ro_recommend
Fork

(Original repository, No fork origin)

Commit

簡単なparserを実装

		@@ -5,7 +5,7 @@ import scala.actors.Actor
5	5	import com.parrotstudio.recommend.ro.model._
6	6
7	7	class RecommendWorker(val index: Int) extends Actor with RecommendMessage {
8		- val parser = """.<div id="more">(.?)</div>.*""".r
	8	+ val parser = """.<title>Angel, alone.孤独な天使.-(.)-</title>.<div id="more">(.?)</div>.*""".r
9	9
10	10	def act {
11	11	loop {

		@@ -28,17 +28,22 @@ class RecommendWorker(val index: Int) extends Actor with RecommendMessage {
28	28
29	29	def parse(p: Page): Article = {
30	30	p.page match {
31		- case parser(art) => Article("本文", art)
32		- case _ => Article("", "")
	31	+ case parser(title, ext) => Article(normalize(title), "本文", normalize(ext))
	32	+ case _ => new Article()
33	33	}
34	34	}
35	35
36	36	def analysis(a: Article): Terms = {
37	37
38		- new Terms()
	38	+ Terms()
39	39	}
40	40
41	41	def createVector(t: Terms) = {
42	42
43	43	}
	44	+
	45	+ private def normalize(text: String): String = {
	46	+ text.trim.replaceAll("<br.?/?>", "\n").replaceAll("<a.+href=.?>", "").replaceAll("</a>", "")
	47	+ }
	48	+
44	49	}
		\ No newline at end of file

		@@ -1,5 +1,9 @@
1	1	package com.parrotstudio.recommend.ro.model
2	2
3		-case class Article(val body: String, val extend: String) {
	3	+case class Article(val title: String, val body: String, val extend: String) {
	4	+
	5	+ def this(title: String, body: String) = this(title, body, "")
	6	+ def this(title: String) = this(title, "")
	7	+ def this() = this("")
4	8
5	9	}