File Coverage

blib/lib/WWW/Google/News/TW.pm

Criterion	Covered	Total	%
statement	58	118	49.1
branch	6	22	27.2
condition			n/a
subroutine	10	11	90.9
pod	2	3	66.6
total	76	154	49.3

package WWW::Google::News::TW;

use utf8;
use strict;
use warnings;

require Exporter;

our @ISA       = qw(Exporter);
our @EXPORT_OK = qw(get_news get_news_for_topic get_news_for_category);
our $VERSION   = '0.12';

use Carp;
use LWP;
use URI::Escape;
use Encode;

sub get_news {
    # Web version: http://news.google.com.tw/news?ned=tw
    # plain text version : http://news.google.com.tw/news?ned=ttw
  my $url = 'http://news.google.com.tw/news?ned=ttw';
  my $ua = LWP::UserAgent->new;
     $ua->agent('Mozilla/5.0');
  my $response = $ua->get($url);
  my $results = {};
  return unless $response->is_success;

  my $re1 =  '
  my $re2 =  '([^<]*)'.
	    '([^<]*)'.
	    '\s?([^<]*)'.
	    '([^<]*)...';

  my $content = $response->decoded_content;
  $content = $response->content if (not defined $content);
  my @sections = split /($re1)/m,$content;
  my $current_section = '';
  foreach my $section (@sections) {
    if ($section =~ m/$re1/m) {
      $current_section = $1;
      $current_section =~ s/ //g; # or put this  (.*?)(?: )? in re1
    } else {
      my @stories = split /($re2)/mi,$section;
      foreach my $story (@stories) {
        if ($story =~ m/$re2/mi) {
          if (!(exists($results->{$current_section}))) {
            $results->{$current_section} = [];
          }
          my $story_h = {};
	  my( $url, $headline, $source, $update_time, $summary ) = ( $1, $2, $3, $4, $5 );
          $story_h->{url} = $url;
          $story_h->{headline} = $headline;
	  $story_h->{source} = $source;
	  $story_h->{source} =~ s/ -//g;
	  $story_h->{update_time} = $update_time;
	  $story_h->{summary} = $summary;
          push(@{$results->{$current_section}},$story_h);
        }
      }
    }
  }
  return $results;
}

sub get_news_for_topic {

    my $topic = uri_escape( $_[0] );

    my @results = ();
    my $url = "http://news.google.com.tw/news?hl=zh-TW&ned=ttw&q=$topic";
    my $ua = LWP::UserAgent->new();
    $ua->agent('Mozilla/5.0');

    my $response = $ua->get($url);
    return unless $response->is_success;

    my $re1 =  '
    my $re2 =  '(.*?)'.
	'([^<]*)'.
	'\s?([^<]*)'.
	'(.*?)...';

    my $content = $response->decoded_content;
    $content = $response->content if (not defined $content);
    my( $section ) = ( $content =~ m/$re1/s );
    $section =~ s/\n//g;
    my @stories = split /($re2)/mi,$section;

    foreach my $story (@stories) {
	if ($story =~ m/$re2/i) {
	    my $story_h = {};

	    my( $url, $headline, $source, $update_time, $summary ) = ( $1, $2, $3, $4, $5 );
	    $source =~ s/ / /g;
	    $source =~ s/\s+/ /g;
	    $update_time =~ s/ / /g;
	    $update_time =~ s/\s+/ /g;
	    $update_time =~ s/-//g;
	    $headline =~ s#<.+?>##gi;
	    $summary =~ s#<.+?>##gi;

	    $story_h->{url} = $url;
	    $story_h->{headline} = $headline;
	    $story_h->{source} = $source;
	    $story_h->{update_time} = $update_time;
	    $story_h->{summary} = $summary;

	    push(@results,$story_h);

	}
    }

    return \@results;

}

sub get_news_for_category {
    # Web version: http://news.google.com.tw/news?ned=tw
    # plain text version : http://news.google.com.tw/news?ned=ttw
  my $topic = $_[0];
  my $url = 'http://news.google.com.tw/news?ned=ttw&topic='.$topic;
  my $ua = LWP::UserAgent->new;
     $ua->agent('Mozilla/5.0');
  my $response = $ua->get($url);
  my $results = [];
  return unless $response->is_success;

  my $re1 = '(.*?)';
  my $re2 =  '([^<]*)'.
	    '([^<]*)'.
	    '\s?([^<]*)'.
	    '([^<]*)....*?'.
	    ']*)>([^<]*)';
  my @sections = split /($re1)/s,$response->content;
  my $current_section = '';
  foreach my $section (@sections) {
    if ($section =~ m/$re1/s) {
      $current_section = $1;
      my @stories = split /($re2)/si,$current_section;
      foreach my $story (@stories) {
        if ($story =~ m/$re2/si) {
          my $story_h = {};
	  my( $url, $headline, $source, $update_time, $summary, $related_url, $related_news) =
	    ( $1, $2, $3, $4, $5, $6, $7 );
          $story_h->{url} = $url;
          $story_h->{headline} = $headline;
	  $story_h->{source} = $source;
	  $story_h->{source} =~ s/ -//g;
	  $story_h->{update_time} = $update_time;
	  $story_h->{summary} = $summary;
	  $story_h->{related_url} = $related_url;
	  $story_h->{related_news} = $related_news;
          push(@{$results},$story_h);
        }
      }
    }
  }
  return $results;
}
1;

__END__