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::Long;
  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, $opt_r);
  33 my (%conv_author_name, %conv_author_email);
  34
  35 sub usage(;$) {
  36         my $msg = shift;
  37         print(STDERR "Error: $msg\n") if $msg;
  38         print STDERR <<END;
  39 Usage: git cvsimport     # fetch/update GIT from CVS
  40        [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
  41        [-p opts-for-cvsps] [-P file] [-C GIT_repository] [-z fuzz] [-i] [-k]
  42        [-u] [-s subst] [-a] [-m] [-M regex] [-S regex] [-L commitlimit]
  43        [-r remote] [CVS_module]
  44 END
  45         exit(1);
  46 }
  47
  48 sub read_author_info($) {
  49         my ($file) = @_;
  50         my $user;
  51         open my $f, '<', "$file" or die("Failed to open $file: $!\n");
  52
  53         while (<$f>) {
  54                 # Expected format is this:
  55                 #   exon=Andreas Ericsson <ae@op5.se>
  56                 if (m/^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/) {
  57                         $user = $1;
  58                         $conv_author_name{$user} = $2;
  59                         $conv_author_email{$user} = $3;
  60                 }
  61                 # However, we also read from CVSROOT/users format
  62                 # to ease migration.
  63                 elsif (/^(\w+):(['"]?)(.+?)\2\s*$/) {
  64                         my $mapped;
  65                         ($user, $mapped) = ($1, $3);
  66                         if ($mapped =~ /^\s*(.*?)\s*<(.*)>\s*$/) {
  67                                 $conv_author_name{$user} = $1;
  68                                 $conv_author_email{$user} = $2;
  69                         }
  70                         elsif ($mapped =~ /^<?(.*)>?$/) {
  71                                 $conv_author_name{$user} = $user;
  72                                 $conv_author_email{$user} = $1;
  73                         }
  74                 }
  75                 # NEEDSWORK: Maybe warn on unrecognized lines?
  76         }
  77         close ($f);
  78 }
  79
  80 sub write_author_info($) {
  81         my ($file) = @_;
  82         open my $f, '>', $file or
  83           die("Failed to open $file for writing: $!");
  84
  85         foreach (keys %conv_author_name) {
  86                 print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
  87         }
  88         close ($f);
  89 }
  90
  91 # convert getopts specs for use by git config
  92 sub read_repo_config {
  93     # Split the string between characters, unless there is a ':'
  94     # So "abc:de" becomes ["a", "b", "c:", "d", "e"]
  95         my @opts = split(/ *(?!:)/, shift);
  96         foreach my $o (@opts) {
  97                 my $key = $o;
  98                 $key =~ s/://g;
  99                 my $arg = 'git config';
 100                 $arg .= ' --bool' if ($o !~ /:$/);
 101
 102         chomp(my $tmp = `$arg --get cvsimport.$key`);
 103                 if ($tmp && !($arg =~ /--bool/ && $tmp eq 'false')) {
 104             no strict 'refs';
 105             my $opt_name = "opt_" . $key;
 106             if (!$$opt_name) {
 107                 $$opt_name = $tmp;
 108             }
 109                 }
 110         }
 111 }
 112
 113 my $opts = "haivmkuo:d:p:r:C:z:s:M:P:A:S:L:";
 114 read_repo_config($opts);
 115 Getopt::Long::Configure( 'no_ignore_case', 'bundling' );
 116
 117 # turn the Getopt::Std specification in a Getopt::Long one,
 118 # with support for multiple -M options
 119 GetOptions( map { s/:/=s/; /M/ ? "$_\@" : $_ } split( /(?!:)/, $opts ) )
 120     or usage();
 121 usage if $opt_h;
 122
 123 if (@ARGV == 0) {
 124                 chomp(my $module = `git config --get cvsimport.module`);
 125                 push(@ARGV, $module) if $? == 0;
 126 }
 127 @ARGV <= 1 or usage("You can't specify more than one CVS module");
 128
 129 if ($opt_d) {
 130         $ENV{"CVSROOT"} = $opt_d;
 131 } elsif (-f 'CVS/Root') {
 132         open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
 133         $opt_d = <$f>;
 134         chomp $opt_d;
 135         close $f;
 136         $ENV{"CVSROOT"} = $opt_d;
 137 } elsif ($ENV{"CVSROOT"}) {
 138         $opt_d = $ENV{"CVSROOT"};
 139 } else {
 140         usage("CVSROOT needs to be set");
 141 }
 142 $opt_s ||= "-";
 143 $opt_a ||= 0;
 144
 145 my $git_tree = $opt_C;
 146 $git_tree ||= ".";
 147
 148 my $remote;
 149 if (defined $opt_r) {
 150         $remote = 'refs/remotes/' . $opt_r;
 151         $opt_o ||= "master";
 152 } else {
 153         $opt_o ||= "origin";
 154         $remote = 'refs/heads';
 155 }
 156
 157 my $cvs_tree;
 158 if ($#ARGV == 0) {
 159         $cvs_tree = $ARGV[0];
 160 } elsif (-f 'CVS/Repository') {
 161         open my $f, '<', 'CVS/Repository' or
 162             die 'Failed to open CVS/Repository';
 163         $cvs_tree = <$f>;
 164         chomp $cvs_tree;
 165         close $f;
 166 } else {
 167         usage("CVS module has to be specified");
 168 }
 169
 170 our @mergerx = ();
 171 if ($opt_m) {
 172         @mergerx = ( qr/\b(?:from|of|merge|merging|merged) ([-\w]+)/i );
 173 }
 174 if (@opt_M) {
 175         push (@mergerx, map { qr/$_/ } @opt_M);
 176 }
 177
 178 # Remember UTC of our starting time
 179 # we'll want to avoid importing commits
 180 # that are too recent
 181 our $starttime = time();
 182
 183 select(STDERR); $|=1; select(STDOUT);
 184
 185
 186 package CVSconn;
 187 # Basic CVS dialog.
 188 # We're only interested in connecting and downloading, so ...
 189
 190 use File::Spec;
 191 use File::Temp qw(tempfile);
 192 use POSIX qw(strftime dup2);
 193
 194 sub new {
 195         my ($what,$repo,$subdir) = @_;
 196         $what=ref($what) if ref($what);
 197
 198         my $self = {};
 199         $self->{'buffer'} = "";
 200         bless($self,$what);
 201
 202         $repo =~ s#/+$##;
 203         $self->{'fullrep'} = $repo;
 204         $self->conn();
 205
 206         $self->{'subdir'} = $subdir;
 207         $self->{'lines'} = undef;
 208
 209         return $self;
 210 }
 211
 212 sub conn {
 213         my $self = shift;
 214         my $repo = $self->{'fullrep'};
 215         if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
 216                 my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
 217
 218                 my ($proxyhost,$proxyport);
 219                 if ($param && ($param =~ m/proxy=([^;]+)/)) {
 220                         $proxyhost = $1;
 221                         # Default proxyport, if not specified, is 8080.
 222                         $proxyport = 8080;
 223                         if ($ENV{"CVS_PROXY_PORT"}) {
 224                                 $proxyport = $ENV{"CVS_PROXY_PORT"};
 225                         }
 226                         if ($param =~ m/proxyport=([^;]+)/) {
 227                                 $proxyport = $1;
 228                         }
 229                 }
 230                 $repo ||= '/';
 231
 232                 # if username is not explicit in CVSROOT, then use current user, as cvs would
 233                 $user=(getlogin() || $ENV{'LOGNAME'} || $ENV{'USER'} || "anonymous") unless $user;
 234                 my $rr2 = "-";
 235                 unless ($port) {
 236                         $rr2 = ":pserver:$user\@$serv:$repo";
 237                         $port=2401;
 238                 }
 239                 my $rr = ":pserver:$user\@$serv:$port$repo";
 240
 241                 unless ($pass) {
 242                         open(H,$ENV{'HOME'}."/.cvspass") and do {
 243                                 # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
 244                                 while (<H>) {
 245                                         chomp;
 246                                         s/^\/\d+\s+//;
 247                                         my ($w,$p) = split(/\s/,$_,2);
 248                                         if ($w eq $rr or $w eq $rr2) {
 249                                                 $pass = $p;
 250                                                 last;
 251                                         }
 252                                 }
 253                         };
 254                 }
 255                 $pass="A" unless $pass;
 256
 257                 my ($s, $rep);
 258                 if ($proxyhost) {
 259
 260                         # Use a HTTP Proxy. Only works for HTTP proxies that
 261                         # don't require user authentication
 262                         #
 263                         # See: http://www.ietf.org/rfc/rfc2817.txt
 264
 265                         $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
 266                         die "Socket to $proxyhost: $!\n" unless defined $s;
 267                         $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
 268                                 or die "Write to $proxyhost: $!\n";
 269                         $s->flush();
 270
 271                         $rep = <$s>;
 272
 273                         # The answer should look like 'HTTP/1.x 2yy ....'
 274                         if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
 275                                 die "Proxy connect: $rep\n";
 276                         }
 277                         # Skip up to the empty line of the proxy server output
 278                         # including the response headers.
 279                         while ($rep = <$s>) {
 280                                 last if (!defined $rep ||
 281                                          $rep eq "\n" ||
 282                                          $rep eq "\r\n");
 283                         }
 284                 } else {
 285                         $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
 286                         die "Socket to $serv: $!\n" unless defined $s;
 287                 }
 288
 289                 $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
 290                         or die "Write to $serv: $!\n";
 291                 $s->flush();
 292
 293                 $rep = <$s>;
 294
 295                 if ($rep ne "I LOVE YOU\n") {
 296                         $rep="<unknown>" unless $rep;
 297                         die "AuthReply: $rep\n";
 298                 }
 299                 $self->{'socketo'} = $s;
 300                 $self->{'socketi'} = $s;
 301         } else { # local or ext: Fork off our own cvs server.
 302                 my $pr = IO::Pipe->new();
 303                 my $pw = IO::Pipe->new();
 304                 my $pid = fork();
 305                 die "Fork: $!\n" unless defined $pid;
 306                 my $cvs = 'cvs';
 307                 $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
 308                 my $rsh = 'rsh';
 309                 $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
 310
 311                 my @cvs = ($cvs, 'server');
 312                 my ($local, $user, $host);
 313                 $local = $repo =~ s/:local://;
 314                 if (!$local) {
 315                     $repo =~ s/:ext://;
 316                     $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
 317                     ($user, $host) = ($1, $2);
 318                 }
 319                 if (!$local) {
 320                     if ($user) {
 321                         unshift @cvs, $rsh, '-l', $user, $host;
 322                     } else {
 323                         unshift @cvs, $rsh, $host;
 324                     }
 325                 }
 326
 327                 unless ($pid) {
 328                         $pr->writer();
 329                         $pw->reader();
 330                         dup2($pw->fileno(),0);
 331                         dup2($pr->fileno(),1);
 332                         $pr->close();
 333                         $pw->close();
 334                         exec(@cvs);
 335                 }
 336                 $pw->writer();
 337                 $pr->reader();
 338                 $self->{'socketo'} = $pw;
 339                 $self->{'socketi'} = $pr;
 340         }
 341         $self->{'socketo'}->write("Root $repo\n");
 342
 343         # Trial and error says that this probably is the minimum set
 344         $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
 345
 346         $self->{'socketo'}->write("valid-requests\n");
 347         $self->{'socketo'}->flush();
 348
 349         chomp(my $rep=$self->readline());
 350         if ($rep !~ s/^Valid-requests\s*//) {
 351                 $rep="<unknown>" unless $rep;
 352                 die "Expected Valid-requests from server, but got: $rep\n";
 353         }
 354         chomp(my $res=$self->readline());
 355         die "validReply: $res\n" if $res ne "ok";
 356
 357         $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
 358         $self->{'repo'} = $repo;
 359 }
 360
 361 sub readline {
 362         my ($self) = @_;
 363         return $self->{'socketi'}->getline();
 364 }
 365
 366 sub _file {
 367         # Request a file with a given revision.
 368         # Trial and error says this is a good way to do it. :-/
 369         my ($self,$fn,$rev) = @_;
 370         $self->{'socketo'}->write("Argument -N\n") or return undef;
 371         $self->{'socketo'}->write("Argument -P\n") or return undef;
 372         # -kk: Linus' version doesn't use it - defaults to off
 373         if ($opt_k) {
 374             $self->{'socketo'}->write("Argument -kk\n") or return undef;
 375         }
 376         $self->{'socketo'}->write("Argument -r\n") or return undef;
 377         $self->{'socketo'}->write("Argument $rev\n") or return undef;
 378         $self->{'socketo'}->write("Argument --\n") or return undef;
 379         $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
 380         $self->{'socketo'}->write("Directory .\n") or return undef;
 381         $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
 382         # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
 383         $self->{'socketo'}->write("co\n") or return undef;
 384         $self->{'socketo'}->flush() or return undef;
 385         $self->{'lines'} = 0;
 386         return 1;
 387 }
 388 sub _line {
 389         # Read a line from the server.
 390         # ... except that 'line' may be an entire file. ;-)
 391         my ($self, $fh) = @_;
 392         die "Not in lines" unless defined $self->{'lines'};
 393
 394         my $line;
 395         my $res=0;
 396         while (defined($line = $self->readline())) {
 397                 # M U gnupg-cvs-rep/AUTHORS
 398                 # Updated gnupg-cvs-rep/
 399                 # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
 400                 # /AUTHORS/1.1///T1.1
 401                 # u=rw,g=rw,o=rw
 402                 # 0
 403                 # ok
 404
 405                 if ($line =~ s/^(?:Created|Updated) //) {
 406                         $line = $self->readline(); # path
 407                         $line = $self->readline(); # Entries line
 408                         my $mode = $self->readline(); chomp $mode;
 409                         $self->{'mode'} = $mode;
 410                         defined (my $cnt = $self->readline())
 411                                 or die "EOF from server after 'Changed'\n";
 412                         chomp $cnt;
 413                         die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
 414                         $line="";
 415                         $res = $self->_fetchfile($fh, $cnt);
 416                 } elsif ($line =~ s/^ //) {
 417                         print $fh $line;
 418                         $res += length($line);
 419                 } elsif ($line =~ /^M\b/) {
 420                         # output, do nothing
 421                 } elsif ($line =~ /^Mbinary\b/) {
 422                         my $cnt;
 423                         die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
 424                         chomp $cnt;
 425                         die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
 426                         $line="";
 427                         $res += $self->_fetchfile($fh, $cnt);
 428                 } else {
 429                         chomp $line;
 430                         if ($line eq "ok") {
 431                                 # print STDERR "S: ok (".length($res).")\n";
 432                                 return $res;
 433                         } elsif ($line =~ s/^E //) {
 434                                 # print STDERR "S: $line\n";
 435                         } elsif ($line =~ /^(Remove-entry|Removed) /i) {
 436                                 $line = $self->readline(); # filename
 437                                 $line = $self->readline(); # OK
 438                                 chomp $line;
 439                                 die "Unknown: $line" if $line ne "ok";
 440                                 return -1;
 441                         } else {
 442                                 die "Unknown: $line\n";
 443                         }
 444                 }
 445         }
 446         return undef;
 447 }
 448 sub file {
 449         my ($self,$fn,$rev) = @_;
 450         my $res;
 451
 452         my ($fh, $name) = tempfile('gitcvs.XXXXXX',
 453                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 454
 455         $self->_file($fn,$rev) and $res = $self->_line($fh);
 456
 457         if (!defined $res) {
 458             print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
 459             truncate $fh, 0;
 460             $self->conn();
 461             $self->_file($fn,$rev) or die "No file command send";
 462             $res = $self->_line($fh);
 463             die "Retry failed" unless defined $res;
 464         }
 465         close ($fh);
 466
 467         return ($name, $res);
 468 }
 469 sub _fetchfile {
 470         my ($self, $fh, $cnt) = @_;
 471         my $res = 0;
 472         my $bufsize = 1024 * 1024;
 473         while ($cnt) {
 474             if ($bufsize > $cnt) {
 475                 $bufsize = $cnt;
 476             }
 477             my $buf;
 478             my $num = $self->{'socketi'}->read($buf,$bufsize);
 479             die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
 480             print $fh $buf;
 481             $res += $num;
 482             $cnt -= $num;
 483         }
 484         return $res;
 485 }
 486
 487
 488 package main;
 489
 490 my $cvs = CVSconn->new($opt_d, $cvs_tree);
 491
 492
 493 sub pdate($) {
 494         my ($d) = @_;
 495         m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
 496                 or die "Unparseable date: $d\n";
 497         my $y=$1; $y-=1900 if $y>1900;
 498         return timegm($6||0,$5,$4,$3,$2-1,$y);
 499 }
 500
 501 sub pmode($) {
 502         my ($mode) = @_;
 503         my $m = 0;
 504         my $mm = 0;
 505         my $um = 0;
 506         for my $x(split(//,$mode)) {
 507                 if ($x eq ",") {
 508                         $m |= $mm&$um;
 509                         $mm = 0;
 510                         $um = 0;
 511                 } elsif ($x eq "u") { $um |= 0700;
 512                 } elsif ($x eq "g") { $um |= 0070;
 513                 } elsif ($x eq "o") { $um |= 0007;
 514                 } elsif ($x eq "r") { $mm |= 0444;
 515                 } elsif ($x eq "w") { $mm |= 0222;
 516                 } elsif ($x eq "x") { $mm |= 0111;
 517                 } elsif ($x eq "=") { # do nothing
 518                 } else { die "Unknown mode: $mode\n";
 519                 }
 520         }
 521         $m |= $mm&$um;
 522         return $m;
 523 }
 524
 525 sub getwd() {
 526         my $pwd = `pwd`;
 527         chomp $pwd;
 528         return $pwd;
 529 }
 530
 531 sub is_sha1 {
 532         my $s = shift;
 533         return $s =~ /^[a-f0-9]{40}$/;
 534 }
 535
 536 sub get_headref ($) {
 537         my $name = shift;
 538         my $r = `git rev-parse --verify '$name' 2>/dev/null`;
 539         return undef unless $? == 0;
 540         chomp $r;
 541         return $r;
 542 }
 543
 544 -d $git_tree
 545         or mkdir($git_tree,0777)
 546         or die "Could not create $git_tree: $!";
 547 chdir($git_tree);
 548
 549 my $last_branch = "";
 550 my $orig_branch = "";
 551 my %branch_date;
 552 my $tip_at_start = undef;
 553
 554 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 555 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 556 $ENV{"GIT_DIR"} = $git_dir;
 557 my $orig_git_index;
 558 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 559
 560 my %index; # holds filenames of one index per branch
 561
 562 unless (-d $git_dir) {
 563         system("git-init");
 564         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 565         system("git-read-tree");
 566         die "Cannot init an empty tree: $?\n" if $?;
 567
 568         $last_branch = $opt_o;
 569         $orig_branch = "";
 570 } else {
 571         open(F, "git-symbolic-ref HEAD |") or
 572                 die "Cannot run git-symbolic-ref: $!\n";
 573         chomp ($last_branch = <F>);
 574         $last_branch = basename($last_branch);
 575         close(F);
 576         unless ($last_branch) {
 577                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 578                 $last_branch = "master";
 579         }
 580         $orig_branch = $last_branch;
 581         $tip_at_start = `git-rev-parse --verify HEAD`;
 582
 583         # Get the last import timestamps
 584         my $fmt = '($ref, $author) = (%(refname), %(author));';
 585         open(H, "git-for-each-ref --perl --format='$fmt' $remote |") or
 586                 die "Cannot run git-for-each-ref: $!\n";
 587         while (defined(my $entry = <H>)) {
 588                 my ($ref, $author);
 589                 eval($entry) || die "cannot eval refs list: $@";
 590                 my ($head) = ($ref =~ m|^$remote/(.*)|);
 591                 $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
 592                 $branch_date{$head} = $1;
 593         }
 594         close(H);
 595         if (!exists $branch_date{$opt_o}) {
 596                 die "Branch '$opt_o' does not exist.\n".
 597                        "Either use the correct '-o branch' option,\n".
 598                        "or import to a new repository.\n";
 599         }
 600 }
 601
 602 -d $git_dir
 603         or die "Could not create git subdir ($git_dir).\n";
 604
 605 # now we read (and possibly save) author-info as well
 606 -f "$git_dir/cvs-authors" and
 607   read_author_info("$git_dir/cvs-authors");
 608 if ($opt_A) {
 609         read_author_info($opt_A);
 610         write_author_info("$git_dir/cvs-authors");
 611 }
 612
 613
 614 #
 615 # run cvsps into a file unless we are getting
 616 # it passed as a file via $opt_P
 617 #
 618 my $cvspsfile;
 619 unless ($opt_P) {
 620         print "Running cvsps...\n" if $opt_v;
 621         my $pid = open(CVSPS,"-|");
 622         my $cvspsfh;
 623         die "Cannot fork: $!\n" unless defined $pid;
 624         unless ($pid) {
 625                 my @opt;
 626                 @opt = split(/,/,$opt_p) if defined $opt_p;
 627                 unshift @opt, '-z', $opt_z if defined $opt_z;
 628                 unshift @opt, '-q'         unless defined $opt_v;
 629                 unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
 630                         push @opt, '--cvs-direct';
 631                 }
 632                 exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
 633                 die "Could not start cvsps: $!\n";
 634         }
 635         ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
 636                                           DIR => File::Spec->tmpdir());
 637         while (<CVSPS>) {
 638             print $cvspsfh $_;
 639         }
 640         close CVSPS;
 641         $? == 0 or die "git-cvsimport: fatal: cvsps reported error\n";
 642         close $cvspsfh;
 643 } else {
 644         $cvspsfile = $opt_P;
 645 }
 646
 647 open(CVS, "<$cvspsfile") or die $!;
 648
 649 ## cvsps output:
 650 #---------------------
 651 #PatchSet 314
 652 #Date: 1999/09/18 13:03:59
 653 #Author: wkoch
 654 #Branch: STABLE-BRANCH-1-0
 655 #Ancestor branch: HEAD
 656 #Tag: (none)
 657 #Log:
 658 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
 659 #Members:
 660 #       README:1.57->1.57.2.1
 661 #       VERSION:1.96->1.96.2.1
 662 #
 663 #---------------------
 664
 665 my $state = 0;
 666
 667 sub update_index (\@\@) {
 668         my $old = shift;
 669         my $new = shift;
 670         open(my $fh, '|-', qw(git-update-index -z --index-info))
 671                 or die "unable to open git-update-index: $!";
 672         print $fh
 673                 (map { "0 0000000000000000000000000000000000000000\t$_\0" }
 674                         @$old),
 675                 (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
 676                         @$new)
 677                 or die "unable to write to git-update-index: $!";
 678         close $fh
 679                 or die "unable to write to git-update-index: $!";
 680         $? and die "git-update-index reported error: $?";
 681 }
 682
 683 sub write_tree () {
 684         open(my $fh, '-|', qw(git-write-tree))
 685                 or die "unable to open git-write-tree: $!";
 686         chomp(my $tree = <$fh>);
 687         is_sha1($tree)
 688                 or die "Cannot get tree id ($tree): $!";
 689         close($fh)
 690                 or die "Error running git-write-tree: $?\n";
 691         print "Tree ID $tree\n" if $opt_v;
 692         return $tree;
 693 }
 694
 695 my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
 696 my (@old,@new,@skipped,%ignorebranch);
 697
 698 # commits that cvsps cannot place anywhere...
 699 $ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
 700
 701 sub commit {
 702         if ($branch eq $opt_o && !$index{branch} &&
 703                 !get_headref("$remote/$branch")) {
 704             # looks like an initial commit
 705             # use the index primed by git-init
 706             $ENV{GIT_INDEX_FILE} = "$git_dir/index";
 707             $index{$branch} = "$git_dir/index";
 708         } else {
 709             # use an index per branch to speed up
 710             # imports of projects with many branches
 711             unless ($index{$branch}) {
 712                 $index{$branch} = tmpnam();
 713                 $ENV{GIT_INDEX_FILE} = $index{$branch};
 714                 if ($ancestor) {
 715                     system("git-read-tree", "$remote/$ancestor");
 716                 } else {
 717                     system("git-read-tree", "$remote/$branch");
 718                 }
 719                 die "read-tree failed: $?\n" if $?;
 720             }
 721         }
 722         $ENV{GIT_INDEX_FILE} = $index{$branch};
 723
 724         update_index(@old, @new);
 725         @old = @new = ();
 726         my $tree = write_tree();
 727         my $parent = get_headref("$remote/$last_branch");
 728         print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
 729
 730         my @commit_args;
 731         push @commit_args, ("-p", $parent) if $parent;
 732
 733         # loose detection of merges
 734         # based on the commit msg
 735         foreach my $rx (@mergerx) {
 736                 next unless $logmsg =~ $rx && $1;
 737                 my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
 738                 if (my $sha1 = get_headref("$remote/$mparent")) {
 739                         push @commit_args, '-p', "$remote/$mparent";
 740                         print "Merge parent branch: $mparent\n" if $opt_v;
 741                 }
 742         }
 743
 744         my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
 745         $ENV{GIT_AUTHOR_NAME} = $author_name;
 746         $ENV{GIT_AUTHOR_EMAIL} = $author_email;
 747         $ENV{GIT_AUTHOR_DATE} = $commit_date;
 748         $ENV{GIT_COMMITTER_NAME} = $author_name;
 749         $ENV{GIT_COMMITTER_EMAIL} = $author_email;
 750         $ENV{GIT_COMMITTER_DATE} = $commit_date;
 751         my $pid = open2(my $commit_read, my $commit_write,
 752                 'git-commit-tree', $tree, @commit_args);
 753
 754         # compatibility with git2cvs
 755         substr($logmsg,32767) = "" if length($logmsg) > 32767;
 756         $logmsg =~ s/[\s\n]+\z//;
 757
 758         if (@skipped) {
 759             $logmsg .= "\n\n\nSKIPPED:\n\t";
 760             $logmsg .= join("\n\t", @skipped) . "\n";
 761             @skipped = ();
 762         }
 763
 764         print($commit_write "$logmsg\n") && close($commit_write)
 765                 or die "Error writing to git-commit-tree: $!\n";
 766
 767         print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
 768         chomp(my $cid = <$commit_read>);
 769         is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
 770         print "Commit ID $cid\n" if $opt_v;
 771         close($commit_read);
 772
 773         waitpid($pid,0);
 774         die "Error running git-commit-tree: $?\n" if $?;
 775
 776         system('git-update-ref', "$remote/$branch", $cid) == 0
 777                 or die "Cannot write branch $branch for update: $!\n";
 778
 779         if ($tag) {
 780                 my ($xtag) = $tag;
 781                 $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
 782                 $xtag =~ tr/_/\./ if ( $opt_u );
 783                 $xtag =~ s/[\/]/$opt_s/g;
 784                 $xtag =~ s/\[//g;
 785
 786                 system('git-tag', '-f', $xtag, $cid) == 0
 787                         or die "Cannot create tag $xtag: $!\n";
 788
 789                 print "Created tag '$xtag' on '$branch'\n" if $opt_v;
 790         }
 791 };
 792
 793 my $commitcount = 1;
 794 while (<CVS>) {
 795         chomp;
 796         if ($state == 0 and /^-+$/) {
 797                 $state = 1;
 798         } elsif ($state == 0) {
 799                 $state = 1;
 800                 redo;
 801         } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
 802                 $patchset = 0+$_;
 803                 $state=2;
 804         } elsif ($state == 2 and s/^Date:\s+//) {
 805                 $date = pdate($_);
 806                 unless ($date) {
 807                         print STDERR "Could not parse date: $_\n";
 808                         $state=0;
 809                         next;
 810                 }
 811                 $state=3;
 812         } elsif ($state == 3 and s/^Author:\s+//) {
 813                 s/\s+$//;
 814                 if (/^(.*?)\s+<(.*)>/) {
 815                     ($author_name, $author_email) = ($1, $2);
 816                 } elsif ($conv_author_name{$_}) {
 817                         $author_name = $conv_author_name{$_};
 818                         $author_email = $conv_author_email{$_};
 819                 } else {
 820                     $author_name = $author_email = $_;
 821                 }
 822                 $state = 4;
 823         } elsif ($state == 4 and s/^Branch:\s+//) {
 824                 s/\s+$//;
 825                 tr/_/\./ if ( $opt_u );
 826                 s/[\/]/$opt_s/g;
 827                 $branch = $_;
 828                 $state = 5;
 829         } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
 830                 s/\s+$//;
 831                 $ancestor = $_;
 832                 $ancestor = $opt_o if $ancestor eq "HEAD";
 833                 $state = 6;
 834         } elsif ($state == 5) {
 835                 $ancestor = undef;
 836                 $state = 6;
 837                 redo;
 838         } elsif ($state == 6 and s/^Tag:\s+//) {
 839                 s/\s+$//;
 840                 if ($_ eq "(none)") {
 841                         $tag = undef;
 842                 } else {
 843                         $tag = $_;
 844                 }
 845                 $state = 7;
 846         } elsif ($state == 7 and /^Log:/) {
 847                 $logmsg = "";
 848                 $state = 8;
 849         } elsif ($state == 8 and /^Members:/) {
 850                 $branch = $opt_o if $branch eq "HEAD";
 851                 if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
 852                         # skip
 853                         print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
 854                         $state = 11;
 855                         next;
 856                 }
 857                 if (!$opt_a && $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
 858                         # skip if the commit is too recent
 859                         # given that the cvsps default fuzz is 300s, we give ourselves another
 860                         # 300s just in case -- this also prevents skipping commits
 861                         # due to server clock drift
 862                         print "skip patchset $patchset: $date too recent\n" if $opt_v;
 863                         $state = 11;
 864                         next;
 865                 }
 866                 if (exists $ignorebranch{$branch}) {
 867                         print STDERR "Skipping $branch\n";
 868                         $state = 11;
 869                         next;
 870                 }
 871                 if ($ancestor) {
 872                         if ($ancestor eq $branch) {
 873                                 print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
 874                                 $ancestor = $opt_o;
 875                         }
 876                         if (defined get_headref("$remote/$branch")) {
 877                                 print STDERR "Branch $branch already exists!\n";
 878                                 $state=11;
 879                                 next;
 880                         }
 881                         my $id = get_headref("$remote/$ancestor");
 882                         if (!$id) {
 883                                 print STDERR "Branch $ancestor does not exist!\n";
 884                                 $ignorebranch{$branch} = 1;
 885                                 $state=11;
 886                                 next;
 887                         }
 888
 889                         system(qw(git update-ref -m cvsimport),
 890                                 "$remote/$branch", $id);
 891                         if($? != 0) {
 892                                 print STDERR "Could not create branch $branch\n";
 893                                 $ignorebranch{$branch} = 1;
 894                                 $state=11;
 895                                 next;
 896                         }
 897                 }
 898                 $last_branch = $branch if $branch ne $last_branch;
 899                 $state = 9;
 900         } elsif ($state == 8) {
 901                 $logmsg .= "$_\n";
 902         } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
 903 #       VERSION:1.96->1.96.2.1
 904                 my $init = ($2 eq "INITIAL");
 905                 my $fn = $1;
 906                 my $rev = $3;
 907                 $fn =~ s#^/+##;
 908                 if ($opt_S && $fn =~ m/$opt_S/) {
 909                     print "SKIPPING $fn v $rev\n";
 910                     push(@skipped, $fn);
 911                     next;
 912                 }
 913                 print "Fetching $fn   v $rev\n" if $opt_v;
 914                 my ($tmpname, $size) = $cvs->file($fn,$rev);
 915                 if ($size == -1) {
 916                         push(@old,$fn);
 917                         print "Drop $fn\n" if $opt_v;
 918                 } else {
 919                         print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
 920                         my $pid = open(my $F, '-|');
 921                         die $! unless defined $pid;
 922                         if (!$pid) {
 923                             exec("git-hash-object", "-w", $tmpname)
 924                                 or die "Cannot create object: $!\n";
 925                         }
 926                         my $sha = <$F>;
 927                         chomp $sha;
 928                         close $F;
 929                         my $mode = pmode($cvs->{'mode'});
 930                         push(@new,[$mode, $sha, $fn]); # may be resurrected!
 931                 }
 932                 unlink($tmpname);
 933         } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
 934                 my $fn = $1;
 935                 $fn =~ s#^/+##;
 936                 push(@old,$fn);
 937                 print "Delete $fn\n" if $opt_v;
 938         } elsif ($state == 9 and /^\s*$/) {
 939                 $state = 10;
 940         } elsif (($state == 9 or $state == 10) and /^-+$/) {
 941                 $commitcount++;
 942                 if ($opt_L && $commitcount > $opt_L) {
 943                         last;
 944                 }
 945                 commit();
 946                 if (($commitcount & 1023) == 0) {
 947                         system("git repack -a -d");
 948                 }
 949                 $state = 1;
 950         } elsif ($state == 11 and /^-+$/) {
 951                 $state = 1;
 952         } elsif (/^-+$/) { # end of unknown-line processing
 953                 $state = 1;
 954         } elsif ($state != 11) { # ignore stuff when skipping
 955                 print "* UNKNOWN LINE * $_\n";
 956         }
 957 }
 958 commit() if $branch and $state != 11;
 959
 960 unless ($opt_P) {
 961         unlink($cvspsfile);
 962 }
 963
 964 # The heuristic of repacking every 1024 commits can leave a
 965 # lot of unpacked data.  If there is more than 1MB worth of
 966 # not-packed objects, repack once more.
 967 my $line = `git-count-objects`;
 968 if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
 969   my ($n_objects, $kb) = ($1, $2);
 970   1024 < $kb
 971     and system("git repack -a -d");
 972 }
 973
 974 foreach my $git_index (values %index) {
 975     if ($git_index ne "$git_dir/index") {
 976         unlink($git_index);
 977     }
 978 }
 979
 980 if (defined $orig_git_index) {
 981         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 982 } else {
 983         delete $ENV{GIT_INDEX_FILE};
 984 }
 985
 986 # Now switch back to the branch we were in before all of this happened
 987 if ($orig_branch) {
 988         print "DONE.\n" if $opt_v;
 989         if ($opt_i) {
 990                 exit 0;
 991         }
 992         my $tip_at_end = `git-rev-parse --verify HEAD`;
 993         if ($tip_at_start ne $tip_at_end) {
 994                 for ($tip_at_start, $tip_at_end) { chomp; }
 995                 print "Fetched into the current branch.\n" if $opt_v;
 996                 system(qw(git-read-tree -u -m),
 997                        $tip_at_start, $tip_at_end);
 998                 die "Fast-forward update failed: $?\n" if $?;
 999         }
1000         else {
1001                 system(qw(git-merge cvsimport HEAD), "$remote/$opt_o");
1002                 die "Could not merge $opt_o into the current branch.\n" if $?;
1003         }
1004 } else {
1005         $orig_branch = "master";
1006         print "DONE; creating $orig_branch branch\n" if $opt_v;
1007         system("git-update-ref", "refs/heads/master", "$remote/$opt_o")
1008                 unless defined get_headref('refs/heads/master');
1009         system("git-symbolic-ref", "$remote/HEAD", "$remote/$opt_o")
1010                 if ($opt_r && $opt_o ne 'HEAD');
1011         system('git-update-ref', 'HEAD', "$orig_branch");
1012         unless ($opt_i) {
1013                 system('git checkout -f');
1014                 die "checkout failed: $?\n" if $?;
1015         }
1016 }