Merge pull request #35 from Lab-317/feature/013-stormmedia

add Stormmedia Parser
Lab-317 · Nov 10, 2014 · 4e7df1b · 4e7df1b
2 parents aa47406 + 31f5bfa
commit 4e7df1b
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 5 deletions.
diff --git a/newsParser/NewsParser.py b/newsParser/NewsParser.py
@@ -3,7 +3,7 @@
 # @Author: balicanta
 # @Date:   2014-10-25 00:09:39
 # @Last Modified by:   DinoLai
-# @Last Modified time: 2014-11-09 03:49:57
+# @Last Modified time: 2014-11-09 04:07:50
 
 import sys
 

diff --git a/newsParser/strategies/StormMediaGroupParseStrategy.py b/newsParser/strategies/StormMediaGroupParseStrategy.py
@@ -0,0 +1,27 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Author: DinoLai
+# @Date:   2014-11-08 23:06:56
+# @Last Modified by:   DinoLai
+# @Last Modified time: 2014-11-09 01:54:31
+
+from AbstractNewsParseStrategy import AbstractNewsParseStrategy
+
+
+class StormMediaGroupParseStrategy(AbstractNewsParseStrategy):
+
+    def isURLMatch(self, url):
+        return ".stormmediagroup.com" in url
+
+    def getTitle(self, beautiful_soup_object):
+        return beautiful_soup_object.select('.innerBigNewsTitle')[0].text
+
+    def getAuthor(self, beautiful_soup_object):
+        return beautiful_soup_object.select('.innerNewsInfo > a')[0].text
+
+    def getContent(self, beautiful_soup_object):
+        #beautiful_soup_object.select('.newsDescBlk')[0].style.decompose()
+        return beautiful_soup_object.select('.newsDescBlk')[0].text
+
+    def getPublishDate(self, beautiful_soup_object):
+        return beautiful_soup_object.select('.innerNewsInfo')[0].text
diff --git a/newsParser/strategies/__init__.py b/newsParser/strategies/__init__.py
@@ -3,13 +3,14 @@
 # @Author: balicanta
 # @Date:   2014-11-01 21:01:48
 # @Last Modified by:   DinoLai
-# @Last Modified time: 2014-11-09 03:59:22
+# @Last Modified time: 2014-11-09 04:07:23
 
 __all__ = [
     "AnntwNewsParseStrategy",
     "CoolLoudParseStrategy",
     "LtnNewsParseStrategy",
     "PeopleNewsParseStrategy",
     "UdnNewsParseStrategy",
-    "WorldYamParseStrategy"
+    "WorldYamParseStrategy",
+    "StormMediaGroupParseStrategy"
     ]
diff --git a/newsParser/tests/NewsParser_Test.py b/newsParser/tests/NewsParser_Test.py
@@ -3,7 +3,7 @@
 # @Author: balicanta
 # @Date:   2014-10-25 09:57:26
 # @Last Modified by:   DinoLai
-# @Last Modified time: 2014-11-09 03:58:04
+# @Last Modified time: 2014-11-09 04:08:11
 
 
 from ..NewsParser import NewsParser
@@ -21,7 +21,9 @@
     {"url": "http://www.coolloud.org.tw/node/80590",
         "title": "澳洲打工遭台商剝削", "author":"王顥中", "content": "青年勞動九五聯盟29日上午召開記者會"},
     {"url": "http://www.peoplenews.tw/news/80a3de53-1c06-4d30-a330-b76e335132f7",
-        "title": "協助弱勢募款", "author":"朱蒲青", "content": "沒有什麼假想敵啦"}
+        "title": "協助弱勢募款", "author":"朱蒲青", "content": "沒有什麼假想敵啦"},
+    {"url": "http://www.stormmediagroup.com/opencms/investigate/detail/1e517551-6735-11e4-a007-ef2804cba5a1/?uuid=1e517551-6735-11e4-a007-ef2804cba5a1",
+        "title": "藍營高層看連勝文", "author":"王彥喬", "content": "當年郝龍斌選前因為貓纜"}
 ]