Merge pull request #20 from alex9smith/handle-api-error

Catch API errors when a query string is invalid

CHANGELOG.md

@@ -1,7 +1,8 @@
 # Changelog
 
-## Unreleased
+## 1.4.0
 Validate `timespan` filter parameter to make sure it's an allowed value
+Catch API errors when a query string is invalid and return them to the user
 
 ## 1.3.3
 Fix a bug in `multi_repeat` which meant any filter using `OR` would fail

gdeltdoc/__init__.py

@@ -1,4 +1,4 @@
 from gdeltdoc.api_client import GdeltDoc
 from gdeltdoc.filters import Filters, near, repeat, multi_repeat, VALID_TIMESPAN_UNITS
 
-__version__ = "1.3.3"
+__version__ = "1.4.0"

gdeltdoc/api_client.py

@@ -154,4 +154,9 @@ def _query(self, mode: str, query_string: str) -> Dict:
         if response.status_code not in [200, 202]:
             raise ValueError("The gdelt api returned a non-successful statuscode. This is the response message: {}".
                              format(response.text))
+
+        # Response is text/html if it's an error and application/json if it's ok
+        if "text/html" in response.headers["content-type"]:
+            raise ValueError(f"The query was not valid. The API error message was: {response.text.strip()}")
+
         return load_json(response.content, self.max_depth_json_parsing)

tests/test_client.py

@@ -100,3 +100,9 @@ def test_vol_has_two_columns(self):
 
     def test_vol_raw_has_three_columns(self):
         self.assertEqual(self.all_results[1].shape[1], 3)
+
+class QueryTestCase(unittest.TestCase):
+
+    def test_handles_invalid_query_string(self):
+        with self.assertRaisesRegex(ValueError, "The query was not valid. The API error message was"):
+            GdeltDoc()._query("artlist", "environment&timespan=mins15")