crawler: better use of chardet

Scan whole doc since beginning of html pages tends to be too regular. Ignore ASCII detection for the same reason.

morss/crawler.py

@@ -64,8 +64,8 @@ def detect_encoding(data, con=None):
     if match:
         return match.groups()[0].lower().decode()
 
-    enc = chardet.detect(data[:1000])['encoding']
-    if enc:
+    enc = chardet.detect(data)['encoding']
+    if enc and enc != 'ascii':
         return enc
 
     return 'utf-8'