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