modern.cf

harvest是一个下载html网页得机器人

# $Id: modern.cf,v 1.2 2002/07/30 by Harald Weinreich <[e-mail]>
#
# Definitions for '{nph-}search.cgi' configuration.
#
# A hash (#) in the first column denotes a comment and is not processed
# even inside a definition.  Hashes not in the first column are left as-is.
#
# Variable substitution occurs on these definitions.  If you want
# a dollar-sign ($) 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"]
# opaqueflag        : 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
# weightflag        : show weight of hit
# csumflag          : show link to indexing data
# errorflag         : Number of spelling errors (therefore it's not really a flag!)
# broker            : Name of broker
# host              : Name of broker host
# attribute         : Space sperated list of attributes to return
# sort              : Sort options ["by-rank"]
# hp_url            : Url of search page

# These defaults mimic the values on the provided query-glimpse-modern.html form.
<Default>
caseflag      : on
wordflag      : on
opaqueflag    : on
descflag      : on
noregexflag   : off
weightflag    : off
csumflag      : off
errorflag     : 0
maxobjflag    : 500
maxlineflag   : 30
maxresultflag : 3000
perpageflag   : 10
sort          : by-rank
charset       : ISO-8859-1
attribute     : headings type file-size
#attribute     : headings type file-size last-modification-time
</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
# $filterhint   a (HTML)-Text with a verbal description of the current broker filter
#
# 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: $html_query</TITLE>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
<style><!--
dt { margin-left:1px; }
dd { margin-left:40px; }
// --></style>
</HEAD>
<BODY bgcolor=white>
<H2>Search Results for: <font color="#800000">$html_query</font></H2>
$filterhint
</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>
</BODY>
</HTML>

</ResultTrailer>

# Calculations to obtain $navigationBar
# Original code by Javier Masa Marin <[e-mail]>
#
<CreateNavBars>
if ($totalPages > 0)
{
  # elements before and after current page in navigation bar
  local($navPart) = 5;

  $prevPage = "";
  $prevPage = "\n\n<a href=\"".&create_link($page-1)."\">&lt;= Prev</a> &middot;" if ($page > 1);

  $nextPage = "";
  $nextPage = "&middot; <a class=\"navBar\" href=\"".&create_link($page+1)."\">Next =&gt;</a>" if ($page < $totalPages);

  $pageList = "";

  local($startP) = ($page <= $navPart) ? 1 : $page - $navPart;
  for ($i = $startP; $i < $page; $i++)
  { $pageList .= "<a href=\"".&create_link($i)."\">$i</a>\n"; }

  $pageList .= "<b><font color=\"#800000\" size=+1>$page</font></b>\n";

  local($endP) = ($page + $navPart >= $totalPages) ? $totalPages : $page + $navPart;
  for ($i = $page+1; $i <= $endP; $i++)
  { $pageList .= "\n<a href=\"".&create_link($i)."\">$i</a>\n"; }

  $navigationBar = "\n$prevPage\n $pageList\n $nextPage" if ($totalPages > 1);
}
</CreateNavBars>

# Output just before beginning the loop over objects returned by the broker.
#
<ResultSetBegin>
<form method="GET" action="/Harvest/cgi-bin/search.cgi">\n
<table border="0" cellspacing=0 cellpadding=4 width="100%">
<tr bgcolor="#E0E0E0">
  <td align=left width="20%" nowrap>
    $totnumber objects found
  </td>
  <td align=center width="60%">
    $navigationBar
  </td>
  <td align=right width="20%">
    Page: $page/$totalPages
  </td>
</tr>
<tr>
<td align=middle colspan=3>
 <input type=hidden name=broker value="$broker">
 <input type=hidden name=hp_url value="$hp_url">
 <input type=hidden name=caseflag value="$RQ{'caseflag'}">
 <input type=hidden name=wordflag value="$RQ{'wordflag'}">
# <input type=hidden name=opaqueflag value="$RQ{'opaqueflag'}">
# <input type=hidden name=descflag value="$RQ{'descflag'}">
# <input type=hidden name=maxresultflag value="$RQ{'maxresultflag'}">
# <input type=hidden name=maxobjflag value="$RQ{'maxobjflag'}">
# <input type=hidden name=maxlineflag value="$RQ{'maxlineflag'}">
 <input type=hidden name=weightflag value="$RQ{'weightflag'}">
 <input type=hidden name=attribute value="@atts">
<a href="$hp_url">New Search:</a> <input type=text name=query value='$RQ{'query'}'> <input type=submit>
</td></tr>
</form>
</table>

<DL>\n
</ResultSetBegin>

# output just after ending the loop over objects returned by the broker
#
<ResultSetEnd>
</DL>
<form method="GET" action="/Harvest/cgi-bin/search.cgi">
<table border="0" width="100%" bgcolor="#E0E0E0">
<tr>
 <td align="left" width="20%" nowrap>
  <input type=hidden name=broker value="$broker">
  <input type=hidden name=hp_url value="$hp_url">
  <input type=hidden name=caseflag value="$RQ{'caseflag'}">
  <input type=hidden name=wordflag value="$RQ{'wordflag'}">
#  <input type=hidden name=opaqueflag value="$RQ{'opaqueflag'}">
#  <input type=hidden name=descflag value="$RQ{'descflag'}">
#  <input type=hidden name=maxresultflag value="$RQ{'maxresultflag'}">
#  <input type=hidden name=maxobjflag value="$RQ{'maxobjflag'}">
#  <input type=hidden name=maxlineflag value="$RQ{'maxlineflag'}">
  <input type=hidden name=weightflag value="$RQ{'weightflag'}">
  <input type=hidden name=attribute value="@atts">
  <a href="$hp_url">Search:</a> <input type=text name=query size=15 value='$RQ{'query'}'> <input type=submit value="Go">
 </td>
 <td align="center" width="60%">
  $navigationBar
 </td>
  <td align="right" width="20%"><a href="$CFG{'HarvestUrl'}"><img alt="Powered by Harvest" border="0" src="$harvestIcon"></a></td>
</table>
</form>

</ResultSetEnd>

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