the road to new texdoc handling:

- allow tags "details" and "language" for docfiles
- allow catalogue-* keys in a TLPOBJ file
- TLPOBJ.pm: add update_from_catalogue which updates some information
- adapt the tlpsrc2tlpdb script to take another argument -catalogue
  in which case the database is build with catalogue information included


git-svn-id: svn://tug.org/texlive/trunk@4568 c570f23f-e606-0410-a88d-b1316a301751

10 files changed, 435 insertions, 82 deletions

diff --git a/Master/tlpkg/TeXLive/TLPDB.pm b/Master/tlpkg/TeXLive/TLPDB.pm
index 5d4d146957b..cc426df6dcb 100644
--- a/Master/tlpkg/TeXLive/TLPDB.pm
+++ b/Master/tlpkg/TeXLive/TLPDB.pm
@@ -245,11 +245,11 @@ sub _generate_listfile {
   } 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
+    # 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;
@@ -265,11 +265,11 @@ sub _generate_listfile {
     }
     # 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;
-			}
+      if (!TeXLive::TLUtils::member($p,@collpkg)) {
+        $s += $self->get_package($p)->total_size;
+      }
     } 
-		$s += $tlp->runsize + $tlp->srcsize + $tlp->docsize;
+    $s += $tlp->runsize + $tlp->srcsize + $tlp->docsize;
     print TMP "*Size: $s\n";
   }
   # dependencies and inclusion of packages
@@ -346,4 +346,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/TeXLive/TLPOBJ.pm b/Master/tlpkg/TeXLive/TLPOBJ.pm
index 814c2d7d314..425bdcd480f 100644
--- a/Master/tlpkg/TeXLive/TLPOBJ.pm
+++ b/Master/tlpkg/TeXLive/TLPOBJ.pm
@@ -8,10 +8,13 @@
 package TeXLive::TLPOBJ;
 
 use TeXLive::TLUtils;
+use TeXLive::TeXCatalogue;
 use File::Path;
 use Cwd;
 use TeXLive::TLTREE;
 use FileHandle;
+use Text::ParseWords;
+use File::Basename;
 
 my $_tmp;
 my $_zipdir;
@@ -37,6 +40,7 @@ sub new {
     binsize     => defined($params{'binsize'}) ? $params{'binsize'} : {},
     depends     => defined($params{'depends'}) ? $params{'depends'} : [],
     revision    => $params{'revision'},
+    cataloguedata   => defined($params{'cataloguedata'}) ? $params{'cataloguedata'} : {},
   };
   $_zipdir = $params{'zipdir'} if defined($params{'zipdir'});
   bless $self, $class;
@@ -61,7 +65,6 @@ sub from_fh {
 
   while (my $line = $fh->getline) {
     &TeXLive::TLUtils::debug("reading line: >>>$line<<<\n");
-    &TeXLive::TLUtils::debug("multi=...$multi...\n");
     $line =~ /^\s*#/ && next;          # skip comment lines
     if ($line =~ /^\s*$/) {
       if (!$started) { next; }
@@ -135,7 +138,24 @@ sub from_fh {
           $self->{"${type}size"} = $size;
         }
         next;
-      } elsif ($line =~ /^(doc|src|run)filescontinued\s+(.*)$/) {
+      } elsif ($line =~ /^docfilescontinued\s+(.*)$/) {
+        # docfiles can have tags
+        my @words = &quotewords('\s+', 0, "$1");
+        # the first entry is the file
+        my $f = shift @words;
+        push @{$self->{'docfiles'}}, $f;
+        while (@words) {
+          $_ = shift @words;
+          if (/^details=(.*)$/) {
+            $self->{'docfiledata'}{$f}{'details'} = "$1";
+          } elsif (/^language=(.*)$/) {
+            $self->{'docfiledata'}{$f}{'href'} = "$1";
+          } else {
+            die "Unknown docfile tag: $line";
+          }
+        }
+        next;
+      } elsif ($line =~ /^(src|run)filescontinued\s+(.*)$/) {
         push @{$self->{$1."files"}}, "$2";
         next;
       } elsif ($line =~ /^binfiles\s+/) {
@@ -168,6 +188,8 @@ sub from_fh {
         push @{$self->{'depends'}}, "$2" unless "$2" eq "";
         $lastcmd = "depend";
         next;
+      } elsif ($line =~ /^catalogue-([^\s]+)\s+(.*)$/) {
+        $self->{'cataloguedata'}{$1} = "$2";
       } else {
         die("Unknown directive ...$line... , please fix it!");
       }
@@ -237,6 +259,7 @@ sub writeout {
   defined($self->{'revision'}) && print $fd "revision $self->{'revision'}\n";
   defined($self->{'catalogue'}) && print $fd "catalogue $self->{'catalogue'}\n";
   defined($self->{'shortdesc'}) && print $fd "shortdesc $self->{'shortdesc'}\n";
+  defined($self->{'license'}) && print $fd "license $self->{'license'}\n";
   if (defined($self->{'longdesc'})) {
     $_tmp = "$self->{'longdesc'}";
     write $fd;
@@ -253,8 +276,15 @@ sub writeout {
   }
   if (defined($self->{'docfiles'}) && (@{$self->{'docfiles'}})) {
     print $fd "docfiles size=$self->{'docsize'}\n";
-    foreach (sort @{$self->{'docfiles'}}) {
-      print $fd " $_\n";
+    foreach my $f (sort @{$self->{'docfiles'}}) {
+      print $fd " $f";
+      if (defined($self->{'docfiledata'}{$f}{'details'})) {
+        print $fd ' details="', $self->{'docfiledata'}{$f}{'details'}, '"';
+      }
+      if (defined($self->{'docfiledata'}{$f}{'language'})) {
+        print $fd ' language="', $self->{'docfiledata'}{$f}{'language'}, '"';
+      }
+      print $fd "\n";
     }
   }
   if (defined($self->{'srcfiles'}) && (@{$self->{'srcfiles'}})) {
@@ -277,6 +307,10 @@ sub writeout {
       }
     }
   }
+  # writeout all the catalogue keys
+  foreach my $k (keys %{$self->cataloguedata}) {
+    print $fd "catalogue-$k ", $self->cataloguedata->{$k}, "\n";
+  }
 }
 
 sub writeout_simple {
@@ -381,24 +415,24 @@ sub _make_zip {
     return;
   }
   my $cwd = &getcwd;
-	if ("$destdir" !~ m@^/@) {
-		# we have an relative zipdir, so we have to make it absolute
-		$destdir = "$cwd/$destdir";
-	}
+  if ("$destdir" !~ m@^/@) {
+    # we have an relative zipdir, so we have to make it absolute
+    $destdir = "$cwd/$destdir";
+  }
   &mkpath("$destdir") if (! -d "$destdir");
   chdir($tltree->svnroot);
-	if ($arch eq "all") {
-		# we add the .tlpobj into the .tlpobj directory
-		my $removetlpobjdir = 0;
-		if (! -d ".tlpobj") {
-  		&mkpath(".tlpobj");
-			$removetlpobjdir = 1;
-		}
-		open(TMP,">.tlpobj/$self->{'name'}.tlpobj") or die "Cannot create .tlpobj/$self->{'name'}.tlpobj";
-		$self->writeout(\*TMP);
-		close(TMP);
-		push @files, ".tlpobj/$self->{'name'}.tlpobj";
-	}
+  if ($arch eq "all") {
+    # we add the .tlpobj into the .tlpobj directory
+    my $removetlpobjdir = 0;
+    if (! -d ".tlpobj") {
+      &mkpath(".tlpobj");
+      $removetlpobjdir = 1;
+    }
+    open(TMP,">.tlpobj/$self->{'name'}.tlpobj") or die "Cannot create .tlpobj/$self->{'name'}.tlpobj";
+    $self->writeout(\*TMP);
+    close(TMP);
+    push @files, ".tlpobj/$self->{'name'}.tlpobj";
+  }
   unlink($zipname);
   print $zipcmd . "$destdir/$zipname" . " > $nul\n" if ($::opt_debug);
   open ZIP, $zipcmd . "$destdir/$zipname" . " > $nul";
@@ -410,11 +444,11 @@ sub _make_zip {
     }
   } @files;
   close ZIP;
-	if ($arch eq "all") {
-		# cleaning up
-		unlink(".tlpobj/$self->{'name'}.tlpobj");
-		rmdir(".tlpobj") if $removetlpobjdir;
-	}
+  if ($arch eq "all") {
+    # cleaning up
+    unlink(".tlpobj/$self->{'name'}.tlpobj");
+    rmdir(".tlpobj") if $removetlpobjdir;
+  }
   print "Done $zipname\n" if ($::opt_debug);
 }
 
@@ -444,6 +478,57 @@ sub total_size {
   return($ret);
 }
 
+sub update_from_catalogue {
+  my ($self, $tlc) = @_;
+  my $tlcname = defined($self->catalogue) ? $self->catalogue : $self->name;
+  if (defined($tlc->entries->{$tlcname})) {
+    my $entry = $tlc->entries->{$tlcname};
+    if (defined($entry->license)) {
+      $self->cataloguedata->{'license'} = $entry->license;
+    }
+    if (defined($entry->version)) {
+      $self->cataloguedata->{'version'} = $entry->version;
+    }
+    if (defined($entry->url)) {
+      $self->cataloguedata->{'url'} = $entry->url;
+    }
+    # 
+    # we need to do the following:
+    # - take the href entry for a documentation file entry in the TC
+    # - remove the 'ctan:' prefix
+    # - remove the <ctan path='...'> part
+    # - match the rest against all docfiles in an intelligent way
+    #
+    # Example:
+    # juramisc.xml contains:
+    # <documentation details='Package documentation' language='de'
+    #   href='ctan:/macros/latex/contrib/juramisc/doc/jmgerdoc.pdf'/>
+    # <ctan path='/macros/latex/contrib/juramisc'/>
+    my @tcdocfiles = keys %{$entry->docs};
+    my @tlpdocfiles = $self->docfiles;
+    foreach my $tcdocfile (@tcdocfiles) {
+      # basename also kills the ctan: prefix!
+      my $tcdocfilebasename = $tcdocfile;
+      # remove the ctan: prefix
+      $tcdocfilebasename =~ s/^ctan://;
+      # remove the ctan path if present
+      my $tcctanpath = $entry->ctan;
+      $tcdocfilebasename =~ s/^$tcctanpath//;
+      foreach my $tlpdocfile (@tlpdocfiles) {
+        if ($tlpdocfile =~ m/$tcdocfilebasename$/) {
+          # update the language/detail tags if present!
+          if (defined($entry->docs->{$tcdocfile}{'details'})) {
+            $self->{'docfiledata'}{$tlpdocfile}{'details'} = $entry->docs->{$tcdocfile}{'details'};
+          }
+          if (defined($entry->docs->{$tcdocfile}{'language'})) {
+            $self->{'docfiledata'}{$tlpdocfile}{'language'} = $entry->docs->{$tcdocfile}{'language'};
+          }
+        }
+      }
+    }
+  }
+}
+
 #
 # member access functions
 #
@@ -589,6 +674,12 @@ sub zipdir {
   if (@_) { $_zipdir = $_[0] }
   return $_zipdir;
 }
+sub cataloguedata {
+  my $self = shift;
+  my %ct = @_;
+  if (@_) { $self->{'cataloguedata'} = \%ct }
+  return $self->{'cataloguedata'};
+}
 
 1;
 
@@ -623,8 +714,10 @@ files describing a self-contained package.
 Please see L<TeXLive::TLPSRC> documentation for the specification. The
 only differences are that the various C<*pattern> keys are invalid, and
 instead there are the respective C<*files> keys described below. Furthermore
-one more I<key> is allowed, C<revision> which specifies the maximum of 
-all last changed revision of files contained in the package.
+some more I<keys> is allowed: C<revision> which specifies the maximum of 
+all last changed revision of files contained in the package, and anything
+starting with C<catalogue-> specifying information coming from the
+TeX Catalogue.
 
 All these keys have in common that the are followed by a list of files
 B<intended> by one space. They differ only in the first line itself
@@ -632,15 +725,32 @@ B<intended> by one space. They differ only in the first line itself
 
 =over
 
-=item C<srcfiles>, C<docfiles>, C<runfiles>
+=item C<srcfiles>, C<runfiles>
 
 each of these three items contains in addition the sum of sizes of the
 single files (currently in byte, but this may change).
 
 	srcfiles size=NNNNNN
-	docfiles size=NNNNNN
 	runfiles size=NNNNNN
 
+=item C<docfiles>
+
+The docfiles line itself is similar to the C<srcfiles> and C<runfiles> lines
+above:
+
+	docfiles size=NNNNNN
+
+But the lines listing the files are allowed to have additional tags:
+
+	/------- excerpt from achemso.tlpobj
+	|...
+	|docfiles size=1702468
+	| texmf-dist/doc/latex/aeguill/README details="Package Readme"
+	| texmf-dist/doc/latex/achemso/achemso.pdf details="Package documentation" language="en"
+	|...
+
+Currently only the tags C<details> and C<language> are allowed.
+
 =item C<runfiles>
 
 Since C<runfiles> are different for the different architectures one C<tlpobj>
@@ -685,7 +795,12 @@ For any of the I<keys> a function
 	$tlpobj->key
 
 is available, which returns the current value when called without an argument,
-and sets the respective value when called with an argument.
+and sets the respective value when called with an argument. For the 
+TeX Catalogue Data the function
+
+	$tlpobj->cataloguedata
+
+returns and takes as argument a hash.
 
 Arguments and return values for C<name>, C<category>, C<shortdesc>,
 C<longdesc>, C<catalogue>, C<revision> are single scalars.
@@ -771,6 +886,12 @@ recomputes the sizes based on the information present in C<$tltree>.
 
 recomputes the revision based on the information present in C<$tltree>.
 
+=item C<update_from_catalogue ($texcatalogue)>
+
+adds information from a C<TeXCatalogue> object 
+(currently license, version, url, and updates docfiles with details and
+languages tags if present in the Catalogue)
+
 =item C<is_arch_dependent>
 
 returns 1 if there are C<binfiles>, otherwise 0.
@@ -799,7 +920,7 @@ adds or removes files for the given type (only for C<run>, C<src>, C<doc>).
 =head1 SEE ALSO
 
 The modules L<TeXLive::TLUtils>, L<TeXLive::TLPSRC>,
-L<TeXLive::TLPDB>, L<TeXLive::TLTREE>.
+L<TeXLive::TLPDB>, L<TeXLive::TLTREE>, L<TeXLive::TeXCatalogue>.
 
 =head1 AUTHORS AND COPYRIGHT
 
@@ -814,4 +935,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/TeXLive/TLPSRC.pm b/Master/tlpkg/TeXLive/TLPSRC.pm
index 7fdf16a3657..cf03fe819f0 100644
--- a/Master/tlpkg/TeXLive/TLPSRC.pm
+++ b/Master/tlpkg/TeXLive/TLPSRC.pm
@@ -664,4 +664,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/TeXLive/TLTREE.pm b/Master/tlpkg/TeXLive/TLTREE.pm
index 47202a9fcd0..bf007b9cfb2 100644
--- a/Master/tlpkg/TeXLive/TLTREE.pm
+++ b/Master/tlpkg/TeXLive/TLTREE.pm
@@ -45,11 +45,11 @@ sub init_from_statusfile {
   $self->_initialize_lines(@lines);
 }
 sub init_from_files {
-	my $self = shift;
-	my $svnroot = $self->{'svnroot'};
-	my @lines = `find $svnroot ! -wholename '*/.svn*'`;
-	@lines = map { s@^$svnroot@@; s@^/@@; "             1 1 dummy $_" } @lines;
-	$self->_initialize_lines(@lines);
+  my $self = shift;
+  my $svnroot = $self->{'svnroot'};
+  my @lines = `find $svnroot ! -wholename '*/.svn*'`;
+  @lines = map { s@^$svnroot@@; s@^/@@; "             1 1 dummy $_" } @lines;
+  $self->_initialize_lines(@lines);
 }
 
 sub _initialize_lines {
@@ -73,7 +73,7 @@ sub _initialize_lines {
       add_path_to_tree($self->{'_dirtree'}, split("[/\\\\]", $dn));
       push @{$self->{'_filesofdir'}{$dn}}, $fn;
     } else {
-		  warn("Ignoring svn status output line:\n    $l\n");
+      warn("Ignoring svn status output line:\n    $l\n");
       #die("Cannot read svn status output line:\n    $l\n");
     }
   }
diff --git a/Master/tlpkg/TeXLive/TLUtils.pm b/Master/tlpkg/TeXLive/TLUtils.pm
index 86f98e2de03..2a939443b81 100644
--- a/Master/tlpkg/TeXLive/TLUtils.pm
+++ b/Master/tlpkg/TeXLive/TLUtils.pm
@@ -133,4 +133,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/TeXLive/TeXCatalogue.pm b/Master/tlpkg/TeXLive/TeXCatalogue.pm
index 25aae3074d9..33ef33e6a7d 100644
--- a/Master/tlpkg/TeXLive/TeXCatalogue.pm
+++ b/Master/tlpkg/TeXLive/TeXCatalogue.pm
@@ -1,21 +1,168 @@
 # $Id$
 # TeXLive::TeXCatalogue - module for accessing the TeX Catalogue
+#
 # Copyright 2007 Norbert Preining
 #
+# Loads of code is taken from the catalogue checking script of Robin Fairbairns
+#
 # This file is licensed under the GNU General Public Licence version 2
 # or any later version.
 
-die "THIS MODULE IS NOT EVEN CLOSE TO BE FINISHED\n";
+use XML::DOM;
 
-package TeXLive::TeXCatalogue;
+package TeXLive::TeXCatalogue::Entry;
 
-use XML::Parser;
+sub new {
+  my $class = shift;
+  my %params = @_;
+  my $self = {
+    xmlfile => $params{'xmlfile'},
+    entry => defined($params{'entry'}) ? $params{'entry'} : {},
+    docs => defined($params{'docs'}) ? $params{'docs'} : {},
+    name => $params{'name'},
+    caption => $params{'caption'},
+    ctan => $params{'ctan'},
+    texlive => $params{'texlive'},
+    version => $params{'version'},
+    url => $params{'url'},
+  };
+  bless $self, $class;
+  if (defined($self->{'xmlfile'})) {
+    $self->initialize();
+  }
+  return $self;
+}
+
+sub initialize {
+  my $self = shift;
+  my $parser = new XML::DOM::Parser;
+  # parse all the files
+  my $cat = $parser->parsefile ($self->{'xmlfile'}) || 
+    die "Failed to parse ", $self->{'xmlfile'};
+  # there's almost always only one, but...
+  my $entryset = $cat -> getElementsByTagName ("entry");
+  my $entry = $entryset->item(0);
+  # entry attributes
+  my $entryid = $entry->getAttributeNode("id");
+  $self->{'entry'}{'id'} = $entryid->getValue;
+  my $entrydate = $entry->getAttributeNode("datestamp");
+  if ($entrydate) {
+    $self->{'entry'}{'date'} = $entrydate->getValue;
+  }
+  my $entrymodder = $entry->getAttributeNode("modifier");
+  if ($entrymodder) {
+    $self->{'entry'}{'modder'} = $entrymodder->getValue;
+  }
+  # parse the rest of the entries
+  # only the first name and caption is allowed
+  # description CANNOT be parsed like this since it contains more childs!!
+  # (<tt> <xref> etc elements
+  foreach my $k qw/name caption/ {
+    $self->{$k} = $cat->getElementsByTagName($k)->item(0)->getFirstChild->toString;
+  }
+  $self->{'license'} = $cat->getElementsByTagName("license")->item(0)->getAttributeNode("type")->getValue;
+  my $versnode = $cat->getElementsByTagName("version")->item(0);
+  if ($versnode) {
+    my $numnode = $versnode->getAttributeNode("number");
+    if ($numnode) {
+      $self->{'version'} = $numnode->getValue;
+    }
+  }
+  my $urlnode = $cat->getElementsByTagName("home")->item(0);
+  if ($urlnode) { 
+    $self->{'url'} = $urlnode->getAttributeNode("href")->getValue;
+  }
+  $ctannode = $cat->getElementsByTagName("ctan")->item(0);
+  if ($ctannode) {
+    $self->{'ctan'} = $ctannode->getAttributeNode("path")->getValue;
+  }
+  my $tlnode = $cat->getElementsByTagName("texlive")->item(0);
+  if ($tlnode) {
+    $self->{'texlive'} = $tlnode->getAttributeNode("location")->getValue;
+  }
+  # parse the documentation entries
+  my $doc = $cat->getElementsByTagName ("documentation");
+  for (my $i=0; $i < $doc->getLength; $i++ ) {
+    my $node = $doc->item($i);
+    my $docfile = $node->getAttributeNode("href")->getValue;
+    my $detailsnode = $node->getAttributeNode("details");
+    if ($detailsnode) { $details = $detailsnode->getValue; }
+    my $languagenode = $node->getAttributeNode("language");
+    if ($languagenode) { $language = $languagenode->getValue; }
+    $self->{'docs'}{$docfile}{'available'} = 1;
+    if ($details) { $self->{'docs'}{$docfile}{'details'} = $details; }
+    if ($language) { $self->{'docs'}{$docfile}{'language'} = $language; }
+  }
+  $cat->dispose;
+}
+
+sub xmlfile {
+  my $self = shift;
+  if (@_) { $self->{'xmlfile'} = shift }
+  return $self->{'xmlfile'};
+}
+sub name {
+  my $self = shift;
+  if (@_) { $self->{'name'} = shift }
+  return $self->{'name'};
+}
+sub license {
+  my $self = shift;
+  if (@_) { $self->{'license'} = shift }
+  return $self->{'license'};
+}
+sub version {
+  my $self = shift;
+  if (@_) { $self->{'version'} = shift }
+  return $self->{'version'};
+}
+sub caption {
+  my $self = shift;
+  if (@_) { $self->{'caption'} = shift }
+  return $self->{'caption'};
+}
+sub url {
+  my $self = shift;
+  if (@_) { $self->{'url'} = shift }
+  return $self->{'url'};
+}
+sub ctan {
+  my $self = shift;
+  if (@_) { $self->{'ctan'} = shift }
+  return $self->{'ctan'};
+}
+sub texlive {
+  my $self = shift;
+  if (@_) { $self->{'texlive'} = shift }
+  return $self->{'texlive'};
+}
+sub docs {
+  my $self = shift;
+  my %newdocs = @_;
+  if (@_) { $self->{'docs'} = \%newdocs }
+  return $self->{'docs'};
+}
+sub entry {
+  my $self = shift;
+  my %newentry = @_;
+  if (@_) { $self->{'entry'} = \%newentry }
+  return $self->{'entry'};
+}
+
+
+################################################################
+#
+# TeXLive::TeXCatalogue
+#
+################################################################
+package TeXLive::TeXCatalogue;
 
 sub new { 
   my $class = shift;
   my %params = @_;
   my $self = {
     location => $params{'location'},
+    entries => defined($params{'entries'}) ? $params{'entries'} : {},
   };
   bless $self, $class;
   if (defined($self->{'location'})) {
@@ -25,10 +172,26 @@ sub new {
 }
 
 sub initialize {
-	my $class = shift;
-	die "Not done yet ...\n";
+  my $self = shift;
+  my $parser = new XML::DOM::Parser;
+  # parse all the files
+  foreach (glob($self->{'location'} . "/entries/?/*.xml")) {
+    my $tce = TeXLive::TeXCatalogue::Entry->new( 'xmlfile' => "$_" );
+    $self->{'entries'}{$tce->name} = $tce;
+  }
 }
 
+sub location {
+  my $self = shift;
+  if (@_) { $self->{'location'} = shift }
+  return $self->{'location'};
+}
+sub entries {
+  my $self = shift;
+  my %newentries = @_;
+  if (@_) { $self->{'entries'} = \%newentries }
+  return $self->{'entries'};
+}
 
 1;
 
@@ -48,6 +211,8 @@ missing
 The L<TeXLive::TeXCatalogue> module provides access to the data stored
 in the TeX Catalogue.
 
+DOCUMENTATION MISSING, SORRY!!!
+
 =head1 AUTHORS AND COPYRIGHT
 
 This script and its documentation were written for the TeX Live
@@ -61,4 +226,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/bin/tlpsrc2tlpdb b/Master/tlpkg/bin/tlpsrc2tlpdb
index 6989de9be76..a496ad7183f 100755
--- a/Master/tlpkg/bin/tlpsrc2tlpdb
+++ b/Master/tlpkg/bin/tlpsrc2tlpdb
@@ -19,6 +19,7 @@ use TeXLive::TLPSRC;
 use TeXLive::TLPOBJ;
 use TeXLive::TLPDB;
 use TeXLive::TLTREE;
+use TeXLive::TeXCatalogue;
 use Getopt::Long;
 use Pod::Usage;
 use File::Path;
@@ -28,29 +29,44 @@ chomp (my $opt_master = `cd $::mydir/../.. && pwd`);
 our $opt_debug = 0;
 my $man = 0;
 my $help = 0;
+my $opt_catalogue = "";
 
 GetOptions("debug!",	# debug mode
-		"master=s" => \$opt_master,	# location of the tree
-		'help|?' => \$help, man => \$man
-	) or pod2usage(2);
+    "master=s" => \$opt_master,	# location of the tree
+    "catalogue=s" => \$opt_catalogue, # location of the TeX Catalogue
+    'help|?' => \$help, man => \$man
+  ) or pod2usage(2);
 
 pod2usage(1) if $help;
 pod2usage(-exitstatus => 0, -verbose => 2) if $man;
 
 if (! -d "$opt_master") {
-	die "Master $opt_master does not exists!";
+  die "Master $opt_master does not exists!";
+}
+
+my $tlc;
+if (! -d "$opt_catalogue") {
+  warn "Catalogue not found, we will continue without using it!";
+} else {
+  $tlc = TeXLive::TeXCatalogue->new( 'location' => "$opt_catalogue" );
 }
 
 my $tltree = TeXLive::TLTREE->new( 'svnroot' => "$opt_master" );
 $tltree->init_from_svn;
 
+
 my $tldb = TeXLive::TLPDB->new;
 
 foreach my $f (@ARGV) {
-	my $tlsrc = new TeXLive::TLPSRC;
-	$tlsrc->from_file($f);
-	my $tlp = $tlsrc->make_tlpobj($tltree);
-	$tldb->add_tlpobj($tlp);
+  my $tlsrc = new TeXLive::TLPSRC;
+  $tlsrc->from_file($f);
+  my $tlp = $tlsrc->make_tlpobj($tltree);
+  # now merge the TeX Catalogue Information, currently only license
+  my $tlcname = defined($tlp->catalogue) ? $tlp->catalogue : $tlp->name;
+  if ($tlc) {
+    $tlp->update_from_catalogue($tlc);
+  }
+  $tldb->add_tlpobj($tlp);
 }
 
 $tldb->writeout;
@@ -79,6 +95,10 @@ Prints the manual page and exits.
 The location given by B<-master> must point to a valid svn repository
 of TeX Live Master direcory. Defaults to ../.. of C<$0>.
 
+=item B<-catalogue>
+The location given by B<-catalogue> must point to a valid TeX Catalogue.
+No default value.
+
 =item B<-debug>
 Give debug messages from the TeX Live modules
 
@@ -87,7 +107,10 @@ Give debug messages from the TeX Live modules
 =head1 DESCRIPTION
 
 B<tlpsrc2tlpdb> converts TeX Live Source Package files (tlpsrc) files into a
-TeX Live Package Database (tlpdb).
+TeX Live Package Database (tlpdb). If a TeX Catalogue is specified additional
+information from the catalogue are merged into the TeX Live Database 
+(see L<TeXLive::TLPOBJ> for the list of included information from
+the TeX Catalogue).
 
 =head1 AUTHORS AND COPYRIGHT
 
@@ -102,4 +125,4 @@ GNU General Public License Version 2 or later.
 ### tab-width: 2
 ### indent-tabs-mode: nil
 ### End:
-# vim:set tabstop=2: #
+# vim:set tabstop=2 expandtab: #
diff --git a/Master/tlpkg/doc/Perl-API.txt b/Master/tlpkg/doc/Perl-API.txt
index 6859acfa6ea..b3e2eb139b5 100644
--- a/Master/tlpkg/doc/Perl-API.txt
+++ b/Master/tlpkg/doc/Perl-API.txt
@@ -86,6 +86,9 @@ Input/Output
 
 		(Compatibility with the scheme used before)
 	
+	$tlp->update_from_catalogue($texcatalogue);
+		merges some information from the TeX Catalogue into 
+		the TLPOBJ
 
 Utilities:
 	$tlp->recompute_sizes($tltree)
diff --git a/Master/tlpkg/doc/specification.txt b/Master/tlpkg/doc/specification.txt
index 18c79d83f90..6233c95e084 100644
--- a/Master/tlpkg/doc/specification.txt
+++ b/Master/tlpkg/doc/specification.txt
@@ -120,6 +120,7 @@ same format as tlpsrc, but the keys are
 	binfiles
 	execute
 	depend
+	catalogue-*	(whatever)
 	??? what am I missing ???
 
 Interpretation
@@ -146,6 +147,10 @@ Interpretation
 	revision
 		maxmimum of all the last_changed_revisions of all
 		contained files
+	
+	catalogue-*
+		can contain arbitrary fields from the TeX Catalogue
+
 
 3) tlpdb file
 ------------
diff --git a/Master/tlpkg/man/TeXLive-TLPOBJ.1 b/Master/tlpkg/man/TeXLive-TLPOBJ.1
index e9d47d3f9cf..5ce64690b55 100644
--- a/Master/tlpkg/man/TeXLive-TLPOBJ.1
+++ b/Master/tlpkg/man/TeXLive-TLPOBJ.1
@@ -128,8 +128,8 @@
 .rm #[ #] #H #V #F C
 .\" ========================================================================
 .\"
-.IX Title "TeXLive::TLPOBJ 3"
-.TH TeXLive::TLPOBJ 3 "2007-06-27" "perl v5.8.8" "User Contributed Perl Documentation"
+.IX Title "TLPOBJ 3"
+.TH TLPOBJ 3 "2007-07-12" "perl v5.8.8" "User Contributed Perl Documentation"
 .SH "NAME"
 TeXLive::TLPOBJ \- TeX Live Package Object access module
 .SH "SYNOPSIS"
@@ -150,23 +150,46 @@ files describing a self-contained package.
 Please see TeXLive::TLPSRC documentation for the specification. The
 only differences are that the various \f(CW*pattern\fR keys are invalid, and
 instead there are the respective \f(CW*files\fR keys described below. Furthermore
-one more \fIkey\fR is allowed, \f(CW\*(C`revision\*(C'\fR which specifies the maximum of 
-all last changed revision of files contained in the package.
+some more \fIkeys\fR is allowed: \f(CW\*(C`revision\*(C'\fR which specifies the maximum of 
+all last changed revision of files contained in the package, and anything
+starting with \f(CW\*(C`catalogue\-\*(C'\fR specifying information coming from the
+TeX Catalogue.
 .PP
 All these keys have in common that the are followed by a list of files
 \&\fBintended\fR by one space. They differ only in the first line itself
 (described below).
-.ie n .IP """srcfiles""\fR, \f(CW""docfiles""\fR, \f(CW""runfiles""" 4
-.el .IP "\f(CWsrcfiles\fR, \f(CWdocfiles\fR, \f(CWrunfiles\fR" 4
-.IX Item "srcfiles, docfiles, runfiles"
+.ie n .IP """srcfiles""\fR, \f(CW""runfiles""" 4
+.el .IP "\f(CWsrcfiles\fR, \f(CWrunfiles\fR" 4
+.IX Item "srcfiles, runfiles"
 each of these three items contains in addition the sum of sizes of the
 single files (currently in byte, but this may change).
 .Sp
-.Vb 3
+.Vb 2
 \&        srcfiles size=NNNNNN
-\&        docfiles size=NNNNNN
 \&        runfiles size=NNNNNN
 .Ve
+.ie n .IP """docfiles""" 4
+.el .IP "\f(CWdocfiles\fR" 4
+.IX Item "docfiles"
+The docfiles line itself is similar to the \f(CW\*(C`srcfiles\*(C'\fR and \f(CW\*(C`runfiles\*(C'\fR lines
+above:
+.Sp
+.Vb 1
+\&        docfiles size=NNNNNN
+.Ve
+.Sp
+But the lines listing the files are allowed to have additional tags:
+.Sp
+.Vb 6
+\&        /\-\-\-\-\-\-\- excerpt from achemso.tlpobj
+\&        |...
+\&        |docfiles size=1702468
+\&        | texmf\-dist/doc/latex/aeguill/README details="Package Readme"
+\&        | texmf\-dist/doc/latex/achemso/achemso.pdf details="Package documentation" language="en"
+\&        |...
+.Ve
+.Sp
+Currently only the tags \f(CW\*(C`details\*(C'\fR and \f(CW\*(C`language\*(C'\fR are allowed.
 .ie n .IP """runfiles""" 4
 .el .IP "\f(CWrunfiles\fR" 4
 .IX Item "runfiles"
@@ -215,7 +238,14 @@ For any of the \fIkeys\fR a function
 .Ve
 .PP
 is available, which returns the current value when called without an argument,
-and sets the respective value when called with an argument.
+and sets the respective value when called with an argument. For the 
+TeX Catalogue Data the function
+.PP
+.Vb 1
+\&        $tlpobj\->cataloguedata
+.Ve
+.PP
+returns and takes as argument a hash.
 .PP
 Arguments and return values for \f(CW\*(C`name\*(C'\fR, \f(CW\*(C`category\*(C'\fR, \f(CW\*(C`shortdesc\*(C'\fR,
 \&\f(CW\*(C`longdesc\*(C'\fR, \f(CW\*(C`catalogue\*(C'\fR, \f(CW\*(C`revision\*(C'\fR are single scalars.
@@ -305,6 +335,12 @@ recomputes the sizes based on the information present in \f(CW$tltree\fR.
 .el .IP "\f(CWrecompute_revision ($tltree)\fR" 4
 .IX Item "recompute_revision ($tltree)"
 recomputes the revision based on the information present in \f(CW$tltree\fR.
+.ie n .IP """update_from_catalogue ($texcatalogue)""" 4
+.el .IP "\f(CWupdate_from_catalogue ($texcatalogue)\fR" 4
+.IX Item "update_from_catalogue ($texcatalogue)"
+adds information from a \f(CW\*(C`TeXCatalogue\*(C'\fR object 
+(currently license, version, url, and updates docfiles with details and
+languages tags if present in the Catalogue)
 .ie n .IP """is_arch_dependent""" 4
 .el .IP "\f(CWis_arch_dependent\fR" 4
 .IX Item "is_arch_dependent"
@@ -328,12 +364,12 @@ adds or removes files from the list of \f(CW\*(C`binfiles\*(C'\fR for the given
 .el .IP "\f(CW{add,remove}_files ($type, $files)\fR" 4
 .IX Item "{add,remove}_files ($type, $files)"
 adds or removes files for the given type (only for \f(CW\*(C`run\*(C'\fR, \f(CW\*(C`src\*(C'\fR, \f(CW\*(C`doc\*(C'\fR).
-.SH "AUTHORS AND COPYRIGHT"
-.IX Header "AUTHORS AND COPYRIGHT"
-This module and its documentation was written by
-Norbert Preining <\fIpreining@logic.at\fR> for the TeX Live distribution and both
-are licensed under the \s-1GNU\s0 General Public License Version 2 or later.
 .SH "SEE ALSO"
 .IX Header "SEE ALSO"
 The modules TeXLive::TLUtils, TeXLive::TLPSRC,
-TeXLive::TLPDB, TeXLive::TLTREE.
+TeXLive::TLPDB, TeXLive::TLTREE, TeXLive::TeXCatalogue.
+.SH "AUTHORS AND COPYRIGHT"
+.IX Header "AUTHORS AND COPYRIGHT"
+This script and its documentation were written for the TeX Live
+distribution (<http://tug.org/texlive>) and both are licensed under the
+\&\s-1GNU\s0 General Public License Version 2 or later.