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);
  33 my (%conv_author_name, %conv_author_email);
  34
  35 sub usage() {
  36         print STDERR <<END;
  37 Usage: ${\basename $0}     # fetch/update GIT from CVS
  38        [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
  39        [-p opts-for-cvsps] [-C GIT_repository] [-z fuzz] [-i] [-k] [-u]
  40        [-s subst] [-m] [-M regex] [-S regex] [CVS_module]
  41 END
  42         exit(1);
  43 }
  44
  45 sub read_author_info($) {
  46         my ($file) = @_;
  47         my $user;
  48         open my $f, '<', "$file" or die("Failed to open $file: $!\n");
  49
  50         while (<$f>) {
  51                 # Expected format is this:
  52                 #   exon=Andreas Ericsson <ae@op5.se>
  53                 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
  54                         $user = $1;
  55                         $conv_author_name{$user} = $2;
  56                         $conv_author_email{$user} = $3;
  57                 }
  58                 # However, we also read from CVSROOT/users format
  59                 # to ease migration.
  60                 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
  61                         my $mapped;
  62                         ($user, $mapped) = ($1, $3);
  63                         if ($mapped =~ /^\s*(.*?)\s*<(.*)>\s*$/) {
  64                                 $conv_author_name{$user} = $1;
  65                                 $conv_author_email{$user} = $2;
  66                         }
  67                         elsif ($mapped =~ /^<?(.*)>?$/) {
  68                                 $conv_author_name{$user} = $user;
  69                                 $conv_author_email{$user} = $1;
  70                         }
  71                 }
  72                 # NEEDSWORK: Maybe warn on unrecognized lines?
  73         }
  74         close ($f);
  75 }
  76
  77 sub write_author_info($) {
  78         my ($file) = @_;
  79         open my $f, '>', $file or
  80           die("Failed to open $file for writing: $!");
  81
  82         foreach (keys %conv_author_name) {
  83                 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
  84         }
  85         close ($f);
  86 }
  87
  88 getopts("hivmkuo:d:p:C:z:s:M:P:A:S:L:") or usage();
  89 usage if $opt_h;
  90
  91 @ARGV <= 1 or usage();
  92
  93 if ($opt_d) {
  94         $ENV{"CVSROOT"} = $opt_d;
  95 } elsif (-f 'CVS/Root') {
  96         open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
  97         $opt_d = <$f>;
  98         chomp $opt_d;
  99         close $f;
 100         $ENV{"CVSROOT"} = $opt_d;
 101 } elsif ($ENV{"CVSROOT"}) {
 102         $opt_d = $ENV{"CVSROOT"};
 103 } else {
 104         die "CVSROOT needs to be set";
 105 }
 106 $opt_o ||= "origin";
 107 $opt_s ||= "-";
 108 my $git_tree = $opt_C;
 109 $git_tree ||= ".";
 110
 111 my $cvs_tree;
 112 if ($#ARGV == 0) {
 113         $cvs_tree = $ARGV[0];
 114 } elsif (-f 'CVS/Repository') {
 115         open my $f, '<', 'CVS/Repository' or
 116             die 'Failed to open CVS/Repository';
 117         $cvs_tree = <$f>;
 118         chomp $cvs_tree;
 119         close $f;
 120 } else {
 121         usage();
 122 }
 123
 124 our @mergerx = ();
 125 if ($opt_m) {
 126         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
 127 }
 128 if ($opt_M) {
 129         push (@mergerx, qr/$opt_M/);
 130 }
 131
 132 select(STDERR); $|=1; select(STDOUT);
 133
 134
 135 package CVSconn;
 136 # Basic CVS dialog.
 137 # We're only interested in connecting and downloading, so ...
 138
 139 use File::Spec;
 140 use File::Temp qw(tempfile);
 141 use POSIX qw(strftime dup2);
 142
 143 sub new {
 144         my ($what,$repo,$subdir) = @_;
 145         $what=ref($what) if ref($what);
 146
 147         my $self = {};
 148         $self->{'buffer'} = "";
 149         bless($self,$what);
 150
 151         $repo =~ s#/+$##;
 152         $self->{'fullrep'} = $repo;
 153         $self->conn();
 154
 155         $self->{'subdir'} = $subdir;
 156         $self->{'lines'} = undef;
 157
 158         return $self;
 159 }
 160
 161 sub conn {
 162         my $self = shift;
 163         my $repo = $self->{'fullrep'};
 164         if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
 165                 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
 166
 167                 my ($proxyhost,$proxyport);
 168                 if ($param && ($param =~ m/proxy=([^;]+)/)) {
 169                         $proxyhost = $1;
 170                         # Default proxyport, if not specified, is 8080.
 171                         $proxyport = 8080;
 172                         if ($ENV{"CVS_PROXY_PORT"}) {
 173                                 $proxyport = $ENV{"CVS_PROXY_PORT"};
 174                         }
 175                         if ($param =~ m/proxyport=([^;]+)/) {
 176                                 $proxyport = $1;
 177                         }
 178                 }
 179
 180                 $user="anonymous" unless defined $user;
 181                 my $rr2 = "-";
 182                 unless ($port) {
 183                         $rr2 = ":pserver:$user\@$serv:$repo";
 184                         $port=2401;
 185                 }
 186                 my $rr = ":pserver:$user\@$serv:$port$repo";
 187
 188                 unless ($pass) {
 189                         open(H,$ENV{'HOME'}."/.cvspass") and do {
 190                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
 191                                 while (<H>) {
 192                                         chomp;
 193                                         s/^\/\d+\s+//;
 194                                         my ($w,$p) = split(/\s/,$_,2);
 195                                         if ($w eq $rr or $w eq $rr2) {
 196                                                 $pass = $p;
 197                                                 last;
 198                                         }
 199                                 }
 200                         };
 201                 }
 202                 $pass="A" unless $pass;
 203
 204                 my ($s, $rep);
 205                 if ($proxyhost) {
 206
 207                         # Use a HTTP Proxy. Only works for HTTP proxies that
 208                         # don't require user authentication
 209                         #
 210                         # See: http://www.ietf.org/rfc/rfc2817.txt
 211
 212                         $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
 213                         die "Socket to $proxyhost: $!\n" unless defined $s;
 214                         $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
 215                                 or die "Write to $proxyhost: $!\n";
 216                         $s->flush();
 217
 218                         $rep = <$s>;
 219
 220                         # The answer should look like 'HTTP/1.x 2yy ....'
 221                         if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
 222                                 die "Proxy connect: $rep\n";
 223                         }
 224                         # Skip up to the empty line of the proxy server output
 225                         # including the response headers.
 226                         while ($rep = <$s>) {
 227                                 last if (!defined $rep ||
 228                                          $rep eq "\n" ||
 229                                          $rep eq "\r\n");
 230                         }
 231                 } else {
 232                         $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
 233                         die "Socket to $serv: $!\n" unless defined $s;
 234                 }
 235
 236                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
 237                         or die "Write to $serv: $!\n";
 238                 $s->flush();
 239
 240                 $rep = <$s>;
 241
 242                 if ($rep ne "I LOVE YOU\n") {
 243                         $rep="<unknown>" unless $rep;
 244                         die "AuthReply: $rep\n";
 245                 }
 246                 $self->{'socketo'} = $s;
 247                 $self->{'socketi'} = $s;
 248         } else { # local or ext: Fork off our own cvs server.
 249                 my $pr = IO::Pipe->new();
 250                 my $pw = IO::Pipe->new();
 251                 my $pid = fork();
 252                 die "Fork: $!\n" unless defined $pid;
 253                 my $cvs = 'cvs';
 254                 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
 255                 my $rsh = 'rsh';
 256                 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
 257
 258                 my @cvs = ($cvs, 'server');
 259                 my ($local, $user, $host);
 260                 $local = $repo =~ s/:local://;
 261                 if (!$local) {
 262                     $repo =~ s/:ext://;
 263                     $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
 264                     ($user, $host) = ($1, $2);
 265                 }
 266                 if (!$local) {
 267                     if ($user) {
 268                         unshift @cvs, $rsh, '-l', $user, $host;
 269                     } else {
 270                         unshift @cvs, $rsh, $host;
 271                     }
 272                 }
 273
 274                 unless ($pid) {
 275                         $pr->writer();
 276                         $pw->reader();
 277                         dup2($pw->fileno(),0);
 278                         dup2($pr->fileno(),1);
 279                         $pr->close();
 280                         $pw->close();
 281                         exec(@cvs);
 282                 }
 283                 $pw->writer();
 284                 $pr->reader();
 285                 $self->{'socketo'} = $pw;
 286                 $self->{'socketi'} = $pr;
 287         }
 288         $self->{'socketo'}->write("Root $repo\n");
 289
 290         # Trial and error says that this probably is the minimum set
 291         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
 292
 293         $self->{'socketo'}->write("valid-requests\n");
 294         $self->{'socketo'}->flush();
 295
 296         chomp(my $rep=$self->readline());
 297         if ($rep !~ s/^Valid-requests\s*//) {
 298                 $rep="<unknown>" unless $rep;
 299                 die "Expected Valid-requests from server, but got: $rep\n";
 300         }
 301         chomp(my $res=$self->readline());
 302         die "validReply: $res\n" if $res ne "ok";
 303
 304         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
 305         $self->{'repo'} = $repo;
 306 }
 307
 308 sub readline {
 309         my ($self) = @_;
 310         return $self->{'socketi'}->getline();
 311 }
 312
 313 sub _file {
 314         # Request a file with a given revision.
 315         # Trial and error says this is a good way to do it. :-/
 316         my ($self,$fn,$rev) = @_;
 317         $self->{'socketo'}->write("Argument -N\n") or return undef;
 318         $self->{'socketo'}->write("Argument -P\n") or return undef;
 319         # -kk: Linus' version doesn't use it - defaults to off
 320         if ($opt_k) {
 321             $self->{'socketo'}->write("Argument -kk\n") or return undef;
 322         }
 323         $self->{'socketo'}->write("Argument -r\n") or return undef;
 324         $self->{'socketo'}->write("Argument $rev\n") or return undef;
 325         $self->{'socketo'}->write("Argument --\n") or return undef;
 326         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
 327         $self->{'socketo'}->write("Directory .\n") or return undef;
 328         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
 329         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
 330         $self->{'socketo'}->write("co\n") or return undef;
 331         $self->{'socketo'}->flush() or return undef;
 332         $self->{'lines'} = 0;
 333         return 1;
 334 }
 335 sub _line {
 336         # Read a line from the server.
 337         # ... except that 'line' may be an entire file. ;-)
 338         my ($self, $fh) = @_;
 339         die "Not in lines" unless defined $self->{'lines'};
 340
 341         my $line;
 342         my $res=0;
 343         while (defined($line = $self->readline())) {
 344                 # M U gnupg-cvs-rep/AUTHORS
 345                 # Updated gnupg-cvs-rep/
 346                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
 347                 # /AUTHORS/1.1///T1.1
 348                 # u=rw,g=rw,o=rw
 349                 # 0
 350                 # ok
 351
 352                 if ($line =~ s/^(?:Created|Updated) //) {
 353                         $line = $self->readline(); # path
 354                         $line = $self->readline(); # Entries line
 355                         my $mode = $self->readline(); chomp $mode;
 356                         $self->{'mode'} = $mode;
 357                         defined (my $cnt = $self->readline())
 358                                 or die "EOF from server after 'Changed'\n";
 359                         chomp $cnt;
 360                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
 361                         $line="";
 362                         $res = $self->_fetchfile($fh, $cnt);
 363                 } elsif ($line =~ s/^ //) {
 364                         print $fh $line;
 365                         $res += length($line);
 366                 } elsif ($line =~ /^M\b/) {
 367                         # output, do nothing
 368                 } elsif ($line =~ /^Mbinary\b/) {
 369                         my $cnt;
 370                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
 371                         chomp $cnt;
 372                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
 373                         $line="";
 374                         $res += $self->_fetchfile($fh, $cnt);
 375                 } else {
 376                         chomp $line;
 377                         if ($line eq "ok") {
 378                                 # print STDERR "S: ok (".length($res).")\n";
 379                                 return $res;
 380                         } elsif ($line =~ s/^E //) {
 381                                 # print STDERR "S: $line\n";
 382                         } elsif ($line =~ /^(Remove-entry|Removed) /i) {
 383                                 $line = $self->readline(); # filename
 384                                 $line = $self->readline(); # OK
 385                                 chomp $line;
 386                                 die "Unknown: $line" if $line ne "ok";
 387                                 return -1;
 388                         } else {
 389                                 die "Unknown: $line\n";
 390                         }
 391                 }
 392         }
 393         return undef;
 394 }
 395 sub file {
 396         my ($self,$fn,$rev) = @_;
 397         my $res;
 398
 399         my ($fh, $name) = tempfile('gitcvs.XXXXXX',
 400                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 401
 402         $self->_file($fn,$rev) and $res = $self->_line($fh);
 403
 404         if (!defined $res) {
 405             print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
 406             truncate $fh, 0;
 407             $self->conn();
 408             $self->_file($fn,$rev) or die "No file command send";
 409             $res = $self->_line($fh);
 410             die "Retry failed" unless defined $res;
 411         }
 412         close ($fh);
 413
 414         return ($name, $res);
 415 }
 416 sub _fetchfile {
 417         my ($self, $fh, $cnt) = @_;
 418         my $res = 0;
 419         my $bufsize = 1024 * 1024;
 420         while ($cnt) {
 421             if ($bufsize > $cnt) {
 422                 $bufsize = $cnt;
 423             }
 424             my $buf;
 425             my $num = $self->{'socketi'}->read($buf,$bufsize);
 426             die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
 427             print $fh $buf;
 428             $res += $num;
 429             $cnt -= $num;
 430         }
 431         return $res;
 432 }
 433
 434
 435 package main;
 436
 437 my $cvs = CVSconn->new($opt_d, $cvs_tree);
 438
 439
 440 sub pdate($) {
 441         my ($d) = @_;
 442         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
 443                 or die "Unparseable date: $d\n";
 444         my $y=$1; $y-=1900 if $y>1900;
 445         return timegm($6||0,$5,$4,$3,$2-1,$y);
 446 }
 447
 448 sub pmode($) {
 449         my ($mode) = @_;
 450         my $m = 0;
 451         my $mm = 0;
 452         my $um = 0;
 453         for my $x(split(//,$mode)) {
 454                 if ($x eq ",") {
 455                         $m |= $mm&$um;
 456                         $mm = 0;
 457                         $um = 0;
 458                 } elsif ($x eq "u") { $um |= 0700;
 459                 } elsif ($x eq "g") { $um |= 0070;
 460                 } elsif ($x eq "o") { $um |= 0007;
 461                 } elsif ($x eq "r") { $mm |= 0444;
 462                 } elsif ($x eq "w") { $mm |= 0222;
 463                 } elsif ($x eq "x") { $mm |= 0111;
 464                 } elsif ($x eq "=") { # do nothing
 465                 } else { die "Unknown mode: $mode\n";
 466                 }
 467         }
 468         $m |= $mm&$um;
 469         return $m;
 470 }
 471
 472 sub getwd() {
 473         my $pwd = `pwd`;
 474         chomp $pwd;
 475         return $pwd;
 476 }
 477
 478 sub is_sha1 {
 479         my $s = shift;
 480         return $s =~ /^[a-f0-9]{40}$/;
 481 }
 482
 483 sub get_headref ($$) {
 484     my $name    = shift;
 485     my $git_dir = shift;
 486
 487     my $f = "$git_dir/refs/heads/$name";
 488     if (open(my $fh, $f)) {
 489             chomp(my $r = <$fh>);
 490             is_sha1($r) or die "Cannot get head id for $name ($r): $!";
 491             return $r;
 492     }
 493     die "unable to open $f: $!" unless $! == POSIX::ENOENT;
 494     return undef;
 495 }
 496
 497 -d $git_tree
 498         or mkdir($git_tree,0777)
 499         or die "Could not create $git_tree: $!";
 500 chdir($git_tree);
 501
 502 my $last_branch = "";
 503 my $orig_branch = "";
 504 my %branch_date;
 505 my $tip_at_start = undef;
 506
 507 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 508 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 509 $ENV{"GIT_DIR"} = $git_dir;
 510 my $orig_git_index;
 511 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 512
 513 my %index; # holds filenames of one index per branch
 514
 515 unless (-d $git_dir) {
 516         system("git-init-db");
 517         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 518         system("git-read-tree");
 519         die "Cannot init an empty tree: $?\n" if $?;
 520
 521         $last_branch = $opt_o;
 522         $orig_branch = "";
 523 } else {
 524         -f "$git_dir/refs/heads/$opt_o"
 525                 or die "Branch '$opt_o' does not exist.\n".
 526                        "Either use the correct '-o branch' option,\n".
 527                        "or import to a new repository.\n";
 528
 529         open(F, "git-symbolic-ref HEAD |") or
 530                 die "Cannot run git-symbolic-ref: $!\n";
 531         chomp ($last_branch = <F>);
 532         $last_branch = basename($last_branch);
 533         close(F);
 534         unless ($last_branch) {
 535                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 536                 $last_branch = "master";
 537         }
 538         $orig_branch = $last_branch;
 539         $tip_at_start = `git-rev-parse --verify HEAD`;
 540
 541         # Get the last import timestamps
 542         my $fmt = '($ref, $author) = (%(refname), %(author));';
 543         open(H, "git-for-each-ref --perl --format='$fmt' refs/heads |") or
 544                 die "Cannot run git-for-each-ref: $!\n";
 545         while (defined(my $entry = <H>)) {
 546                 my ($ref, $author);
 547                 eval($entry) || die "cannot eval refs list: $@";
 548                 my ($head) = ($ref =~ m|^refs/heads/(.*)|);
 549                 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
 550                 $branch_date{$head} = $1;
 551         }
 552         close(H);
 553 }
 554
 555 -d $git_dir
 556         or die "Could not create git subdir ($git_dir).\n";
 557
 558 # now we read (and possibly save) author-info as well
 559 -f "$git_dir/cvs-authors" and
 560   read_author_info("$git_dir/cvs-authors");
 561 if ($opt_A) {
 562         read_author_info($opt_A);
 563         write_author_info("$git_dir/cvs-authors");
 564 }
 565
 566
 567 #
 568 # run cvsps into a file unless we are getting
 569 # it passed as a file via $opt_P
 570 #
 571 unless ($opt_P) {
 572         print "Running cvsps...\n" if $opt_v;
 573         my $pid = open(CVSPS,"-|");
 574         die "Cannot fork: $!\n" unless defined $pid;
 575         unless ($pid) {
 576                 my @opt;
 577                 @opt = split(/,/,$opt_p) if defined $opt_p;
 578                 unshift @opt, '-z', $opt_z if defined $opt_z;
 579                 unshift @opt, '-q'         unless defined $opt_v;
 580                 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
 581                         push @opt, '--cvs-direct';
 582                 }
 583                 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
 584                 die "Could not start cvsps: $!\n";
 585         }
 586         my ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
 587                                              DIR => File::Spec->tmpdir());
 588         while (<CVSPS>) {
 589             print $cvspsfh $_;
 590         }
 591         close CVSPS;
 592         close $cvspsfh;
 593         $opt_P = $cvspsfile;
 594 }
 595
 596
 597 open(CVS, "<$opt_P") or die $!;
 598
 599 ## cvsps output:
 600 #---------------------
 601 #PatchSet 314
 602 #Date: 1999/09/18 13:03:59
 603 #Author: wkoch
 604 #Branch: STABLE-BRANCH-1-0
 605 #Ancestor branch: HEAD
 606 #Tag: (none)
 607 #Log:
 608 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
 609 #Members:
 610 #       README:1.57->1.57.2.1
 611 #       VERSION:1.96->1.96.2.1
 612 #
 613 #---------------------
 614
 615 my $state = 0;
 616
 617 sub update_index (\@\@) {
 618         my $old = shift;
 619         my $new = shift;
 620         open(my $fh, '|-', qw(git-update-index -z --index-info))
 621                 or die "unable to open git-update-index: $!";
 622         print $fh
 623                 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
 624                         @$old),
 625                 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
 626                         @$new)
 627                 or die "unable to write to git-update-index: $!";
 628         close $fh
 629                 or die "unable to write to git-update-index: $!";
 630         $? and die "git-update-index reported error: $?";
 631 }
 632
 633 sub write_tree () {
 634         open(my $fh, '-|', qw(git-write-tree))
 635                 or die "unable to open git-write-tree: $!";
 636         chomp(my $tree = <$fh>);
 637         is_sha1($tree)
 638                 or die "Cannot get tree id ($tree): $!";
 639         close($fh)
 640                 or die "Error running git-write-tree: $?\n";
 641         print "Tree ID $tree\n" if $opt_v;
 642         return $tree;
 643 }
 644
 645 my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
 646 my (@old,@new,@skipped,%ignorebranch);
 647
 648 # commits that cvsps cannot place anywhere...
 649 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
 650
 651 sub commit {
 652         if ($branch eq $opt_o && !$index{branch} && !get_headref($branch, $git_dir)) {
 653             # looks like an initial commit
 654             # use the index primed by git-init-db
 655             $ENV{GIT_INDEX_FILE} = '.git/index';
 656             $index{$branch} = '.git/index';
 657         } else {
 658             # use an index per branch to speed up
 659             # imports of projects with many branches
 660             unless ($index{$branch}) {
 661                 $index{$branch} = tmpnam();
 662                 $ENV{GIT_INDEX_FILE} = $index{$branch};
 663                 if ($ancestor) {
 664                     system("git-read-tree", $ancestor);
 665                 } else {
 666                     system("git-read-tree", $branch);
 667                 }
 668                 die "read-tree failed: $?\n" if $?;
 669             }
 670         }
 671         $ENV{GIT_INDEX_FILE} = $index{$branch};
 672
 673         update_index(@old, @new);
 674         @old = @new = ();
 675         my $tree = write_tree();
 676         my $parent = get_headref($last_branch, $git_dir);
 677         print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
 678
 679         my @commit_args;
 680         push @commit_args, ("-p", $parent) if $parent;
 681
 682         # loose detection of merges
 683         # based on the commit msg
 684         foreach my $rx (@mergerx) {
 685                 next unless $logmsg =~ $rx && $1;
 686                 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
 687                 if (my $sha1 = get_headref($mparent, $git_dir)) {
 688                         push @commit_args, '-p', $mparent;
 689                         print "Merge parent branch: $mparent\n" if $opt_v;
 690                 }
 691         }
 692
 693         my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
 694         $ENV{GIT_AUTHOR_NAME} = $author_name;
 695         $ENV{GIT_AUTHOR_EMAIL} = $author_email;
 696         $ENV{GIT_AUTHOR_DATE} = $commit_date;
 697         $ENV{GIT_COMMITTER_NAME} = $author_name;
 698         $ENV{GIT_COMMITTER_EMAIL} = $author_email;
 699         $ENV{GIT_COMMITTER_DATE} = $commit_date;
 700         my $pid = open2(my $commit_read, my $commit_write,
 701                 'git-commit-tree', $tree, @commit_args);
 702
 703         # compatibility with git2cvs
 704         substr($logmsg,32767) = "" if length($logmsg) > 32767;
 705         $logmsg =~ s/[\s\n]+\z//;
 706
 707         if (@skipped) {
 708             $logmsg .= "\n\n\nSKIPPED:\n\t";
 709             $logmsg .= join("\n\t", @skipped) . "\n";
 710             @skipped = ();
 711         }
 712
 713         print($commit_write "$logmsg\n") && close($commit_write)
 714                 or die "Error writing to git-commit-tree: $!\n";
 715
 716         print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
 717         chomp(my $cid = <$commit_read>);
 718         is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
 719         print "Commit ID $cid\n" if $opt_v;
 720         close($commit_read);
 721
 722         waitpid($pid,0);
 723         die "Error running git-commit-tree: $?\n" if $?;
 724
 725         system("git-update-ref refs/heads/$branch $cid") == 0
 726                 or die "Cannot write branch $branch for update: $!\n";
 727
 728         if ($tag) {
 729                 my ($in, $out) = ('','');
 730                 my ($xtag) = $tag;
 731                 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
 732                 $xtag =~ tr/_/\./ if ( $opt_u );
 733                 $xtag =~ s/[\/]/$opt_s/g;
 734
 735                 my $pid = open2($in, $out, 'git-mktag');
 736                 print $out "object $cid\n".
 737                     "type commit\n".
 738                     "tag $xtag\n".
 739                     "tagger $author_name <$author_email>\n"
 740                     or die "Cannot create tag object $xtag: $!\n";
 741                 close($out)
 742                     or die "Cannot create tag object $xtag: $!\n";
 743
 744                 my $tagobj = <$in>;
 745                 chomp $tagobj;
 746
 747                 if ( !close($in) or waitpid($pid, 0) != $pid or
 748                      $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 749                     die "Cannot create tag object $xtag: $!\n";
 750                 }
 751
 752
 753                 open(C,">$git_dir/refs/tags/$xtag")
 754                         or die "Cannot create tag $xtag: $!\n";
 755                 print C "$tagobj\n"
 756                         or die "Cannot write tag $xtag: $!\n";
 757                 close(C)
 758                         or die "Cannot write tag $xtag: $!\n";
 759
 760                 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
 761         }
 762 };
 763
 764 my $commitcount = 1;
 765 while (<CVS>) {
 766         chomp;
 767         if ($state == 0 and /^-+$/) {
 768                 $state = 1;
 769         } elsif ($state == 0) {
 770                 $state = 1;
 771                 redo;
 772         } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
 773                 $patchset = 0+$_;
 774                 $state=2;
 775         } elsif ($state == 2 and s/^Date:\s+//) {
 776                 $date = pdate($_);
 777                 unless ($date) {
 778                         print STDERR "Could not parse date: $_\n";
 779                         $state=0;
 780                         next;
 781                 }
 782                 $state=3;
 783         } elsif ($state == 3 and s/^Author:\s+//) {
 784                 s/\s+$//;
 785                 if (/^(.*?)\s+<(.*)>/) {
 786                     ($author_name, $author_email) = ($1, $2);
 787                 } elsif ($conv_author_name{$_}) {
 788                         $author_name = $conv_author_name{$_};
 789                         $author_email = $conv_author_email{$_};
 790                 } else {
 791                     $author_name = $author_email = $_;
 792                 }
 793                 $state = 4;
 794         } elsif ($state == 4 and s/^Branch:\s+//) {
 795                 s/\s+$//;
 796                 s/[\/]/$opt_s/g;
 797                 $branch = $_;
 798                 $state = 5;
 799         } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
 800                 s/\s+$//;
 801                 $ancestor = $_;
 802                 $ancestor = $opt_o if $ancestor eq "HEAD";
 803                 $state = 6;
 804         } elsif ($state == 5) {
 805                 $ancestor = undef;
 806                 $state = 6;
 807                 redo;
 808         } elsif ($state == 6 and s/^Tag:\s+//) {
 809                 s/\s+$//;
 810                 if ($_ eq "(none)") {
 811                         $tag = undef;
 812                 } else {
 813                         $tag = $_;
 814                 }
 815                 $state = 7;
 816         } elsif ($state == 7 and /^Log:/) {
 817                 $logmsg = "";
 818                 $state = 8;
 819         } elsif ($state == 8 and /^Members:/) {
 820                 $branch = $opt_o if $branch eq "HEAD";
 821                 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
 822                         # skip
 823                         print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
 824                         $state = 11;
 825                         next;
 826                 }
 827                 if (exists $ignorebranch{$branch}) {
 828                         print STDERR "Skipping $branch\n";
 829                         $state = 11;
 830                         next;
 831                 }
 832                 if ($ancestor) {
 833                         if ($ancestor eq $branch) {
 834                                 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
 835                                 $ancestor = $opt_o;
 836                         }
 837                         if (-f "$git_dir/refs/heads/$branch") {
 838                                 print STDERR "Branch $branch already exists!\n";
 839                                 $state=11;
 840                                 next;
 841                         }
 842                         unless (open(H,"$git_dir/refs/heads/$ancestor")) {
 843                                 print STDERR "Branch $ancestor does not exist!\n";
 844                                 $ignorebranch{$branch} = 1;
 845                                 $state=11;
 846                                 next;
 847                         }
 848                         chomp(my $id = <H>);
 849                         close(H);
 850                         unless (open(H,"> $git_dir/refs/heads/$branch")) {
 851                                 print STDERR "Could not create branch $branch: $!\n";
 852                                 $ignorebranch{$branch} = 1;
 853                                 $state=11;
 854                                 next;
 855                         }
 856                         print H "$id\n"
 857                                 or die "Could not write branch $branch: $!";
 858                         close(H)
 859                                 or die "Could not write branch $branch: $!";
 860                 }
 861                 $last_branch = $branch if $branch ne $last_branch;
 862                 $state = 9;
 863         } elsif ($state == 8) {
 864                 $logmsg .= "$_\n";
 865         } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
 866 #       VERSION:1.96->1.96.2.1
 867                 my $init = ($2 eq "INITIAL");
 868                 my $fn = $1;
 869                 my $rev = $3;
 870                 $fn =~ s#^/+##;
 871                 if ($opt_S && $fn =~ m/$opt_S/) {
 872                     print "SKIPPING $fn v $rev\n";
 873                     push(@skipped, $fn);
 874                     next;
 875                 }
 876                 print "Fetching $fn   v $rev\n" if $opt_v;
 877                 my ($tmpname, $size) = $cvs->file($fn,$rev);
 878                 if ($size == -1) {
 879                         push(@old,$fn);
 880                         print "Drop $fn\n" if $opt_v;
 881                 } else {
 882                         print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
 883                         my $pid = open(my $F, '-|');
 884                         die $! unless defined $pid;
 885                         if (!$pid) {
 886                             exec("git-hash-object", "-w", $tmpname)
 887                                 or die "Cannot create object: $!\n";
 888                         }
 889                         my $sha = <$F>;
 890                         chomp $sha;
 891                         close $F;
 892                         my $mode = pmode($cvs->{'mode'});
 893                         push(@new,[$mode, $sha, $fn]); # may be resurrected!
 894                 }
 895                 unlink($tmpname);
 896         } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
 897                 my $fn = $1;
 898                 $fn =~ s#^/+##;
 899                 push(@old,$fn);
 900                 print "Delete $fn\n" if $opt_v;
 901         } elsif ($state == 9 and /^\s*$/) {
 902                 $state = 10;
 903         } elsif (($state == 9 or $state == 10) and /^-+$/) {
 904                 $commitcount++;
 905                 if ($opt_L && $commitcount > $opt_L) {
 906                         last;
 907                 }
 908                 commit();
 909                 if (($commitcount & 1023) == 0) {
 910                         system("git repack -a -d");
 911                 }
 912                 $state = 1;
 913         } elsif ($state == 11 and /^-+$/) {
 914                 $state = 1;
 915         } elsif (/^-+$/) { # end of unknown-line processing
 916                 $state = 1;
 917         } elsif ($state != 11) { # ignore stuff when skipping
 918                 print "* UNKNOWN LINE * $_\n";
 919         }
 920 }
 921 commit() if $branch and $state != 11;
 922
 923 # The heuristic of repacking every 1024 commits can leave a
 924 # lot of unpacked data.  If there is more than 1MB worth of
 925 # not-packed objects, repack once more.
 926 my $line = `git-count-objects`;
 927 if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
 928   my ($n_objects, $kb) = ($1, $2);
 929   1024 < $kb
 930     and system("git repack -a -d");
 931 }
 932
 933 foreach my $git_index (values %index) {
 934     if ($git_index ne '.git/index') {
 935         unlink($git_index);
 936     }
 937 }
 938
 939 if (defined $orig_git_index) {
 940         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 941 } else {
 942         delete $ENV{GIT_INDEX_FILE};
 943 }
 944
 945 # Now switch back to the branch we were in before all of this happened
 946 if ($orig_branch) {
 947         print "DONE.\n" if $opt_v;
 948         if ($opt_i) {
 949                 exit 0;
 950         }
 951         my $tip_at_end = `git-rev-parse --verify HEAD`;
 952         if ($tip_at_start ne $tip_at_end) {
 953                 for ($tip_at_start, $tip_at_end) { chomp; }
 954                 print "Fetched into the current branch.\n" if $opt_v;
 955                 system(qw(git-read-tree -u -m),
 956                        $tip_at_start, $tip_at_end);
 957                 die "Fast-forward update failed: $?\n" if $?;
 958         }
 959         else {
 960                 system(qw(git-merge cvsimport HEAD), "refs/heads/$opt_o");
 961                 die "Could not merge $opt_o into the current branch.\n" if $?;
 962         }
 963 } else {
 964         $orig_branch = "master";
 965         print "DONE; creating $orig_branch branch\n" if $opt_v;
 966         system("git-update-ref", "refs/heads/master", "refs/heads/$opt_o")
 967                 unless -f "$git_dir/refs/heads/master";
 968         system('git-update-ref', 'HEAD', "$orig_branch");
 969         unless ($opt_i) {
 970                 system('git checkout');
 971                 die "checkout failed: $?\n" if $?;
 972         }
 973 }