Start _post method; add constants for extractor methods

webarchiv.py

@@ -7,6 +7,11 @@ from requests import HTTPError
 _datetime_format_string = '%Y%m%d%H%M%S'
 
 
+EXTRACTOR_TEXT = 1
+EXTRACTOR_CSS = 2
+EXTRACTOR_BINARY = 3
+
+
 class SessionTimeoutError(Exception):
     pass
 
@@ -102,27 +107,39 @@ class WebarchivSession:
         print(self._error_template.format(status_code=e.response.status_code,
                                           response_text=e.response.text),
               file=sys.stderr)
-
+        
+    def _handle_response_errors(self, r):
+        if r.status_code == 403:
+            print('Forbidden. Invalid Token or ApiKey transmitted', file=sys.stderr)
+            return r
+        elif r.status_code == 400:
+            print('Bad request', file=sys.stderr)
+            return r
+        elif r.status_code == 410:
+            print('The requested API Version (via X-API-VERSION Header) is not available', file=sys.stderr)
+        return r
+    
     def _get(self, op, auto_connect=True, **kwargs, ):
         kwargs['params'] = self._add_api_key_and_token(kwargs.pop('params', {}))
-        # kwargs = self._add_api_key_and_token(kwargs)
         r = requests.get(self.base_url.format(op), **kwargs)
         if r.ok:
             return r
+        elif r.status_code == 403 and auto_connect:
+            self.connect()
+            return self._get(op=op, auto_connect=False, **kwargs)
         else:
-            if r.status_code == 403:
-                if auto_connect:
-                    self.connect()
-                    return self._get(op=op, auto_connect=False, **kwargs)
-                else:
-                    print('Forbidden. Invalid Token or ApiKey transmitted', file=sys.stderr)
-                    return r
-            elif r.status_code == 400:
-                print('Bad request', file=sys.stderr)
-                return r
-            elif r.status_code == 410:
-                print('The requested API Version (via X-API-VERSION Header) is not available', file=sys.stderr)
+            return self._handle_response_errors(r)
+    
+    def _post(self, op, auto_connect=True, **kwargs):
+        kwargs['json'] = self._add_api_key_and_token(kwargs.pop('json', {}))
+        r = requests.post(self.base_url.format(op), **kwargs)
+        if r.ok:
             return r
+        elif r.status_code == 403 and auto_connect:
+            self.connect()
+            return self._post(op=op, auto_connect=False, **kwargs)
+        else:
+            return self._handle_response_errors(r)
 
     def fulltext_search(self, query_string, from_=None, to_=None):
         """
@@ -227,9 +244,6 @@ class WebarchivSession:
     def waitForResponse(self, response):
         """
         Polls until the server responds with a result
-
-        :param response: String to search for
-        :return: response
         """
         if response.status_code == 400:
             return response
@@ -242,12 +256,11 @@ class WebarchivSession:
 
     def status_query(self, resp):
         """
-        this is the pollingrequest for the given typen of request
-
-        :param resp: String to search for
-        :return: response
+        this is the polling request for the given type of request
         """
-        requestid = resp.json()['requestid']
+        j = resp.json()
+        context = j['context']
+        requestid = j['requestid']
         type_ = resp.json()['type']
         if type_ == 1:
             r = self._get(op='/search/status/fulltext', params={'requestid': requestid})