Fix handling of files with cached rulesets (#251)

* Fix handling of files with cached rulesets

* Fixed RulesPrivider

* Py2.6 test

Author: Yuxael

mwscan/ruleset.py

@@ -51,7 +51,7 @@ def get_rules(self):
                 return rawrules.encode('ascii', 'ignore')
         except NameError:
             pass # py3
-        
+
         return rawrules
 
 
@@ -85,7 +85,7 @@ def _get_cache_timestamp_content(self, cachefile):
             mtime = time.gmtime(mtime)
             mtime = time.strftime('%a, %d %b %Y %H:%M:%S GMT', mtime)
 
-            with open(cachefile) as fh:
+            with open(cachefile, 'rb') as fh:
                 cachedcontent = fh.read()
 
         return mtime, cachedcontent
@@ -114,21 +114,23 @@ def _httpget(self, url):
 
         if resp.status_code == 200:
             with open(cachefile, 'wb') as fh:
-                fh.write(resp.content)
+                # py3 vs py2
+                if type(resp.content) is bytes:
+                    ruleset = resp.content.decode('utf-8', 'ignore')
+                else:
+                    ruleset = resp.content
 
-            # py3 vs py2
-            if type(resp.content) is bytes:
-                return resp.content.decode('utf-8', errors='ignore')
-            else:
-                return resp.content
+                fh.write(ruleset.encode('utf-8', 'ignore'))
+
+            return ruleset
 
         if resp.status_code == 304:
             logging.debug(
                 'Upstream {0} is the same as our cache (HTTP 304)'.format(url))
 
         # Upstream hasn't changed (304) or has err'd
         if cachedcontent is not None:
-            return cachedcontent
+            return cachedcontent.decode('utf-8', 'ignore')
 
         raise RuntimeError("No cache @ {0} and invalid response for {1}: {2}".format(
             cachefile, url, resp.status_code))
@@ -170,8 +172,8 @@ class Files(RulesProvider):
     def get_rules(self):
         path = self._args.rules
         logging.info("Loading {0}".format(self._args.rules))
-        with open(path) as fh:
-            return fh.read()
+        with open(path, 'rb') as fh:
+            return fh.read().decode('utf-8', 'ignore')
 
 
 class NBS(RulesProvider):