modern.cf

harvest是一个下载html网页得机器人

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


# PER-OBJECT DEFINITIONS
# ======================
#
# 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
#    $objectType  Type of object if not HTML
#    $objectSize  Size of object
#    $objectDate  Last modification date of object
#
<PrintObject>
<DT>$objectnum <A HREF="$url">$description</A>
&nbsp;&nbsp;<FONT size=-1 color=#606060>$objectType</FONT>
&nbsp;&nbsp;<FONT size=-1 color=#606060>$objectSize</FONT>
&nbsp;$objectWeight
<DD>$attributes
<DD>$opaque
<?$cs_urlX>
<DD>
<B>indexing data:</B>
<A HREF="$cs_a://$cs_h/Harvest/cgi-bin/displaySOIF.cgi?object=$cs_p&query=$html_query">formatted</A>
&middot;
<A HREF="$cs_a://$cs_h/Harvest/cgi-bin/displaySOIF.cgi?object=$cs_p&style=plain">plain</A>
</?$cs_urlX>
<DD><FONT size=-1 color=#006000>$url</font>
&nbsp;<FONT size=-1 color=#606060>$objectUpdate</FONT>
<BR>
<BR>
</PrintObject>

# This definition is eval'd for each opaque (i.e. matched) line retruned by
# the broker.  It is intended to be used to remove SOIF attributes
# and the 'Matched line' string from the output.
# The results of these operations are stored in $opaque.
#
<MatchedLineSub>
#1. Completely remove lines with @FILE
s/^(.*)\@FILE(.*)$/ /;
#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/(.*)/... $1/;              #OK, add dots.
}
#highlight search words.
for ($i=0; $i<=$#searchwords; $i++) {
   s/(\b$searchwords[$i]\b)/<font color=#800000>$1<\/font>/ig;
}
</MatchedLineSub>

# PerObjectFunction is eval'd before every object is printed out.
#
<PerObjectFunction>
# init output variables
$objectType="";
$objectSize="";
$objectUpdate="";
$objectWeight = "";

$opaquePerObjectCount = 1;
$lastOpaqueObject     = "";

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

# highlight search words.
for ($i=0; $i<=$#searchwords; $i++) {
   s/(\b$searchwords[$i]\b)/<b>$1<\/b>/ig;
}

# Create the HTML code to print the weight balls
if ($maxWeight > 0 && $weightflag)
{
  $nWeight = int($weight * 5 / $maxWeight);
  for ($idxWeight = 1; $idxWeight <= $nWeight; $idxWeight++)
       { $objectWeight .= "<img alt=\"*\" src=\"$weightIcon\">"; }
#  $objectWeight = "<font size=-1 color=#606060>".int($weight * 100 / $maxWeight)."%</font>";
}
# format matched lines
if ($opaque ne '') {
   $opaque = "<strong>matches:</strong> $opaque..."
}
</PerObjectFunction>

# PerAttributeFunction is eval'd before the Attributes are printed out.
#
<PerAttributeFunction>
# Remove all empty lines
$val =~ s/^(\s*)//mg;

# Now prepare the object's type and size to display at the description.
if ($att eq "type") {                           ## show file type
   $objectType = "[".$val."]" if !($val =~ /HTML|HTTP-Query/);
   $att="";                                     # avoid <FormatAttribute>
} elsif ($att eq "file-size") {                 ## calculate 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
   $att="";
} elsif ($att eq "last-modification-time") {    ## calculate last modification date
   local(@date,@months,$month,$minute,$year);
   @date = localtime($val);
   @months = ('Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec');
   $month = $months[$date[4]];
   $minute = $date[1];
   $minute = "0$minute" if $minute < 10;
   $year = $date[5]+1900;
   $objectUpdate = "$date[3]-$month-$year"; # $date[2]:$minute";
   $att="";
} else {                                        ## other attributes
   # 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)*/$1.../;
   #highlight search words.
   for ($i=0; $i<=$#searchwords; $i++) {
      $val =~ s/(\b$searchwords[$i]\b)/<font color=#800000>$1<\/font>/ig;
   }
}
</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' or the number of returned objects equals the
# 'maxobjflag' value of the query.html form.
#
<TruncateWarning>
<P><STRONG>WARNING: The search results were truncated at $nopaquelines
matched lines and $nreturned found objects.</STRONG><P>\n
</TruncateWarning>

# A warning message printed only when the broker returned 0 results.
#
<EmptySetWarning>
<H2>Your query <font color="#800000">$html_query</font> did not match any document.</H2>
#<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.
<P>
Suggestions:
<UL>
<LI>Make sure all words are spelled correctly
<LI>Try different, less or more general keywords
<LI>Select the "particial word search" on the search page to allow searches in words.
<LI>Allow "spelling errors" on the search page.
</UL>
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>
<H2>No query entered.</H2>
Please enter one or several search words.<BR><BR>\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>

# Error message if broker is not available.
#
<BrokerDown>
<H2>Search engine temporarily not available.</H2>
Sorry, the Broker at <STRONG>$host, port $port</STRONG> is unavailable.<P>
Please try again later.<P>\n
</BrokerDown>

# Message returned if the broker sends back a
#      111 - PARSE ERROR
# reply.
#
<ParseError>
</pre>
<H2>Your query <font color="#800000">$html_query</font> does not have the proper syntax.</H2>
<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>