purge linklist

parser.py

@@ -42,21 +42,26 @@ class LinkList():
             feed = feedparser.parse(url)
             #print(feed["bozo"]) FIXME: If bozo==1 => error
             self.links=[i["link"] for i in feed["items"]]
-            return
-        if selector is None:
-            #s=newspaper.build(url,language="en",memoize_articles=False) #memoize puede salvarme la vida
-            s = newspaper.Source(url,memoize_articles=False)
-            s.download()
-            s.parse()
-            s.set_categories()
-            s.download_categories()
-            s.parse_categories()
-            s.generate_articles()
-
-            self.links=[a.url for a in s.articles]
         else:
-            f=FrontPage(url,selector)
-            self.links=f.links
+            if selector is None:
+                #s=newspaper.build(url,language="en",memoize_articles=False) #memoize puede salvarme la vida
+                s = newspaper.Source(url,memoize_articles=False)
+                s.download()
+                s.parse()
+                s.set_categories()
+                s.download_categories()
+                s.parse_categories()
+                s.generate_articles()
+    
+                self.links=[a.url for a in s.articles]
+            else:
+                f=FrontPage(url,selector)
+                self.links=f.links
+
+        self.links=self.purgeLinks(self.links)
+
+    def purgeLinks(self,l):
+        return [ link for link in l if not "presslist" in link.lower() and not "videolist" in link.lower() ]
 
 class News():
     r=re.compile(r"hours? ago|yesterday|today|last week|month",flags=re.IGNORECASE)
@@ -121,7 +126,13 @@ l=None
 #n=News("http://newsroom.fb.com/news/2016/04/news-feed-fyi-more-articles-you-want-to-spend-time-viewing/")
 
 #l=LinkList("https://www.blogger.com/feeds/10861780/posts/default",rss=True)
-n=News("http://googleblog.blogspot.com/2016/05/translate-where-you-need-it-in-any-app.html")
+#n=News("http://googleblog.blogspot.com/2016/05/translate-where-you-need-it-in-any-app.html") #FIXME
+
+#l=LinkList("https://apps.shareholder.com/rss/rss.aspx?channels=14&companyid=ONE",rss=True)
+#n=News("http://investor.shareholder.com/jpmorganchase/press/releasedetail.cfm?ReleaseID=970074")
+
+l=LinkList("http://www.samsungmobilepress.com/press/pressList.asp")
+
 if l is not None:
     for a in l.links:
         print(a)