zquery-gils.pl.in

harvest是一个下载html网页得机器人

#!@PERL@ -w
#
# Query a z39.50 server with zoomsh.
# 5dec1999/kjl
#
# Usage:
#   for production:
#     put it into your /cgi-bin/ and use it via your browser as
#	http://localhost/cgi-bin/zquery.pl
#   for limited debugging:
#     zquery.pl [querystring]
#
# Installation:
# Edit path to point to the directory where your zoomsh binary is and
# put this script to your cgi-bin directory of the http daemon.
# Additionally, you can modify print_html_* () to customize the appearance
# of the html pages.
#
# Some notes about parsing output from zoomsh:
# read_hash_zoomsh () reads the output from zoomsh and builds a hash
# containing data received from z39.50 server. It uses a simple and
# inefficient parse mechanism without validity check of the returned data.
# It works well enough to experiment with the search system.
#
# If you want to roll your own parser, here are some non special cases:
#
# - lines are continued at next line with leading whitespaces like this:
#   title: This is a very long title line which does not fit in one line and
#     it will be continued at next line, like this
#
# - long lines can look like this:
#   linkage: http://do.main/path/page.html
#	       or
#   linkage:
#     http://do.main/path/page.html
#
# - lines longer than 77 characters are broken into two lines by putting a
#   "=" at position 77 and continued at next line.
#

use CGI 'param';
use CGI::Carp;
use CGI::Carp qw(fatalsToBrowser);
use URI::Escape;
use integer;
use strict;

$::DEBUG = param ('DEBUG');
$::DEBUG = 0 unless ($::DEBUG);

#############################
# configuration starts here #
#############################
# where am I installed?
# this script will find out its location when started as cgi script.
# If you want to use it under mod_perl, move this script to your perl
# modules directory and change the path.
##########################################
$::ME = $ENV { 'SCRIPT_NAME' };
$::ME = "/Harvest/cgi-bin/zquery.pl" unless $::ME;

# path to raw zebra data
#$::DATA_STORE = "http://tengu.arco.de/Harvest/brokers/tengu.arco.de/objects/";

# prefix to supplemental html pages
$::HTML_BASE = "/Harvest/brokers/zquery";

# make sure zoomsh is in path
$ENV{'PATH'} = "@prefix@/bin";

# where is the zebrasrv and the database to use
$::ZURL = "localhost:9999/harvest";	 # how to contact z39.50 server

# some default values
# you can override these with the values of your query form
$::OP	    = "and";	# default boolean operator for queries
#$::OP	     = "or";	 # "and" and "or" allowed
$::N_LINES  = 3;	# default number of lines per hit to show
$::PER_PAGE = 10;	# default number of hits per page to show
# end of basic configuration

# exit after this number of seconds
alarm (60);

# parse arguments
# no arguments
# we must be called as a cgi script
$::query = param ('query') if ($#ARGV == -1);

# we found some arguments
$::query = join (" ", @ARGV) if ($#ARGV >= 0);

if ($::query) {

    # page to show
    $::page = param ('page');
    $::page = 0 if (! $::page);

    # number of hits to show per page
    $::per_page = param ('per_page');
    $::per_page = $::PER_PAGE if (! $::per_page);

    # number of lines per entry to show
    $::n_lines = param ('n_lines');
    $::n_lines = $::N_LINES if (! $::n_lines);

    # how to combine query words
    $::op = param ('op');
    $::op = $::OP if (! $::op);

    # build query string in zoomsh notation
    $::z_query = build_query ($::query, $::op);

    $::start = $::page * $::per_page;
    $::zoomsh =  "zoomsh \"connect $::ZURL\" \"search $::z_query\"";
    $::zoomsh .= " \"show $::start $::per_page\" quit";

    ($::num_hit, $::aref) = read_hash_zoomsh ();
    @::hits = @{$::aref};

    # output
    print_html_header ("Search result for \"$::query\"");
    print_html_query_form ($::query);
    print_z3950_query ($::z_query)		  if ($::DEBUG);
    print_num_hit ($::query, $::num_hit);
    print_hit ($::page, $::per_page, \@::hits)	  if ($::num_hit && $::num_hit > 0);

    # print pager navigation links
    print_html_page_index (uri_escape ($::query), $::num_hit,
			   $::page, $::per_page)
	if ($::num_hit && $::num_hit > $::per_page);

} else { # if ($query)
    print_html_header ("Z39.50 Search");
    print_html_query_form ();
}

print_html_footer ();
#############################################
# execution ends here			    #
#############################################

#############################################
# subroutines				    #
#############################################

#############################################
# html header				    #
#   args: title				    #
#############################################
sub print_html_header {

    my $title = shift;

    print << "EOF";
Content-type: text/html


<html>
<head>
<title>$title
</title>
<style><!--
div.pager A { color: black; }
//--></style>
</head>
<body
 bgcolor="#ffffff"
 text="#000000"
 link="#0000ff"
 vlink="#000080"
 alink="#ff0000"
>

EOF
}


#############################################
# html end tags				    #
#############################################
sub print_html_footer {

    print << "EOF";
<hr size=3 width=90% noshade>
<center>
<b>
<a href="$::HTML_BASE/help.html"><nobr>[Help formulating queries]</nobr></a>
<a href="$::HTML_BASE/register.html"><nobr>[Register new pages]</nobr></a>
<a href="$::HTML_BASE/contact.html"><nobr>[Contact information]</nobr></a>
<a href="$::HTML_BASE/about.html"><nobr>[About this search system]</nobr></a>
</b>
</center>
<br>
<br>
</body>
</html>
EOF
}


#############################################
# query form				    #
#   args: default querystring value	    #
#############################################
sub print_html_query_form {

    my $query = shift;

    $::op = $::OP unless ($::op);
    $::n_lines = $::N_LINES unless ($::n_lines);
    $::per_page = $::PER_PAGE unless ($::per_page);

    print << "EOF";
<h1>Harvest Search</h1>

<form action=$::ME method=get>

<center><nobr>
<input type=text name=query
EOF

    if ($query) {
	$query =~ s/\"/&quot;/g;
	print " value=\"$query\"";
    }

    print << "EOF";
 size=40 maxsize=200>
<input type=submit name=what value=Search>
</nobr></center>

<input type=hidden name=op value=$::op>
<input type=hidden name=n_lines value=$::n_lines>
<input type=hidden name=per_page value=$::per_page>
<input type=hidden name=DEBUG value=$::DEBUG>
</form>

EOF
}


#############################################
# print z39.50 query sent to server	    #
#############################################
sub print_z3950_query {

    my $query = shift;

    if ($query) {
	print "Query sent to Z39.50 server: <b>$query</b><br>\n";
    } else {
	print "No query sent to Z39.50 server.<br>\n";
    }
}


#############################################
# print number of hits for a query	    #
# args: querystring, number_of_hits	    #
#############################################
sub print_num_hit {

    my $query = shift;
    my $num_hit = shift;

    print "<b>$query:</b> ";

    if (! $num_hit) {
	print "no match";
    } elsif ($num_hit == 1) {
	print "$num_hit match";
    } else {
	print "$num_hit matches";
    }

    print " found\n";
    print "<br><br>\n";
}


#############################################
# print hits				    #
# arguments:				    #
#   number of page			    #
#   hits per page			    #
#   hash of hit arrays			    #
#############################################
sub print_hit {

    my ($page, $per_page, $aref, @hits, $i, $entry, $key); #, $obj_data);
    my ($rank, $title, $dateOfLastModification, $bytes, $linkage, $sampleText);
    my (@date);

    $page = shift;
    $per_page = shift;
    $aref = shift;
    @hits = @{$aref};

    $entry = $page * $per_page;

    for $i (0 .. $#hits) {

	$entry++;

	$rank			= $hits [$i] { 'rank' };
	$linkage		= $hits [$i] { 'linkage' };
	$dateOfLastModification = $hits [$i] { 'dateOfLastModification' };
	$bytes			= $hits [$i] { 'bytes' };
	$title			= $hits [$i] { 'title' };
	$sampleText		= $hits [$i] { 'sampleText' };

	@date = gmtime ($dateOfLastModification);
	$date[4]++;
	$date[5] += 1900;
	$dateOfLastModification = $date[5] . "-" . $date[4] . "-" . $date[3];

	print << "EOF";
<b>$entry. <a href="$linkage">$title</a></b>
<br>
$sampleText
<br>
$linkage
<br>
Document size: <b>$bytes Bytes</b>,