|
|
@@ -52,10 +52,12 @@ class News():
|
|
|
TEXT=""
|
|
|
HTML=""
|
|
|
ERROR=False
|
|
|
+ SOURCE=""
|
|
|
|
|
|
- def __init__(self,url,lang="en"):
|
|
|
+ def __init__(self,url,source,lang="en"):
|
|
|
url=sanitizeUrl(url)
|
|
|
self.URL=url
|
|
|
+ self.SOURCE=source
|
|
|
|
|
|
a = Article(url, language=lang,keep_article_html=True)
|
|
|
a.download()
|
|
|
@@ -73,6 +75,8 @@ class News():
|
|
|
|
|
|
def get(self):
|
|
|
return {
|
|
|
+ "url": self.URL,
|
|
|
+ "source": self.SOURCE,
|
|
|
"authors": self.AUTHORS,
|
|
|
"scrap_date": self.SCRAP_DATE,
|
|
|
"publish_date": self.PUBLISH_DATE,
|
|
|
@@ -125,6 +129,6 @@ for s in d.sources():
|
|
|
for a in l.links:
|
|
|
print(a)
|
|
|
print(len(l.links))
|
|
|
- n=News(l.links[0])
|
|
|
+ n=News(l.links[0],s["_id"])
|
|
|
print(n.get())
|
|
|
break
|