/ 
WWW-ContentRetrieval-0.07
/ WWW::ContentRetrieval

NAME

WWW::ContentRetrieval - WWW robot plus text analyzer

SYNOPSIS

  use WWW::ContentRetrieval;
  use Data::Dumper;
  $robot = WWW::ContentRetrieval->new($desc,
				 {
				     TIMEOUT    => 3,
				     HTTP_PROXY => 'http://fooproxy:2345/',
				 });
  print Dumper $robot->retrieve( $query );

DESCRIPTION

WWW::ContentRetrieval combines the power of a www robot and a text analyzer. It can fetch a series of web pages with some attributes in common, for example, a product catalogue. Users write down a description file and WWW::ContentRetrieval can do fetching and extract desired data. This can be applied to do price comparison or meta search, for instance.

METHODS

new

  $s =
    new WWW::ContentRetrieval(
			      $desc,
			      {
				  TIMEOUT    => 3,
				  # default is 10 seconds.

				  HTTP_PROXY => 'http://fooproxy:2345/',

				  DEBUG      => 1,
				  # non-zero to print out debugging msgs
			      });

retrieve

$s->retrieve($query) returns an anonymous array of retrieved data.

You may use Data::Dumper to see it.

EXPORT

genDescTmpl

generates a description template.

Users can do it in a command line,

perl -MWWW::ContentRetrieval -e'print genDescTmpl'

DESC FILE TUTORIAL

OVERVIEW

Currently, this module uses Perl's native anonymous array and hash for users to write down site descriptions. Let's see an example.

Suppose the product's query url of "foobar technology" be http://foo.bar/query.pl?encoding=UTF8&product=blahblahblah, then the description is like the following:

$items = <<'ITEMS';

 match=<a href="(.+?)">(.+)</a>
 site=$1
 url=$2
 replace(url)=s/http/ptth/;

 match=<img src="(.+?)">
 photo="http://foo.bar/".$1

ITEMS

$desc ={
   NAME => "foobar tech.",
   NEXT => [
    'query.pl' => 'detail.pl',
   ],
   POLICY => [
     'http://foo.bar/foobar_product.pl' => \&extraction_callback,
     'http://foo.bar/product_detail.pl' => \$items,
   ],
   METHOD => 'GET',
   QHANDL => 'http://foo.bar/query.pl',
   PARAM => [
    ['encoding', 'UTF8'],
   ],
   KEY => 'product',
};

NAME

The name of the site.

NEXT

NEXT is an anonymous array containing pairs of (this pattern => next pattern). If the current url matches /this pattern/, then text is searched for urls that match /next pattern/ and these urls will be queued for next retrieval.

POLICY

POLICY stores information for a certain page's extraction. It is composed of pairs of (this url's pattern => callback function) or (this url's pattern => reference to retrieval settings). If the current url matches /this pattern/, then this modules will invoke the corresponding callback or extract data according to the retrieval settings given by users.

In simple cases, users only need to write down the retrieval settings instead of a callback function. Retrieval settings contains lines of instructions in a /key=value/ format. Here's an example.

$setting=<<'SETTING';
match=<a href="(.+?)">(.+?)</a>
url=$1
desc="<".$2.">"
replace(url)=s/http/ptth/;
SETTING

Then the module will try to match the pattern in the retrieved page, and assigns the keys with matched values. And, replace follows a substitution matcher, which can transform the specified extracted data.

If users have to write callback functions for more complex cases, here is the guideline:

WWW::ContentRetrieval passes two parameters to a callback function: a reference to page's content and page's url.

E.g.

sub my_callback{
    my ($textref, $thisurl) = @_;
    while( $$textref =~ /blahblah/g ){
         do some blahblahs here.
    }
    return an array of hashes, with keys and extracted information.
}

N.B. Callback's return value should be like the following

[
 {
  PRODUCT => "foobar",
  PRICE => 256,
 },
 {
  ...
  }
];

If users need WWW::ContentRetrieval to retrieve next page, e.g. dealing with several pages of search results, push an anonymous hash with only one entry: _DTLURL

{
 _DTLURL => next url,
}

See also t/extract.t, t/robot.t

METHOD

Request method: GET, POST, or PLAIN.

QHANDL

Query Handler, Url of the query script.

PARAM

Constant script parameters, excluding user's queries.

KEY

Key to user's query strings, e.g. product names

TO DO

  • A small language for site description

SEE ALSO

WWW::ContentRetrieval::Spider, WWW::ContentRetrieval::Extract

COPYRIGHT

xern <xern@cpan.org>

This module is free software; you can redistribute it or modify it under the same terms as Perl itself.