git-cvsimport.perl

   1 #!/usr/bin/perl -w
   2
   3 # This tool is copyright (c) 2005, Matthias Urlichs.
   4 # It is released under the Gnu Public License, version 2.
   5 #
   6 # The basic idea is to aggregate CVS check-ins into related changes.
   7 # Fortunately, "cvsps" does that for us; all we have to do is to parse
   8 # its output.
   9 #
  10 # Checking out the files is done by a single long-running CVS connection
  11 # / server process.
  12 #
  13 # The head revision is on branch "origin" by default.
  14 # You can change that with the '-o' option.
  15
  16 use strict;
  17 use warnings;
  18 use Getopt::Std;
  19 use File::Spec;
  20 use File::Temp qw(tempfile 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;
  28
  29 $SIG{'PIPE'}="IGNORE";
  30 $ENV{'TZ'}="UTC";
  31
  32 our ($opt_h,$opt_o,$opt_v,$opt_k,$opt_u,$opt_d,$opt_p,$opt_C,$opt_z,$opt_i,$opt_P, $opt_s,$opt_m,$opt_M,$opt_A,$opt_S,$opt_L, $opt_a);
  33 my (%conv_author_name, %conv_author_email);
  34
  35 sub usage(;$) {
  36         my $msg = shift;
  37         print(STDERR "Error: $msg\n") if $msg;
  38         print STDERR <<END;
  39 Usage: ${\basename $0}     # fetch/update GIT from CVS
  40        [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
  41        [-p opts-for-cvsps] [-P file] [-C GIT_repository] [-z fuzz] [-i] [-k]
  42        [-u] [-s subst] [-a] [-m] [-M regex] [-S regex] [-L commitlimit]
  43        [CVS_module]
  44 END
  45         exit(1);
  46 }
  47
  48 sub read_author_info($) {
  49         my ($file) = @_;
  50         my $user;
  51         open my $f, '<', "$file" or die("Failed to open $file: $!\n");
  52
  53         while (<$f>) {
  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} = $3;
  60                 }
  61                 # However, we also read from CVSROOT/users format
  62                 # to ease migration.
  63                 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
  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} = $1;
  73                         }
  74                 }
  75                 # NEEDSWORK: Maybe warn on unrecognized lines?
  76         }
  77         close ($f);
  78 }
  79
  80 sub 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) {
  86                 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
  87         }
  88         close ($f);
  89 }
  90
  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 !~ /:$/);
 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
 117 my $opts = "haivmkuo:d:p:C:z:s:M:P:A:S:L:";
 118 read_repo_config($opts);
 119 getopts($opts) or usage();
 120 usage if $opt_h;
 121
 122 @ARGV <= 1 or usage("You can't specify more than one CVS module");
 123
 124 if ($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");
 136 }
 137 $opt_o ||= "origin";
 138 $opt_s ||= "-";
 139 $opt_a ||= 0;
 140
 141 my $git_tree = $opt_C;
 142 $git_tree ||= ".";
 143
 144 my $cvs_tree;
 145 if ($#ARGV == 0) {
 146         $cvs_tree = $ARGV[0];
 147 } elsif (-f 'CVS/Repository') {
 148         open my $f, '<', 'CVS/Repository' or
 149             die 'Failed to open CVS/Repository';
 150         $cvs_tree = <$f>;
 151         chomp $cvs_tree;
 152         close $f;
 153 } else {
 154         usage("CVS module has to be specified");
 155 }
 156
 157 our @mergerx = ();
 158 if ($opt_m) {
 159         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
 160 }
 161 if ($opt_M) {
 162         push (@mergerx, qr/$opt_M/);
 163 }
 164
 165 # Remember UTC of our starting time
 166 # we'll want to avoid importing commits
 167 # that are too recent
 168 our $starttime = time();
 169
 170 select(STDERR); $|=1; select(STDOUT);
 171
 172
 173 package CVSconn;
 174 # Basic CVS dialog.
 175 # We're only interested in connecting and downloading, so ...
 176
 177 use File::Spec;
 178 use File::Temp qw(tempfile);
 179 use POSIX qw(strftime dup2);
 180
 181 sub new {
 182         my ($what,$repo,$subdir) = @_;
 183         $what=ref($what) if ref($what);
 184
 185         my $self = {};
 186         $self->{'buffer'} = "";
 187         bless($self,$what);
 188
 189         $repo =~ s#/+$##;
 190         $self->{'fullrep'} = $repo;
 191         $self->conn();
 192
 193         $self->{'subdir'} = $subdir;
 194         $self->{'lines'} = undef;
 195
 196         return $self;
 197 }
 198
 199 sub conn {
 200         my $self = shift;
 201         my $repo = $self->{'fullrep'};
 202         if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
 203                 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
 204
 205                 my ($proxyhost,$proxyport);
 206                 if ($param && ($param =~ m/proxy=([^;]+)/)) {
 207                         $proxyhost = $1;
 208                         # Default proxyport, if not specified, is 8080.
 209                         $proxyport = 8080;
 210                         if ($ENV{"CVS_PROXY_PORT"}) {
 211                                 $proxyport = $ENV{"CVS_PROXY_PORT"};
 212                         }
 213                         if ($param =~ m/proxyport=([^;]+)/) {
 214                                 $proxyport = $1;
 215                         }
 216                 }
 217
 218                 $user="anonymous" unless defined $user;
 219                 my $rr2 = "-";
 220                 unless ($port) {
 221                         $rr2 = ":pserver:$user\@$serv:$repo";
 222                         $port=2401;
 223                 }
 224                 my $rr = ":pserver:$user\@$serv:$port$repo";
 225
 226                 unless ($pass) {
 227                         open(H,$ENV{'HOME'}."/.cvspass") and do {
 228                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
 229                                 while (<H>) {
 230                                         chomp;
 231                                         s/^\/\d+\s+//;
 232                                         my ($w,$p) = split(/\s/,$_,2);
 233                                         if ($w eq $rr or $w eq $rr2) {
 234                                                 $pass = $p;
 235                                                 last;
 236                                         }
 237                                 }
 238                         };
 239                 }
 240                 $pass="A" unless $pass;
 241
 242                 my ($s, $rep);
 243                 if ($proxyhost) {
 244
 245                         # Use a HTTP Proxy. Only works for HTTP proxies that
 246                         # don't require user authentication
 247                         #
 248                         # See: http://www.ietf.org/rfc/rfc2817.txt
 249
 250                         $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
 251                         die "Socket to $proxyhost: $!\n" unless defined $s;
 252                         $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
 253                                 or die "Write to $proxyhost: $!\n";
 254                         $s->flush();
 255
 256                         $rep = <$s>;
 257
 258                         # The answer should look like 'HTTP/1.x 2yy ....'
 259                         if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
 260                                 die "Proxy connect: $rep\n";
 261                         }
 262                         # Skip up to the empty line of the proxy server output
 263                         # including the response headers.
 264                         while ($rep = <$s>) {
 265                                 last if (!defined $rep ||
 266                                          $rep eq "\n" ||
 267                                          $rep eq "\r\n");
 268                         }
 269                 } else {
 270                         $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
 271                         die "Socket to $serv: $!\n" unless defined $s;
 272                 }
 273
 274                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
 275                         or die "Write to $serv: $!\n";
 276                 $s->flush();
 277
 278                 $rep = <$s>;
 279
 280                 if ($rep ne "I LOVE YOU\n") {
 281                         $rep="<unknown>" unless $rep;
 282                         die "AuthReply: $rep\n";
 283                 }
 284                 $self->{'socketo'} = $s;
 285                 $self->{'socketi'} = $s;
 286         } else { # local or ext: Fork off our own cvs server.
 287                 my $pr = IO::Pipe->new();
 288                 my $pw = IO::Pipe->new();
 289                 my $pid = fork();
 290                 die "Fork: $!\n" unless defined $pid;
 291                 my $cvs = 'cvs';
 292                 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
 293                 my $rsh = 'rsh';
 294                 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
 295
 296                 my @cvs = ($cvs, 'server');
 297                 my ($local, $user, $host);
 298                 $local = $repo =~ s/:local://;
 299                 if (!$local) {
 300                     $repo =~ s/:ext://;
 301                     $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
 302                     ($user, $host) = ($1, $2);
 303                 }
 304                 if (!$local) {
 305                     if ($user) {
 306                         unshift @cvs, $rsh, '-l', $user, $host;
 307                     } else {
 308                         unshift @cvs, $rsh, $host;
 309                     }
 310                 }
 311
 312                 unless ($pid) {
 313                         $pr->writer();
 314                         $pw->reader();
 315                         dup2($pw->fileno(),0);
 316                         dup2($pr->fileno(),1);
 317                         $pr->close();
 318                         $pw->close();
 319                         exec(@cvs);
 320                 }
 321                 $pw->writer();
 322                 $pr->reader();
 323                 $self->{'socketo'} = $pw;
 324                 $self->{'socketi'} = $pr;
 325         }
 326         $self->{'socketo'}->write("Root $repo\n");
 327
 328         # Trial and error says that this probably is the minimum set
 329         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
 330
 331         $self->{'socketo'}->write("valid-requests\n");
 332         $self->{'socketo'}->flush();
 333
 334         chomp(my $rep=$self->readline());
 335         if ($rep !~ s/^Valid-requests\s*//) {
 336                 $rep="<unknown>" unless $rep;
 337                 die "Expected Valid-requests from server, but got: $rep\n";
 338         }
 339         chomp(my $res=$self->readline());
 340         die "validReply: $res\n" if $res ne "ok";
 341
 342         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
 343         $self->{'repo'} = $repo;
 344 }
 345
 346 sub readline {
 347         my ($self) = @_;
 348         return $self->{'socketi'}->getline();
 349 }
 350
 351 sub _file {
 352         # Request a file with a given revision.
 353         # Trial and error says this is a good way to do it. :-/
 354         my ($self,$fn,$rev) = @_;
 355         $self->{'socketo'}->write("Argument -N\n") or return undef;
 356         $self->{'socketo'}->write("Argument -P\n") or return undef;
 357         # -kk: Linus' version doesn't use it - defaults to off
 358         if ($opt_k) {
 359             $self->{'socketo'}->write("Argument -kk\n") or return undef;
 360         }
 361         $self->{'socketo'}->write("Argument -r\n") or return undef;
 362         $self->{'socketo'}->write("Argument $rev\n") or return undef;
 363         $self->{'socketo'}->write("Argument --\n") or return undef;
 364         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
 365         $self->{'socketo'}->write("Directory .\n") or return undef;
 366         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
 367         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
 368         $self->{'socketo'}->write("co\n") or return undef;
 369         $self->{'socketo'}->flush() or return undef;
 370         $self->{'lines'} = 0;
 371         return 1;
 372 }
 373 sub _line {
 374         # Read a line from the server.
 375         # ... except that 'line' may be an entire file. ;-)
 376         my ($self, $fh) = @_;
 377         die "Not in lines" unless defined $self->{'lines'};
 378
 379         my $line;
 380         my $res=0;
 381         while (defined($line = $self->readline())) {
 382                 # M U gnupg-cvs-rep/AUTHORS
 383                 # Updated gnupg-cvs-rep/
 384                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
 385                 # /AUTHORS/1.1///T1.1
 386                 # u=rw,g=rw,o=rw
 387                 # 0
 388                 # ok
 389
 390                 if ($line =~ s/^(?:Created|Updated) //) {
 391                         $line = $self->readline(); # path
 392                         $line = $self->readline(); # Entries line
 393                         my $mode = $self->readline(); chomp $mode;
 394                         $self->{'mode'} = $mode;
 395                         defined (my $cnt = $self->readline())
 396                                 or die "EOF from server after 'Changed'\n";
 397                         chomp $cnt;
 398                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
 399                         $line="";
 400                         $res = $self->_fetchfile($fh, $cnt);
 401                 } elsif ($line =~ s/^ //) {
 402                         print $fh $line;
 403                         $res += length($line);
 404                 } elsif ($line =~ /^M\b/) {
 405                         # output, do nothing
 406                 } elsif ($line =~ /^Mbinary\b/) {
 407                         my $cnt;
 408                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
 409                         chomp $cnt;
 410                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
 411                         $line="";
 412                         $res += $self->_fetchfile($fh, $cnt);
 413                 } else {
 414                         chomp $line;
 415                         if ($line eq "ok") {
 416                                 # print STDERR "S: ok (".length($res).")\n";
 417                                 return $res;
 418                         } elsif ($line =~ s/^E //) {
 419                                 # print STDERR "S: $line\n";
 420                         } elsif ($line =~ /^(Remove-entry|Removed) /i) {
 421                                 $line = $self->readline(); # filename
 422                                 $line = $self->readline(); # OK
 423                                 chomp $line;
 424                                 die "Unknown: $line" if $line ne "ok";
 425                                 return -1;
 426                         } else {
 427                                 die "Unknown: $line\n";
 428                         }
 429                 }
 430         }
 431         return undef;
 432 }
 433 sub file {
 434         my ($self,$fn,$rev) = @_;
 435         my $res;
 436
 437         my ($fh, $name) = tempfile('gitcvs.XXXXXX',
 438                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 439
 440         $self->_file($fn,$rev) and $res = $self->_line($fh);
 441
 442         if (!defined $res) {
 443             print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
 444             truncate $fh, 0;
 445             $self->conn();
 446             $self->_file($fn,$rev) or die "No file command send";
 447             $res = $self->_line($fh);
 448             die "Retry failed" unless defined $res;
 449         }
 450         close ($fh);
 451
 452         return ($name, $res);
 453 }
 454 sub _fetchfile {
 455         my ($self, $fh, $cnt) = @_;
 456         my $res = 0;
 457         my $bufsize = 1024 * 1024;
 458         while ($cnt) {
 459             if ($bufsize > $cnt) {
 460                 $bufsize = $cnt;
 461             }
 462             my $buf;
 463             my $num = $self->{'socketi'}->read($buf,$bufsize);
 464             die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
 465             print $fh $buf;
 466             $res += $num;
 467             $cnt -= $num;
 468         }
 469         return $res;
 470 }
 471
 472
 473 package main;
 474
 475 my $cvs = CVSconn->new($opt_d, $cvs_tree);
 476
 477
 478 sub pdate($) {
 479         my ($d) = @_;
 480         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
 481                 or die "Unparseable date: $d\n";
 482         my $y=$1; $y-=1900 if $y>1900;
 483         return timegm($6||0,$5,$4,$3,$2-1,$y);
 484 }
 485
 486 sub pmode($) {
 487         my ($mode) = @_;
 488         my $m = 0;
 489         my $mm = 0;
 490         my $um = 0;
 491         for my $x(split(//,$mode)) {
 492                 if ($x eq ",") {
 493                         $m |= $mm&$um;
 494                         $mm = 0;
 495                         $um = 0;
 496                 } elsif ($x eq "u") { $um |= 0700;
 497                 } elsif ($x eq "g") { $um |= 0070;
 498                 } elsif ($x eq "o") { $um |= 0007;
 499                 } elsif ($x eq "r") { $mm |= 0444;
 500                 } elsif ($x eq "w") { $mm |= 0222;
 501                 } elsif ($x eq "x") { $mm |= 0111;
 502                 } elsif ($x eq "=") { # do nothing
 503                 } else { die "Unknown mode: $mode\n";
 504                 }
 505         }
 506         $m |= $mm&$um;
 507         return $m;
 508 }
 509
 510 sub getwd() {
 511         my $pwd = `pwd`;
 512         chomp $pwd;
 513         return $pwd;
 514 }
 515
 516 sub is_sha1 {
 517         my $s = shift;
 518         return $s =~ /^[a-f0-9]{40}$/;
 519 }
 520
 521 sub get_headref ($$) {
 522     my $name    = shift;
 523     my $git_dir = shift;
 524
 525     my $f = "$git_dir/refs/heads/$name";
 526     if (open(my $fh, $f)) {
 527             chomp(my $r = <$fh>);
 528             is_sha1($r) or die "Cannot get head id for $name ($r): $!";
 529             return $r;
 530     }
 531     die "unable to open $f: $!" unless $! == POSIX::ENOENT;
 532     return undef;
 533 }
 534
 535 -d $git_tree
 536         or mkdir($git_tree,0777)
 537         or die "Could not create $git_tree: $!";
 538 chdir($git_tree);
 539
 540 my $last_branch = "";
 541 my $orig_branch = "";
 542 my %branch_date;
 543 my $tip_at_start = undef;
 544
 545 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 546 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 547 $ENV{"GIT_DIR"} = $git_dir;
 548 my $orig_git_index;
 549 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 550
 551 my %index; # holds filenames of one index per branch
 552
 553 unless (-d $git_dir) {
 554         system("git-init");
 555         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 556         system("git-read-tree");
 557         die "Cannot init an empty tree: $?\n" if $?;
 558
 559         $last_branch = $opt_o;
 560         $orig_branch = "";
 561 } else {
 562         -f "$git_dir/refs/heads/$opt_o"
 563                 or die "Branch '$opt_o' does not exist.\n".
 564                        "Either use the correct '-o branch' option,\n".
 565                        "or import to a new repository.\n";
 566
 567         open(F, "git-symbolic-ref HEAD |") or
 568                 die "Cannot run git-symbolic-ref: $!\n";
 569         chomp ($last_branch = <F>);
 570         $last_branch = basename($last_branch);
 571         close(F);
 572         unless ($last_branch) {
 573                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 574                 $last_branch = "master";
 575         }
 576         $orig_branch = $last_branch;
 577         $tip_at_start = `git-rev-parse --verify HEAD`;
 578
 579         # Get the last import timestamps
 580         my $fmt = '($ref, $author) = (%(refname), %(author));';
 581         open(H, "git-for-each-ref --perl --format='$fmt' refs/heads |") or
 582                 die "Cannot run git-for-each-ref: $!\n";
 583         while (defined(my $entry = <H>)) {
 584                 my ($ref, $author);
 585                 eval($entry) || die "cannot eval refs list: $@";
 586                 my ($head) = ($ref =~ m|^refs/heads/(.*)|);
 587                 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
 588                 $branch_date{$head} = $1;
 589         }
 590         close(H);
 591 }
 592
 593 -d $git_dir
 594         or die "Could not create git subdir ($git_dir).\n";
 595
 596 # now we read (and possibly save) author-info as well
 597 -f "$git_dir/cvs-authors" and
 598   read_author_info("$git_dir/cvs-authors");
 599 if ($opt_A) {
 600         read_author_info($opt_A);
 601         write_author_info("$git_dir/cvs-authors");
 602 }
 603
 604
 605 #
 606 # run cvsps into a file unless we are getting
 607 # it passed as a file via $opt_P
 608 #
 609 my $cvspsfile;
 610 unless ($opt_P) {
 611         print "Running cvsps...\n" if $opt_v;
 612         my $pid = open(CVSPS,"-|");
 613         my $cvspsfh;
 614         die "Cannot fork: $!\n" unless defined $pid;
 615         unless ($pid) {
 616                 my @opt;
 617                 @opt = split(/,/,$opt_p) if defined $opt_p;
 618                 unshift @opt, '-z', $opt_z if defined $opt_z;
 619                 unshift @opt, '-q'         unless defined $opt_v;
 620                 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
 621                         push @opt, '--cvs-direct';
 622                 }
 623                 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
 624                 die "Could not start cvsps: $!\n";
 625         }
 626         ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
 627                                           DIR => File::Spec->tmpdir());
 628         while (<CVSPS>) {
 629             print $cvspsfh $_;
 630         }
 631         close CVSPS;
 632         close $cvspsfh;
 633 } else {
 634         $cvspsfile = $opt_P;
 635 }
 636
 637 open(CVS, "<$cvspsfile") or die $!;
 638
 639 ## cvsps output:
 640 #---------------------
 641 #PatchSet 314
 642 #Date: 1999/09/18 13:03:59
 643 #Author: wkoch
 644 #Branch: STABLE-BRANCH-1-0
 645 #Ancestor branch: HEAD
 646 #Tag: (none)
 647 #Log:
 648 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
 649 #Members:
 650 #       README:1.57->1.57.2.1
 651 #       VERSION:1.96->1.96.2.1
 652 #
 653 #---------------------
 654
 655 my $state = 0;
 656
 657 sub update_index (\@\@) {
 658         my $old = shift;
 659         my $new = shift;
 660         open(my $fh, '|-', qw(git-update-index -z --index-info))
 661                 or die "unable to open git-update-index: $!";
 662         print $fh
 663                 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
 664                         @$old),
 665                 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
 666                         @$new)
 667                 or die "unable to write to git-update-index: $!";
 668         close $fh
 669                 or die "unable to write to git-update-index: $!";
 670         $? and die "git-update-index reported error: $?";
 671 }
 672
 673 sub write_tree () {
 674         open(my $fh, '-|', qw(git-write-tree))
 675                 or die "unable to open git-write-tree: $!";
 676         chomp(my $tree = <$fh>);
 677         is_sha1($tree)
 678                 or die "Cannot get tree id ($tree): $!";
 679         close($fh)
 680                 or die "Error running git-write-tree: $?\n";
 681         print "Tree ID $tree\n" if $opt_v;
 682         return $tree;
 683 }
 684
 685 my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
 686 my (@old,@new,@skipped,%ignorebranch);
 687
 688 # commits that cvsps cannot place anywhere...
 689 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
 690
 691 sub commit {
 692         if ($branch eq $opt_o && !$index{branch} && !get_headref($branch, $git_dir)) {
 693             # looks like an initial commit
 694             # use the index primed by git-init
 695             $ENV{GIT_INDEX_FILE} = '.git/index';
 696             $index{$branch} = '.git/index';
 697         } else {
 698             # use an index per branch to speed up
 699             # imports of projects with many branches
 700             unless ($index{$branch}) {
 701                 $index{$branch} = tmpnam();
 702                 $ENV{GIT_INDEX_FILE} = $index{$branch};
 703                 if ($ancestor) {
 704                     system("git-read-tree", $ancestor);
 705                 } else {
 706                     system("git-read-tree", $branch);
 707                 }
 708                 die "read-tree failed: $?\n" if $?;
 709             }
 710         }
 711         $ENV{GIT_INDEX_FILE} = $index{$branch};
 712
 713         update_index(@old, @new);
 714         @old = @new = ();
 715         my $tree = write_tree();
 716         my $parent = get_headref($last_branch, $git_dir);
 717         print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
 718
 719         my @commit_args;
 720         push @commit_args, ("-p", $parent) if $parent;
 721
 722         # loose detection of merges
 723         # based on the commit msg
 724         foreach my $rx (@mergerx) {
 725                 next unless $logmsg =~ $rx && $1;
 726                 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
 727                 if (my $sha1 = get_headref($mparent, $git_dir)) {
 728                         push @commit_args, '-p', $mparent;
 729                         print "Merge parent branch: $mparent\n" if $opt_v;
 730                 }
 731         }
 732
 733         my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
 734         $ENV{GIT_AUTHOR_NAME} = $author_name;
 735         $ENV{GIT_AUTHOR_EMAIL} = $author_email;
 736         $ENV{GIT_AUTHOR_DATE} = $commit_date;
 737         $ENV{GIT_COMMITTER_NAME} = $author_name;
 738         $ENV{GIT_COMMITTER_EMAIL} = $author_email;
 739         $ENV{GIT_COMMITTER_DATE} = $commit_date;
 740         my $pid = open2(my $commit_read, my $commit_write,
 741                 'git-commit-tree', $tree, @commit_args);
 742
 743         # compatibility with git2cvs
 744         substr($logmsg,32767) = "" if length($logmsg) > 32767;
 745         $logmsg =~ s/[\s\n]+\z//;
 746
 747         if (@skipped) {
 748             $logmsg .= "\n\n\nSKIPPED:\n\t";
 749             $logmsg .= join("\n\t", @skipped) . "\n";
 750             @skipped = ();
 751         }
 752
 753         print($commit_write "$logmsg\n") && close($commit_write)
 754                 or die "Error writing to git-commit-tree: $!\n";
 755
 756         print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
 757         chomp(my $cid = <$commit_read>);
 758         is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
 759         print "Commit ID $cid\n" if $opt_v;
 760         close($commit_read);
 761
 762         waitpid($pid,0);
 763         die "Error running git-commit-tree: $?\n" if $?;
 764
 765         system("git-update-ref refs/heads/$branch $cid") == 0
 766                 or die "Cannot write branch $branch for update: $!\n";
 767
 768         if ($tag) {
 769                 my ($in, $out) = ('','');
 770                 my ($xtag) = $tag;
 771                 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
 772                 $xtag =~ tr/_/\./ if ( $opt_u );
 773                 $xtag =~ s/[\/]/$opt_s/g;
 774
 775                 my $pid = open2($in, $out, 'git-mktag');
 776                 print $out "object $cid\n".
 777                     "type commit\n".
 778                     "tag $xtag\n".
 779                     "tagger $author_name <$author_email>\n"
 780                     or die "Cannot create tag object $xtag: $!\n";
 781                 close($out)
 782                     or die "Cannot create tag object $xtag: $!\n";
 783
 784                 my $tagobj = <$in>;
 785                 chomp $tagobj;
 786
 787                 if ( !close($in) or waitpid($pid, 0) != $pid or
 788                      $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 789                     die "Cannot create tag object $xtag: $!\n";
 790                 }
 791
 792
 793                 open(C,">$git_dir/refs/tags/$xtag")
 794                         or die "Cannot create tag $xtag: $!\n";
 795                 print C "$tagobj\n"
 796                         or die "Cannot write tag $xtag: $!\n";
 797                 close(C)
 798                         or die "Cannot write tag $xtag: $!\n";
 799
 800                 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
 801         }
 802 };
 803
 804 my $commitcount = 1;
 805 while (<CVS>) {
 806         chomp;
 807         if ($state == 0 and /^-+$/) {
 808                 $state = 1;
 809         } elsif ($state == 0) {
 810                 $state = 1;
 811                 redo;
 812         } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
 813                 $patchset = 0+$_;
 814                 $state=2;
 815         } elsif ($state == 2 and s/^Date:\s+//) {
 816                 $date = pdate($_);
 817                 unless ($date) {
 818                         print STDERR "Could not parse date: $_\n";
 819                         $state=0;
 820                         next;
 821                 }
 822                 $state=3;
 823         } elsif ($state == 3 and s/^Author:\s+//) {
 824                 s/\s+$//;
 825                 if (/^(.*?)\s+<(.*)>/) {
 826                     ($author_name, $author_email) = ($1, $2);
 827                 } elsif ($conv_author_name{$_}) {
 828                         $author_name = $conv_author_name{$_};
 829                         $author_email = $conv_author_email{$_};
 830                 } else {
 831                     $author_name = $author_email = $_;
 832                 }
 833                 $state = 4;
 834         } elsif ($state == 4 and s/^Branch:\s+//) {
 835                 s/\s+$//;
 836                 s/[\/]/$opt_s/g;
 837                 $branch = $_;
 838                 $state = 5;
 839         } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
 840                 s/\s+$//;
 841                 $ancestor = $_;
 842                 $ancestor = $opt_o if $ancestor eq "HEAD";
 843                 $state = 6;
 844         } elsif ($state == 5) {
 845                 $ancestor = undef;
 846                 $state = 6;
 847                 redo;
 848         } elsif ($state == 6 and s/^Tag:\s+//) {
 849                 s/\s+$//;
 850                 if ($_ eq "(none)") {
 851                         $tag = undef;
 852                 } else {
 853                         $tag = $_;
 854                 }
 855                 $state = 7;
 856         } elsif ($state == 7 and /^Log:/) {
 857                 $logmsg = "";
 858                 $state = 8;
 859         } elsif ($state == 8 and /^Members:/) {
 860                 $branch = $opt_o if $branch eq "HEAD";
 861                 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
 862                         # skip
 863                         print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
 864                         $state = 11;
 865                         next;
 866                 }
 867                 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
 868                         # skip if the commit is too recent
 869                         # that the cvsps default fuzz is 300s, we give ourselves another
 870                         # 300s just in case -- this also prevents skipping commits
 871                         # due to server clock drift
 872                         print "skip patchset $patchset: $date too recent\n" if $opt_v;
 873                         $state = 11;
 874                         next;
 875                 }
 876                 if (exists $ignorebranch{$branch}) {
 877                         print STDERR "Skipping $branch\n";
 878                         $state = 11;
 879                         next;
 880                 }
 881                 if ($ancestor) {
 882                         if ($ancestor eq $branch) {
 883                                 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
 884                                 $ancestor = $opt_o;
 885                         }
 886                         if (-f "$git_dir/refs/heads/$branch") {
 887                                 print STDERR "Branch $branch already exists!\n";
 888                                 $state=11;
 889                                 next;
 890                         }
 891                         unless (open(H,"$git_dir/refs/heads/$ancestor")) {
 892                                 print STDERR "Branch $ancestor does not exist!\n";
 893                                 $ignorebranch{$branch} = 1;
 894                                 $state=11;
 895                                 next;
 896                         }
 897                         chomp(my $id = <H>);
 898                         close(H);
 899                         unless (open(H,"> $git_dir/refs/heads/$branch")) {
 900                                 print STDERR "Could not create branch $branch: $!\n";
 901                                 $ignorebranch{$branch} = 1;
 902                                 $state=11;
 903                                 next;
 904                         }
 905                         print H "$id\n"
 906                                 or die "Could not write branch $branch: $!";
 907                         close(H)
 908                                 or die "Could not write branch $branch: $!";
 909                 }
 910                 $last_branch = $branch if $branch ne $last_branch;
 911                 $state = 9;
 912         } elsif ($state == 8) {
 913                 $logmsg .= "$_\n";
 914         } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
 915 #       VERSION:1.96->1.96.2.1
 916                 my $init = ($2 eq "INITIAL");
 917                 my $fn = $1;
 918                 my $rev = $3;
 919                 $fn =~ s#^/+##;
 920                 if ($opt_S && $fn =~ m/$opt_S/) {
 921                     print "SKIPPING $fn v $rev\n";
 922                     push(@skipped, $fn);
 923                     next;
 924                 }
 925                 print "Fetching $fn   v $rev\n" if $opt_v;
 926                 my ($tmpname, $size) = $cvs->file($fn,$rev);
 927                 if ($size == -1) {
 928                         push(@old,$fn);
 929                         print "Drop $fn\n" if $opt_v;
 930                 } else {
 931                         print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
 932                         my $pid = open(my $F, '-|');
 933                         die $! unless defined $pid;
 934                         if (!$pid) {
 935                             exec("git-hash-object", "-w", $tmpname)
 936                                 or die "Cannot create object: $!\n";
 937                         }
 938                         my $sha = <$F>;
 939                         chomp $sha;
 940                         close $F;
 941                         my $mode = pmode($cvs->{'mode'});
 942                         push(@new,[$mode, $sha, $fn]); # may be resurrected!
 943                 }
 944                 unlink($tmpname);
 945         } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
 946                 my $fn = $1;
 947                 $fn =~ s#^/+##;
 948                 push(@old,$fn);
 949                 print "Delete $fn\n" if $opt_v;
 950         } elsif ($state == 9 and /^\s*$/) {
 951                 $state = 10;
 952         } elsif (($state == 9 or $state == 10) and /^-+$/) {
 953                 $commitcount++;
 954                 if ($opt_L && $commitcount > $opt_L) {
 955                         last;
 956                 }
 957                 commit();
 958                 if (($commitcount & 1023) == 0) {
 959                         system("git repack -a -d");
 960                 }
 961                 $state = 1;
 962         } elsif ($state == 11 and /^-+$/) {
 963                 $state = 1;
 964         } elsif (/^-+$/) { # end of unknown-line processing
 965                 $state = 1;
 966         } elsif ($state != 11) { # ignore stuff when skipping
 967                 print "* UNKNOWN LINE * $_\n";
 968         }
 969 }
 970 commit() if $branch and $state != 11;
 971
 972 unless ($opt_P) {
 973         unlink($cvspsfile);
 974 }
 975
 976 # The heuristic of repacking every 1024 commits can leave a
 977 # lot of unpacked data.  If there is more than 1MB worth of
 978 # not-packed objects, repack once more.
 979 my $line = `git-count-objects`;
 980 if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
 981   my ($n_objects, $kb) = ($1, $2);
 982   1024 < $kb
 983     and system("git repack -a -d");
 984 }
 985
 986 foreach my $git_index (values %index) {
 987     if ($git_index ne '.git/index') {
 988         unlink($git_index);
 989     }
 990 }
 991
 992 if (defined $orig_git_index) {
 993         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 994 } else {
 995         delete $ENV{GIT_INDEX_FILE};
 996 }
 997
 998 # Now switch back to the branch we were in before all of this happened
 999 if ($orig_branch) {
1000         print "DONE.\n" if $opt_v;
1001         if ($opt_i) {
1002                 exit 0;
1003         }
1004         my $tip_at_end = `git-rev-parse --verify HEAD`;
1005         if ($tip_at_start ne $tip_at_end) {
1006                 for ($tip_at_start, $tip_at_end) { chomp; }
1007                 print "Fetched into the current branch.\n" if $opt_v;
1008                 system(qw(git-read-tree -u -m),
1009                        $tip_at_start, $tip_at_end);
1010                 die "Fast-forward update failed: $?\n" if $?;
1011         }
1012         else {
1013                 system(qw(git-merge cvsimport HEAD), "refs/heads/$opt_o");
1014                 die "Could not merge $opt_o into the current branch.\n" if $?;
1015         }
1016 } else {
1017         $orig_branch = "master";
1018         print "DONE; creating $orig_branch branch\n" if $opt_v;
1019         system("git-update-ref", "refs/heads/master", "refs/heads/$opt_o")
1020                 unless -f "$git_dir/refs/heads/master";
1021         system('git-update-ref', 'HEAD', "$orig_branch");
1022         unless ($opt_i) {
1023                 system('git checkout');
1024                 die "checkout failed: $?\n" if $?;
1025         }
1026 }