X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fhttp_getter%2Fhttp_getter.pl.in;h=e9d1f4a3e04609137ad68b8db5658759be797f81;hb=4e683370099ea7aaea877bc345ca8821938d49ed;hp=9c71fce076a1e4cd6e69f95f2713e269c1fd5d3d;hpb=9b03d3f69b91921f10e764d1a465880248c14e55;p=helm.git

diff --git a/helm/http_getter/http_getter.pl.in b/helm/http_getter/http_getter.pl.in
index 9c71fce07..e9d1f4a3e 100755
--- a/helm/http_getter/http_getter.pl.in
+++ b/helm/http_getter/http_getter.pl.in
@@ -30,7 +30,7 @@ my $VERSION = "@VERSION@";
 # various variables
 my ($HELM_LIB_PATH);
 my $cgi_dir = "@HELM_CGI_DIR@";
-my (%map, %rdf_map);
+my (%xml_map, %rdf_map, %xslt_map);
 
 # First of all, let's load HELM configuration
 use Env;
@@ -43,23 +43,28 @@ if (defined ($HELM_LIB_DIR)) {
    $HELM_LIB_PATH = $DEFAULT_HELM_LIB_DIR."/configuration.pl";
 }
 
-# <ZACK>: TODO temporary, move this setting to configuration file
-# set the cache mode, may be "gzipped" or "normal"
-my $cachemode = $ENV{'HTTP_GETTER_CACHE_MODE'} || 'gzipped';
-if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) {
- die "Invalid HTTP_GETTER_CACHE_MODE environment variable, must be".
- "'normal' or 'gzipped'\n";
-}
-# </ZACK>
-
 # next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm, $indexname
 require $HELM_LIB_PATH;
 
 # TEMP: TODO put these vars in configuration file configuration.xml
 # <move_to_conf_file>
-$helm_rdf_dir = "/usr/local/helm/rdf_library";
-$rdf_dbm = "/usr/local/helm/rdf_urls_of_uris";
-$rdf_indexname = "rdf_index.txt";
+my $cachemode = $ENV{'HTTP_GETTER_CACHE_MODE'} ||
+    'gzipped';
+if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) {
+ die "Invalid HTTP_GETTER_CACHE_MODE environment variable, must be".
+ "'normal' or 'gzipped'\n";
+}
+
+$helm_rdf_dir = $ENV{'HTTP_GETTER_RDF_DIR'} || $helm_rdf_dir;
+$rdf_dbm = $ENV{'HTTP_GETTER_RDF_DBM'} || $rdf_dbm;
+$xslt_dbm = $ENV{'HTTP_GETTER_XSLT_DBM'} || $xslt_dbm;
+$rdf_indexname = $ENV{'HTTP_GETTER_RDF_INDEXNAME'} || $rdf_indexname;
+$xslt_indexname = $ENV{'HTTP_GETTER_XSLT_INDEXNAME'} || $xslt_indexname;
+$servers_file = $ENV{'HTTP_GETTER_SERVERS_FILE'} || $servers_file;
+$uris_dbm = $ENV{'HTTP_GETTER_URIS_DBM'} || $uris_dbm;
+$dtdbaseurl = $ENV{'HTTP_GETTER_DTD_BASE_URL'} || $dtdbaseurl;
+$getterport = $ENV{'HTTP_GETTER_PORT'} || $getterport;
+  
 # </move_to_conf_file>
 
 # Let's override the configuration file
@@ -73,13 +78,15 @@ use LWP::UserAgent;
 use DB_File;
 use Compress::Zlib;
 use CGI;
+use URI;
 use URI::Escape;
 
 #CSC: mancano i controlli sulle condizioni di errore di molte funzioni
 #CSC: ==> non e' robusto
 #CSC: altra roba da sistemare segnata con CSC
 
-my $d = new HTTP::Daemon LocalPort => 8081;
+my $d = new HTTP::Daemon LocalPort => $getterport
+ or die "Error: port $getterport not available.";
 my $myownurl = $d->url;
 
 # Let's patch the returned URL
@@ -90,22 +97,28 @@ $myownurl  =~ s/http:\/\/(.*):(.*)/$1/;
 ($myownurl) = gethostbyname($myownurl);
 $myownurl = "http://".$myownurl.":".$myownport;
 
-tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # open dbs
-tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
+tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # open dbs
+tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
+tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
 
 print "\n";
 print "HTTP Getter $VERSION\n"; # print hello information
 print "Please contact me at: <URL:", $myownurl."/", ">\n";
 print "\n";
+print "servers_file: $servers_file\n";
 print "helm_dir: $helm_dir\n";
 print "helm_rdf_dir: $helm_rdf_dir\n";
 print "style_dir: $style_dir\n";
 print "dtd_dir: $dtd_dir\n";
-print "urls_of_uris db: $uris_dbm.db\n";
-print "rdf db: $rdf_dbm.db\n";
+print "urls_of_uris db: $uris_dbm\n";
+print "rdf db: $rdf_dbm\n";
+print "xslt db: $xslt_dbm\n";
 print "cache mode: $cachemode\n";
 print "indexname: $indexname\n";
 print "rdf_indexname: $rdf_indexname\n";
+print "xslt_indexname: $xslt_indexname\n";
+print "dtdbaseurl: $dtdbaseurl\n";
+print "getterport: $getterport\n";
 print "\n";
 
 $SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
@@ -144,12 +157,20 @@ while (my $c = $d->accept) {
             }
 
             my $filename = $inputuri;
+            my $prefix;
             if (not isRdfUri($inputuri)) { # standad cic: or theory: uri
              $filename =~ s/^cic:(.*)/$1/;
              $filename =~ s/^theory:(.*)/$1/;
             } else { # rdf uri
+             # The "helm:rdf/<<...>>//cic:" prefix is kept, but quoted
+             # (so that a "helm:rdf/<<...>>//cic:" dir is created
+             $prefix = $filename;
              $filename =~ s/^(.*)\/\/cic:(.*)/$2/;
              $filename =~ s/^(.*)\/\/theory:(.*)/$2/;
+             $prefix =~ s/\Q$filename\E//;
+             $prefix =~ s/_/__/g;
+             $prefix =~ s/\//_/g;
+             $filename = $prefix.$filename;
             }
 
             my $url = resolve ($inputuri); # resolve uri in url
@@ -170,7 +191,7 @@ while (my $c = $d->accept) {
               if (not isRdfUri ($inputuri)) { # save in uri std cache dir
                $filename = $helm_dir.$filename.$extension;
               } else { # save in rdf metadata cache dir
-               $filename = $helm_rdf_dir.$filename.$extension;
+               $filename = $helm_rdf_dir."/".$filename.$extension;
               }
 
               print_request($inputuri,$url,$filename);
@@ -189,7 +210,7 @@ while (my $c = $d->accept) {
         } elsif ($http_method eq 'GET' and $http_path eq "/register") {
           my $inputurl = $cgi->param('url');
           print "Register requested...\n";
-          $map{$inputuri}=$inputurl;
+          $xml_map{$inputuri}=$inputurl;
 
           # Now let's clean the cache
           my $cicfilename = $inputuri;
@@ -201,13 +222,13 @@ while (my $c = $d->accept) {
           unlink ($helm_dir.$cicfilename.".xml.gz");
 
           kill(USR1,getppid()); # signal changes to parent
-          untie %map;
+          untie %xml_map;
           print "done.\n";
           html_nice_answer($c,"Register done");
 
                                   # "/resolve" works with rdf uri
         } elsif ($http_method eq 'GET' and $http_path eq "/resolve") {
-          #my $outputurl = $map{$inputuri};
+          #my $outputurl = $xml_map{$inputuri};
           my $outputurl = resolve($inputuri);
           $outputurl = "" if (not defined($outputurl));
           $cont = "<?xml version=\"1.0\" ?>\n\n";
@@ -219,6 +240,7 @@ while (my $c = $d->accept) {
           answer($c,$cont,"text/xml","");
 
         } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
+
             my $filename = $inputuri;
             $filename = $dtd_dir."/".$filename;
             print "DTD: $inputuri ==> ($filename)\n";
@@ -238,42 +260,48 @@ while (my $c = $d->accept) {
                die "Could not find DTD!";
             }
 
-        } elsif ($http_method eq 'GET' and $http_path eq "/getstyleconf") {
-            my $filename = $inputuri;
-            $filename = $style_dir."/config/".$filename;
-            if (stat($filename)) {
-               open(FD, $filename) or die "Cannot open $filename\n";
-               $cont = "";
-               while(<FD>) {
-	        s/DOCTYPE (.*) SYSTEM\s+"/DOCTYPE $1 SYSTEM "$myownurl\/getstyleconf?uri=/g;
-	        $cont .= $_;
-	       }
-               close(FD);
-               answer($c,$cont,"text/plain","");
-            } else {
-               die "Could not find Style Configuration File!";
-            }
-
         } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
-            my $filename = $inputuri;
-            $filename = $style_dir."/".$filename;
-            print "XSLT: $inputuri ==> ($filename)\n";
-            if (stat($filename)) {
-               print "Using local copy\n";
-               open(FD, $filename) or die "Cannot open $filename\n";
-               $cont = "";
-               while(<FD>) {
-                # Vary bad heuristic for substituion of absolute URLs
-                # for relative ones
-                s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
-                s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
-                $cont .= $_;
-               }
-               close(FD);
-               answer($c,$cont,"text/xml","");
-            } else {
-               die "Could not find XSLT!";
-            }
+
+          $url = $xslt_map{$inputuri};
+          print "Downloading XSLT ...\n";
+          helm_wget($tmp_dir, $url);
+          print "XSLT downloaded!\n";
+          $inputuri =~ s/.*\///;
+          my $filename = $tmp_dir . "/" . $inputuri;
+          my $cont;
+          die "Could not find XSLT!\n" unless stat($filename);
+          open(FD, $filename) or die "Can't open xslt local copy: '$filename'\n";
+          while(<FD>) {
+            s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g;
+            s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g;
+            $cont .= $_;
+          }
+          close(FD);
+          answer($c, $cont, "text/xml", "");
+          unlink $filename or
+            print "Can't unlink XSLT temp file '$filename'";
+
+# <old_version>
+#             my $filename = $inputuri;
+#             $filename = $style_dir."/".$filename;
+#             print "XSLT: $inputuri ==> ($filename)\n";
+#             if (stat($filename)) {
+#                print "Using local copy\n";
+#                open(FD, $filename) or die "Cannot open $filename\n";
+#                $cont = "";
+#                while(<FD>) {
+#                 # Vary bad heuristic for substituion of absolute URLs
+#                 # for relative ones
+#                 s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
+#                 s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
+#                 $cont .= $_;
+#                }
+#                close(FD);
+#                answer($c,$cont,"text/xml","");
+#             } else {
+#                die "Could not find XSLT!";
+#             }
+# </old_version>
 
                                     # "/update" works with rdf uri
         } elsif ($http_method eq 'GET' and $http_path eq "/update") {
@@ -284,7 +312,21 @@ while (my $c = $d->accept) {
            print " done\n";
            html_nice_answer($c,"Update done");
 
-                                 # "/ls" does not work with rdf uris
+                                 # "/getalluris" does not work with rdf uris
+        } elsif ($http_method eq 'GET' and $http_path eq "/getalluris") {
+            # send back all the keys in xml
+            print "BASEURI $baseuri, FORMAT $outype\n";
+            $cont = getalluris();
+            answer($c,$cont,"text/xml","");
+
+                                 # "/getallrdfuris"
+        } elsif ($http_method eq 'GET' and $http_path eq "/getallrdfuris") {
+            # send back all the keys in xml
+            my $class = $cgi->param('class');
+            print "BASEURI $baseuri, FORMAT $outype, CLASS $class\n";
+            $cont = getallrdfuris($class);
+            answer($c,$cont,"text/xml","");
+
         } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
             # send back keys that begin with a given uri
            my ($uritype,$uripattern,$uriprefix);
@@ -329,6 +371,17 @@ while (my $c = $d->accept) {
            print "Help requested!\n";
            html_nice_answer($c,"HTTP Getter Version: $VERSION");
 
+        } elsif ($http_method eq 'GET' and $http_path eq "/getempty") {
+          # return an empty, but valid, xml document
+          my $emptyxml = <<EOXML;
+<?xml version="1.0"?>
+<!DOCTYPE empty [
+  <!ELEMENT empty EMPTY>
+]>
+<empty />
+EOXML
+          answer($c, $emptyxml, "", "");
+
         } elsif ($http_method eq 'GET' and $http_path =~ /\.cgi$/) {
           # cgi handling
            print "CGI requested!\n";
@@ -398,11 +451,54 @@ sub resolve { # resolve an uri in a url, work both with standard cic: or theory:
   return ($rdf_map{$uri});
  } else { # standard cic: or theory: uri, resolve using std uri db
   print "IS NOT A RDF URI\n";
-  print "I WILL RETURN '$map{$uri}'\n";
-  return ($map{$uri});
+  print "I WILL RETURN '$xml_map{$uri}'\n";
+  return ($xml_map{$uri});
  }
 }
 
+sub getalluris { # get all the keys whose prefix is cic
+ my $content = "";
+ my ($uri);
+ my $debug=1; # for debug
+
+  $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
+  $content .= "<!DOCTYPE alluris SYSTEM ";
+  $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
+  $content .= "<alluris>\n";
+  foreach $uri (sort (keys(%xml_map))) { # select matching uris
+   if ($uri =~ /^cic:/ && not $uri =~ /.types$/) {
+    print "GETALLURI: $uri\n" if defined($debug);
+    $content .= "\t<uri value=\"$uri\"/>\n";
+   }
+  }
+  $content .= "</alluris>\n";
+ return $content;
+}
+
+sub getallrdfuris {
+ my $class = $_[0];
+ my $content = "";
+ my ($uri);
+ my $debug=1; # for debug
+
+  $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
+  $content .= "<!DOCTYPE allrdfuris SYSTEM ";
+  $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
+  $content .= "<allrdfuris>\n";
+  foreach $uri (sort (keys(%rdf_map))) {
+   if ($class eq "forward" &&
+        $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/forward/
+       ||
+        $class eq "backward" &&
+         $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/backward/) {
+      print "GETALLRDFURI: $uri\n" if defined($debug);
+      $content .= "\t<uri value=\"$uri\"/>\n";
+   }
+  }
+  $content .= "</allrdfuris>\n";
+ return $content;
+}
+
 sub finduris { # find uris for cic and theory trees generation
  my ($uritype,$uripattern,$format) = @_;
  my $content = "";
@@ -416,7 +512,7 @@ sub finduris { # find uris for cic and theory trees generation
  
  if (($uritype eq "cic") or ($uritype eq "theory")) {
    # get info only of one type: cic or theory
-  foreach (keys(%map)) { # select matching uris
+  foreach (keys(%xml_map)) { # select matching uris
    $uri = $_;
    if ($uri =~ /^$uritype:$uripattern(\/|$|\.)/) {
     if ($uri =~ /^$uritype:$uripattern\//) { # directory match
@@ -433,9 +529,9 @@ sub finduris { # find uris for cic and theory trees generation
 
     if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
      $basepart = $localpart;
-     $basepart =~ s/^([^.]*\.[^.]*)(\.types)?(\.ann)?/$1/;
-                                              # remove exts .types or
-                                              # .types.ann
+     $basepart =~ s/^([^.]*\.[^.]*)((\.body)|(\.types))?(\.ann)?/$1/;
+                                              # remove exts .types, .body,
+                                              # .types.ann or .body.ann
      $flags = $objects{$basepart}; # get old flags
      if ($localpart =~ /\.ann$/) {
       $flags = add_flag("ann","YES",$flags);
@@ -449,6 +545,13 @@ sub finduris { # find uris for cic and theory trees generation
      } else {
       $flags = add_flag("types","NO",$flags);
      }
+     if ($localpart =~ /\.body$/) {
+      $flags = add_flag("body","YES",$flags);
+     } elsif ($localpart =~ /\.body\.ann$/) {
+      $flags = add_flag("body","ANN",$flags);
+     } else {
+      $flags = add_flag("body","NO",$flags);
+     }
      $objects{$basepart} = $flags; # save new flags
     } else { # exists at least one slash, a DIR
      ($dirname) = split (/\//, $localpart);
@@ -480,9 +583,10 @@ sub finduris { # find uris for cic and theory trees generation
    $content .= "\t<object name=\"$key\">\n";
    $flags = $objects{$key};
    $flags =~ s/^<(.*)>$/$1/;
-   my ($annflag,$typesflag) = split /,/,$flags;
+   my ($annflag,$typesflag,$bodyflag) = split /,/,$flags;
    $content .= "\t\t<ann value=\"$annflag\" />\n";
    $content .= "\t\t<types value=\"$typesflag\" />\n";
+   $content .= "\t\t<body value=\"$bodyflag\" />\n";
    $content .= "\t</object>\n";
   }
   $content .= "</ls>\n";
@@ -493,17 +597,18 @@ sub finduris { # find uris for cic and theory trees generation
 }
 
 sub add_flag {
-# manage string like: "<ann_flag,type_flag>"
+# manage string like: "<ann_flag,type_flag,body_flag>"
 # "ann_flag" may be one of "ann_YES", "ann_NO"
 # "type_flag" may be one of "types_NO", "types_YES", "types_ANN"
+# "body_flag" may be one of "body_NO", "body_YES", "body_ANN"
 # when adding a flag the max between the current flag and the new flag
-# is taken, the orders are ann_NO < ann_YES and types_NO < types_YES <
-# types_ANN
+# is taken, the orders are ann_NO < ann_YES, types_NO < types_YES <
+# types_ANN and body_NO < body_YES < body_ANN
  my ($flagtype,$newflag,$str) = @_;
- $str = "<,>" if ($str eq "");
- ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ".
+ $str = "<,,>" if ($str eq "");
+ ($str =~ s/^<(.*,.*,.*)>$/$1/) or die "Internal error: ".
    "wrong string format for flag adding in $str";
- my ($annflag,$typeflag) = split /,/,$str;
+ my ($annflag,$typeflag,$bodyflag) = split /,/,$str;
  if ($flagtype eq "ann") { # ANN flag handling
   if ($newflag eq "YES") {
    $annflag = "YES";
@@ -523,10 +628,21 @@ sub add_flag {
   } else {
    die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
   }
+ } elsif ($flagtype eq "body") { # BODY flag handling
+  if ($newflag eq "ANN") {
+   $bodyflag = "ANN";
+  } elsif ($newflag eq "YES") {
+   $bodyflag = "YES" unless ($bodyflag eq "ANN");
+  } elsif ($newflag eq "NO") {
+   $bodyflag = "NO"
+     unless (($bodyflag eq "ANN") or ($bodyflag eq "YES"));
+  } else {
+   die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
+  }
  } else {
   die "Internal error: unsupported flagtype \"$flagtype\"";
  }
- $str = "<$annflag,$typeflag>";
+ $str = "<$annflag,$typeflag,$bodyflag>";
 }
 
 #CSC: Too much powerful: creates even /home, /home/users/, ...
@@ -553,12 +669,6 @@ sub print_request
  print "</download_request>\n";
 }
 
-sub callback
-{
- my ($data) = @_;
- $cont .= $data;
-}
-
 sub gunzip { # gunzip a file and return the deflated content
 	my ($filename) = @_;
 
@@ -595,7 +705,7 @@ sub download {
  my ($gz, $buffer);
 
 # print "DOWNLOAD subs receives url: \"$url\"\n";
-$ print "DOWNLOAD subs receives filename: \"$filename\"\n";
+# print "DOWNLOAD subs receives filename: \"$filename\"\n";
 
  my $resourcetype;	# retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
  if ($filename =~ /\.xml$/) {	# set retrieve mode
@@ -677,7 +787,7 @@ $ print "DOWNLOAD subs receives filename: \"$filename\"\n";
    # $cont now contained uncompressed data
  }
  if ($patch_dtd eq "yes") {
-    $cont =~ s/DOCTYPE (.*) SYSTEM\s+"http:\/\/www.cs.unibo.it\/helm\/dtd\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
+    $cont =~ s/DOCTYPE (.*) SYSTEM\s+"$dtdbaseurl\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
  }
  if ($format eq "gz") {
   gzip($cont,"$basefname.tmp");
@@ -709,6 +819,7 @@ sub answer {
  $res->push_header("Pragma" => "no-cache");
  $res->push_header("Expires" => "0");
  $c->send_response($res);
+ $c->force_last_request();
 }
 
 sub html_nice_answer {
@@ -718,9 +829,14 @@ sub html_nice_answer {
  answer($c,$content,"text/html","");
 }
 
+sub callback {
+ my ($data) = @_;
+ $cont .= $data;
+}
+
 sub helm_wget {
-#retrieve a file from an url and write it to a temp dir
-#used for retrieve resource index from servers
+# retrieve a file from an url and write it to a temp dir
+# used for retrieve resource index from servers
  $cont = "";
  my ($prefix, $URL) = @_;
  my $ua = LWP::UserAgent->new;
@@ -734,30 +850,35 @@ sub helm_wget {
 }
 
 sub update {
- untie %map;
- tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664);
+# retie dbs
+ untie %xml_map;
+ tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664);
  untie %rdf_map;
- tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
+ tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
+ untie %xslt_map;
+ tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
 }
 
 sub update_dbs {
 # rebuild dbs fetching resource indexes from servers.
-# Rebuild urls_of_uris db (1) and rdf db (2)
+# Rebuild urls_of_uris db (1), rdf db (2) and xslt db (3)
  my (
   $server, $idxfile, $rdf_idxfile, $uri, $url, $comp, $line,
   @servers,
-  %urls_of_uris, %rdf # local proxy of $map and $rdf_map
+  %urls_of_uris, %rdf, %xslt # local proxy of dbs
  );
 
- untie %map;
+ untie %xml_map;
  untie %rdf_map;
- foreach my $file ($uris_dbm, $rdf_dbm) { # remove old db file
-  if (stat $file.".db") { # remove if exists
-   unlink ($file.".db") or die "can't unlink old db file: $file.db\n";
+ untie %xslt_map;
+ foreach my $file ($uris_dbm, $rdf_dbm, $xslt_dbm) { # remove old db file
+  if (stat $file) { # remove if exists
+   unlink ($file) or die "can't unlink old db file: $file\n";
   }
  }
- tie(%urls_of_uris, 'DB_File', $uris_dbm.".db", O_RDWR|O_CREAT, 0664);
- tie(%rdf, 'DB_File', $rdf_dbm.".db", O_RDWR|O_CREAT, 0664);
+ tie(%urls_of_uris, 'DB_File', $uris_dbm, O_RDWR|O_CREAT, 0664);
+ tie(%rdf, 'DB_File', $rdf_dbm, O_RDWR|O_CREAT, 0664);
+ tie(%xslt, 'DB_File', $xslt_dbm, O_RDWR|O_CREAT, 0664);
 
  open (SRVS, "< $servers_file") or
   die "cannot open servers file: $servers_file\n";
@@ -769,12 +890,15 @@ sub update_dbs {
   chomp $server;
   helm_wget($tmp_dir, $server."/".$indexname); # get index
   helm_wget($tmp_dir, $server."/".$rdf_indexname); # get rdf index
+  helm_wget($tmp_dir, $server."/".$xslt_indexname); # get xslt index
   $idxfile = $tmp_dir."/".$indexname;
   $rdf_idxfile = $tmp_dir."/".$rdf_indexname;
+  $xslt_idxfile = $tmp_dir . "/" . $xslt_indexname;
 
-  open (INDEX, "< $idxfile") or # (1) REBUILD URLS_OF_URIS DB
+                  # (1) REBUILD URLS_OF_URIS DB
+  open (INDEX, "< $idxfile") or
    die "cannot open temporary index file: $idxfile\n";
-  while ($line = <INDEX>) {  # parse index and add entry to urls_of_uris db
+  while (my $line = <INDEX>) {  # parse index and add entry to urls_of_uris db
    chomp $line;
    ($uri,$comp) = split /[ \t]+/, $line;
              # build url:
@@ -788,9 +912,10 @@ sub update_dbs {
    $urls_of_uris{$uri} = $url;
   }
 
-  open (RDF_INDEX, "< $rdf_idxfile") or # (2) REBUILD RDF DB
-   die "cannot open temporary rdf index file: $rdf_idxfile\n";
-  while ($line = <RDF_INDEX>) {  # parse index and add entry to rdf db
+                  # (2) REBUILD RDF DB
+  open (RDF_INDEX, "< $rdf_idxfile") or
+   die "cannot open temporary RDF index file: $rdf_idxfile\n";
+  while (my $line = <RDF_INDEX>) {  # parse index and add entry to rdf db
    chomp $line;
    ($uri,$comp) = split /[ \t]+/, $line; # comp is "gz" or nothing
              # build url:
@@ -804,18 +929,36 @@ sub update_dbs {
    $rdf{$uri} = $url;
   }
 
+                  # (3) REBUILD XSLT DB
+  open (XSLT_INDEX, "< $xslt_idxfile") or
+    die "cannot open temporary XSLT index file: $xslt_idxfile\n";
+  while (my $line = <XSLT_INDEX>) { # parse index and add entry to xslt db
+    chomp $line;
+    my $stylesheet = $line;
+    my $url = $server . "/" . $stylesheet;
+    $xslt{$stylesheet} = $url;
+  }
+
   close INDEX;  # close indexes
   close RDF_INDEX;
-  die "cannot unlink temporary file: $idxfile\n"  # remove temp files
-   if (unlink $idxfile) != 1;
-  die "cannot unlink temporary file: $rdf_idxfile\n"
-   if (unlink $rdf_idxfile) != 1;
+  close XSLT_INDEX;
+
+                  # remove temp files
+  foreach my $file ($idxfile, $rdf_idxfile, $xslt_idxfile) {
+    print "cannot unlink temporary file: $file\n" if (unlink $file) != 1;
+  }
 
  }
 
  untie(%urls_of_uris); # untie local proxies
  untie(%rdf);
- tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # retie global ones
- tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
+ untie(%xslt);
+ tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # retie global ones
+ tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
+ tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
+
 } # update_dbs
 
+# vim modline: do not remove!
+# vim: set ft=perl:
+