# $Id: modern.cf,v 1.1 2002/07/05 by Harald Weinreich <harald@weinreichs.de>
#
# Definitions for '{nph-}search.cgi' configuration.
#
# A hash (#) in the first column denotes a comment and is not processed
# even inside a definition.  Hashes not in the first column are left as-is.
#
# Variable substitution occurs on these definitions.  If you want
# a dollar-sign ($) to occur in the output, escape it with backslash (\).
# Other metacharacters (quotes, asterisks, etc) probably don't need to be
# escaped.  Printf-like special characters are allowed: \n \t \r etc.
#
# The ending newline is chopped off from each definiton.  So
#
#   <FooBar>
#   abcxyz
#   </FooBar>
#
# becomes "abcxyz".  You may have blank lines in between the beginning
# and ending tags, or use the newline character \n.
#

#
# CGI defaults.
#
# In this section you can give default values for any attribute that would
# normally be passed in from the CGI query form.  The following attributes
# are currently available:
#
# brokerqueryconfig : Name of a broker specific configuration file
# lifetime          : Maximum lifetime of a query (see also <Lifetime> section)
# caseflag          : Whether the query is casesensitive ["on" or "off"]
# wordflag          : Match on word boundaries ["on" or "off"]
# opaqflag          : Show matched lines ["on" or "off"]
# descflag          : Return object description ["on" or "off"]
# noregexflag       : Don't do regular expressions ["on" or "off"]
# maxresultflag     : Maximum number of results to return
# maxobjflag        : Maximum number of objects to return
# maxlineflag       : Maximum number of matched lines
# errors            : Number of spelling errors
# broker            : Name of broker
# host              : Name of broker host
# attributes        : Space sperated list of attributes to return
# sort              : Sort options ["by-rank"]

# These defaults mimic the values on the provided query-glimpse-modern.html form.
<Default>
caseflag      : on
wordflag      : on
errorflag     : 0
opaqueflag    : on
descflag      : on
maxobjflag    : 500
maxlineflag   : 30
maxresultflag : 3000
perpageflag   : 25
sort          : by-rank
</Default>

# 'GLOBAL' VARIABLES: are defined globally within the {nph-}search.cgi program,
#                     but not necessarily set at all times.
#
# $query	the user query string
# $html_query	the user query string, special HTML characters escaped
# $bquery	the query string sent to the broker
# $host		the broker hostname
# $port		the broker port
# $hp_url	the URL of the broker query (home) page
# $maxresult	the maximum number of matched lines the broker will return
# $nobjects	a running count of the number of objects returned
# $nopaquelines a running count of the number of opaque (matched) lines
#
# definitions here can be output in other definitions.  Each definition
# here is placed into the %CFG associative array.  For example,
# to print out the Timeout defined below, write $CFG{'Timeout'}.
#

# The URL to the Harvest Project home page.
#
<HarvestUrl>
http://harvest.sourceforge.net/
</HarvestUrl>

# The amount of time to wait for the broker results.
# Can use aritmetic # here also, eg: 5*60
#
# We use a conservative 5 minutes here.
#
<Timeout>
300
</Timeout>

# InitFunction can be some perl code which gets eval'd before they
# query is sent.  Use this for any special hackings.
#
<InitFunction>
$cs_urlX = $csumflag ? 'Y' : undef;
</InitFunction>

# The MIME content type of the query results.  Should be "text/html"
# if we are going to return HTML tags.
#
<ContentType>
text/html
</ContentType>


##############################################################################
# RESULT SECTION
#
# The output of '{nph-}search.cgi' consists of the following tags:
#
#    ResultHeader
#    CreateNavBars
#    ResultSetBegin
#    ( Errors from 'broker' )
#    foreach object {
#        PrintObject
#    }
#    ResultSetEnd
#    EndBrokerResults
#    ( TruncateWarning )
#    ResultTrailer
#
#
##############################################################################

# First output for the HTML containing the results.  Should probably include
# <TITLE> tag and the user query.
#
<ResultHeader>
<HEAD>
<TITLE>Search Results for: $html_query</TITLE>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
<style><!--
.rNavBar {
  color: #6699cc;
  font-size: small;
  font-family: arial, helvetica, sans-serif;
}
.rMoreInfo {
  color: #6699cc;
  font-size: small;
  font-family: arial, helvetica, sans-serif;
  font-style: italic;
  text-decoration: none;
}
dt { margin-left:1px; }
dd { margin-left:40px; }
// --></style>
</HEAD>
<BODY bgcolor=white>
<H2>Search Results for: <font color="#800000">$html_query</font></H2>
<HR>
</ResultHeader>

# Final HTML section for the output.  Should probably contain a pointer
# to the broker home-page ($hp_url) and a pointer to the Harvest
# home page.
#
<ResultTrailer>
#----------------------------------------------------------------------
# JMM - 20011020
# We close the HTML page
#---------------------------------------------------------------------,
<table border="0" cellpadding="0" cellspacing="0" width="100%">
<tr align="center">
  <td width="100%">
    $rNavigationBar
  </td>
  <td><a href="$CFG{'HarvestUrl'}"><img
         alt="Powered by harvest" border="0" src="$rHarvestIcon"></a></td>
</tr>
</table>
</body>
</html>
#---------------------------------------------------------------------'
</ResultTrailer>


#----------------------------------------------------------------------
# JMM 20011020
# Calculations to obtain $rResultsBar and $rNavigationBar
# We use them in <ResultSetBegin>
#---------------------------------------------------------------------,
<CreateNavBars>

if ($rTotHtmlFiles > 0)
{
  $rCurrentPage = (!$current_page) ? 1 : $current_page;

  $rPrevPage = "";
  if ($rCurrentPage > 1)
  {
    $rPrevPage = $rCurrentPage - 1 ;
    $rPrevPage = "\n\n<a class=\"rNavBar\" href=\"$tmp_www/$^T-${$}-$rPrevPage.html\">&lt;&lt; Prev</a> ·";
  }

  $rNextPage = "";
  if ($rCurrentPage < $rTotHtmlFiles)
  {
    $rNextPage = $rCurrentPage + 1;
    $rNextPage = "\n· <a class=\"rNavBar\" href=\"$tmp_www/$^T-${$}-$rNextPage.html\">Next &gt;&gt;</a>";
  }

  $rIniP = $rCurrentPage - $rNavPart;
  $rIniP = 1 if ($rIniP < 1);
  $rEndP = $rCurrentPage + $rNavPart;
  $rEndP = $rTotHtmlFiles if ($rEndP >= $rTotHtmlFiles);

  $rPag10s = "";
for ($rI = $rIniP; $rI < $rCurrentPage; $rI++)
{
$rPag10s .= "\n<a class=\"rNavBar\" href=\"$tmp_www/$^T-${$}-$rI.html\">$rI</a> ";
}
$rPag10s .= "\n[<b>$rCurrentPage</b>] ";
for ($rI = $rCurrentPage+1; $rI <= $rEndP; $rI++)
{
$rPag10s .= "\n<a class=\"rNavBar\" href=\"$tmp_www/$^T-${$}-$rI.html\">$rI</a> ";
}

$rResultsBar = "<br><b>$totnumber</b> objects - Page: $rCurrentPage/$rTotHtmlFiles",
$rResultsBar .= " - <a href=\"$rSearchPage\">New Search</a>";

$rNavigationBar  = "\n<!-- Navigation bar -->\n";
$rNavigationBar .= "$rPrevPage $rPag10s $rNextPage" if ($rTotHtmlFiles > 1);
}
else
{
$rResultsBar = " - <a href=\"$rSearchPage\">New Search</a>";
}

</CreateNavBars>
#---------------------------------------------------------------------'




# Output just before beginning the loop over objects returned by the
# broker.
#
<ResultSetBegin>
#----------------------------------------------------------------------
# JMM - 20011020
# We put a navigation bar
#---------------------------------------------------------------------,
#<DL>\n
<p>Search for: <b>$html_query</b>
$rResultsBar
<br>$rNavigationBar
<hr>
#---------------------------------------------------------------------'
</ResultSetBegin>



# output just after ending the loop over objects returned by the broker
#
<ResultSetEnd>
</DL>\n
</ResultSetEnd>

# EndBrokerResults is printed when the broker results end normally.
#
<EndBrokerResults>
<STRONG>$msg</STRONG><BR>\n
</EndBrokerResults>

# FailBrokerResults is printed when the broker results end in error
#
<FailBrokerResults>
\n<STRONG>$msg</STRONG><BR>\n
</FailBrokerResults>


# PER-OBJECT DEFINITIONS
#
# NOTES: In order for us to just use "$opaque" and have it print all the
#        matched lines properly, we must be in <PRE> mode.
#
#        Rather than just putting $cs_url in a HREF, we break it up and
#        stick the components around the 'DisplayObject' CGI program
#        which formats the SOIF into nice-looking HTML.
#
# VARIABLES:
#
#    $url	Object url: http://www.cia.gov:3333/Spies/KGB/secret.html
#    $A		URL Access: http
#    $H		URL Host  : www.cia.gov:3333
#    $P		URL Path  : /Spies/KGB/secret.html
#    $D		URL Dir   : /Spies/KGB/
#    $F		URL File  : secret.html
#    $cs_url	URL to the SOIF object in the broker databse
#    $cs_[ahp]	elements of $cs_url as above with $url
#    $desc	Description attribute of the matched object
#    $opaque	A matched line (or all matched lines in obj-at-a-time mode)
#    $usermsg	A user message
#    $attributes Requested attributes

<PrintObject>
<DT>$objectnum <A HREF="$url">$description</A>&nbsp;<FONT size=-1 color=#606060>$objectType&nbsp;&nbsp;$objectSize</FONT>
#----------------------------------------------------------------------
# JMM - 20011020
# We print "+info" to access detailed metadata and the balls
# identifying the weight
#---------------------------------------------------------------------,
<a class="rMoreInfo" href="$rDisplaySoifPage?object=$cs_p&query=$html_query">+info</a>
<br>$rWeightBalls
#---------------------------------------------------------------------'
<DD>$attributes
<DD><strong>matches:</strong> $opaque...
<?$cs_urlX>
<DD><A HREF="$cs_a://$cs_h/Harvest/cgi-bin/DisplayObject.cgi?object=$cs_p">indexing data</A>
</?$cs_urlX>
<DD><FONT size=-1 color=#006000>$url</font>
<BR><BR>
</PrintObject>

# This definition is eval'd for each opaque (matched) line retruned by
# the broker.  It is intended to be used to remove SOIF attributes
# and the 'Matched line' string from the output.  See the perl(1) manual
# page or book for information on building Perl regular expressions.
#
#  Add this to have URLs in mached lines be links
#
#    s/([a-z]*:\/\/\S*)/<A HREF="\1">\1<\/A>/;
#
<MatchedLineSub>
#1. Completely remove lines with @FILE
s/^(.*)\@FILE(.*)$/ /;
#2. Remove "Matched line #"
s/^(.*)\# (.*)$/$2/;
#3. Remove "{12}:" etc.
s/^(.*)\}:(.*)$/$2/;
#4. Remove leading and trailing whitespaces
s/^\s+//;
s/\s+$//;
#5. Limit length to 10 words per matched line.
s/^(((\s*)(\S*)){10}).*$/$1/;
#6. show 4 matched lines, set maximum length and avoid double matched lines.
if ($opaquePerObjectCount>4 || length($_)>120 || $lastOpaqueObject eq $_ ) {
   $_ = "";                     #no output
} else {
   $lastOpaqueObject=$_;
   $opaquePerObjectCount++;
   s/(.*)/... $1/;              #OK, add dots.
}
#highlight search words.
@worte=split(/ +/,$query);
for ($i=0; $i<=$#worte; $i++) {
   s/(\b$worte[$i]\b)/<font color=#800000>$1<\/font>/ig if !($worte[$i] =~ /AND|OR|NOT|\(|\)/);
}
</MatchedLineSub>

# PerObjectFunction is eval'd before every object is printed out.
#
<PerObjectFunction>
#----------------------------------------------------------------------
# JMM - 20012020
# We create the HTML code to print the balls
#---------------------------------------------------------------------,
$rWeightBalls = "";

if ($rGreaterWeight > 1)
{
  $nWeight = int($rWeight * 5 /$rGreaterWeight);
  $rWeightBalls = "<br>" if ($nWeight>0);

  for ($idxWeight=1; $idxWeight <= $nWeight; $idxWeight++)
  {
    $rWeightBalls .= "<img alt=\"*\" src=\"$rWeightIcon\"> ";
  }
  $rWeightBalls = '<font color="red">'.$rWeightBalls.'</font>';
}
#---------------------------------------------------------------------'

$opaquePerObjectCount = 1;
$objectType="";
$objectSize="";

$lastOpaqueObject     = "";
$description = "<I>File:</I>&nbsp;&nbsp;$F";
$description = $desc if (length($desc) > 5);

#highlight search words.
@worte=split(/ +/,$query);
for ($i=0; $i<=$#worte; $i++) {
   $description =~ s/(\b$worte[$i]\b)/<b>$1<\/b>/ig if !($worte[$i] =~ /AND|OR|NOT|\(|\)/);
}
</PerObjectFunction>

# PerAttributeFunction is eval'd before the Attributes are printed out.
<PerAttributeFunction>
# Remove all empty lines
$val =~ s/^(\s*)//mg;
# insert commas to separate headings
$val =~ s/\n(.)/, $1/mg;
# If more than 120 Characters, display only the first 120 plus "..."
$val =~ s/^((.|\n){120})(.|\n)*/$1.../;

# Now prepare the object's type and size to display at the description.
if ($att eq "type") {
   $objectType = "[".$val."]" if !($val =~ /HTML|HTTP-Query/);
   $att="";
} elsif ($att eq "file-size") {
   $val = int($val/1024);
   $objectSize = $val." KByte" if ($val > 150);  # more than 150 KByte
   $val = int($val/10.24);
   $objectSize = ($val/100)." MByte" if ($val > 100); # more than one meg
   $att="";
} else {
   #highlight search words.
   @worte=split(/ +/,$query);
   for ($i=0; $i<=$#worte; $i++) {
      $val =~ s/(\b$worte[$i]\b)/<font color=#800000>$1<\/font>/ig
          if !($worte[$i] =~ /AND|OR|NOT|\(|\)/);
   }
}
</PerAttributeFunction>

# Format Requested Attributes.  Before this is eval'd, $att and $val
# should be set.
#
<FormatAttribute>
<STRONG>$att:</STRONG> $val<BR>
</FormatAttribute>


# ======================================================================
# ERROR, STATUS and WARNING functions
# ======================================================================

# The message printed to the browser and logged to the HTTP server log
# when the processes is killed.  If the Timeout time is reached, the
# process dies from SIGALRM.  The short name of the offending signal
# is placed in $sig.
#
<sigdie>
Killed by SIG$sig...\n
</sigdie>

# How to format the object number.  Use a printf format specification
# to left/right justify, or whatever.  Do not include quotes around
# the format string.
#
<ObjectNumPrintf>
%2d.
</ObjectNumPrintf>

# A warning message printed only when the broker might have truncated
# the result set.  Only printed if the number of matched lines equals
# the 'maxresultflag' value of the query.html form.
#
<TruncateWarning>
<P><STRONG>WARNING: The result set was truncated at
$nopaquelines matched lines.</STRONG><P>\n
</TruncateWarning>

# A warning message printed only when the broker returned 0 results.
#
<EmptySetWarning>
<P>Your query either <em>did not match</em> any information in this
Broker, or you <em>may</em> have specified a query that is not supported by
this Broker's search subsystem.  Please refer to the
<A HREF="/Harvest/brokers/queryhelp.html">help on formulating queries</A>
for further assistance.<P>\n
</EmptySetWarning>

# Error Message returned if there is no query string sent.
#
<NoQuery>
No query information to decode. Please enter a search word.\n
</NoQuery>

# Message returned if the broker sends back a
#      111 - Broker is too heavily loaded
# reply.
#
<BrokerLoad>
</pre><P>
Sorry, the search broker at <STRONG>$host, port $port</STRONG> is currently too
heavily loaded to process your request.<P>
Please try again later.<P>
</BrokerLoad>

# Message returned if the broker sends back a
#      111 - PARSE ERROR
# reply.
#
<ParseError>
</pre>
<P>Sorry, your query does not have the proper syntax:
<PRE>    $html_query</PRE>
<P>Please refer to the
<A HREF="/Harvest/brokers/queryhelp.html">help in formulating queries</A>
for details on the proper syntax.
<P>
Common syntax mistakes include:
<UL>
<LI>
<STRONG>Not using quotes around phrases or regular expressions</STRONG>.<BR>
For example, the <em>phrase</em>:
<CODE>resource discovery</CODE> should be
<CODE>"resource discovery"</CODE>.
The <em>regular expression</em>:
<CODE>res.* disc.*</CODE> should be
<CODE>"res.* disc.*"</CODE>.
Note that a <em>single dot</em> (``<CODE>.</CODE>'')
in a keyword needs quotes (e.g. <CODE>"3.0"</CODE>).
<LI>
<STRONG>Missing punctuation</STRONG>.<BR>
Structured queries need a colon and optional parentheses.  For example,
<CODE>Type:PostScript</CODE> or, <CODE>(Type : PostScript)</CODE>
</UL>\n
</ParseError>