# $Id$
# TeXLive::TLPDB.pm - module for using tlpdb files
# Copyright 2007 Norbert Preining
#
# This file is licensed under the GNU General Public License version 2
# or any later version.

package TeXLive::TLPDB;

=pod

=head1 NAME

C<TeXLive::TLPDB> -- A database of TeX Live Packages

=head1 SYNOPSIS

  use TeXLive::TLPDB;

  TeXLive::TLPDB->new ();
  TeXLive::TLPDB->new (root => "/path/to/texlive/installation/root");

  $tlpdb->root("/path/to/root/of/texlive/installation");
  $tlpdb->from_file($filename);
  $tlpdb->writeout;
  $tlpdb->writeout(FILEHANDLE);
  $tlpdb->save;
  $tlpdb->available_architectures();
  $tlpdb->add_tlpcontainer($pkg, $ziploc [, $archrefs [, $dest ]] );
  $tlpdb->add_tlpobj($tlpobj);
  $tlpdb->get_package("packagename");
  $tlpdb->list_packages;
  $tlpdb->updmap_cfg_lines;
  $tlpdb->fmtutil_cnf_lines;
  $tlpdb->language_dat_lines;
  $tlpdb->package_revision("packagename");

  TeXLive::TLPDB->listdir([$dir]);
  $tlpdb->generate_listfiles([$destdir]);

=head1 DESCRIPTION

=cut

use TeXLive::TLConfig qw($CategoriesRegexp $DefaultCategory $InfraLocation
      $DatabaseName);
use TeXLive::TLUtils qw(dirname);
use TeXLive::TLPOBJ;

my $_listdir;

=pod

=over 6

=item C<< TeXLive::TLPDB->new >>

=item C<< TeXLive::TLPDB->new( [root => "$path"] ) >>

C<< TeXLive::TLPDB->new >> creates a new C<TLPDB> object. If the
argument C<root> is given it will be initialized from the respective
location within $path. If
C<$path> begins with C<http://> or C<ftp://>, the program C<wget>
is used to download the file.

=cut

sub new { 
  my $class = shift;
  my %params = @_;
  my $self = {
    root => $params{'root'},
    tlps => $params{'tlps'}
  };
  $_listdir = $params{'listdir'} if defined($params{'listdir'});
  bless $self, $class;
  if (defined($self->{'root'})) {
    $self->from_file("$self->{'root'}/$InfraLocation/$DatabaseName");
  }
  return $self;
}

=pod

=item C<< $tlpdb->add_tlpobj($tlpobj) >>

The C<add_tlpobj> adds an object of the type TLPOBJ to the TLPDB.

=cut

sub add_tlpobj {
  my ($self,$tlp) = @_;
  $self->{'tlps'}{$tlp->name} = $tlp;
}

=pod

=item C<< $tlpdb->from_file($filename) >>

The C<from_file> function initializes the C<TLPDB> in case the
location was not given at generation time.  If C<$filename> begins
with C<http://> or C<ftp://>, the program C<wget> is used to download
the file.

=cut

sub from_file {
  my ($self, $path) = @_;
  if (@_ != 2) {
    die("Need a filename for initialization!");
  }
  my $root_from_path = dirname(dirname($path));
  if (defined($self->{'root'})) {
    if ($self->{'root'} ne $root_from_path) {
      printf STDERR "root=$self->{'root'}, root_from_path=$root_from_path\n";
      printf STDERR "Initialisation from different location as originally given.\nHope you are sure!\n";
    }
  } else {
    $self->root($root_from_path);
  }
  if ($path =~ /^(http|ftp):\/\//) {
    open(TMP, "wget -nv --output-document=- $path|") 
        || die("Cannot open tlpdb file: $path"); 
  } else {
    open(TMP,"<$path") || die("Cannot open tlpdb file: $path");
  }
  my $found = 0;
  do {
    my $tlp = TeXLive::TLPOBJ->new;
    &TeXLive::TLUtils::debug("creating tlp from $path...");
    $found = $tlp->from_fh(\*TMP,1);
    if ($found) {
      $self->add_tlpobj($tlp);
    }
  } until (!$found);
}

=pod

=item C<< $tlpdb->writeout >>

=item C<< $tlpdb->writeout(FILEHANDLE) >>

The C<writeout> function writes the database to C<STDOUT>, or 
the file handle given as argument.

=cut

sub writeout {
  my $self = shift;
  my $fd = (@_ ? $_[0] : STDOUT);
  foreach (sort keys %{$self->{'tlps'}}) {
    &TeXLive::TLUtils::debug("tlpname = $_");
    &TeXLive::TLUtils::debug("foo: " . $self->{'tlps'}{$_}->name);
    $self->{'tlps'}{$_}->writeout($fd);
    print $fd "\n";
  }
}

=pod

=item C<< $tlpdb->save >>

The C<save> functions saves the C<TLPDB> to the file which has been set
as location. If the location is undefined, die.

=cut

sub save {
  my $self = shift;
  my $path = "$self->{'root'}/$InfraLocation/$DatabaseName";
  open(FOO,">$path") || die("Cannot open $path for writing: $!");
  $self->writeout(\*FOO);
  close(FOO);
}

=pod

=item C<< $tlpdb->available_architectures >>

The C<available_architectures> functions returns the list of available 
architectures by checking for binfiles in the "bin-tex" TLPOBJ. If this
package does not exist, die.

=cut

sub available_architectures {
  my $self = shift;
  my @packs = $self->list_packages;
  my @archs;
  map { s@^$tltree/@@ ; push @nf, $_; } @files;
  map { s/^bin-tex\.// ; push @archs, $_ ; } grep(/^bin-tex\.(.*)$/, @packs);
  return @archs;
}

=pod

=item C<< $tlpdb->add_tlpcontainer($pkg, $ziploc [, $archrefs [, $dest ]] ) >>

Installs the package C<$pkg> from the container files in C<$ziploc>. If
C<$archrefs> is given then it must be a reference to a list of 
architectures to be installed. If the normal (arch=all) package is
architecture dependent then all arch packages in this list are installed.
If C<$dest> is given then the files are
installed into it, otherwise into the location of the TLPDB.

Note that this procedure does NOT check for dependencies. So if your package
adds new dependencies they are not necessarily fulfilled.

=cut

sub add_tlpcontainer {
  my ($self, $package, $ziplocation, $archrefs, $dest) = @_;
  my @archs;
  require Cwd;
  if (defined($archrefs)) {
    @archs = @$archrefs;
  }
  my $cwd = getcwd();
  if ($ziplocation !~ m,^/,) {
    $ziplocation = "$cwd/$ziplocation";
  }
  my $tlpobj = $self->_add_tlpcontainer($package, $ziplocation, "all", $dest);
  if ($tlpobj->is_arch_dependent) {
    foreach (@$archrefs) {
      $self->_add_tlpcontainer($package, $ziplocation, $_, $dest);
    }
  }
}

sub _add_tlpcontainer {
  my ($self, $package, $ziplocation, $arch, $dest) = @_;
  my $unpackprog;
  my $args;
  require Cwd;
  # WARNING: If you change the location of the texlive.tlpdb this
  # has to be changed, too!!
  if (not(defined($dest))) { 
    $dest = $self->{'root'};
  }
  my $container = "$ziplocation/$package";
  if ($arch ne "all") {
    $container .= ".$arch";
  }
  if (-r "$container.zip") {
    $container .= ".zip";
    $unpackprog="unzip";
    $args="-o -qq $container -d $dest";
  } elsif (-r "$container.lzma") {
    $container .= ".lzma";
    $unpackprog="NO_IDEA_HOW_TO_UNPACK_LZMA";
    $args="NO IDEA WHAT ARGS IT NEEDS";
    die("lzma is checked for but not implemented, please edit TLPDB.pm\n");
  } else {
    die "Cannot find a package $container (.zip or .lzma) in $ziplocation\n";
  }
  warn "Huuu, this needs testing and error checking!\n";
  warn "Should we use -a -- adapt line endings etc?\n";
  `$unpackprog $args`;
  # we only create/add tlpobj for arch eq "all"
  if ($arch eq "all") {
    my $tlpobj = new TeXLive::TLPOBJ;
    $tlpobj->from_file("$dest/$TeXLive::TLConfig::InfraLocation/tlpobj/$package.tlpobj");
    $self->add_tlpobj($tlpobj);
    return $tlpobj;
  }
}


=pod

=item C<< $tlpdb->get_package("packagename") >> 

The C<get_package> function returns a reference to a C<TLPOBJ> object
in case its name the the argument name coincide.

=cut

sub get_package {
  my ($self,$pkg) = @_;
  if (defined($self->{'tlps'}{$pkg})) {
    return($self->{'tlps'}{$pkg});
  } else {
    return(undef);
  }
}

=pod

=item C<< $tlpdb->list_packages >>

The C<list_packages> function returns the list of all included packages.

=cut

sub list_packages {
  my $self = shift;
  return (sort keys %{$self->{'tlps'}});
}

=pod

=item C<< $tlpdb->package_revision("packagename") >>

The C<package_revision> function returns the revision number of the
package named in the first argument.

=cut

sub package_revision {
  my ($self,$pkg) = @_;
  if (defined($self->{'tlps'}{$pkg})) {
    return($self->{'tlps'}{$pkg}->revision);
  } else {
    return(undef);
  }
}

=pod

=item C<< $tlpdb->generate_packagelist >>

The C<generate_packagelist> lists all available TL Packages together
with their revisions. It furthermore lists all available architectures 
as packages with revision number -1.

=cut

sub generate_packagelist {
  my $self = shift;
  my $fd = (@_ ? $_[0] : STDOUT);
  foreach (sort keys %{$self->{'tlps'}}) {
    print $fd $self->{'tlps'}{$_}->name, " ", $self->{'tlps'}{$_}->revision, "\n";
  }
  foreach ($self->available_architectures) {
    print $fd "$_ -1\n";
  }
}

=pod

=item C<< $tlpdb->generate_listfiles >>

=item C<< $tlpdb->generate_listfiles($destdir) >>

The C<generate_listfiles> generates the list files for the old 
installers. This function will probably go away.

=cut

sub generate_listfiles {
  my ($self,$destdir) = @_;
  if (not(defined($destdir))) {
    $destdir = TeXLive::TLPDB->listdir;
  }
  foreach (sort keys %{$self->{'tlps'}}) {
    $tlp = $self->{'tlps'}{$_};
    $self->_generate_listfile($tlp, $destdir);
  }
}

sub _generate_listfile {
  my ($self,$tlp,$destdir) = @_;
  my $listname = $tlp->name;
  my @files = $tlp->all_files;
  @files = TeXLive::TLUtils::sort_uniq(@files);
  &mkpath("$destdir") if (! -d "$destdir");
  my (@lop, @lot);
  foreach my $d ($tlp->depends) {
    my $subtlp = $self->get_package($d);
    if (defined($subtlp)) {
      if ($subtlp->is_meta_package) {
        push @lot, $d;
      } else {
        push @lop, $d;
      }
    } else {
      # speudo dependencies on $Package.ARCH can be ignored
      if ($d !~ m/\.ARCH$/) {
        warn "Strange: $tlp->name depends on $d, but this package does not exists!\n";
      }
    }
  }
  open(TMP, ">$destdir/$listname") or die "Cannot open $destdir/$listname!";
  # title and size information for collections and schemes in the
  # first two lines, marked with *
	if ($tlp->category eq "Collection") {
    print TMP "*Title: ", $tlp->shortdesc, "\n";
    # collections references Packages, we have to collect the sizes of
    # all the Package-tlps included
    # What is unclear for me is HOW the size is computed for bin-*
    # packages. The collection-basicbin contains quite a lot of
    # bin-files, but the sizes for the different archs differ.
    # I guess we have to take the maximum?
    my $s = 0;
    foreach my $p (@lop) {
      my $subtlp = $self->get_package($p);
      if (!defined($subtlp)) {
        warn "Strange: $listname references $p, but it is not in tlpdb";
      }
      $s += $subtlp->total_size;
    }
    # in case the collection itself ships files ...
    $s += $tlp->runsize + $tlp->srcsize + $tlp->docsize;
    print TMP "*Size: $s\n";
  } elsif ($tlp->category eq "Scheme") {
    print TMP "*Title: ", $tlp->shortdesc, "\n";
    my $s = 0;
    # schemes size includes ONLY those packages which are directly
    # included and direclty included files, not the size of the
    # included collections. But if a package is included in one of
    # the called for collections AND listed directly, we don't want
    # to count its size two times
    my (@inccol,@incpkg,@collpkg);
    # first we add all the packages tlps that are directly included
    @incpkg = @lop;
    # now we select all collections, and for all collections we
    # again select all packages of type Documentation and Package
    foreach my $c (@lot) {
      my $coll = $self->get_package($c);
      foreach my $d ($coll->depends) {
        my $subtlp = $self->get_package($d);
        if (defined($subtlp)) {
          if (!($subtlp->is_meta_package)) {
            TeXLive::TLUtils::push_uniq(\@collpkg,$d);
          }
        } else {
          warn "Strange: $coll->name depends on $d, but this does not exists!\n";
        }
      }
    }
    # finally go through all packages and add the ->total_size
    foreach my $p (@incpkg) {
      if (!TeXLive::TLUtils::member($p,@collpkg)) {
        $s += $self->get_package($p)->total_size;
      }
    } 
    $s += $tlp->runsize + $tlp->srcsize + $tlp->docsize;
    print TMP "*Size: $s\n";
  }
  # dependencies and inclusion of packages
  foreach my $t (@lot) {
    # strange, schemes mark included collections via -, while collections
    # themself mark deps on other collections with +. collection are
    # never referenced in Packages
    if ($listname =~ m/^scheme/) {
      print TMP "-";
    } else {
      print TMP "+";
    }
    print TMP "$t\n";
  }
  foreach my $t (@lop) { print TMP "+$t\n"; }
  # included files
  foreach my $f (@files) { print TMP "$f\n"; }
  # also print the listfile itself
  print TMP "$destdir/$listname\n";
  # execute statements
  foreach my $e ($tlp->executes) {
    print TMP "!$e\n";
  }
  # finish
  close(TMP);
}

=pod

=item C<< $tlpdb->root([ "/path/to/installation" ]) >>

The function C<root> allows to read and set the root of the
installation. 

=cut

sub root {
  my $self = shift;
  if (@_) { $self->{'root'} = shift }
  return $self->{'root'};
}

=pod

=item C<< $tlpdb->listdir >>

The function C<listdir> allows to read and set the packages variable
specifiying where generated list files are created.

=cut
sub listdir {
  my $self = shift;
  if (@_) { $_listdir = $_[0] }
  return $_listdir;
}

=back


=item C<< $tlpdb->fmtutil_cnf_lines >>

The function C<fmtutil_cnf_lines> returns the list of a fmtutil.cnf file
containing only those formats present in the installation.

=cut
sub fmtutil_cnf_lines {
  my $self = shift;
  my %fmtcnffiles;
  foreach my $p ($self->list_packages) {
    my $obj = $self->get_package ($p);
    die "No TeX Live package named $p, strange.\n" if ! $obj;
    foreach my $e ($obj->executes) {
      if ($e =~ m/BuildFormat (.*)$/) {
        $fmtcnffiles{$1} = 1;
      } 
      # others are ignored here
    }
  }
  my @formatlines;
  open(INFILE,"<$self->{'root'}/texmf/fmtutil/fmtutil-hdr.cnf")
    or die("Cannot open $self->{'root'}/texmf/fmtutil/fmtutil-hdr.cnf\n");
  my @tmp = <INFILE>;
  close (INFILE);
  push @formatlines, @tmp;
  foreach my $f (keys %fmtcnffiles) {
    open(INFILE,"<$self->{'root'}/texmf/fmtutil/format.$f.cnf")
      or die("Cannot open $self->{'root'}/texmf/fmtutil/format.$f.cnf");
    @tmp = <INFILE>;
    close(INFILE);
    push @formatlines, @tmp;
  }
  return(@formatlines);
}

=item C<< $tlpdb->updmap_cfg_lines >>

The function C<updmap_cfg_lines> returns the list of a updmap.cfg file
containing only those maps present in the installation.

=cut
sub updmap_cfg_lines {
  my $self = shift;
  my %maps;
  foreach my $p ($self->list_packages) {
    my $obj = $self->get_package ($p);
    die "No TeX Live package named $p, strange.\n" if ! $obj;
    foreach my $e ($obj->executes) {
      if ($e =~ m/addMap (.*)$/) {
        $maps{$1} = 1;
      } elsif ($e =~ m/addMixedMap (.*)$/) {
        $maps{$1} = 2;
      }
      # others are ignored here
    }
  }
  my @updmaplines;
  open(UPDHDR,"<$self->{'root'}/texmf/web2c/updmap-hdr.cfg") 
    or die("Cannot find $self->{'root'}/texmf/web2c/updmap-hdr.cfg!\n");
  my @tmp = <UPDHDR>;
  close(UPDHDR);
  push @updmaplines, @tmp;
  foreach (sort keys %maps) {
    if ($maps{$_} == 2) {
      push @updmaplines, "MixedMap $_\n";
    } else {
      push @updmaplines, "Map $_\n";
    }
  }
  return(@updmaplines);
}

=item C<< $tlpdb->language_dat_lines >>

The function C<language_dat_lines> returns the list of a language.dat file
containing only those hyphenation patterns present in the installation.

=cut
sub language_dat_lines {
  my $self = shift;
  my %langfiles;
  foreach my $p ($self->list_packages) {
    my $obj = $self->get_package ($p);
    die "No TeX Live package named $p, strange.\n" if ! $obj;
    foreach my $e ($obj->executes) {
      if ($e =~ m/BuildLanguageDat (.*)$/) {
      $langfiles{$1} = 1;
      }
      # others are ignored here
    }
  }
  my @langlines;
  open(LANHDR,"<$self->{'root'}/texmf/tex/generic/config/language.us") 
    or die("Cannot find $self{'root'}/texmf/tex/generic/config/language.us!\n");
  my @tmp = <LANHDR>;
  close(LANHDR);
  push @langlines, @tmp;
  foreach my $f (keys %langfiles) {
    open(INFILE,"<$self->{'root'}/texmf/tex/generic/config/language.$f.dat")
      or die("Cannot open $self->{'root'}/texmf/tex/generic/config/language.$f.cnf");
    @tmp = <INFILE>;
    close(INFILE);
    push @langlines, @tmp;
  }
  return(@langlines);
}


=back

=pod

=head1 SEE ALSO

The modules L<TeXLive::TLPSRC>, L<TeXLive::TLPOBJ>, 
L<TeXLive::TLTREE>, L<TeXLive::TLUtils> and the
document L<Perl-API.txt> and the specification in the TeX Live
repository trunk/Master/tlpkg/doc/.

=head1 AUTHORS AND COPYRIGHT

This script and its documentation were written for the TeX Live
distribution (L<http://tug.org/texlive>) and both are licensed under the
GNU General Public License Version 2 or later.

=cut

1;

### Local Variables:
### perl-indent-level: 2
### tab-width: 2
### indent-tabs-mode: nil
### End:
# vim:set tabstop=2 expandtab: #