|
@@ -11,6 +11,7 @@ class News():
|
|
|
r_noise=re.compile(r"posted|Product|Google",flags=re.IGNORECASE)
|
|
r_noise=re.compile(r"posted|Product|Google",flags=re.IGNORECASE)
|
|
|
|
|
|
|
|
URL=""
|
|
URL=""
|
|
|
|
|
+ LOWERURL=""
|
|
|
AUTHORS=[]
|
|
AUTHORS=[]
|
|
|
SCRAP_DATE=None
|
|
SCRAP_DATE=None
|
|
|
PUBLISH_DATE=None
|
|
PUBLISH_DATE=None
|
|
@@ -29,6 +30,7 @@ class News():
|
|
|
def __init__(self,url,sourceid, sourcename,sourcecategory,kw,lang="en",html=None):
|
|
def __init__(self,url,sourceid, sourcename,sourcecategory,kw,lang="en",html=None):
|
|
|
url=sanitizeUrl(url)
|
|
url=sanitizeUrl(url)
|
|
|
self.URL=url
|
|
self.URL=url
|
|
|
|
|
+ self.LOWERURL=url.lower()
|
|
|
self.SOURCENAME=sourcename
|
|
self.SOURCENAME=sourcename
|
|
|
self.SOURCE=sourceid
|
|
self.SOURCE=sourceid
|
|
|
self.CATEGORY=sourcecategory
|
|
self.CATEGORY=sourcecategory
|
|
@@ -83,6 +85,7 @@ class News():
|
|
|
def get(self):
|
|
def get(self):
|
|
|
return {
|
|
return {
|
|
|
"url": self.URL,
|
|
"url": self.URL,
|
|
|
|
|
+ "lowerurl": self.LOWERURL,
|
|
|
"source": self.SOURCE,
|
|
"source": self.SOURCE,
|
|
|
"source_name": self.SOURCENAME,
|
|
"source_name": self.SOURCENAME,
|
|
|
"authors": self.AUTHORS,
|
|
"authors": self.AUTHORS,
|