pls_search

harvest是一个下载html网页得机器人

        }
        $formatPrice = sprintf("%.2f", $price);
        
        #
        # Format field widths for display.
        #
        eval '&formatFieldWidths';
        
        # NOTE: SKIPPING PRICING FOR REMOTE DOCUMENTS!!
        if (! $remoteDocUrl) {      
            $numCharTrans += length($hl);
            $numHitLines++;
        }
        


        # Display the title info for this record.
        #
        eval '&printTitleRow($index, $remoteDocUrl, $maxTitleLen)';

        $index++;       
     }          # foreach results line

    # Generate billing information.

    eval '&bill_hl';
}



##############################################################################
#
#   Name:           billHl
#   Description:    Writes Billing Record For Hitlist
#   Created:        --/--/--    Terry Hardgrave
#   Changed:        3/9/95      Herman Vandermolen
#                       Add a blank to the call to bill_one to represent
#                       the dbgroup of the account holder.
#                   3/20/95     Herman Vandermolen
#                       Don't do billing here. NOw done at dt leevl.
#   Returns:        (none)
#
##############################################################################

sub bill_hl
    {
    local ($acct, $dbid, $docid, $temp, $price, $nc);

    $dbid = "_HEADLINE";

    $docid = $numHitLines;

    $acct = $account;

                       # format price

    $temp  = $numHitLines * 0.01;
    $price = sprintf("%8.2f", $temp);

#   $numCharTrans is set in sub query

#   do "ibill.pl";
#   Note: Pass by parameters resulted in null values -- abandoned
#   $blank = ' ';
#   &bill_one($billfile,$acct,$blank,$blank,$blank,$dbid,$docid,$numCharTrans,$price);
}


##############################################################################
#
#   Name:           genPrice 
#   Description:    Generates a Price
#   Created:        --/--/--    Terry Hardgrave
#   Changed:        3/6/95      Herman Vandermolen
#                       Improve calculation of price per KC.
#   Returns:        (none)
#
##############################################################################

sub gen_price
{
    local ($qprice, $mode);

    if($price eq "") {
        $qprice = $price{$dbid};
        $mode = $mode{$dbid};

        if ($mode eq "KC") {                    # Pay per character
            $price = int(($size * $qprice) / 10) / 100;
        }
        else {                                  # Pay per document
            $price = $qprice;
        }
    }

    if (    ($price eq "") ||
            ($price < 0) ||
            ($price > 1000.00) ) {
        $price = -1;
    }

# print "price = $price, qprice = $qprice, size = $size\n";
}

##############################################################################
#
#  Name:        printColumnHeadings
#  Description: Print text for headings of search results columns.
#  Created:     ??
#  Changed:     2/7/95      Bryan Slavin
#                   Created versions without price information display;
#                   commented out original versions.
#  Changed:     2/11/95     Larry Cynkin
#                   Made into a routine instead of stuck in calling routine.
#               2/12/95     Larry Cynkin
#                   Eliminated bad $platmode check, since now checked in
#                   isKnownPlatmode().
#               3/15/95     Saad Mufti
#                   Get rid of PRICE heading
#               3/29/95     Herman Vandermolen
#                   Make display of price column configurable. 
#               4/6/95      Bryan Slavin
#                   Change format and calculate the number of tabs to insert
#                   to maintain field columns
#
##############################################################################

sub printColumnHeadings
{
    ($maxTitleLen) = @_;

    #
    # Calculate the number of tabs needed to pad output.
    #
    $titlePad = "";
    $titlePad = " " x ($maxTitleLen - 7);

    if ($bPriceColumn == 1) {
        print " VAL " . "DOCUMENT" . "$titlePad" . "PRICE   " . "DB      " . "  SIZE" . "<HR>";
    }
    else {
        print " VAL " . "DOCUMENT" . "$titlePad" . "DB      " . "  SIZE" . "<HR>";
    }
}


##############################################################################
#
#  Name:        printTitleRow
#  Description: Print a line of hitlist info, for one hit record.
#  Created:     ??
#  Changed:     2/7/95      Bryan Slavin
#                   Created versions without price information display;
#                   commented out original versions.
#               2/12/95     Larry Cynkin
#                   Made into a routine instead of stuck in calling routine.
#                   Eliminated bad $platmode check, since now checked in
#                   isKnownPlatmode().
#               2/16/95     Bryan Slavin
#                   Made $nodeName variable so user does not have to see the
#                   port number.
#               2/21/95     Larry Cynkin
#                   Added platmode to output so document has enough information
#                   to return to search/operations form.
#               02/27/95    Naval Deshbandhu
#                   make URL path relative to /plweb 
#               3/3/95      Herman Vandermolen
#                   Display .. instead of : as node name & port separator.
#                   Mosaic doesn't like the ':' character.
#               3/3/95      Herman Vandermolen
#                   Add billing dbgroup to HREF.
#               3/4/95      Saad Mufti
#                   Add the URL for remote documents as an argument to the
#                   script pointed to by the HREF.
#               3/15/95     Saad Mufti
#                   Take out printing of price.
#               3/29/95     Herman Vandermolen
#                   Make display of price column configurable. 
#               3/29/95     Herman Vandermolen
#                   Change argument order in HREF to make docID more 
#                   visible for administrators. 
#               4/4/95      Herman Vandermolen
#                   Set price to $0.00 if price is negative.
#               4/6/95      Bryan Slavin
#                   Change format of output and pad hitlist title
#
##############################################################################

sub printTitleRow
{
    local($index, $remoteDocUrl, $maxTitleLen) = @_;        # if non-NULL, contains the URL for the
                                    # remote document that corresponds to the
                                    # doc id for this document

    ## encode special characters in remote document URL
    #$remoteDocUrl =~
    #    s/([ ~@#&\!\^\*\(\)\>\<\?\+\/=:'])/sprintf("%%%lx",ord($1))/eg;
    $remoteDocUrl =~ s/([\t])/sprintf("%%%lx",ord($1))/eg; # encode tabs

    #
    # Note: idoc is executed from local node so local accounting can be invoked,
    # i.e., HREF starts with "idoc.pl?...", rather than
    # "http://$xnode/plweb-cgi/idoc.pl?...".
        

    #########################################################
    # NOTE: If both the node AND the port are desired to be displayed,
    #       comment out the line below.
    #
    ($nodeName, $nodePort) = split(/:/,$xnode);

    print "PLWeb Result: $remoteDocUrl\t$relevance\t$chCount\n";
}


##############################################################################
#
#  Name:        prepHitInfo
#  Description: Do necessary preprocessing of search result information.
#  Outputs:     Possibly changes current $hl.  Sets $short_dbalias.
#  Created:     ??
#  Changed:     2/11/95     Larry Cynkin
#                   Made into a routine instead of stuck in calling routine.
#
##############################################################################

sub prepHitInfo
{
    #
    # Handle records with no title information
    #
    if (!($hl =~ /\S/)) {
        $hl = "** UNTITLED **";
    }
    
    #
    # Eliminate any carriage return.
    #
    if ($hl =~ /\n$/) {
        chop($hl);
    }
    
    #
    # Converts $dbalias to have only 7 characters when printing out
    #
#   if (length($dbalias) > 7) {
#       $short_dbalias = substr($dbalias, 0, 7);
#   }
#   else {
#       $short_dbalias = $dbalias;
#   }
}


##############################################################################
#
#  Name:        formatFieldWidths
#  Description: Pad fields with spaces to create proper output width.
#  Outputs:     (none)
#  Created:     ??
#  Changed:     2/12/95     Larry Cynkin
#                   Made into a routine instead of stuck in calling routine.
#                   Changed to use 'x' operator instead of looping for each
#                   added character.
#
##############################################################################


sub formatFieldWidths
{
    # Field widths for display
    $DATABASE_LEN  = 6; 
    $RELEVANCE_LEN = 4;
    $CH_COUNT_LEN  = 6;
    $PRICE_LEN     = 5;

    local($myLength);

    if (($myLength = length($dbalias)) < $DATABASE_LEN) {
        $short_dbalias = $dbalias . (' ' x ($DATABASE_LEN - $myLength)) ;
    }
    else {
        $short_dbalias = substr($dbalias,0,6);
    }
    
    if (($myLength = length($relevance)) < $RELEVANCE_LEN) {
        $relevance = (' ' x  ($RELEVANCE_LEN - $myLength)) . $relevance;
    }
    
    if (($myLength = length($chCount)) < $CH_COUNT_LEN) {
        $chCount = (' ' x  ($CH_COUNT_LEN - $myLength)) . $chCount;
    }
    
    if (($myLength = length($formatPrice)) < $PRICE_LEN) {
        $formatPrice = (' ' x  ($PRICE_LEN - $myLength)) . $formatPrice;
    }
}


##############################################################################
#
#  Name:        isKnownPlatmode
#  Description: Determine if $platmode is recognized.
#  Parms:       platmode
#  Outputs:     Return value is true if recognized, false if not.
#  Created:     2/12/95     Larry Cynkin
#  Changed:     
#
##############################################################################

sub isKnownPlatmode
{
    local ($myPlatmode) = @_;

    $myPlatmode eq "unix" || $myPlatmode eq "mac" || $myPlatmode eq "msw";
}