26867751d76f0a86bf406274bd916191d3a5bdbd
   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
  16use strict;
  17use warnings;
  18use Getopt::Std;
  19use File::Spec;
  20use File::Temp qw(tempfile tmpnam);
  21use File::Path qw(mkpath);
  22use File::Basename qw(basename dirname);
  23use Time::Local;
  24use IO::Socket;
  25use IO::Pipe;
  26use POSIX qw(strftime dup2 ENOENT);
  27use IPC::Open2;
  28
  29$SIG{'PIPE'}="IGNORE";
  30$ENV{'TZ'}="UTC";
  31
  32our ($opt_h,$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);
  33my (%conv_author_name, %conv_author_email);
  34
  35sub usage() {
  36        print STDERR <<END;
  37Usage: ${\basename $0}     # fetch/update GIT from CVS
  38       [-o branch-for-HEAD] [-h] [-v] [-d CVSROOT] [-A author-conv-file]
  39       [-p opts-for-cvsps] [-C GIT_repository] [-z fuzz] [-i] [-k] [-u]
  40       [-s subst] [-m] [-M regex] [-S regex] [CVS_module]
  41END
  42        exit(1);
  43}
  44
  45sub read_author_info($) {
  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
  77sub 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
  88getopts("hivmkuo:d:p:C:z:s:M:P:A:S:L:") or usage();
  89usage if $opt_h;
  90
  91@ARGV <= 1 or usage();
  92
  93if ($opt_d) {
  94        $ENV{"CVSROOT"} = $opt_d;
  95} elsif (-f 'CVS/Root') {
  96        open my $f, '<', 'CVS/Root' or die 'Failed to open CVS/Root';
  97        $opt_d = <$f>;
  98        chomp $opt_d;
  99        close $f;
 100        $ENV{"CVSROOT"} = $opt_d;
 101} elsif ($ENV{"CVSROOT"}) {
 102        $opt_d = $ENV{"CVSROOT"};
 103} else {
 104        die "CVSROOT needs to be set";
 105}
 106$opt_o ||= "origin";
 107$opt_s ||= "-";
 108my $git_tree = $opt_C;
 109$git_tree ||= ".";
 110
 111my $cvs_tree;
 112if ($#ARGV == 0) {
 113        $cvs_tree = $ARGV[0];
 114} elsif (-f 'CVS/Repository') {
 115        open my $f, '<', 'CVS/Repository' or 
 116            die 'Failed to open CVS/Repository';
 117        $cvs_tree = <$f>;
 118        chomp $cvs_tree;
 119        close $f;
 120} else {
 121        usage();
 122}
 123
 124our @mergerx = ();
 125if ($opt_m) {
 126        @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
 127}
 128if ($opt_M) {
 129        push (@mergerx, qr/$opt_M/);
 130}
 131
 132# Remember UTC of our starting time
 133# we'll want to avoid importing commits
 134# that are too recent
 135our $starttime = time();
 136
 137select(STDERR); $|=1; select(STDOUT);
 138
 139
 140package CVSconn;
 141# Basic CVS dialog.
 142# We're only interested in connecting and downloading, so ...
 143
 144use File::Spec;
 145use File::Temp qw(tempfile);
 146use POSIX qw(strftime dup2);
 147
 148sub new {
 149        my ($what,$repo,$subdir) = @_;
 150        $what=ref($what) if ref($what);
 151
 152        my $self = {};
 153        $self->{'buffer'} = "";
 154        bless($self,$what);
 155
 156        $repo =~ s#/+$##;
 157        $self->{'fullrep'} = $repo;
 158        $self->conn();
 159
 160        $self->{'subdir'} = $subdir;
 161        $self->{'lines'} = undef;
 162
 163        return $self;
 164}
 165
 166sub conn {
 167        my $self = shift;
 168        my $repo = $self->{'fullrep'};
 169        if ($repo =~ s/^:pserver(?:([^:]*)):(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?//) {
 170                my ($param,$user,$pass,$serv,$port) = ($1,$2,$3,$4,$5);
 171
 172                my ($proxyhost,$proxyport);
 173                if ($param && ($param =~ m/proxy=([^;]+)/)) {
 174                        $proxyhost = $1;
 175                        # Default proxyport, if not specified, is 8080.
 176                        $proxyport = 8080;
 177                        if ($ENV{"CVS_PROXY_PORT"}) {
 178                                $proxyport = $ENV{"CVS_PROXY_PORT"};
 179                        }
 180                        if ($param =~ m/proxyport=([^;]+)/) {
 181                                $proxyport = $1;
 182                        }
 183                }
 184
 185                $user="anonymous" unless defined $user;
 186                my $rr2 = "-";
 187                unless ($port) {
 188                        $rr2 = ":pserver:$user\@$serv:$repo";
 189                        $port=2401;
 190                }
 191                my $rr = ":pserver:$user\@$serv:$port$repo";
 192
 193                unless ($pass) {
 194                        open(H,$ENV{'HOME'}."/.cvspass") and do {
 195                                # :pserver:cvs@mea.tmt.tele.fi:/cvsroot/zmailer Ah<Z
 196                                while (<H>) {
 197                                        chomp;
 198                                        s/^\/\d+\s+//;
 199                                        my ($w,$p) = split(/\s/,$_,2);
 200                                        if ($w eq $rr or $w eq $rr2) {
 201                                                $pass = $p;
 202                                                last;
 203                                        }
 204                                }
 205                        };
 206                }
 207                $pass="A" unless $pass;
 208
 209                my ($s, $rep);
 210                if ($proxyhost) {
 211
 212                        # Use a HTTP Proxy. Only works for HTTP proxies that
 213                        # don't require user authentication
 214                        #
 215                        # See: http://www.ietf.org/rfc/rfc2817.txt
 216
 217                        $s = IO::Socket::INET->new(PeerHost => $proxyhost, PeerPort => $proxyport);
 218                        die "Socket to $proxyhost: $!\n" unless defined $s;
 219                        $s->write("CONNECT $serv:$port HTTP/1.1\r\nHost: $serv:$port\r\n\r\n")
 220                                or die "Write to $proxyhost: $!\n";
 221                        $s->flush();
 222
 223                        $rep = <$s>;
 224
 225                        # The answer should look like 'HTTP/1.x 2yy ....'
 226                        if (!($rep =~ m#^HTTP/1\.. 2[0-9][0-9]#)) {
 227                                die "Proxy connect: $rep\n";
 228                        }
 229                        # Skip up to the empty line of the proxy server output
 230                        # including the response headers.
 231                        while ($rep = <$s>) {
 232                                last if (!defined $rep ||
 233                                         $rep eq "\n" ||
 234                                         $rep eq "\r\n");
 235                        }
 236                } else {
 237                        $s = IO::Socket::INET->new(PeerHost => $serv, PeerPort => $port);
 238                        die "Socket to $serv: $!\n" unless defined $s;
 239                }
 240
 241                $s->write("BEGIN AUTH REQUEST\n$repo\n$user\n$pass\nEND AUTH REQUEST\n")
 242                        or die "Write to $serv: $!\n";
 243                $s->flush();
 244
 245                $rep = <$s>;
 246
 247                if ($rep ne "I LOVE YOU\n") {
 248                        $rep="<unknown>" unless $rep;
 249                        die "AuthReply: $rep\n";
 250                }
 251                $self->{'socketo'} = $s;
 252                $self->{'socketi'} = $s;
 253        } else { # local or ext: Fork off our own cvs server.
 254                my $pr = IO::Pipe->new();
 255                my $pw = IO::Pipe->new();
 256                my $pid = fork();
 257                die "Fork: $!\n" unless defined $pid;
 258                my $cvs = 'cvs';
 259                $cvs = $ENV{CVS_SERVER} if exists $ENV{CVS_SERVER};
 260                my $rsh = 'rsh';
 261                $rsh = $ENV{CVS_RSH} if exists $ENV{CVS_RSH};
 262
 263                my @cvs = ($cvs, 'server');
 264                my ($local, $user, $host);
 265                $local = $repo =~ s/:local://;
 266                if (!$local) {
 267                    $repo =~ s/:ext://;
 268                    $local = !($repo =~ s/^(?:([^\@:]+)\@)?([^:]+)://);
 269                    ($user, $host) = ($1, $2);
 270                }
 271                if (!$local) {
 272                    if ($user) {
 273                        unshift @cvs, $rsh, '-l', $user, $host;
 274                    } else {
 275                        unshift @cvs, $rsh, $host;
 276                    }
 277                }
 278
 279                unless ($pid) {
 280                        $pr->writer();
 281                        $pw->reader();
 282                        dup2($pw->fileno(),0);
 283                        dup2($pr->fileno(),1);
 284                        $pr->close();
 285                        $pw->close();
 286                        exec(@cvs);
 287                }
 288                $pw->writer();
 289                $pr->reader();
 290                $self->{'socketo'} = $pw;
 291                $self->{'socketi'} = $pr;
 292        }
 293        $self->{'socketo'}->write("Root $repo\n");
 294
 295        # Trial and error says that this probably is the minimum set
 296        $self->{'socketo'}->write("Valid-responses ok error Valid-requests Mode M Mbinary E Checked-in Created Updated Merged Removed\n");
 297
 298        $self->{'socketo'}->write("valid-requests\n");
 299        $self->{'socketo'}->flush();
 300
 301        chomp(my $rep=$self->readline());
 302        if ($rep !~ s/^Valid-requests\s*//) {
 303                $rep="<unknown>" unless $rep;
 304                die "Expected Valid-requests from server, but got: $rep\n";
 305        }
 306        chomp(my $res=$self->readline());
 307        die "validReply: $res\n" if $res ne "ok";
 308
 309        $self->{'socketo'}->write("UseUnchanged\n") if $rep =~ /\bUseUnchanged\b/;
 310        $self->{'repo'} = $repo;
 311}
 312
 313sub readline {
 314        my ($self) = @_;
 315        return $self->{'socketi'}->getline();
 316}
 317
 318sub _file {
 319        # Request a file with a given revision.
 320        # Trial and error says this is a good way to do it. :-/
 321        my ($self,$fn,$rev) = @_;
 322        $self->{'socketo'}->write("Argument -N\n") or return undef;
 323        $self->{'socketo'}->write("Argument -P\n") or return undef;
 324        # -kk: Linus' version doesn't use it - defaults to off
 325        if ($opt_k) {
 326            $self->{'socketo'}->write("Argument -kk\n") or return undef;
 327        }
 328        $self->{'socketo'}->write("Argument -r\n") or return undef;
 329        $self->{'socketo'}->write("Argument $rev\n") or return undef;
 330        $self->{'socketo'}->write("Argument --\n") or return undef;
 331        $self->{'socketo'}->write("Argument $self->{'subdir'}/$fn\n") or return undef;
 332        $self->{'socketo'}->write("Directory .\n") or return undef;
 333        $self->{'socketo'}->write("$self->{'repo'}\n") or return undef;
 334        # $self->{'socketo'}->write("Sticky T1.0\n") or return undef;
 335        $self->{'socketo'}->write("co\n") or return undef;
 336        $self->{'socketo'}->flush() or return undef;
 337        $self->{'lines'} = 0;
 338        return 1;
 339}
 340sub _line {
 341        # Read a line from the server.
 342        # ... except that 'line' may be an entire file. ;-)
 343        my ($self, $fh) = @_;
 344        die "Not in lines" unless defined $self->{'lines'};
 345
 346        my $line;
 347        my $res=0;
 348        while (defined($line = $self->readline())) {
 349                # M U gnupg-cvs-rep/AUTHORS
 350                # Updated gnupg-cvs-rep/
 351                # /daten/src/rsync/gnupg-cvs-rep/AUTHORS
 352                # /AUTHORS/1.1///T1.1
 353                # u=rw,g=rw,o=rw
 354                # 0
 355                # ok
 356
 357                if ($line =~ s/^(?:Created|Updated) //) {
 358                        $line = $self->readline(); # path
 359                        $line = $self->readline(); # Entries line
 360                        my $mode = $self->readline(); chomp $mode;
 361                        $self->{'mode'} = $mode;
 362                        defined (my $cnt = $self->readline())
 363                                or die "EOF from server after 'Changed'\n";
 364                        chomp $cnt;
 365                        die "Duh: Filesize $cnt" if $cnt !~ /^\d+$/;
 366                        $line="";
 367                        $res = $self->_fetchfile($fh, $cnt);
 368                } elsif ($line =~ s/^ //) {
 369                        print $fh $line;
 370                        $res += length($line);
 371                } elsif ($line =~ /^M\b/) {
 372                        # output, do nothing
 373                } elsif ($line =~ /^Mbinary\b/) {
 374                        my $cnt;
 375                        die "EOF from server after 'Mbinary'" unless defined ($cnt = $self->readline());
 376                        chomp $cnt;
 377                        die "Duh: Mbinary $cnt" if $cnt !~ /^\d+$/ or $cnt<1;
 378                        $line="";
 379                        $res += $self->_fetchfile($fh, $cnt);
 380                } else {
 381                        chomp $line;
 382                        if ($line eq "ok") {
 383                                # print STDERR "S: ok (".length($res).")\n";
 384                                return $res;
 385                        } elsif ($line =~ s/^E //) {
 386                                # print STDERR "S: $line\n";
 387                        } elsif ($line =~ /^(Remove-entry|Removed) /i) {
 388                                $line = $self->readline(); # filename
 389                                $line = $self->readline(); # OK
 390                                chomp $line;
 391                                die "Unknown: $line" if $line ne "ok";
 392                                return -1;
 393                        } else {
 394                                die "Unknown: $line\n";
 395                        }
 396                }
 397        }
 398        return undef;
 399}
 400sub file {
 401        my ($self,$fn,$rev) = @_;
 402        my $res;
 403
 404        my ($fh, $name) = tempfile('gitcvs.XXXXXX', 
 405                    DIR => File::Spec->tmpdir(), UNLINK => 1);
 406
 407        $self->_file($fn,$rev) and $res = $self->_line($fh);
 408
 409        if (!defined $res) {
 410            print STDERR "Server has gone away while fetching $fn $rev, retrying...\n";
 411            truncate $fh, 0;
 412            $self->conn();
 413            $self->_file($fn,$rev) or die "No file command send";
 414            $res = $self->_line($fh);
 415            die "Retry failed" unless defined $res;
 416        }
 417        close ($fh);
 418
 419        return ($name, $res);
 420}
 421sub _fetchfile {
 422        my ($self, $fh, $cnt) = @_;
 423        my $res = 0;
 424        my $bufsize = 1024 * 1024;
 425        while ($cnt) {
 426            if ($bufsize > $cnt) {
 427                $bufsize = $cnt;
 428            }
 429            my $buf;
 430            my $num = $self->{'socketi'}->read($buf,$bufsize);
 431            die "Server: Filesize $cnt: $num: $!\n" if not defined $num or $num<=0;
 432            print $fh $buf;
 433            $res += $num;
 434            $cnt -= $num;
 435        }
 436        return $res;
 437}
 438
 439
 440package main;
 441
 442my $cvs = CVSconn->new($opt_d, $cvs_tree);
 443
 444
 445sub pdate($) {
 446        my ($d) = @_;
 447        m#(\d{2,4})/(\d\d)/(\d\d)\s(\d\d):(\d\d)(?::(\d\d))?#
 448                or die "Unparseable date: $d\n";
 449        my $y=$1; $y-=1900 if $y>1900;
 450        return timegm($6||0,$5,$4,$3,$2-1,$y);
 451}
 452
 453sub pmode($) {
 454        my ($mode) = @_;
 455        my $m = 0;
 456        my $mm = 0;
 457        my $um = 0;
 458        for my $x(split(//,$mode)) {
 459                if ($x eq ",") {
 460                        $m |= $mm&$um;
 461                        $mm = 0;
 462                        $um = 0;
 463                } elsif ($x eq "u") { $um |= 0700;
 464                } elsif ($x eq "g") { $um |= 0070;
 465                } elsif ($x eq "o") { $um |= 0007;
 466                } elsif ($x eq "r") { $mm |= 0444;
 467                } elsif ($x eq "w") { $mm |= 0222;
 468                } elsif ($x eq "x") { $mm |= 0111;
 469                } elsif ($x eq "=") { # do nothing
 470                } else { die "Unknown mode: $mode\n";
 471                }
 472        }
 473        $m |= $mm&$um;
 474        return $m;
 475}
 476
 477sub getwd() {
 478        my $pwd = `pwd`;
 479        chomp $pwd;
 480        return $pwd;
 481}
 482
 483sub is_sha1 {
 484        my $s = shift;
 485        return $s =~ /^[a-f0-9]{40}$/;
 486}
 487
 488sub get_headref ($$) {
 489    my $name    = shift;
 490    my $git_dir = shift; 
 491    
 492    my $f = "$git_dir/refs/heads/$name";
 493    if (open(my $fh, $f)) {
 494            chomp(my $r = <$fh>);
 495            is_sha1($r) or die "Cannot get head id for $name ($r): $!";
 496            return $r;
 497    }
 498    die "unable to open $f: $!" unless $! == POSIX::ENOENT;
 499    return undef;
 500}
 501
 502-d $git_tree
 503        or mkdir($git_tree,0777)
 504        or die "Could not create $git_tree: $!";
 505chdir($git_tree);
 506
 507my $last_branch = "";
 508my $orig_branch = "";
 509my %branch_date;
 510my $tip_at_start = undef;
 511
 512my $git_dir = $ENV{"GIT_DIR"} || ".git";
 513$git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 514$ENV{"GIT_DIR"} = $git_dir;
 515my $orig_git_index;
 516$orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 517
 518my %index; # holds filenames of one index per branch
 519
 520unless (-d $git_dir) {
 521        system("git-init-db");
 522        die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 523        system("git-read-tree");
 524        die "Cannot init an empty tree: $?\n" if $?;
 525
 526        $last_branch = $opt_o;
 527        $orig_branch = "";
 528} else {
 529        -f "$git_dir/refs/heads/$opt_o"
 530                or die "Branch '$opt_o' does not exist.\n".
 531                       "Either use the correct '-o branch' option,\n".
 532                       "or import to a new repository.\n";
 533
 534        open(F, "git-symbolic-ref HEAD |") or
 535                die "Cannot run git-symbolic-ref: $!\n";
 536        chomp ($last_branch = <F>);
 537        $last_branch = basename($last_branch);
 538        close(F);
 539        unless ($last_branch) {
 540                warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 541                $last_branch = "master";
 542        }
 543        $orig_branch = $last_branch;
 544        $tip_at_start = `git-rev-parse --verify HEAD`;
 545
 546        # Get the last import timestamps
 547        my $fmt = '($ref, $author) = (%(refname), %(author));';
 548        open(H, "git-for-each-ref --perl --format='$fmt' refs/heads |") or
 549                die "Cannot run git-for-each-ref: $!\n";
 550        while (defined(my $entry = <H>)) {
 551                my ($ref, $author);
 552                eval($entry) || die "cannot eval refs list: $@";
 553                my ($head) = ($ref =~ m|^refs/heads/(.*)|);
 554                $author =~ /^.*\s(\d+)\s[-+]\d{4}$/;
 555                $branch_date{$head} = $1;
 556        }
 557        close(H);
 558}
 559
 560-d $git_dir
 561        or die "Could not create git subdir ($git_dir).\n";
 562
 563# now we read (and possibly save) author-info as well
 564-f "$git_dir/cvs-authors" and
 565  read_author_info("$git_dir/cvs-authors");
 566if ($opt_A) {
 567        read_author_info($opt_A);
 568        write_author_info("$git_dir/cvs-authors");
 569}
 570
 571
 572#
 573# run cvsps into a file unless we are getting
 574# it passed as a file via $opt_P
 575#
 576unless ($opt_P) {
 577        print "Running cvsps...\n" if $opt_v;
 578        my $pid = open(CVSPS,"-|");
 579        die "Cannot fork: $!\n" unless defined $pid;
 580        unless ($pid) {
 581                my @opt;
 582                @opt = split(/,/,$opt_p) if defined $opt_p;
 583                unshift @opt, '-z', $opt_z if defined $opt_z;
 584                unshift @opt, '-q'         unless defined $opt_v;
 585                unless (defined($opt_p) && $opt_p =~ m/--no-cvs-direct/) {
 586                        push @opt, '--cvs-direct';
 587                }
 588                exec("cvsps","--norc",@opt,"-u","-A",'--root',$opt_d,$cvs_tree);
 589                die "Could not start cvsps: $!\n";
 590        }
 591        my ($cvspsfh, $cvspsfile) = tempfile('gitXXXXXX', SUFFIX => '.cvsps',
 592                                             DIR => File::Spec->tmpdir());
 593        while (<CVSPS>) {
 594            print $cvspsfh $_;
 595        }
 596        close CVSPS;
 597        close $cvspsfh;
 598        $opt_P = $cvspsfile;
 599}
 600
 601
 602open(CVS, "<$opt_P") or die $!;
 603
 604## cvsps output:
 605#---------------------
 606#PatchSet 314
 607#Date: 1999/09/18 13:03:59
 608#Author: wkoch
 609#Branch: STABLE-BRANCH-1-0
 610#Ancestor branch: HEAD
 611#Tag: (none)
 612#Log:
 613#    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
 614#Members:
 615#       README:1.57->1.57.2.1
 616#       VERSION:1.96->1.96.2.1
 617#
 618#---------------------
 619
 620my $state = 0;
 621
 622sub update_index (\@\@) {
 623        my $old = shift;
 624        my $new = shift;
 625        open(my $fh, '|-', qw(git-update-index -z --index-info))
 626                or die "unable to open git-update-index: $!";
 627        print $fh
 628                (map { "0 0000000000000000000000000000000000000000\t$_\0" }
 629                        @$old),
 630                (map { '100' . sprintf('%o', $_->[0]) . " $_->[1]\t$_->[2]\0" }
 631                        @$new)
 632                or die "unable to write to git-update-index: $!";
 633        close $fh
 634                or die "unable to write to git-update-index: $!";
 635        $? and die "git-update-index reported error: $?";
 636}
 637
 638sub write_tree () {
 639        open(my $fh, '-|', qw(git-write-tree))
 640                or die "unable to open git-write-tree: $!";
 641        chomp(my $tree = <$fh>);
 642        is_sha1($tree)
 643                or die "Cannot get tree id ($tree): $!";
 644        close($fh)
 645                or die "Error running git-write-tree: $?\n";
 646        print "Tree ID $tree\n" if $opt_v;
 647        return $tree;
 648}
 649
 650my ($patchset,$date,$author_name,$author_email,$branch,$ancestor,$tag,$logmsg);
 651my (@old,@new,@skipped,%ignorebranch);
 652
 653# commits that cvsps cannot place anywhere...
 654$ignorebranch{'#CVSPS_NO_BRANCH'} = 1;
 655
 656sub commit {
 657        if ($branch eq $opt_o && !$index{branch} && !get_headref($branch, $git_dir)) {
 658            # looks like an initial commit
 659            # use the index primed by git-init-db
 660            $ENV{GIT_INDEX_FILE} = '.git/index';
 661            $index{$branch} = '.git/index';
 662        } else {
 663            # use an index per branch to speed up
 664            # imports of projects with many branches
 665            unless ($index{$branch}) {
 666                $index{$branch} = tmpnam();
 667                $ENV{GIT_INDEX_FILE} = $index{$branch};
 668                if ($ancestor) {
 669                    system("git-read-tree", $ancestor);
 670                } else {
 671                    system("git-read-tree", $branch);
 672                }
 673                die "read-tree failed: $?\n" if $?;
 674            }
 675        }
 676        $ENV{GIT_INDEX_FILE} = $index{$branch};
 677
 678        update_index(@old, @new);
 679        @old = @new = ();
 680        my $tree = write_tree();
 681        my $parent = get_headref($last_branch, $git_dir);
 682        print "Parent ID " . ($parent ? $parent : "(empty)") . "\n" if $opt_v;
 683
 684        my @commit_args;
 685        push @commit_args, ("-p", $parent) if $parent;
 686
 687        # loose detection of merges
 688        # based on the commit msg
 689        foreach my $rx (@mergerx) {
 690                next unless $logmsg =~ $rx && $1;
 691                my $mparent = $1 eq 'HEAD' ? $opt_o : $1;
 692                if (my $sha1 = get_headref($mparent, $git_dir)) {
 693                        push @commit_args, '-p', $mparent;
 694                        print "Merge parent branch: $mparent\n" if $opt_v;
 695                }
 696        }
 697
 698        my $commit_date = strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date));
 699        $ENV{GIT_AUTHOR_NAME} = $author_name;
 700        $ENV{GIT_AUTHOR_EMAIL} = $author_email;
 701        $ENV{GIT_AUTHOR_DATE} = $commit_date;
 702        $ENV{GIT_COMMITTER_NAME} = $author_name;
 703        $ENV{GIT_COMMITTER_EMAIL} = $author_email;
 704        $ENV{GIT_COMMITTER_DATE} = $commit_date;
 705        my $pid = open2(my $commit_read, my $commit_write,
 706                'git-commit-tree', $tree, @commit_args);
 707
 708        # compatibility with git2cvs
 709        substr($logmsg,32767) = "" if length($logmsg) > 32767;
 710        $logmsg =~ s/[\s\n]+\z//;
 711
 712        if (@skipped) {
 713            $logmsg .= "\n\n\nSKIPPED:\n\t";
 714            $logmsg .= join("\n\t", @skipped) . "\n";
 715            @skipped = ();
 716        }
 717
 718        print($commit_write "$logmsg\n") && close($commit_write)
 719                or die "Error writing to git-commit-tree: $!\n";
 720
 721        print "Committed patch $patchset ($branch $commit_date)\n" if $opt_v;
 722        chomp(my $cid = <$commit_read>);
 723        is_sha1($cid) or die "Cannot get commit id ($cid): $!\n";
 724        print "Commit ID $cid\n" if $opt_v;
 725        close($commit_read);
 726
 727        waitpid($pid,0);
 728        die "Error running git-commit-tree: $?\n" if $?;
 729
 730        system("git-update-ref refs/heads/$branch $cid") == 0
 731                or die "Cannot write branch $branch for update: $!\n";
 732
 733        if ($tag) {
 734                my ($in, $out) = ('','');
 735                my ($xtag) = $tag;
 736                $xtag =~ s/\s+\*\*.*$//; # Remove stuff like ** INVALID ** and ** FUNKY **
 737                $xtag =~ tr/_/\./ if ( $opt_u );
 738                $xtag =~ s/[\/]/$opt_s/g;
 739                
 740                my $pid = open2($in, $out, 'git-mktag');
 741                print $out "object $cid\n".
 742                    "type commit\n".
 743                    "tag $xtag\n".
 744                    "tagger $author_name <$author_email>\n"
 745                    or die "Cannot create tag object $xtag: $!\n";
 746                close($out)
 747                    or die "Cannot create tag object $xtag: $!\n";
 748
 749                my $tagobj = <$in>;
 750                chomp $tagobj;
 751
 752                if ( !close($in) or waitpid($pid, 0) != $pid or
 753                     $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 754                    die "Cannot create tag object $xtag: $!\n";
 755                }
 756                
 757
 758                open(C,">$git_dir/refs/tags/$xtag")
 759                        or die "Cannot create tag $xtag: $!\n";
 760                print C "$tagobj\n"
 761                        or die "Cannot write tag $xtag: $!\n";
 762                close(C)
 763                        or die "Cannot write tag $xtag: $!\n";
 764
 765                print "Created tag '$xtag' on '$branch'\n" if $opt_v;
 766        }
 767};
 768
 769my $commitcount = 1;
 770while (<CVS>) {
 771        chomp;
 772        if ($state == 0 and /^-+$/) {
 773                $state = 1;
 774        } elsif ($state == 0) {
 775                $state = 1;
 776                redo;
 777        } elsif (($state==0 or $state==1) and s/^PatchSet\s+//) {
 778                $patchset = 0+$_;
 779                $state=2;
 780        } elsif ($state == 2 and s/^Date:\s+//) {
 781                $date = pdate($_);
 782                unless ($date) {
 783                        print STDERR "Could not parse date: $_\n";
 784                        $state=0;
 785                        next;
 786                }
 787                $state=3;
 788        } elsif ($state == 3 and s/^Author:\s+//) {
 789                s/\s+$//;
 790                if (/^(.*?)\s+<(.*)>/) {
 791                    ($author_name, $author_email) = ($1, $2);
 792                } elsif ($conv_author_name{$_}) {
 793                        $author_name = $conv_author_name{$_};
 794                        $author_email = $conv_author_email{$_};
 795                } else {
 796                    $author_name = $author_email = $_;
 797                }
 798                $state = 4;
 799        } elsif ($state == 4 and s/^Branch:\s+//) {
 800                s/\s+$//;
 801                s/[\/]/$opt_s/g;
 802                $branch = $_;
 803                $state = 5;
 804        } elsif ($state == 5 and s/^Ancestor branch:\s+//) {
 805                s/\s+$//;
 806                $ancestor = $_;
 807                $ancestor = $opt_o if $ancestor eq "HEAD";
 808                $state = 6;
 809        } elsif ($state == 5) {
 810                $ancestor = undef;
 811                $state = 6;
 812                redo;
 813        } elsif ($state == 6 and s/^Tag:\s+//) {
 814                s/\s+$//;
 815                if ($_ eq "(none)") {
 816                        $tag = undef;
 817                } else {
 818                        $tag = $_;
 819                }
 820                $state = 7;
 821        } elsif ($state == 7 and /^Log:/) {
 822                $logmsg = "";
 823                $state = 8;
 824        } elsif ($state == 8 and /^Members:/) {
 825                $branch = $opt_o if $branch eq "HEAD";
 826                if (defined $branch_date{$branch} and $branch_date{$branch} >= $date) {
 827                        # skip
 828                        print "skip patchset $patchset: $date before $branch_date{$branch}\n" if $opt_v;
 829                        $state = 11;
 830                        next;
 831                }
 832                if ( $starttime - 300 - (defined $opt_z ? $opt_z : 300) <= $date) {
 833                        # skip if the commit is too recent
 834                        # that the cvsps default fuzz is 300s, we give ourselves another
 835                        # 300s just in case -- this also prevents skipping commits
 836                        # due to server clock drift
 837                        print "skip patchset $patchset: $date too recent\n" if $opt_v;
 838                        $state = 11;
 839                        next;
 840                }
 841                if (exists $ignorebranch{$branch}) {
 842                        print STDERR "Skipping $branch\n";
 843                        $state = 11;
 844                        next;
 845                }
 846                if ($ancestor) {
 847                        if ($ancestor eq $branch) {
 848                                print STDERR "Branch $branch erroneously stems from itself -- changed ancestor to $opt_o\n";
 849                                $ancestor = $opt_o;
 850                        }
 851                        if (-f "$git_dir/refs/heads/$branch") {
 852                                print STDERR "Branch $branch already exists!\n";
 853                                $state=11;
 854                                next;
 855                        }
 856                        unless (open(H,"$git_dir/refs/heads/$ancestor")) {
 857                                print STDERR "Branch $ancestor does not exist!\n";
 858                                $ignorebranch{$branch} = 1;
 859                                $state=11;
 860                                next;
 861                        }
 862                        chomp(my $id = <H>);
 863                        close(H);
 864                        unless (open(H,"> $git_dir/refs/heads/$branch")) {
 865                                print STDERR "Could not create branch $branch: $!\n";
 866                                $ignorebranch{$branch} = 1;
 867                                $state=11;
 868                                next;
 869                        }
 870                        print H "$id\n"
 871                                or die "Could not write branch $branch: $!";
 872                        close(H)
 873                                or die "Could not write branch $branch: $!";
 874                }
 875                $last_branch = $branch if $branch ne $last_branch;
 876                $state = 9;
 877        } elsif ($state == 8) {
 878                $logmsg .= "$_\n";
 879        } elsif ($state == 9 and /^\s+(.+?):(INITIAL|\d+(?:\.\d+)+)->(\d+(?:\.\d+)+)\s*$/) {
 880#       VERSION:1.96->1.96.2.1
 881                my $init = ($2 eq "INITIAL");
 882                my $fn = $1;
 883                my $rev = $3;
 884                $fn =~ s#^/+##;
 885                if ($opt_S && $fn =~ m/$opt_S/) {
 886                    print "SKIPPING $fn v $rev\n";
 887                    push(@skipped, $fn);
 888                    next;
 889                }
 890                print "Fetching $fn   v $rev\n" if $opt_v;
 891                my ($tmpname, $size) = $cvs->file($fn,$rev);
 892                if ($size == -1) {
 893                        push(@old,$fn);
 894                        print "Drop $fn\n" if $opt_v;
 895                } else {
 896                        print "".($init ? "New" : "Update")." $fn: $size bytes\n" if $opt_v;
 897                        my $pid = open(my $F, '-|');
 898                        die $! unless defined $pid;
 899                        if (!$pid) {
 900                            exec("git-hash-object", "-w", $tmpname)
 901                                or die "Cannot create object: $!\n";
 902                        }
 903                        my $sha = <$F>;
 904                        chomp $sha;
 905                        close $F;
 906                        my $mode = pmode($cvs->{'mode'});
 907                        push(@new,[$mode, $sha, $fn]); # may be resurrected!
 908                }
 909                unlink($tmpname);
 910        } elsif ($state == 9 and /^\s+(.+?):\d+(?:\.\d+)+->(\d+(?:\.\d+)+)\(DEAD\)\s*$/) {
 911                my $fn = $1;
 912                $fn =~ s#^/+##;
 913                push(@old,$fn);
 914                print "Delete $fn\n" if $opt_v;
 915        } elsif ($state == 9 and /^\s*$/) {
 916                $state = 10;
 917        } elsif (($state == 9 or $state == 10) and /^-+$/) {
 918                $commitcount++;
 919                if ($opt_L && $commitcount > $opt_L) {
 920                        last;
 921                }
 922                commit();
 923                if (($commitcount & 1023) == 0) {
 924                        system("git repack -a -d");
 925                }
 926                $state = 1;
 927        } elsif ($state == 11 and /^-+$/) {
 928                $state = 1;
 929        } elsif (/^-+$/) { # end of unknown-line processing
 930                $state = 1;
 931        } elsif ($state != 11) { # ignore stuff when skipping
 932                print "* UNKNOWN LINE * $_\n";
 933        }
 934}
 935commit() if $branch and $state != 11;
 936
 937# The heuristic of repacking every 1024 commits can leave a
 938# lot of unpacked data.  If there is more than 1MB worth of
 939# not-packed objects, repack once more.
 940my $line = `git-count-objects`;
 941if ($line =~ /^(\d+) objects, (\d+) kilobytes$/) {
 942  my ($n_objects, $kb) = ($1, $2);
 943  1024 < $kb
 944    and system("git repack -a -d");
 945}
 946
 947foreach my $git_index (values %index) {
 948    if ($git_index ne '.git/index') {
 949        unlink($git_index);
 950    }
 951}
 952
 953if (defined $orig_git_index) {
 954        $ENV{GIT_INDEX_FILE} = $orig_git_index;
 955} else {
 956        delete $ENV{GIT_INDEX_FILE};
 957}
 958
 959# Now switch back to the branch we were in before all of this happened
 960if ($orig_branch) {
 961        print "DONE.\n" if $opt_v;
 962        if ($opt_i) {
 963                exit 0;
 964        }
 965        my $tip_at_end = `git-rev-parse --verify HEAD`;
 966        if ($tip_at_start ne $tip_at_end) {
 967                for ($tip_at_start, $tip_at_end) { chomp; }
 968                print "Fetched into the current branch.\n" if $opt_v;
 969                system(qw(git-read-tree -u -m),
 970                       $tip_at_start, $tip_at_end);
 971                die "Fast-forward update failed: $?\n" if $?;
 972        }
 973        else {
 974                system(qw(git-merge cvsimport HEAD), "refs/heads/$opt_o");
 975                die "Could not merge $opt_o into the current branch.\n" if $?;
 976        }
 977} else {
 978        $orig_branch = "master";
 979        print "DONE; creating $orig_branch branch\n" if $opt_v;
 980        system("git-update-ref", "refs/heads/master", "refs/heads/$opt_o")
 981                unless -f "$git_dir/refs/heads/master";
 982        system('git-update-ref', 'HEAD', "$orig_branch");
 983        unless ($opt_i) {
 984                system('git checkout');
 985                die "checkout failed: $?\n" if $?;
 986        }
 987}