# - Google.pl


use Irssi qw(command_bind);
use LWP::UserAgent;
use Data::Dumper::Names;
use strict;
use vars qw($VERSION %IRSSI);

$VERSION = '1.10';
%IRSSI = (
    authors     => 'Oddbjørn Kvalsund, with mods by Tamara Temple',
    contact     => 'oddbjorn.kvalsund@hiof.no, tamara@tamaratemple.com',
    name        => 'Google',
    description => 'This script queries google.com and returns the results.',
    license     => 'Public Domain',
);

## Usage:
## /google [-p, prints to current window] [-<number>, number of searchresults returned] [-debug, turn on debugging mode] search-criteria1 search-criteria2 ...
##
## History:
## - Sun May 19 2002
##   Version 0.1 - Initial release
## -------------------------------

#-------------------------------------------------
my $nr_sites = 3; # Search-results returned
my $prefix = ""; # Message printed before results
my $debug = 0; # debugging turned off
#-------------------------------------------------

sub DebugPrint {
	my ($debugmsg) = @_;
	Irssi::print("%Y[$IRSSI{name}]%n " . $debugmsg) if $debug;
}

sub cmd_google {

        my ($data, $server, $witem) = @_;
        my $url = "";
	my $nr_sites = 10;
	my $i = 0;
	my (@lines, @pages);
	my $mode = "quiet";
	
	# Check for debug flag
	if ($data =~ /-debug/i) { $debug = 1; } else {$debug = 0}
	$data =~ s/-debug//ig; # remove debug flag from $data
	DebugPrint("Debugging turned on") if $debug;

	# If user supplied nr_sites, activate his setting
	if ( $data =~ /-(\d)+\s/ ) { $nr_sites = $1 }
	$data =~ s/-\d+//g; # remove nr_sites from $data
	DebugPrint("Number of sites: $nr_sites") if $debug;

	# Switch to public mode
	# and return error msg if invalid window
	if ( $data =~ /-p/ ) {
		$mode = "public";
		if ( ! $witem ) {
		  Irssi::active_win()->print("Must be run run in a valid window (CHANNEL|QUERY)");
		  return;
		}
	}
	$data =~ s/-p//g; # remove -p from $data
	DebugPrint("Mode of display: $mode") if $debug;

	# Format the query-string
	$data =~ URLEncode($data);
	my $query = $data;
	DebugPrint("Query: $query") if $debug;

	# Initialize LWP
	my $ua = new LWP::UserAgent;
	$ua->agent("Mozilla/5.0 " . $ua->agent);
	$ua->from('someone@example.com');
	$ua->default_header('Accept-Language'=> "en");
	$ua->timeout(10);
	DebugPrint(Dumper($ua)) if $debug;

	# Do the actual seach
        my $res = $ua->get("http://www.google.com/search?hl=en&q=$query");
		DebugPrint(irssi_safe(Dumper($res))) if $debug;
        my $content = $res->content;
		DebugPrint(irssi_safe(Dumper($content))) if $debug;

	# Replace <br> with newlines
	# and remove tags
        $content =~ s/\<br[^>]*\>/\n/g;
        #$content =~ s/\<.+?\>/ /sg;

	# Make array @pages of all search-results
        @lines = split("\n", $content);
        @pages = grep (/<h3 class="r">/, @lines);
		DebugPrint("Number of pages: ".$#pages+1) if $debug;

	# Remove empty entries in @pages
	for ($i=0;$i<=$#pages;$i++) {
		$pages[$i] =~ s/\s+.*//g;
		if ($pages[$i] =~ /(^\n|\s+\n)/){ splice(@pages, $i, 1) };
		if ($pages[$i] !~ /\./){ splice(@pages, $i, 1) };
	}

	# Strip extraneous html and words
	for ($i=0;$i<=$#pages;$i++) {
		$pages[$i] =~ s/^.*<h3 class="r">(.*)<\/h3>.*$/\1/g;
		$pages[$i] =~ s/^<a href="([^"]+)"[^>]*>/\1 /;
		$pages[$i] =~ s/<[^>]+>//g;
	}
	
	# Remove pages which aren't actually links
	@pages = grep(/https?:\/\//, @pages);

	if($nr_sites > $#pages) { $nr_sites = $#pages + 1};

	# Print pages to current window if public-mode specified
	# else display a private notice of returned pages
	if ( $mode eq "public") {
	  if ($prefix ne "") { $witem->command("/SAY $prefix") } ;
          for ($i=0; $i<$nr_sites; $i++) {
			$witem->command("/SAY $pages[$i]");
          }
	}
	else {
	  for ($i=0; $i<$nr_sites; $i++) {
		Irssi::active_win()->print("$pages[$i]");
	  }
	}
}

Irssi::command_bind google => \&cmd_google;

sub cmd_lmgtfy {
	my ($data, $server, $witem) = @_;
	return 0 if (!$witem);
	my $codeddata = URLEncode($data);
	my $result = sprintf("/say http://lmgtfy.com/?q=%s",$codeddata);
	$witem->command($result);
	return 1;
}

Irssi::command_bind('lmgtfy', 'cmd_lmgtfy');

sub URLEncode {
	my $theURL = $_[0];
	$theURL =~ s/([\W])/uc(sprintf("%%%02x",ord($1)))/eg;
	return $theURL;
}

sub irssi_safe {
	my $s = $_[0];
	$s =~ s/%/%%/g;
	return $s;
}