493
MU
435 } else {
436 die "Unknown: $line\n";
437 }
438 }
439 }
39ba7d54 440 return undef;
a57a9493
MU
441}
442sub file {
86d11cf2 443 my ($self,$fn,$rev) = @_;
a57a9493
MU
444 my $res;
445
a6080a0a 446 my ($fh, $name) = tempfile('gitcvs.XXXXXX',
2eb6d82e
SV
447 DIR => File::Spec->tmpdir(), UNLINK => 1);
448
449 $self->_file($fn,$rev) and $res = $self->_line($fh);
450
451 if (!defined $res) {
39ba7d54
MM
452 print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
453 truncate $fh, 0;
2eb6d82e 454 $self->conn();
39ba7d54 455 $self->_file($fn,$rev) or die "No file command send";
2eb6d82e 456 $res = $self->_line($fh);
39ba7d54 457 die "Retry failed" unless defined $res;
a57a9493 458 }
c619ad51 459 close ($fh);
a57a9493 460
2eb6d82e 461 return ($name, $res);
a57a9493 462}
55cad842
ML
463sub _fetchfile {
464 my ($self, $fh, $cnt) = @_;
61efa5e3 465 my $res = 0;
55cad842 466 my $bufsize = 1024 * 1024;
86d11cf2 467 while ($cnt) {
55cad842
ML
468 if ($bufsize > $cnt) {
469 $bufsize = $cnt;
470 }
471 my $buf;
472 my $num = $self->{'socketi'}->read($buf,$bufsize);
473 die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
474 print $fh $buf;
475 $res += $num;
476 $cnt -= $num;
477 }
478 return $res;
479}
a57a9493
MU
480
481
482package main;
483
2a3e1a85 484my $cvs = CVSconn->new($opt_d, $cvs_tree);
a57a9493
MU
485
486
487sub pdate($) {
86d11cf2 488 my ($d) = @_;
a57a9493
MU
489 m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
490 or die "Unparseable date: $d\n";
491 my $y=$1; $y-=1900 if $y>1900;
492 return timegm($6||0,$5,$4,$3,$2-1,$y);
9718a00b
TM
493}
494
a57a9493 495sub pmode($) {
86d11cf2 496 my ($mode) = @_;
a57a9493
MU
497 my $m = 0;
498 my $mm = 0;
499 my $um = 0;
500 for my $x(split(//,$mode)) {
86d11cf2 501 if ($x eq ",") {
a57a9493
MU
502 $m |= $mm&$um;
503 $mm = 0;
504 $um = 0;
86d11cf2
JH
505 } elsif ($x eq "u") { $um |= 0700;
506 } elsif ($x eq "g") { $um |= 0070;
507 } elsif ($x eq "o") { $um |= 0007;
508 } elsif ($x eq "r") { $mm |= 0444;
509 } elsif ($x eq "w") { $mm |= 0222;
510 } elsif ($x eq "x") { $mm |= 0111;
511 } elsif ($x eq "=") { # do nothing
a57a9493
MU
512 } else { die "Unknown mode: $mode\n";
513 }
514 }
515 $m |= $mm&$um;
516 return $m;
517}
d4f8b390 518
a57a9493
MU
519sub getwd() {
520 my $pwd = `pwd`;
521 chomp $pwd;
522 return $pwd;
d4f8b390
LT
523}
524
e73aefe4
JK
525sub is_sha1 {
526 my $s = shift;
527 return $s =~ /^[a-f0-9]{40}$/;
528}
db4b6582 529
9da0dabc
JK
530sub get_headref ($) {
531 my $name = shift;
532 my $r = `git rev-parse --verify '$name' 2>/dev/null`;
533 return undef unless $? == 0;
534 chomp $r;
535 return $r;
db4b6582
ML
536}
537
a57a9493
MU
538-d $git_tree
539 or mkdir($git_tree,0777)
540 or die "Could not create $git_tree: $!";
541chdir($git_tree);
d4f8b390 542
a57a9493 543my $last_branch = "";
46541669 544my $orig_branch = "";
a57a9493 545my %branch_date;
8a5f2eac 546my $tip_at_start = undef;
a57a9493
MU
547
548my $git_dir = $ENV{"GIT_DIR"} || ".git";
549$git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
550$ENV{"GIT_DIR"} = $git_dir;
79ee456c
SV
551my $orig_git_index;
552$orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
8f732649
ML
553
554my %index; # holds filenames of one index per branch
061303f0 555
86d11cf2 556unless (-d $git_dir) {
5c94f87e 557 system("git-init");
a57a9493
MU
558 die "Cannot init the GIT db at $git_tree: $?\n" if $?;
559 system("git-read-tree");
560 die "Cannot init an empty tree: $?\n" if $?;
561
562 $last_branch = $opt_o;
46541669 563 $orig_branch = "";
a57a9493 564} else {
8366a10a
PR
565 open(F, "git-symbolic-ref HEAD |") or
566 die "Cannot run git-symbolic-ref: $!\n";
567 chomp ($last_branch = <F>);
568 $last_branch = basename($last_branch);
569 close(F);
86d11cf2 570 unless ($last_branch) {
46541669
MU
571 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
572 $last_branch = "master";
573 }
574 $orig_branch = $last_branch;
8a5f2eac 575 $tip_at_start = `git-rev-parse --verify HEAD`;
a57a9493
MU
576
577 # Get the last import timestamps
1f24c587 578 my $fmt = '($ref, $author) = (%(refname), %(author));';
8b7f5fc1 579 open(H, "git-for-each-ref --perl --format='$fmt' $remote |") or
1f24c587 580 die "Cannot run git-for-each-ref: $!\n";
86d11cf2 581 while (defined(my $entry = <H>)) {
1f24c587
AW
582 my ($ref, $author);
583 eval($entry) || die "cannot eval refs list: $@";
8b7f5fc1 584 my ($head) = ($ref =~ m|^$remote/(.*)|);
1f24c587
AW
585 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
586 $branch_date{$head} = $1;
a57a9493 587 }
1f24c587 588 close(H);
7ca055f7
SS
589 if (!exists $branch_date{$opt_o}) {
590 die "Branch '$opt_o' does not exist.\n".
591 "Either use the correct '-o branch' option,\n".
592 "or import to a new repository.\n";
593 }
a57a9493
MU
594}
595
596-d $git_dir
597 or die "Could not create git subdir ($git_dir).\n";
598
ffd97f3a
AE
599# now we read (and possibly save) author-info as well
600-f "$git_dir/cvs-authors" and
601 read_author_info("$git_dir/cvs-authors");
602if ($opt_A) {
603 read_author_info($opt_A);
604 write_author_info("$git_dir/cvs-authors");
605}
606
2f57c697
ML
607
608#
609# run cvsps into a file unless we are getting
610# it passed as a file via $opt_P
611#
4083c2fc 612my $cvspsfile;
2f57c697
ML
613unless ($opt_P) {
614 print "Running cvsps...\n" if $opt_v;
615 my $pid = open(CVSPS,"-|");
4083c2fc 616 my $cvspsfh;
2f57c697 617 die "Cannot fork: $!\n" unless defined $pid;
86d11cf2 618 unless ($pid) {
2f57c697
ML
619 my @opt;
620 @opt = split(/,/,$opt_p) if defined $opt_p;
621 unshift @opt, '-z', $opt_z if defined $opt_z;
622 unshift @opt, '-q' unless defined $opt_v;
623 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
624 push @opt, '--cvs-direct';
625 }
626 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
627 die "Could not start cvsps: $!\n";
df73e9c6 628 }
4083c2fc
ML
629 ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
630 DIR => File::Spec->tmpdir());
2f57c697
ML
631 while (<CVSPS>) {
632 print $cvspsfh $_;
211dcac6 633 }
2f57c697 634 close CVSPS;
3a969ef1 635 $? == 0 or die "git-cvsimport: fatal: cvsps reported error\n";
2f57c697 636 close $cvspsfh;
4083c2fc
ML
637} else {
638 $cvspsfile = $opt_P;
a57a9493
MU
639}
640
4083c2fc 641open(CVS, "<$cvspsfile") or die $!;
2f57c697 642
a57a9493
MU
643## cvsps output:
644#---------------------
645#PatchSet 314
646#Date: 1999/09/18 13:03:59
647#Author: wkoch
648#Branch: STABLE-BRANCH-1-0
649#Ancestor branch: HEAD
650#Tag: (none)
651#Log:
652# See ChangeLog: Sat Sep 18 13:03:28 CEST 1999 Werner Koch
653#Members:
654# README:1.57->1.57.2.1
655# VERSION:1.96->1.96.2.1
656#
657#---------------------
658
659my $state = 0;
660
e73aefe4
JK
661sub update_index (\@\@) {
662 my $old = shift;
663 my $new = shift;
6a1871e1
JK
664 open(my $fh, '|-', qw(git-update-index -z --index-info))
665 or die "unable to open git-update-index: $!";
666 print $fh
667 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
e73aefe4 668 @$old),
6a1871e1 669 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
e73aefe4 670 @$new)
6a1871e1
JK
671 or die "unable to write to git-update-index: $!";
672 close $fh
673 or die "unable to write to git-update-index: $!";
674 $? and die "git-update-index reported error: $?";
e73aefe4 675}
a57a9493 676
e73aefe4
JK
677sub write_tree () {
678 open(my $fh, '-|', qw(git-write-tree))
679 or die "unable to open git-write-tree: $!";
680 chomp(my $tree = <$fh>);
681 is_sha1($tree)
682 or die "Cannot get tree id ($tree): $!";
683 close($fh)
a57a9493
MU
684 or die "Error running git-write-tree: $?\n";
685 print "Tree ID $tree\n" if $opt_v;
e73aefe4
JK
686 return $tree;
687}
a57a9493 688
86d11cf2
JH
689my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
690my (@old,@new,@skipped,%ignorebranch);
71b08148
ML
691
692# commits that cvsps cannot place anywhere...
693$ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
694
e73aefe4 695sub commit {
9da0dabc
JK
696 if ($branch eq $opt_o && !$index{branch} &&
697 !get_headref("$remote/$branch")) {
c5f448b0 698 # looks like an initial commit
5c94f87e 699 # use the index primed by git-init
23fcdc79
MM
700 $ENV{GIT_INDEX_FILE} = "$git_dir/index";
701 $index{$branch} = "$git_dir/index";
c5f448b0
ML
702 } else {
703 # use an index per branch to speed up
704 # imports of projects with many branches
705 unless ($index{$branch}) {
706 $index{$branch} = tmpnam();
707 $ENV{GIT_INDEX_FILE} = $index{$branch};
708 if ($ancestor) {
8b7f5fc1 709 system("git-read-tree", "$remote/$ancestor");
c5f448b0 710 } else {
8b7f5fc1 711 system("git-read-tree", "$remote/$branch");
c5f448b0
ML
712 }
713 die "read-tree failed: $?\n" if $?;
714 }
715 }
716 $ENV{GIT_INDEX_FILE} = $index{$branch};
717
e73aefe4
JK
718 update_index(@old, @new);
719 @old = @new = ();
720 my $tree = write_tree();
9da0dabc 721 my $parent = get_headref("$remote/$last_branch");
e73aefe4
JK
722 print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
723
724 my @commit_args;
725 push @commit_args, ("-p", $parent) if $parent;
726
727 # loose detection of merges
728 # based on the commit msg
729 foreach my $rx (@mergerx) {
730 next unless $logmsg =~ $rx && $1;
731 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
9da0dabc 732 if (my $sha1 = get_headref("$remote/$mparent")) {
e73aefe4
JK
733 push @commit_args, '-p', $mparent;
734 print "Merge parent branch: $mparent\n" if $opt_v;
db4b6582 735 }
a57a9493 736 }
e73aefe4
JK
737
738 my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
62bf0d96
JK
739 $ENV{GIT_AUTHOR_NAME} = $author_name;
740 $ENV{GIT_AUTHOR_EMAIL} = $author_email;
741 $ENV{GIT_AUTHOR_DATE} = $commit_date;
742 $ENV{GIT_COMMITTER_NAME} = $author_name;
743 $ENV{GIT_COMMITTER_EMAIL} = $author_email;
744 $ENV{GIT_COMMITTER_DATE} = $commit_date;
e73aefe4 745 my $pid = open2(my $commit_read, my $commit_write,
e73aefe4 746 'git-commit-tree', $tree, @commit_args);
e371046b
MU
747
748 # compatibility with git2cvs
749 substr($logmsg,32767) = "" if length($logmsg) > 32767;
750 $logmsg =~ s/[\s\n]+\z//;
751
5179c8a5
ML
752 if (@skipped) {
753 $logmsg .= "\n\n\nSKIPPED:\n\t";
754 $logmsg .= join("\n\t", @skipped) . "\n";
f396f01f 755 @skipped = ();
5179c8a5
ML
756 }
757
e73aefe4 758 print($commit_write "$logmsg\n") && close($commit_write)
a57a9493 759 or die "Error writing to git-commit-tree: $!\n";
2a3e1a85 760
e73aefe4
JK
761 print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
762 chomp(my $cid = <$commit_read>);
763 is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
a57a9493 764 print "Commit ID $cid\n" if $opt_v;
e73aefe4 765 close($commit_read);
2a3e1a85
MU
766
767 waitpid($pid,0);
768 die "Error running git-commit-tree: $?\n" if $?;
a57a9493 769
8b7f5fc1 770 system("git-update-ref $remote/$branch $cid") == 0
a57a9493
MU
771 or die "Cannot write branch $branch for update: $!\n";
772
86d11cf2 773 if ($tag) {
86d11cf2 774 my ($xtag) = $tag;
0d821d4d
PA
775 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
776 $xtag =~ tr/_/\./ if ( $opt_u );
34c99da2 777 $xtag =~ s/[\/]/$opt_s/g;
a6080a0a 778
ee834cf0 779 system('git-tag', '-f', $xtag, $cid) == 0
0d821d4d 780 or die "Cannot create tag $xtag: $!\n";
0d821d4d
PA
781
782 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
a57a9493 783 }
a57a9493
MU
784};
785
06918348 786my $commitcount = 1;
86d11cf2 787while (<CVS>) {
a57a9493 788 chomp;
86d11cf2 789 if ($state == 0 and /^-+$/) {
a57a9493 790 $state = 1;
86d11cf2 791 } elsif ($state == 0) {
a57a9493
MU
792 $state = 1;
793 redo;
86d11cf2 794 } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
a57a9493
MU
795 $patchset = 0+$_;
796 $state=2;
86d11cf2 797 } elsif ($state == 2 and s/^Date:\s+//) {
a57a9493 798 $date = pdate($_);
86d11cf2 799 unless ($date) {
a57a9493
MU
800 print STDERR "Could not parse date: $_\n";
801 $state=0;
802 next;
803 }
804 $state=3;
86d11cf2 805 } elsif ($state == 3 and s/^Author:\s+//) {
a57a9493 806 s/\s+$//;
94c23343
JH
807 if (/^(.*?)\s+<(.*)>/) {
808 ($author_name, $author_email) = ($1, $2);
ffd97f3a
AE
809 } elsif ($conv_author_name{$_}) {
810 $author_name = $conv_author_name{$_};
811 $author_email = $conv_author_email{$_};
94c23343
JH
812 } else {
813 $author_name = $author_email = $_;
814 }
a57a9493 815 $state = 4;
86d11cf2 816 } elsif ($state == 4 and s/^Branch:\s+//) {
a57a9493 817 s/\s+$//;
a0554224 818 tr/_/\./ if ( $opt_u );
fbfd60d6 819 s/[\/]/$opt_s/g;
a57a9493
MU
820 $branch = $_;
821 $state = 5;
86d11cf2 822 } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
a57a9493
MU
823 s/\s+$//;
824 $ancestor = $_;
0fa2824f 825 $ancestor = $opt_o if $ancestor eq "HEAD";
a57a9493 826 $state = 6;
86d11cf2 827 } elsif ($state == 5) {
a57a9493
MU
828 $ancestor = undef;
829 $state = 6;
830 redo;
86d11cf2 831 } elsif ($state == 6 and s/^Tag:\s+//) {
a57a9493 832 s/\s+$//;
86d11cf2 833 if ($_ eq "(none)") {
a57a9493
MU
834 $tag = undef;
835 } else {
836 $tag = $_;
837 }
838 $state = 7;
86d11cf2 839 } elsif ($state == 7 and /^Log:/) {
a57a9493
MU
840 $logmsg = "";
841 $state = 8;
86d11cf2 842 } elsif ($state == 8 and /^Members:/) {
a57a9493 843 $branch = $opt_o if $branch eq "HEAD";
86d11cf2 844 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
a57a9493 845 # skip
9da07f34 846 print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
a57a9493
MU
847 $state = 11;
848 next;
849 }
ded9f400 850 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
6211988f 851 # skip if the commit is too recent
77190eb9 852 # given that the cvsps default fuzz is 300s, we give ourselves another
6211988f
ML
853 # 300s just in case -- this also prevents skipping commits
854 # due to server clock drift
855 print "skip patchset $patchset: $date too recent\n" if $opt_v;
856 $state = 11;
857 next;
858 }
71b08148
ML
859 if (exists $ignorebranch{$branch}) {
860 print STDERR "Skipping $branch\n";
861 $state = 11;
862 next;
863 }
86d11cf2
JH
864 if ($ancestor) {
865 if ($ancestor eq $branch) {
71b08148
ML
866 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
867 $ancestor = $opt_o;
868 }
0750d751 869 if (defined get_headref("$remote/$branch")) {
a57a9493
MU
870 print STDERR "Branch $branch already exists!\n";
871 $state=11;
872 next;
873 }
0750d751
JK
874 my $id = get_headref("$remote/$ancestor");
875 if (!$id) {
a57a9493 876 print STDERR "Branch $ancestor does not exist!\n";
71b08148 877 $ignorebranch{$branch} = 1;
a57a9493
MU
878 $state=11;
879 next;
880 }
0750d751
JK
881
882 system(qw(git update-ref -m cvsimport),
883 "$remote/$branch", $id);
884 if($? != 0) {
885 print STDERR "Could not create branch $branch\n";
71b08148 886 $ignorebranch{$branch} = 1;
a57a9493
MU
887 $state=11;
888 next;
889 }
a57a9493 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/$opt_o")
0750d751 1001 unless defined get_headref('refs/heads/master');
06baffd3
AW
1002 system("git-symbolic-ref", "$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}