tweetmap.xq

module namespace tweetmap = "tweetmap";


declare function tweetmap:fetch-tweets($keyword as xs:string) as node()* {
  let $baseuri := "http://search.twitter.com/search.atom?geocode=52.1583,4.49306,200km&rpp=100&q="
  let $uri := concat($baseuri, $keyword)
  return doc($uri)//*:feed//*:entry
};

declare function tweetmap:get-geo($placename as xs:string) as xs:string {
  let $uri := replace("http://where.yahooapis.com/v1/places.q("XXX");count=1?format=xml&appid=oWI.U034EMJO6CnsA4vFEvJ5oYEKsoD0qLIRfcoEKOLLTqv9bh6jBHZn9GhdRA--","XXX",$placename),
      $centroid := exactly-one(doc($uri)//*:centroid)
  return concat($centroid//*:latitude, ",", $centroid//*:longitude)
};

declare function tweetmap:safe-geo($placename as xs:string) as xs:string { (: protect Yahoo against strange locations :)
  let $safename := translate($placename,";:=+&amp;{}[]<>!@#$%^*=+\/|'&quot;","     ()()()")
  return if (contains($placename,":")) then substring-after($placename,":") else tweetmap:get-geo($safename)
};

declare function tweetmap:order-tweets($keyword as xs:string) as node()* {
  let $tweets := tweetmap:fetch-tweets($keyword)
  let $retweets := for $tweet in $tweets
	where starts-with($tweet//*:title , "RT")
	return $tweet

  let $topics := distinct-values(for $tweet in $retweets return substring($tweet//*:title, 1,40))
  let $groups := for $topic in $topics
	let $group := for $tweet in $retweets
		where starts-with($tweet//*:title, $topic)
		return $tweet
	order by count($group)
	return $group
  
  let $newtweets := for $group in $groups
	let $tweet := exactly-one($group)
	(:$group/@rtcount=count($group):)
	(:order by count(group) descending:)
	return $tweet
	
  return ($newtweets, $tweets except $retweets) (: note the use of set operator "except" :)
};

declare function tweetmap:get-tweets($keyword as xs:string) as node() {
<result keyword="{$keyword}">{
  let $tweets := tweetmap:order-tweets($keyword)[position() <=9] (: TODO: restrict to the first 9 :)
  for $tweet in $tweets
  let 
      $location := exactly-one($tweet//*:location),
      $geo := tweetmap:safe-geo($location),
      $uri := $tweet//*:author//*:uri,
      $txt := $tweet//*:title/text(),
      $nr := (for $item at $pos in $tweets where $item//*:location=$tweet//*:location return $pos)[position() =1]
   return <tweet nr="{$nr}" geo="{$geo}" user="{$uri}" cnt="">{$txt}</tweet>
}</result>
};