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         print STDERR <<END;
  37 Usage: ${\basename $0}     # fetch/update GIT from CVS
  38        [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
  39        [-p opts-for-cvsps] [-P file] [-C GIT_repository] [-z fuzz] [-i] [-k]
  40        [-u] [-s subst] [-a] [-m] [-M regex] [-S regex] [-L commitlimit]
  41        [CVS_module]
  42 END
  43         exit(1);
  44 }
  45
  46 sub read_author_info($) {
  47         my ($file) = @_;
  48         my $user;
  49         open my $f, '<', "$file" or die("Failed to open $file: $!\n");
  50
  51         while (<$f>) {
  52                 # Expected format is this:
  53                 #   exon=Andreas Ericsson <ae@op5.se>
  54                 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
  55                         $user = $1;
  56                         $conv_author_name{$user} = $2;
  57                         $conv_author_email{$user} = $3;
  58                 }
  59                 # However, we also read from CVSROOT/users format
  60                 # to ease migration.
  61                 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
  62                         my $mapped;
  63                         ($user, $mapped) = ($1, $3);
  64                         if ($mapped =~ /^\s*(.*?)\s*<(.*)>\s*$/) {
  65                                 $conv_author_name{$user} = $1;
  66                                 $conv_author_email{$user} = $2;
  67                         }
  68                         elsif ($mapped =~ /^<?(.*)>?$/) {
  69                                 $conv_author_name{$user} = $user;
  70                                 $conv_author_email{$user} = $1;
  71                         }
  72                 }
  73                 # NEEDSWORK: Maybe warn on unrecognized lines?
  74         }
  75         close ($f);
  76 }
  77
  78 sub write_author_info($) {
  79         my ($file) = @_;
  80         open my $f, '>', $file or
  81           die("Failed to open $file for writing: $!");
  82
  83         foreach (keys %conv_author_name) {
  84                 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
  85         }
  86         close ($f);
  87 }
  88
  89 # convert getopts specs for use by git-repo-config
  90 sub read_repo_config {
  91     # Split the string between characters, unless there is a ':'
  92     # So "abc:de" becomes ["a", "b", "c:", "d", "e"]
  93         my @opts = split(/ *(?!:)/, shift);
  94         foreach my $o (@opts) {
  95                 my $key = $o;
  96                 $key =~ s/://g;
  97                 my $arg = 'git-repo-config';
  98                 $arg .= ' --bool' if ($o !~ /:$/);
  99
 100         chomp(my $tmp = `$arg --get cvsimport.$key`);
 101                 if ($tmp && !($arg =~ /--bool/ && $tmp eq 'false')) {
 102             no strict 'refs';
 103             my $opt_name = "opt_" . $key;
 104             if (!$$opt_name) {
 105                 $$opt_name = $tmp;
 106             }
 107                 }
 108         }
 109     if (@ARGV == 0) {
 110         chomp(my $module = `git-repo-config --get cvsimport.module`);
 111         push(@ARGV, $module);
 112     }
 113 }
 114
 115 my $opts = "haivmkuo:d:p:C:z:s:M:P:A:S:L:";
 116 read_repo_config($opts);
 117 getopts($opts) or usage();
 118 usage if $opt_h;
 119
 120 @ARGV <= 1 or usage();
 121
 122 if ($opt_d) {
 123         $ENV{"CVSROOT"} = $opt_d;
 124 } elsif (-f 'CVS/Root') {
 125         open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
 126         $opt_d = <$f>;
 127         chomp $opt_d;
 128         close $f;
 129         $ENV{"CVSROOT"} = $opt_d;
 130 } elsif ($ENV{"CVSROOT"}) {
 131         $opt_d = $ENV{"CVSROOT"};
 132 } else {
 133         die "CVSROOT needs to be set";
 134 }
 135 $opt_o ||= "origin";
 136 $opt_s ||= "-";
 137 $opt_a ||= 0;
 138
 139 my $git_tree = $opt_C;
 140 $git_tree ||= ".";
 141
 142 my $cvs_tree;
 143 if ($#ARGV == 0) {
 144         $cvs_tree = $ARGV[0];
 145 } elsif (-f 'CVS/Repository') {
 146         open my $f, '<', 'CVS/Repository' or
 147             die 'Failed to open CVS/Repository';
 148         $cvs_tree = <$f>;
 149         chomp $cvs_tree;
 150         close $f;
 151 } else {
 152         usage();
 153 }
 154
 155 our @mergerx = ();
 156 if ($opt_m) {
 157         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
 158 }
 159 if ($opt_M) {
 160         push (@mergerx, qr/$opt_M/);
 161 }
 162
 163 # Remember UTC of our starting time
 164 # we'll want to avoid importing commits
 165 # that are too recent
 166 our $starttime = time();
 167
 168 select(STDERR); $|=1; select(STDOUT);
 169
 170
 171 package CVSconn;
 172 # Basic CVS dialog.
 173 # We're only interested in connecting and downloading, so ...
 174
 175 use File::Spec;
 176 use File::Temp qw(tempfile);
 177 use POSIX qw(strftime dup2);
 178
 179 sub new {
 180         my ($what,$repo,$subdir) = @_;
 181         $what=ref($what) if ref($what);
 182
 183         my $self = {};
 184         $self->{'buffer'} = "";
 185         bless($self,$what);
 186
 187         $repo =~ s#/+$##;
 188         $self->{'fullrep'} = $repo;
 189         $self->conn();
 190
 191         $self->{'subdir'} = $subdir;
 192         $self->{'lines'} = undef;
 193
 194         return $self;
 195 }
 196
 197 sub conn {
 198         my $self = shift;
 199         my $repo = $self->{'fullrep'};
 200         if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
 201                 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
 202
 203                 my ($proxyhost,$proxyport);
 204                 if ($param && ($param =~ m/proxy=([^;]+)/)) {
 205                         $proxyhost = $1;
 206                         # Default proxyport, if not specified, is 8080.
 207                         $proxyport = 8080;
 208                         if ($ENV{"CVS_PROXY_PORT"}) {
 209                                 $proxyport = $ENV{"CVS_PROXY_PORT"};
 210                         }
 211                         if ($param =~ m/proxyport=([^;]+)/) {
 212                                 $proxyport = $1;
 213                         }
 214                 }
 215
 216                 $user="anonymous" unless defined $user;
 217                 my $rr2 = "-";
 218                 unless ($port) {
 219                         $rr2 = ":pserver:$user\@$serv:$repo";
 220                         $port=2401;
 221                 }
 222                 my $rr = ":pserver:$user\@$serv:$port$repo";
 223
 224                 unless ($pass) {
 225                         open(H,$ENV{'HOME'}."/.cvspass") and do {
 226                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
 227                                 while (<H>) {
 228                                         chomp;
 229                                         s/^\/\d+\s+//;
 230                                         my ($w,$p) = split(/\s/,$_,2);
 231                                         if ($w eq $rr or $w eq $rr2) {
 232                                                 $pass = $p;
 233                                                 last;
 234                                         }
 235                                 }
 236                         };
 237                 }
 238                 $pass="A" unless $pass;
 239
 240                 my ($s, $rep);
 241                 if ($proxyhost) {
 242
 243                         # Use a HTTP Proxy. Only works for HTTP proxies that
 244                         # don't require user authentication
 245                         #
 246                         # See: http://www.ietf.org/rfc/rfc2817.txt
 247
 248                         $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
 249                         die "Socket to $proxyhost: $!\n" unless defined $s;
 250                         $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
 251                                 or die "Write to $proxyhost: $!\n";
 252                         $s->flush();
 253
 254                         $rep = <$s>;
 255
 256                         # The answer should look like 'HTTP/1.x 2yy ....'
 257                         if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
 258                                 die "Proxy connect: $rep\n";
 259                         }
 260                         # Skip up to the empty line of the proxy server output
 261                         # including the response headers.
 262                         while ($rep = <$s>) {
 263                                 last if (!defined $rep ||
 264                                          $rep eq "\n" ||
 265                                          $rep eq "\r\n");
 266                         }
 267                 } else {
 268                         $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
 269                         die "Socket to $serv: $!\n" unless defined $s;
 270                 }
 271
 272                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
 273                         or die "Write to $serv: $!\n";
 274                 $s->flush();
 275
 276                 $rep = <$s>;
 277
 278                 if ($rep ne "I LOVE YOU\n") {
 279                         $rep="<unknown>" unless $rep;
 280                         die "AuthReply: $rep\n";
 281                 }
 282                 $self->{'socketo'} = $s;
 283                 $self->{'socketi'} = $s;
 284         } else { # local or ext: Fork off our own cvs server.
 285                 my $pr = IO::Pipe->new();
 286                 my $pw = IO::Pipe->new();
 287                 my $pid = fork();
 288                 die "Fork: $!\n" unless defined $pid;
 289                 my $cvs = 'cvs';
 290                 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
 291                 my $rsh = 'rsh';
 292                 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
 293
 294                 my @cvs = ($cvs, 'server');
 295                 my ($local, $user, $host);
 296                 $local = $repo =~ s/:local://;
 297                 if (!$local) {
 298                     $repo =~ s/:ext://;
 299                     $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
 300                     ($user, $host) = ($1, $2);
 301                 }
 302                 if (!$local) {
 303                     if ($user) {
 304                         unshift @cvs, $rsh, '-l', $user, $host;
 305                     } else {
 306                         unshift @cvs, $rsh, $host;
 307                     }
 308                 }
 309
 310                 unless ($pid) {
 311                         $pr->writer();
 312                         $pw->reader();
 313                         dup2($pw->fileno(),0);
 314                         dup2($pr->fileno(),1);
 315                         $pr->close();
 316                         $pw->close();
 317                         exec(@cvs);
 318                 }
 319                 $pw->writer();
 320                 $pr->reader();
 321                 $self->{'socketo'} = $pw;
 322                 $self->{'socketi'} = $pr;
 323         }
 324         $self->{'socketo'}->write("Root $repo\n");
 325
 326         # Trial and error says that this probably is the minimum set
 327         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
 328
 329         $self->{'socketo'}->write("valid-requests\n");
 330         $self->{'socketo'}->flush();
 331
 332         chomp(my $rep=$self->readline());
 333         if ($rep !~ s/^Valid-requests\s*//) {
 334                 $rep="<unknown>" unless $rep;
 335                 die "Expected Valid-requests from server, but got: $rep\n";
 336         }
 337         chomp(my $res=$self->readline());
 338         die "validReply: $res\n" if $res ne "ok";
 339
 340         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
 341         $self->{'repo'} = $repo;
 342 }
 343
 344 sub readline {
 345         my ($self) = @_;
 346         return $self->{'socketi'}->getline();
 347 }
 348
 349 sub _file {
 350         # Request a file with a given revision.
 351         # Trial and error says this is a good way to do it. :-/
 352         my ($self,$fn,$rev) = @_;
 353         $self->{'socketo'}->write("Argument -N\n") or return undef;
 354         $self->{'socketo'}->write("Argument -P\n") or return undef;
 355         # -kk: Linus' version doesn't use it - defaults to off
 356         if ($opt_k) {
 357             $self->{'socketo'}->write("Argument -kk\n") or return undef;
 358         }
 359         $self->{'socketo'}->write("Argument -r\n") or return undef;
 360         $self->{'socketo'}->write("Argument $rev\n") or return undef;
 361         $self->{'socketo'}->write("Argument --\n") or return undef;
 362         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
 363         $self->{'socketo'}->write("Directory .\n") or return undef;
 364         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
 365         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
 366         $self->{'socketo'}->write("co\n") or return undef;
 367         $self->{'socketo'}->flush() or return undef;
 368         $self->{'lines'} = 0;
 369         return 1;
 370 }
 371 sub _line {
 372         # Read a line from the server.
 373         # ... except that 'line' may be an entire file. ;-)
 374         my ($self, $fh) = @_;
 375         die "Not in lines" unless defined $self->{'lines'};
 376
 377         my $line;
 378         my $res=0;
 379         while (defined($line = $self->readline())) {
 380                 # M U gnupg-cvs-rep/AUTHORS
 381                 # Updated gnupg-cvs-rep/
 382                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
 383                 # /AUTHORS/1.1///T1.1
 384                 # u=rw,g=rw,o=rw
 385                 # 0
 386                 # ok
 387
 388                 if ($line =~ s/^(?:Created|Updated) //) {
 389                         $line = $self->readline(); # path
 390                         $line = $self->readline(); # Entries line
 391                         my $mode = $self->readline(); chomp $mode;
 392                         $self->{'mode'} = $mode;
 393                         defined (my $cnt = $self->readline())
 394                                 or die "EOF from server after 'Changed'\n";
 395                         chomp $cnt;
 396                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
 397                         $line="";
 398                         $res = $self->_fetchfile($fh, $cnt);
 399                 } elsif ($line =~ s/^ //) {
 400                         print $fh $line;
 401                         $res += length($line);
 402                 } elsif ($line =~ /^M\b/) {
 403                         # output, do nothing
 404                 } elsif ($line =~ /^Mbinary\b/) {
 405                         my $cnt;
 406                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
 407                         chomp $cnt;
 408                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
 409                         $line="";
 410                         $res += $self->_fetchfile($fh, $cnt);
 411                 } else {
 412                         chomp $line;
 413                         if ($line eq "ok") {
 414                                 # print STDERR "S: ok (".length($res).")\n";
 415                                 return $res;
 416                         } elsif ($line =~ s/^E //) {
 417                                 # print STDERR "S: $line\n";
 418                         } elsif ($line =~ /^(Remove-entry|Removed) /i) {
 419                                 $line = $self->readline(); # filename
 420                                 $line = $self->readline(); # OK
 421                                 chomp $line;
 422                                 die "Unknown: $line" if $line ne "ok";
 423                                 return -1;
 424                         } else {
 425                                 die "Unknown: $line\n";
 426                         }
 427                 }
 428         }
 429         return undef;
 430 }
 431 sub file {
 432         my ($self,$fn,$rev) = @_;
 433         my $res;
 434
 435         my ($fh, $name) = tempfile('gitcvs.XXXXXX',
 436                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 437
 438         $self->_file($fn,$rev) and $res = $self->_line($fh);
 439
 440         if (!defined $res) {
 441             print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
 442             truncate $fh, 0;
 443             $self->conn();
 444             $self->_file($fn,$rev) or die "No file command send";
 445             $res = $self->_line($fh);
 446             die "Retry failed" unless defined $res;
 447         }
 448         close ($fh);
 449
 450         return ($name, $res);
 451 }
 452 sub _fetchfile {
 453         my ($self, $fh, $cnt) = @_;
 454         my $res = 0;
 455         my $bufsize = 1024 * 1024;
 456         while ($cnt) {
 457             if ($bufsize > $cnt) {
 458                 $bufsize = $cnt;
 459             }
 460             my $buf;
 461             my $num = $self->{'socketi'}->read($buf,$bufsize);
 462             die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
 463             print $fh $buf;
 464             $res += $num;
 465             $cnt -= $num;
 466         }
 467         return $res;
 468 }
 469
 470
 471 package main;
 472
 473 my $cvs = CVSconn->new($opt_d, $cvs_tree);
 474
 475
 476 sub pdate($) {
 477         my ($d) = @_;
 478         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
 479                 or die "Unparseable date: $d\n";
 480         my $y=$1; $y-=1900 if $y>1900;
 481         return timegm($6||0,$5,$4,$3,$2-1,$y);
 482 }
 483
 484 sub pmode($) {
 485         my ($mode) = @_;
 486         my $m = 0;
 487         my $mm = 0;
 488         my $um = 0;
 489         for my $x(split(//,$mode)) {
 490                 if ($x eq ",") {
 491                         $m |= $mm&$um;
 492                         $mm = 0;
 493                         $um = 0;
 494                 } elsif ($x eq "u") { $um |= 0700;
 495                 } elsif ($x eq "g") { $um |= 0070;
 496                 } elsif ($x eq "o") { $um |= 0007;
 497                 } elsif ($x eq "r") { $mm |= 0444;
 498                 } elsif ($x eq "w") { $mm |= 0222;
 499                 } elsif ($x eq "x") { $mm |= 0111;
 500                 } elsif ($x eq "=") { # do nothing
 501                 } else { die "Unknown mode: $mode\n";
 502                 }
 503         }
 504         $m |= $mm&$um;
 505         return $m;
 506 }
 507
 508 sub getwd() {
 509         my $pwd = `pwd`;
 510         chomp $pwd;
 511         return $pwd;
 512 }
 513
 514 sub is_sha1 {
 515         my $s = shift;
 516         return $s =~ /^[a-f0-9]{40}$/;
 517 }
 518
 519 sub get_headref ($$) {
 520     my $name    = shift;
 521     my $git_dir = shift;
 522
 523     my $f = "$git_dir/refs/heads/$name";
 524     if (open(my $fh, $f)) {
 525             chomp(my $r = <$fh>);
 526             is_sha1($r) or die "Cannot get head id for $name ($r): $!";
 527             return $r;
 528     }
 529     die "unable to open $f: $!" unless $! == POSIX::ENOENT;
 530     return undef;
 531 }
 532
 533 -d $git_tree
 534         or mkdir($git_tree,0777)
 535         or die "Could not create $git_tree: $!";
 536 chdir($git_tree);
 537
 538 my $last_branch = "";
 539 my $orig_branch = "";
 540 my %branch_date;
 541 my $tip_at_start = undef;
 542
 543 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 544 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 545 $ENV{"GIT_DIR"} = $git_dir;
 546 my $orig_git_index;
 547 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 548
 549 my %index; # holds filenames of one index per branch
 550
 551 unless (-d $git_dir) {
 552         system("git-init");
 553         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 554         system("git-read-tree");
 555         die "Cannot init an empty tree: $?\n" if $?;
 556
 557         $last_branch = $opt_o;
 558         $orig_branch = "";
 559 } else {
 560         -f "$git_dir/refs/heads/$opt_o"
 561                 or die "Branch '$opt_o' does not exist.\n".
 562                        "Either use the correct '-o branch' option,\n".
 563                        "or import to a new repository.\n";
 564
 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);
 570         unless ($last_branch) {
 571                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 572                 $last_branch = "master";
 573         }
 574         $orig_branch = $last_branch;
 575         $tip_at_start = `git-rev-parse --verify HEAD`;
 576
 577         # Get the last import timestamps
 578         my $fmt = '($ref, $author) = (%(refname), %(author));';
 579         open(H, "git-for-each-ref --perl --format='$fmt' refs/heads |") or
 580                 die "Cannot run git-for-each-ref: $!\n";
 581         while (defined(my $entry = <H>)) {
 582                 my ($ref, $author);
 583                 eval($entry) || die "cannot eval refs list: $@";
 584                 my ($head) = ($ref =~ m|^refs/heads/(.*)|);
 585                 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
 586                 $branch_date{$head} = $1;
 587         }
 588         close(H);
 589 }
 590
 591 -d $git_dir
 592         or die "Could not create git subdir ($git_dir).\n";
 593
 594 # now we read (and possibly save) author-info as well
 595 -f "$git_dir/cvs-authors" and
 596   read_author_info("$git_dir/cvs-authors");
 597 if ($opt_A) {
 598         read_author_info($opt_A);
 599         write_author_info("$git_dir/cvs-authors");
 600 }
 601
 602
 603 #
 604 # run cvsps into a file unless we are getting
 605 # it passed as a file via $opt_P
 606 #
 607 my $cvspsfile;
 608 unless ($opt_P) {
 609         print "Running cvsps...\n" if $opt_v;
 610         my $pid = open(CVSPS,"-|");
 611         my $cvspsfh;
 612         die "Cannot fork: $!\n" unless defined $pid;
 613         unless ($pid) {
 614                 my @opt;
 615                 @opt = split(/,/,$opt_p) if defined $opt_p;
 616                 unshift @opt, '-z', $opt_z if defined $opt_z;
 617                 unshift @opt, '-q'         unless defined $opt_v;
 618                 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
 619                         push @opt, '--cvs-direct';
 620                 }
 621                 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
 622                 die "Could not start cvsps: $!\n";
 623         }
 624         ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
 625                                           DIR => File::Spec->tmpdir());
 626         while (<CVSPS>) {
 627             print $cvspsfh $_;
 628         }
 629         close CVSPS;
 630         close $cvspsfh;
 631 } else {
 632         $cvspsfile = $opt_P;
 633 }
 634
 635 open(CVS, "<$cvspsfile") or die $!;
 636
 637 ## cvsps output:
 638 #---------------------
 639 #PatchSet 314
 640 #Date: 1999/09/18 13:03:59
 641 #Author: wkoch
 642 #Branch: STABLE-BRANCH-1-0
 643 #Ancestor branch: HEAD
 644 #Tag: (none)
 645 #Log:
 646 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
 647 #Members:
 648 #       README:1.57->1.57.2.1
 649 #       VERSION:1.96->1.96.2.1
 650 #
 651 #---------------------
 652
 653 my $state = 0;
 654
 655 sub update_index (\@\@) {
 656         my $old = shift;
 657         my $new = shift;
 658         open(my $fh, '|-', qw(git-update-index -z --index-info))
 659                 or die "unable to open git-update-index: $!";
 660         print $fh
 661                 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
 662                         @$old),
 663                 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
 664                         @$new)
 665                 or die "unable to write to git-update-index: $!";
 666         close $fh
 667                 or die "unable to write to git-update-index: $!";
 668         $? and die "git-update-index reported error: $?";
 669 }
 670
 671 sub write_tree () {
 672         open(my $fh, '-|', qw(git-write-tree))
 673                 or die "unable to open git-write-tree: $!";
 674         chomp(my $tree = <$fh>);
 675         is_sha1($tree)
 676                 or die "Cannot get tree id ($tree): $!";
 677         close($fh)
 678                 or die "Error running git-write-tree: $?\n";
 679         print "Tree ID $tree\n" if $opt_v;
 680         return $tree;
 681 }
 682
 683 my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
 684 my (@old,@new,@skipped,%ignorebranch);
 685
 686 # commits that cvsps cannot place anywhere...
 687 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
 688
 689 sub commit {
 690         if ($branch eq $opt_o && !$index{branch} && !get_headref($branch, $git_dir)) {
 691             # looks like an initial commit
 692             # use the index primed by git-init
 693             $ENV{GIT_INDEX_FILE} = '.git/index';
 694             $index{$branch} = '.git/index';
 695         } else {
 696             # use an index per branch to speed up
 697             # imports of projects with many branches
 698             unless ($index{$branch}) {
 699                 $index{$branch} = tmpnam();
 700                 $ENV{GIT_INDEX_FILE} = $index{$branch};
 701                 if ($ancestor) {
 702                     system("git-read-tree", $ancestor);
 703                 } else {
 704                     system("git-read-tree", $branch);
 705                 }
 706                 die "read-tree failed: $?\n" if $?;
 707             }
 708         }
 709         $ENV{GIT_INDEX_FILE} = $index{$branch};
 710
 711         update_index(@old, @new);
 712         @old = @new = ();
 713         my $tree = write_tree();
 714         my $parent = get_headref($last_branch, $git_dir);
 715         print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
 716
 717         my @commit_args;
 718         push @commit_args, ("-p", $parent) if $parent;
 719
 720         # loose detection of merges
 721         # based on the commit msg
 722         foreach my $rx (@mergerx) {
 723                 next unless $logmsg =~ $rx && $1;
 724                 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
 725                 if (my $sha1 = get_headref($mparent, $git_dir)) {
 726                         push @commit_args, '-p', $mparent;
 727                         print "Merge parent branch: $mparent\n" if $opt_v;
 728                 }
 729         }
 730
 731         my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
 732         $ENV{GIT_AUTHOR_NAME} = $author_name;
 733         $ENV{GIT_AUTHOR_EMAIL} = $author_email;
 734         $ENV{GIT_AUTHOR_DATE} = $commit_date;
 735         $ENV{GIT_COMMITTER_NAME} = $author_name;
 736         $ENV{GIT_COMMITTER_EMAIL} = $author_email;
 737         $ENV{GIT_COMMITTER_DATE} = $commit_date;
 738         my $pid = open2(my $commit_read, my $commit_write,
 739                 'git-commit-tree', $tree, @commit_args);
 740
 741         # compatibility with git2cvs
 742         substr($logmsg,32767) = "" if length($logmsg) > 32767;
 743         $logmsg =~ s/[\s\n]+\z//;
 744
 745         if (@skipped) {
 746             $logmsg .= "\n\n\nSKIPPED:\n\t";
 747             $logmsg .= join("\n\t", @skipped) . "\n";
 748             @skipped = ();
 749         }
 750
 751         print($commit_write "$logmsg\n") && close($commit_write)
 752                 or die "Error writing to git-commit-tree: $!\n";
 753
 754         print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
 755         chomp(my $cid = <$commit_read>);
 756         is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
 757         print "Commit ID $cid\n" if $opt_v;
 758         close($commit_read);
 759
 760         waitpid($pid,0);
 761         die "Error running git-commit-tree: $?\n" if $?;
 762
 763         system("git-update-ref refs/heads/$branch $cid") == 0
 764                 or die "Cannot write branch $branch for update: $!\n";
 765
 766         if ($tag) {
 767                 my ($in, $out) = ('','');
 768                 my ($xtag) = $tag;
 769                 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
 770                 $xtag =~ tr/_/\./ if ( $opt_u );
 771                 $xtag =~ s/[\/]/$opt_s/g;
 772
 773                 my $pid = open2($in, $out, 'git-mktag');
 774                 print $out "object $cid\n".
 775                     "type commit\n".
 776                     "tag $xtag\n".
 777                     "tagger $author_name <$author_email>\n"
 778                     or die "Cannot create tag object $xtag: $!\n";
 779                 close($out)
 780                     or die "Cannot create tag object $xtag: $!\n";
 781
 782                 my $tagobj = <$in>;
 783                 chomp $tagobj;
 784
 785                 if ( !close($in) or waitpid($pid, 0) != $pid or
 786                      $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 787                     die "Cannot create tag object $xtag: $!\n";
 788                 }
 789
 790
 791                 open(C,">$git_dir/refs/tags/$xtag")
 792                         or die "Cannot create tag $xtag: $!\n";
 793                 print C "$tagobj\n"
 794                         or die "Cannot write tag $xtag: $!\n";
 795                 close(C)
 796                         or die "Cannot write tag $xtag: $!\n";
 797
 798                 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
 799         }
 800 };
 801
 802 my $commitcount = 1;
 803 while (<CVS>) {
 804         chomp;
 805         if ($state == 0 and /^-+$/) {
 806                 $state = 1;
 807         } elsif ($state == 0) {
 808                 $state = 1;
 809                 redo;
 810         } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
 811                 $patchset = 0+$_;
 812                 $state=2;
 813         } elsif ($state == 2 and s/^Date:\s+//) {
 814                 $date = pdate($_);
 815                 unless ($date) {
 816                         print STDERR "Could not parse date: $_\n";
 817                         $state=0;
 818                         next;
 819                 }
 820                 $state=3;
 821         } elsif ($state == 3 and s/^Author:\s+//) {
 822                 s/\s+$//;
 823                 if (/^(.*?)\s+<(.*)>/) {
 824                     ($author_name, $author_email) = ($1, $2);
 825                 } elsif ($conv_author_name{$_}) {
 826                         $author_name = $conv_author_name{$_};
 827                         $author_email = $conv_author_email{$_};
 828                 } else {
 829                     $author_name = $author_email = $_;
 830                 }
 831                 $state = 4;
 832         } elsif ($state == 4 and s/^Branch:\s+//) {
 833                 s/\s+$//;
 834                 s/[\/]/$opt_s/g;
 835                 $branch = $_;
 836                 $state = 5;
 837         } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
 838                 s/\s+$//;
 839                 $ancestor = $_;
 840                 $ancestor = $opt_o if $ancestor eq "HEAD";
 841                 $state = 6;
 842         } elsif ($state == 5) {
 843                 $ancestor = undef;
 844                 $state = 6;
 845                 redo;
 846         } elsif ($state == 6 and s/^Tag:\s+//) {
 847                 s/\s+$//;
 848                 if ($_ eq "(none)") {
 849                         $tag = undef;
 850                 } else {
 851                         $tag = $_;
 852                 }
 853                 $state = 7;
 854         } elsif ($state == 7 and /^Log:/) {
 855                 $logmsg = "";
 856                 $state = 8;
 857         } elsif ($state == 8 and /^Members:/) {
 858                 $branch = $opt_o if $branch eq "HEAD";
 859                 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
 860                         # skip
 861                         print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
 862                         $state = 11;
 863                         next;
 864                 }
 865                 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
 866                         # skip if the commit is too recent
 867                         # that the cvsps default fuzz is 300s, we give ourselves another
 868                         # 300s just in case -- this also prevents skipping commits
 869                         # due to server clock drift
 870                         print "skip patchset $patchset: $date too recent\n" if $opt_v;
 871                         $state = 11;
 872                         next;
 873                 }
 874                 if (exists $ignorebranch{$branch}) {
 875                         print STDERR "Skipping $branch\n";
 876                         $state = 11;
 877                         next;
 878                 }
 879                 if ($ancestor) {
 880                         if ($ancestor eq $branch) {
 881                                 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
 882                                 $ancestor = $opt_o;
 883                         }
 884                         if (-f "$git_dir/refs/heads/$branch") {
 885                                 print STDERR "Branch $branch already exists!\n";
 886                                 $state=11;
 887                                 next;
 888                         }
 889                         unless (open(H,"$git_dir/refs/heads/$ancestor")) {
 890                                 print STDERR "Branch $ancestor does not exist!\n";
 891                                 $ignorebranch{$branch} = 1;
 892                                 $state=11;
 893                                 next;
 894                         }
 895                         chomp(my $id = <H>);
 896                         close(H);
 897                         unless (open(H,"> $git_dir/refs/heads/$branch")) {
 898                                 print STDERR "Could not create branch $branch: $!\n";
 899                                 $ignorebranch{$branch} = 1;
 900                                 $state=11;
 901                                 next;
 902                         }
 903                         print H "$id\n"
 904                                 or die "Could not write branch $branch: $!";
 905                         close(H)
 906                                 or die "Could not write branch $branch: $!";
 907                 }
 908                 $last_branch = $branch if $branch ne $last_branch;
 909                 $state = 9;
 910         } elsif ($state == 8) {
 911                 $logmsg .= "$_\n";
 912         } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
 913 #       VERSION:1.96->1.96.2.1
 914                 my $init = ($2 eq "INITIAL");
 915                 my $fn = $1;
 916                 my $rev = $3;
 917                 $fn =~ s#^/+##;
 918                 if ($opt_S && $fn =~ m/$opt_S/) {
 919                     print "SKIPPING $fn v $rev\n";
 920                     push(@skipped, $fn);
 921                     next;
 922                 }
 923                 print "Fetching $fn   v $rev\n" if $opt_v;
 924                 my ($tmpname, $size) = $cvs->file($fn,$rev);
 925                 if ($size == -1) {
 926                         push(@old,$fn);
 927                         print "Drop $fn\n" if $opt_v;
 928                 } else {
 929                         print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
 930                         my $pid = open(my $F, '-|');
 931                         die $! unless defined $pid;
 932                         if (!$pid) {
 933                             exec("git-hash-object", "-w", $tmpname)
 934                                 or die "Cannot create object: $!\n";
 935                         }
 936                         my $sha = <$F>;
 937                         chomp $sha;
 938                         close $F;
 939                         my $mode = pmode($cvs->{'mode'});
 940                         push(@new,[$mode, $sha, $fn]); # may be resurrected!
 941                 }
 942                 unlink($tmpname);
 943         } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
 944                 my $fn = $1;
 945                 $fn =~ s#^/+##;
 946                 push(@old,$fn);
 947                 print "Delete $fn\n" if $opt_v;
 948         } elsif ($state == 9 and /^\s*$/) {
 949                 $state = 10;
 950         } elsif (($state == 9 or $state == 10) and /^-+$/) {
 951                 $commitcount++;
 952                 if ($opt_L && $commitcount > $opt_L) {
 953                         last;
 954                 }
 955                 commit();
 956                 if (($commitcount & 1023) == 0) {
 957                         system("git repack -a -d");
 958                 }
 959                 $state = 1;
 960         } elsif ($state == 11 and /^-+$/) {
 961                 $state = 1;
 962         } elsif (/^-+$/) { # end of unknown-line processing
 963                 $state = 1;
 964         } elsif ($state != 11) { # ignore stuff when skipping
 965                 print "* UNKNOWN LINE * $_\n";
 966         }
 967 }
 968 commit() if $branch and $state != 11;
 969
 970 unless ($opt_P) {
 971         unlink($cvspsfile);
 972 }
 973
 974 # The heuristic of repacking every 1024 commits can leave a
 975 # lot of unpacked data.  If there is more than 1MB worth of
 976 # not-packed objects, repack once more.
 977 my $line = `git-count-objects`;
 978 if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
 979   my ($n_objects, $kb) = ($1, $2);
 980   1024 < $kb
 981     and system("git repack -a -d");
 982 }
 983
 984 foreach my $git_index (values %index) {
 985     if ($git_index ne '.git/index') {
 986         unlink($git_index);
 987     }
 988 }
 989
 990 if (defined $orig_git_index) {
 991         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 992 } else {
 993         delete $ENV{GIT_INDEX_FILE};
 994 }
 995
 996 # Now switch back to the branch we were in before all of this happened
 997 if ($orig_branch) {
 998         print "DONE.\n" if $opt_v;
 999         if ($opt_i) {
1000                 exit 0;
1001         }
1002         my $tip_at_end = `git-rev-parse --verify HEAD`;
1003         if ($tip_at_start ne $tip_at_end) {
1004                 for ($tip_at_start, $tip_at_end) { chomp; }
1005                 print "Fetched into the current branch.\n" if $opt_v;
1006                 system(qw(git-read-tree -u -m),
1007                        $tip_at_start, $tip_at_end);
1008                 die "Fast-forward update failed: $?\n" if $?;
1009         }
1010         else {
1011                 system(qw(git-merge cvsimport HEAD), "refs/heads/$opt_o");
1012                 die "Could not merge $opt_o into the current branch.\n" if $?;
1013         }
1014 } else {
1015         $orig_branch = "master";
1016         print "DONE; creating $orig_branch branch\n" if $opt_v;
1017         system("git-update-ref", "refs/heads/master", "refs/heads/$opt_o")
1018                 unless -f "$git_dir/refs/heads/master";
1019         system('git-update-ref', 'HEAD', "$orig_branch");
1020         unless ($opt_i) {
1021                 system('git checkout');
1022                 die "checkout failed: $?\n" if $?;
1023         }
1024 }