David 10 лет назад
Родитель
Сommit
179af9fa5d
1 измененных файлов с 8 добавлено и 1 удалено
  1. 8 1
      news.py

+ 8 - 1
news.py

@@ -1,6 +1,7 @@
 import re
 import re
 import htmlmin
 import htmlmin
 import datetime
 import datetime
+import html
 
 
 from functions import sanitizeUrl
 from functions import sanitizeUrl
 from newspaper import Article
 from newspaper import Article
@@ -60,7 +61,7 @@ class News():
         self.PUBLISH_DATE=a.publish_date
         self.PUBLISH_DATE=a.publish_date
         self.SUMMARY=a.summary
         self.SUMMARY=a.summary
         self.TEXT=a.text
         self.TEXT=a.text
-        self.TITLE=a.title
+        self.TITLE=self.fix_title(a.title)
         self.HTML=htmlmin.minify(a.article_html,remove_empty_space=True)
         self.HTML=htmlmin.minify(a.article_html,remove_empty_space=True)
 #        if html is None:
 #        if html is None:
 #            self.HTML=htmlmin.minify(a.article_html,remove_empty_space=True)
 #            self.HTML=htmlmin.minify(a.article_html,remove_empty_space=True)
@@ -93,6 +94,12 @@ class News():
                 "title": self.TITLE
                 "title": self.TITLE
                 }
                 }
 
 
+    def fix_title(self,n):
+        while(html.unescape(n)!=n):
+            n=html.unescape(n)
+        n=n.replace("&Dquot",'"')
+        return n
+
     def dedup(self,val):
     def dedup(self,val):
         return list(set(val))
         return list(set(val))