Merge pull request #17 from mborho/master

Charset correction by requests-lib only after regexping corrupt chars.

metadoc/__init__.py

@@ -12,6 +12,7 @@
 import requests
 import urllib.parse
 import os
+import re
 import sys
 import logging
 
@@ -186,10 +187,15 @@ def _request_url(self):
         if req.status_code != 200:
           raise Exception('Requesting article body failed with {} status code.'.format(req.status_code))
 
-        # check for encoding conflicts (e.g. t3n.de)
-        enc_apparent = req.apparent_encoding.lower()
-        if req.encoding.lower() != enc_apparent and \
-           enc_apparent != "windows-1254":
-            logger.info("Switching html encoding: {} -> {}".format(req.encoding, enc_apparent))
-            req.encoding = enc_apparent
+        if self._check_invalid_encoding(req.text):
+            # check for encoding conflicts (e.g. t3n.de)
+            enc_apparent = req.apparent_encoding.lower()
+            if req.encoding.lower() != enc_apparent and \
+               enc_apparent != "windows-1254":
+                logger.info("Switching html encoding: {} -> {}".format(req.encoding, enc_apparent))
+                req.encoding = enc_apparent
         return req.text
+
+    def _check_invalid_encoding(self, html):
+        r=r'(ü|ä|ö|ü)'
+        return True if re.search(r, html, re.I|re.M) else False

tests/test_module.py

@@ -61,3 +61,23 @@ def test_no_html(self):
   @asynctest.ignore_loop
   def test_check_result(self):
       self.metadoc._check_result({})
+
+  @asynctest.ignore_loop
+  def test_invalid_charset_check(self):
+      s = "Von da an beginnt fär die meisten jedoch der hektische Teil."
+      assert self.metadoc._check_invalid_encoding(s) == True
+      s = "Von da an beginnt für die meisten jedoch der hektische Teil."
+      assert self.metadoc._check_invalid_encoding(s) == True
+      s = "Von da an beginnt för die meisten jedoch der hektische Teil."
+      assert self.metadoc._check_invalid_encoding(s) == True
+      s = "Von da an beginnt für die meisten jedoch der hektische Teil."
+      assert self.metadoc._check_invalid_encoding(s) == True
+
+      s = "DE PÊRA"
+      assert self.metadoc._check_invalid_encoding(s) == False
+
+  @asynctest.ignore_loop
+  def test_invalid_t3n(self):
+      metadoc = Metadoc(url="https://t3n.de/news/remote-work-home-office-heimarbeit-erfahrungsbericht-1018248/", html=None)
+      result = metadoc.query()
+      assert result["title"] ==  "Remote Workers Life: „Das Home-Office löst viele Probleme, schafft aber auch neue“"