NAME
WWW::Search - Virtual base class for WWW searches
DESCRIPTION
This class is the parent for all access method supported by the WWW::Search library. This library implements a Perl API to web-based search engines.
Current search engines supported include AltaVista (both web and news), Dejanews, Excite (web only), HotBot (web only), Infoseek (e-mail, web, and news) and Lycos.
Search results are limited and there is a pause between each request for results to avoid overloading either the client or the server.
Sample program
Using the library should be straightforward: Here's a sample program:
my($search) = new WWW::Search('AltaVista');
$search->native_query(WWW::Search::escape_query($query));
my($result);
while ($result = $search->next_result()) {
print $result->url, "\n";
};Results are objects of WWW::SearchResult (see WWW::SearchResult) .
SEE ALSO
For more details see LWP.
For specific search engines, see WWW::Search::TheEngineName (replacing TheEngineName with a particular search engine).
For details about the results of a search, see WWW::SearchResult.
METHODS AND FUNCTIONS
new
To create a new WWW::Search, call $search = new WWW::Search('SearchEngineName'); where SearchEngineName is replaced with a particular search engine. For example: $search = new WWW::Search('AltaVista');
If no search engine is specified a default will be chosen for you.
The next step is usually: $search->native_query('search-engine-specific+query+string');
native_query
Specify a query (and optional options) to the current search object. The query and options must be escaped; call "escape_query" in WWW::Search to escape a plain query. The actual search is not actually begun until results or next_result is called.
Example:
$search->native_query('search-engine-specific+query+string',
{ option1 => 'able', option2 => 'baker' } );The hash of options following the query string is optional. Both the query string and the hash of options are interpreted in search-engine-specific manner.
Details about how the search string and option hash are interpreted in the search-engine-specific manual pages (WWW::Search::SearchEngineName).
After native_query, the next step is usually:
@results = $search->results();or
while ($result = $search->next_result()) {
# do_something;
};results
Return all the results of a query as a reference to array of SearchResult objects.
Example: @results = $search->results(); foreach $result (@results) { print $result->url(), "\n"; };
On error, results() will return undef and set response() to the HTTP response code.
next_result
Return each result of a query as a SearchResult object.
Example: while ($result = $search->next_result()) { print $result->url(), "\n"; };
On error, results() will return undef and set response() to the HTTP response code.
response
Return the HTTP Response code for the last query (see HTTP::Response). If the query returns undef, errors could be reported like this:
my($response) = $search->response();
if ($response->is_success) {
print "no search results\n";
} else {
print "error: " . $response->as_string() . "\n";
};Note: even if the back-end does not involve the web it should return HTTP::Response-style codes.
seek_result($offset)
Set which result next_result should return (like lseek in Unix). Results are zero-indexed.
The only guaranteed valid offset is 0 which will replay the results from the beginning. In particular, seeking past the end of the current cached results probably won't do what you might think it should.
Results are cached, so this doesn't re-issue the query or cause IO (unless you go off the end of the results). To re-do the query, create a new search object.
Example: $search->seek_result(0);
maximum_to_retrieve
The maximum number of hits to return (approximately). Queries resulting in more than this many hits will return the first hits, up to this limit.
Defaults to 500.
Example: $max = $seach->maximum_to_retrieve(100);
escape_query
Escape a query. Before queries are made special characters must be escaped so that a proper URL can be formed.
This is like escaping a URL but all non-alphanumeric characters are escaped and and spaces are converted to "+"'s.
Example: $escaped = Search::escape_query('+lsam +replication'); (Returns "%22lsam+replication%22").
See also unescape_query.
unescape_query
Unescape a query. See escape_query for details.
Example: $unescaped = Search::unescape_query('%22lsam+replication%22'); (Returns "+lsam +replication").
See also unescape_query.
http_proxy
Set-up an HTTP proxy (Perhaps for connections from behind a firewall.)
This routine should be called before the first retrival is attempted.
Example:
$search->http_proxy("http://gateway:8080");setup_search (PRIVATE)
This internal routine does generic Search setup. It calls native_setup_search to do back-end specific setup.
user_agent($NON_ROBOT) (PRIVATE)
This internal routine creates a user-agent for dervived classes that query the web. If $NON_ROBOT, a normal user-agent (rather than a robot-style user-agent) is used.
Back-ends should use robot-style user-agents whereever possible. Also, back-ends should call user_agent_delay every page retrival to avoid swamping search-engines.
user_agent_delay (PRIVATE)
Derived classes should call this between requests to remote servers to avoid overloading them with many, fast back-to-back requests.
retrieve_some (PRIVATE)
An internal routine to interface with native_retrieve_some. Checks for overflow.
IMPLEMENTING NEW BACK-ENDS
WWW::Search supports back-ends to separate search engines. Each back-end is implemented as a subclass of WWW::Search. WWW::Search::AltaVista provides a good sample back-end.
A back-end usually has two routines, native_retrieve_some and native_setup_search.
native_retrieve_some is the core of a back-end. It will be called periodically to fetch URLs. Each call it should fetch a page with about 10 or so hits and add them to the cache. It should return the number of hits found or undef when there are no more hits.
Internally, native_retrieve_some typically will parse the HTML, extract the links and descriptions, then find the ``next'' button and save the URL. See the code for the AltaVista implementation for an example.
native_setup_search is invoked before the search. It is passed a single argument: the escaped, native version of the query.
The front- and back-ends share a single object (a hash) The back-end can change any hash element beginning with underscore, and {response} (an HTTP::Response code) and {cache} (the array of WWW::SearchResult objects caching all results).
If you implement a new back-end, please let the authors know.
BUGS AND DESIRED FEATURES
The bugs are there for you to find (some people call them Easter Eggs).
Desired features:
- A portable query language. A portable language would easily allow you to move queries easily between different search engines. A good query abstraction is non-trivial and won't be done anytime soon at ISI, so if you want to take a shot at it, please let me know.
AUTHOR
WWW::Search is written by John Heidemann, <johnh@isi.edu>.
Back-ends and applications for WWW::Search have been done by John Heidemann, Wm. L. Scheding, Cesare Feroldi de Rosa, and GLen Pringle.
COPYRIGHT
Copyright (c) 1996 University of Southern California. All rights reserved.
Redistribution and use in source and binary forms are permitted provided that the above copyright notice and this paragraph are duplicated in all such forms and that any documentation, advertising materials, and other materials related to such distribution and use acknowledge that the software was developed by the University of Southern California, Information Sciences Institute. The name of the University may not be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.