#!@PERL_BINARY@

# Copyright (C) 2000, HELM Team.
# 
# This file is part of HELM, an Hypertextual, Electronic
# Library of Mathematics, developed at the Computer Science
# Department, University of Bologna, Italy.
# 
# HELM is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
# 
# HELM is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with HELM; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
# 
# For details, see the HELM World-Wide-Web page,
# http://cs.unibo.it/helm/.

my $VERSION = "@VERSION@";

# First of all, let's load HELM configuration
use Env;
my $HELM_LIB_DIR = $ENV{"HELM_LIB_DIR"};
# this should be the only fixed constant
my $DEFAULT_HELM_LIB_DIR = "@HELM_LIB_DIR@";
if (defined ($HELM_LIB_DIR)) {
   $HELM_LIB_PATH = $HELM_LIB_DIR."/configuration.pl";
} else {
   $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
require $HELM_LIB_PATH;

# Let's override the configuration file
$style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"}));
$dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"}));

use HTTP::Daemon;
use HTTP::Status;
use HTTP::Request;
use LWP::UserAgent;
use DB_File;
use Compress::Zlib;
use CGI;
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 $myownurl = $d->url;

# Let's patch the returned URL
$myownurl =~ s/\/$//; # chop the final slash
my $myownport = $myownurl;
$myownport =~ s/http:\/\/(.*):(.*)/$2/;
$myownurl  =~ s/http:\/\/(.*):(.*)/$1/;
($myownurl) = gethostbyname($myownurl);
$myownurl = "http://".$myownurl.":".$myownport;

tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664);
print "Please contact me at: <URL:", $myownurl."/", ">\n";
print "helm_dir: $helm_dir\n";
print "style_dir: $style_dir\n";
print "dtd_dir: $dtd_dir\n";
print "urls_of_uris.db: $uris_dbm.db\n";
print "cache mode: $cachemode\n";

$SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
$SIG{USR1} = \&update; # sent by the child to make the parent update
while (my $c = $d->accept) {
 if (fork() == 0) {
    while (my $r = $c->get_request) {
        #CSC: mancano i controlli di sicurezza
        
        my $inputuri = uri_unescape($r->url); 
        $inputuri =~ s/^[^?]*\?uri=(.*)/$1/;
        print "\nRequest: ".$r->url."\n\n";
        my $http_method = $r->method;
        my $http_path = $r->url->path;
        my $http_query = uri_unescape($r->url->query);
        my $cgi = new CGI("$http_query");

	print "\nUnescaped query: ".$http_query."\n";

        if ($http_method eq 'GET' and $http_path eq "/getciconly") {
            # finds the uri, url and filename
            my $cicuri = $inputuri;

            my $cicfilename = $cicuri;
            $cicfilename =~ s/cic:(.*)/$1/;
            $cicfilename =~ s/theory:(.*)/$1/;

            my $cicurl   = $map{$cicuri};
            my $extension;
            if ($cicurl =~ /\.xml$/) { # non gzipped file
              $extension = ".xml";
            } elsif ($cicurl =~ /\.xml\.gz$/) { # gzipped file
              $extension = ".xml.gz";
            } else { # error: unknown extension
             die "unexpected extension in url: $cicurl, might be '.xml'".
              "or '.xml.gz'";
            }
            $cicfilename = $helm_dir.$cicfilename.$extension;

            if (!defined($cicurl)) {
             print "\nNOT FOUND!!!!!\n";
             $c->send_error(RC_NOT_FOUND)
            } else {
               print_request("cic",$cicuri,$cicurl,$cicfilename);

               # Retrieves the file
               my $ciccontent = download(0,"cic",$cicurl,$cicfilename);

               # Answering the client
               answer($c,$ciccontent);
            }
        } elsif ($http_method eq 'GET' and $http_path eq "/get") {
            # finds the uris, urls and filenames
            my $cicuri = $inputuri,
               $typesuri = $inputuri,
               $annuri = $inputuri;
            my $annsuffix;
            if ($inputuri =~ /\.types$/) {
               $cicuri    =~ s/(.*)\.types$/$1/;
               undef($annuri);
            } elsif ($inputuri =~ /\.types\.ann$/) {
               $cicuri    =~ s/(.*)\.types\.ann$/$1/;
               $typesuri  =~ s/(.*)\.ann$/$1/;
               $annsuffix = ".types.ann";
            } elsif ($inputuri =~ /\.ann$/) {
               $cicuri  =~ s/(.*)\.ann$/$1/;
               undef($typesuri);
               $annsuffix = ".ann";
            } else {
               undef($typesuri);
               undef($annuri);
            }

            my $cicfilename = $cicuri;
            $cicfilename =~ s/cic:(.*)/$1/;
            $cicfilename =~ s/theory:(.*)/$1/;
            $cicfilename = $helm_dir.$cicfilename;

            my $cicurl   = $map{$cicuri};
            my $typesurl = $map{$typesuri} if (defined($typesuri));
            my $annurl   = $map{$annuri}  if (defined($annuri));
            my ($cicext, $typesext, $annext);
            if ($cicurl =~ /\.xml$/) { # normal file
             $cicext = ".xml";
            } elsif ($cicurl =~ /\.xml\.gz$/) { # gzipped file
             $cicext = ".xml.gz";
            } else {
             die "unexpected extension in url: $cicurl;".
             "might be '.xml' or '.xml.gz'";
            }
            if (defined($typesuri)) { # extension selection for types file
             if ($typesurl =~ /\.xml$/) { # normal file
              $typesext = ".types.xml";
             } elsif ($typesurl =~ /\.xml\.gz$/) { # gzipped file
              $typesext = ".types.xml.gz";
             } else {
              die "unexpected extension in url: $typesurl;".
              "might be '.xml' or '.xml.gz'";
             }
            }
            if (defined($annuri)) { # extension selection for annotation file
             if ($annurl =~ /\.xml$/) { # normal file
              $annext = ".xml";
             } elsif ($annurl =~ /\.xml\.gz$/) { # gzipped file
              $annext = ".xml.gz";
             } else {
              die "unexpected extension in url: $annurl".
              "might be '.xml' or '.xml.gz'";
             }
            }
            my $typesfilename = $cicfilename.$typesext if $typesuri;
            my $annfilename  = $cicfilename.$annsuffix.$annext if $annuri;
            $cicfilename .= $cicext;

            if (!defined($cicurl) ||
               (!defined($typesurl) && $typesuri) ||
               (!defined($annuri) && $annuri))
            {
             print "\nNOT FOUND!!!!!\n";
             $c->send_error(RC_NOT_FOUND)
            } else {
               print_request("cic",$cicuri,$cicurl,$cicfilename);
               print_request("types",$typesuri,$typesurl,$typesfilename)
                if ($typesuri);
               print_request("ann",$annuri,$annurl,$annfilename)
                if ($annuri);
 
               # Retrieves the files

               my $ciccontent = download(1,"cic",$cicurl,$cicfilename);
               my $typescontent =
                download(1,"types",$typesurl,$typesfilename) if ($typesuri);
               my $anncontent =
                download(1,"ann",$annurl,$annfilename) if ($annuri);
 
               # Merging the files together
 
               my $merged = <<EOT;
<?xml version="1.0" encoding="UTF-8"?>
<cicxml uri="$cicuri">
$ciccontent
$typescontent
$anncontent
</cicxml>
EOT

               # Answering the client
               answer($c,$merged);
            }
         } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
            my $filename = $inputuri;
            $filename = $dtd_dir."/".$filename;
            print "DTD: $inputuri ==> ($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/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
	        $cont .= $_;
	       }
               close(FD);
               answer($c,$cont);
            } else {
               die "Could not find DTD!";
            }
        } elsif ($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,$cont);
            } else {
               die "Could not find Style Configuration File!";
            }
        } elsif ($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,$cont);
            } else {
               die "Could not find XSLT!";
            }
        } elsif ($http_method eq 'GET' and $http_path eq "/conf") {
            my $quoted_html_link = $html_link;
            $quoted_html_link =~ s/&/&amp;/g;
            $quoted_html_link =~ s/</&lt;/g;
            $quoted_html_link =~ s/>/&gt;/g;
            $quoted_html_link =~ s/'/&apos;/g;
            $quoted_html_link =~ s/"/&quot;/g;
            print "\nConfiguration requested, returned #$quoted_html_link#\n";
	    $cont = "<?xml version=\"1.0\"?><html_link>$quoted_html_link</html_link>";
            answer($c,$cont);
        } elsif ($http_method eq 'GET' and $http_path eq "/update") {
            # rebuild urls_of_uris.db
           print "Update requested...\n";
           mk_urls_of_uris();
           kill(USR1,getppid()); # signal changes to parent
           print " done\n";
           answer($c,"<html><body><h1>Update done</h1></body></html>");
        } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
            # send back keys that begin with a given uri
           my ($uritype,$uripattern,$uriprefix);
           my $baseuri = $cgi->param('baseuri');
           chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/"
           my $outype = $cgi->param('format'); # output type, might be 'txt' or 'xml'
           $uripattern = $baseuri;
           $uripattern =~ s/^.*:(.*)/$1/;
           if ($baseuri =~ /^cic:/) {
            $uritype = "cic";
           } elsif ($baseuri =~ /^theory:/) {
            $uritype = "theory";
           } else {
            $uritype = "invalid";
           }
           if ($uritype ne "invalid") { # uri is valid
            if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type
             print "Invalid output type specified: $outype\n";
             answer($c,"<html><body><h1>Invalid output type, may be ".
              "\"txt\" or \"xml\"</h1></body></html>");
            } else { # valid output
             print "BASEURI $baseuri, FORMAT $outype\n";
             $cont = finduris($uritype,$uripattern,$outype);
             answer($c,$cont);
            }
           } else { # invalid uri
            print "Invalid uri: $baseuri, may begin with 'cic:', ".
             "'theory:' or '*:'\n";
            answer($c,"<html><body><h1>Invalid uri , may begin with ".
             "\"cic:\", \"theory:\" or \"*:\"</h1></body></html>");
           }
        } elsif ($http_method eq 'GET' and $http_path eq "/version") {
           print "Version requested!";
           answer($c,"<html><body><h1>HTTP Getter Version ".
            $VERSION."</h1></body></html>");
        } else {
            print "\n";
            print "INVALID REQUEST!!!!!\n";
            print "(PATH: ",$http_path,", ";
            print "QUERY: ",$http_query,")\n";
            $c->send_error(RC_FORBIDDEN)
        }
        print "\nRequest solved: ".$r->url."\n\n";
    }
    $c->close;
    undef($c);
    print "\nCONNECTION CLOSED\n\n";
    exit;
  } # fork
}

#================================

sub finduris { # find uris for cic and theory trees generation
 my ($uritype,$uripattern,$format) = @_;
 my $content = "";
 my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key);
 my (@itemz,@already_pushed_dir);
 my (%objects,%dirs); # map uris to suffixes' flags

 print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ".
  "format: $format\n\n";
 
 if (($uritype eq "cic") or ($uritype eq "theory")) {
   # get info only of one type: cic or theory
  foreach (keys(%map)) { # select matching uris
   $uri = $_;
   if ($uri =~ /^$uritype:$uripattern\//) {
    $localpart = $uri;
    $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/;

    if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
     $basepart = $localpart;
     $basepart =~ s/^(.*)\.types(\.ann)?/$1/; # remove exts .types or
                                              # .types.ann
     $flags = $objects{$basepart}; # get old flags
     if ($localpart =~ /\.ann$/) {
      $flags = add_flag("ann","YES",$flags);
     } else {
      $flags = add_flag("ann","NO",$flags);
     }
     if ($localpart =~ /\.types$/) {
      $flags = add_flag("types","YES",$flags);
     } elsif ($localpart =~ /\.types\.ann$/) {
      $flags = add_flag("types","ANN",$flags);
     } else {
      $flags = add_flag("types","NO",$flags);
     }
     $objects{$basepart} = $flags; # save new flags
    } else { # exists at least one slash, a DIR
     ($dirname) = split (/\//, $localpart);
     $dirs{$dirname} = ""; # no flags requirement for dir
    }
   }
  }
 } else {
  die "getter internal error: unsupported uritype: \"$uritype\"";
 }
                            # now generate OUTPUT:
                            # output will be saved in $content
 if ($format eq "txt") { 
  foreach $key (sort(keys %dirs)) {
   $content .= "dir, " . $key . "\n";
  }
  foreach $key (sort(keys %objects)) {
   $content .= "object, $key, " . $objects{$key} . "\n";
  }
 } elsif ($format eq "xml") {
  $content .= "<ls>\n";
  foreach $key (sort(keys %dirs)) {
   $content .= "\t<section>$key</section>\n";
  }
  foreach $key (sort(keys %objects)) {
   $content .= "\t<object name=\"$key\">\n";
   $flags = $objects{$key};
   $flags =~ s/^<(.*)>$/$1/;
   my ($annflag,$typesflag) = split /,/,$flags;
   $content .= "\t\t<ann value=\"$annflag\" />\n";
   $content .= "\t\t<types value=\"$typesflag\" />\n";
   $content .= "\t</object>\n";
  }
  $content .= "</ls>\n";
 } else { # may not enter this branch
  die "Getter internal error: invalid format received by finduris sub";
 }
 return $content;
}

sub add_flag {
# manage string like: "<ann_flag,type_flag>"
# "ann_flag" may be one of "ann_YES", "ann_NO"
# "type_flag" may be one of "types_NO", "types_YES", "types_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
 my ($flagtype,$newflag,$str) = @_;
 $str = "<,>" if ($str eq "");
 ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ".
   "wrong string format for flag adding in $str";
 my ($annflag,$typeflag) = split /,/,$str;
 if ($flagtype eq "ann") { # ANN flag handling
  if ($newflag eq "YES") {
   $annflag = "YES";
  } elsif ($newflag eq "NO") {
   $annflag = "NO" unless ($annflag eq "YES");
  } else {
   die "Internal error: annflag must be \"YES\" or \"NO\"";
  }
 } elsif ($flagtype eq "types") { # TYPES flag handling
  if ($newflag eq "ANN") {
   $typeflag = "ANN";
  } elsif ($newflag eq "YES") {
   $typeflag = "YES" unless ($typeflag eq "ANN");
  } elsif ($newflag eq "NO") {
   $typeflag = "NO"
     unless (($typeflag eq "ANN") or ($typeflag eq "YES"));
  } else {
   die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
  }
 } else {
  die "Internal error: unsupported flagtype \"$flagtype\"";
 }
 $str = "<$annflag,$typeflag>";
}

#CSC: Too much powerful: creates even /home, /home/users/, ...
#CSC: Does not raise errors if could not create dirs/files
sub mkdirs
{
 my ($pathname) = @_;
 my @dirs = split /\//,$pathname;
 my $tmp;
 foreach $dir (@dirs) {
  $tmp = ((defined($tmp)) ?  $tmp."\/".$dir : "");
  mkdir($tmp,0777);
 }
 rmdir($tmp);
}

sub print_request
{
 my ($str,$uri,$url,$filename) = @_;
 print $str."uri: $uri\n";
 print $str."url: $url\n";
 print $str."filename: $filename\n\n";
}

sub callback
{
 my ($data) = @_;
 $cont .= $data;
}

sub gunzip { # gunzip a file and return the deflated content
	my ($filename) = @_;

	my ($gz, $buffer, $cont);

	print "deflating $filename ...\n";
	$gz = gzopen($filename, "r")
		or die "Cannot open gzip'ed file $filename: $gzerrno";
	$cont = "";
	while ( $gz->gzread($buffer) > 0 ) {
		$cont .= $buffer;
	}
	die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
	$gz->gzclose();

	return $cont;
}

sub gzip {	# gzip the content argument and save it to filename argument
	my ($cont, $filename) = @_;

	my ($gz, $cont);

	$gz = gzopen($filename, "w")
		or die "Cannot gzopen for writing file $filename: $gzerrno";
	$gz->gzwrite($cont) or die "error writing: $gzerrno\n" ;
	$gz->gzclose();
}

sub download
{
 my ($remove_headers,$str,$url,$filename) = @_;
 my ($gz, $buffer);

 my $resourcetype;	# retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
 if ($filename =~ /\.xml$/) {	# set retrieve mode
	 $resourcetype = "normal";
 } elsif ($filename =~ /\.xml\.gz$/) {
	 $resourcetype = "gzipped";
 } else {
	 die "Unsupported download extension, might be '.gz' or '.xml'\n";
 }
 my $basefname = $filename;
 $basefname =~ s/\.gz$//;	# get base resource name removing trailing .gz
 $cont = ""; # modified by side-effect by the callback function

 my $localfname="";
 if (stat($basefname)) {
 	$localfname=$basefname;
 } elsif (stat($basefname.".gz")) {
 	$localfname=$basefname.".gz";
 }
 if ($localfname ne "") { # we already have local copy of requested file
      # check both possible cache entry: gzipped or normal
  print "Using local copy for the $str file\n";
  if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
   $cont = gunzip($localfname);
  } elsif ($localfname =~ /\.xml$/) { # just return cached file
   open(FD, $localfname) or die "Cannot open $localfname";
   while(<FD>) { $cont .= $_; }
   close(FD);
  } else { # error
   die "Internal error: unexpected file name $localfname,"
   ."must end with '.gz' or '.xml.gz'\n";
  }
 } else { # download file from net
    print "Downloading the $str file\n"; # download file
    my $ua = LWP::UserAgent->new;
    my $request = HTTP::Request->new(GET => "$url");
    my $response = $ua->request($request, \&callback);
               
    # cache retrieved file to disk
# <ZACK/> TODO: inefficent, I haven't yet undestood how to deflate
#    in memory gzipped file, without call "gzopen"
 print "Storing the $str file\n";
 mkdirs($filename);
 open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
 print FD $cont;
 close(FD);

 # handle cache conversion normal->gzipped or gzipped->normal as user choice
 if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
   # cache the file as is
  rename "$filename.tmp", $filename; 
 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
   # cache the file as is
   # and update the $cont variabile with deflated content
  rename "$filename.tmp", $filename; 
  $cont = gunzip($filename);
 } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
   # deflate cache entry
   # and update $cont
  open(FD, "> $basefname") or die "cannot open $basefname\n";
  $cont = gunzip($filename.".tmp");
  print FD $cont;
  close(FD);
  unlink "$filename.tmp"; # delete old gzipped file
 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
   # compress cache entry
  gzip($cont, $basefname.".gz");
  unlink "$filename.tmp"; # delete old uncompressed file
 } else {
  die "Internal error, unsopported cachemode, resourcetype couple\n";
 }
 # $cont now contained uncompressed data

 }
 if ($remove_headers) {
    $cont =~ s/<\?xml [^?]*\?>//sg;
    $cont =~ s/<!DOCTYPE [^>]*>//sg;
 } else {
    $cont =~ s/DOCTYPE (.*) SYSTEM\s+"http:\/\/www.cs.unibo.it\/helm\/dtd\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
 }
 return $cont;
}

sub answer
{
 my ($c,$cont) = @_;
 my $res = new HTTP::Response;
 $res->content($cont);
 $c->send_response($res);
}

sub helm_wget {
#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;
 my $request = HTTP::Request->new(GET => "$URL");
 my $response = $ua->request($request, \&callback);
 my ($filename) = reverse (split "/", $URL); # get filename part of the URL
 open (TEMP, "> $prefix/$filename")
  || die "Cannot open temporary file: $prefix/$filename\n";
 print TEMP $cont;
 close TEMP;
}

sub update {
 untie %map;
 tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664);
}

sub mk_urls_of_uris {
#rebuild $uris_dbm.db fetching resource indexes from servers
 my (
  $server, $idxfile, $uri, $url, $comp, $line,
  @servers,
  %urls_of_uris
 );

 untie %map;
 if (stat $uris_dbm.".db") { # remove old db file
  unlink($uris_dbm.".db") or
   die "cannot unlink old db file: $uris_dbm.db\n";
 }
 tie(%urls_of_uris, 'DB_File', $uris_dbm.".db", O_RDWR|O_CREAT, 0664);

 open (SRVS, "< $servers_file") or
  die "cannot open servers file: $servers_file\n";
 @servers = <SRVS>;
 close (SRVS);
 while ($server = pop @servers) { #cicle on servers in reverse order
  print "processing server: $server ...\n";
  chomp $server;
  helm_wget($tmp_dir, $server."/".$indexname); #get index
  $idxfile = $tmp_dir."/".$indexname;
  open (INDEX, "< $idxfile") or
   die "cannot open temporary index file: $idxfile\n";
  while ($line = <INDEX>) { #parse index and add entry to urls_of_uris
   chomp $line;
   ($uri,$comp) = split /[ \t]+/, $line;
             # build url:
   if ($comp =~ /gz/) { 
    $url = $uri . ".xml" . ".gz";
   } else {
    $url = $uri . ".xml";
   }
   $url =~ s/cic:/$server/;
   $url =~ s/theory:/$server/;
   $urls_of_uris{$uri} = $url;
  }
  close INDEX;
  die "cannot unlink temporary file: $idxfile\n"
   if (unlink $idxfile) != 1;
 }

 untie(%urls_of_uris);
 tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664);
}