Merge pull request #365 from robertcv/fixes/guardian_limit

Guardian request limit exceeded

orangecontrib/text/guardian.py

@@ -29,6 +29,10 @@
 ARTICLES_PER_PAGE = 10
 
 
+class APILimitError(Exception):
+    pass
+
+
 class TheGuardianCredentials:
     """ The Guardian API credentials. """
     def __init__(self, key):
@@ -93,6 +97,9 @@ def _search(self, query, from_date, to_date, page=1):
         data = self._build_query(query, from_date, to_date, page)
 
         response = requests.get(BASE_URL, data)
+        if response.status_code == 429:
+            raise APILimitError("API limit exceeded")
+
         parsed = json.loads(response.text)
 
         if page == 1:   # store number of pages

orangecontrib/text/tests/test_guardian.py

@@ -2,6 +2,7 @@
 import os
 
 from datetime import date, datetime
+from unittest import mock
 
 from orangecontrib.text import guardian
 
@@ -10,7 +11,9 @@
 
 
 class TestCredentials(unittest.TestCase):
-    def test_valid(self):
+    @mock.patch('requests.get')
+    def test_valid(self, mock_get):
+        mock_get().status_code = 200
         credentials = guardian.TheGuardianCredentials(API_KEY)
         self.assertTrue(credentials.valid)
 
@@ -19,15 +22,26 @@ def test_equal(self):
         self.assertEquals(credentials, credentials)
 
 
+def skip_limit_exceeded(fun):
+    def wrapper(*args, **kwargs):
+        try:
+            fun(*args, **kwargs)
+        except guardian.APILimitError as err:
+            raise unittest.SkipTest(str(err))
+    return wrapper
+
+
 class TestGuardian(unittest.TestCase):
     def setUp(self):
         c = guardian.TheGuardianCredentials(API_KEY)
         self.api = guardian.TheGuardianAPI(c)
 
+    @skip_limit_exceeded
     def test_search(self):
         corp = self.api.search('Slovenia', max_documents=20)
         self.assertEqual(len(corp), 20)
 
+    @skip_limit_exceeded
     def test_search_from_to_date(self):
         from_date = date(2016, 12, 1)
         to_date = date(2016, 12, 31)
@@ -41,13 +55,53 @@ def test_search_from_to_date(self):
             self.assertGreaterEqual(dt.date(), from_date)
             self.assertLessEqual(dt.date(), to_date)
 
+    @skip_limit_exceeded
     def test_breaking(self):
         self.api.should_break = lambda: True
         corp = self.api.search('Slovenia', max_documents=20)
         self.assertEqual(len(corp), 10)
         self.api.should_break = lambda: False
 
+    @skip_limit_exceeded
     def test_accumulate(self):
         self.api.search('Slovenia', max_documents=10, accumulate=True)
         corp = self.api.search('Ljubljana', max_documents=10, accumulate=True)
         self.assertEqual(len(corp), 20)
+
+    @mock.patch('requests.get')
+    def test_api_limit_error(self, mock_get):
+        mock_get().status_code = 429
+        self.assertRaises(guardian.APILimitError, self.api.search, 'Slovenia')
+
+    @mock.patch('requests.get')
+    def test_search_mock_data(self, mock_get):
+        mock_get().text = """
+        {
+          "response": {
+            "pages": 1,
+            "results": [
+              {
+                "type": "article",
+                "sectionName": "World news",
+                "webPublicationDate": "2018-07-05T23:27:25Z",
+                "webUrl": "https://www.theguardian.com/world/2018/jul/06",
+                "fields": {
+                  "headline": "Rohingya refugees reject UN-Myanmar repatriati",
+                  "trailText": "Leaders say agreement does not address concer",
+                  "body": "<p><strong><strong><strong></strong></strong></str",
+                  "wordcount": "512",
+                  "lang": "en",
+                  "bodyText": "Rohingya community leaders have rejected an."
+                },
+                "tags": [
+                  {
+                    "webTitle": "Myanmar"
+                  }
+                ]
+              }
+            ]
+          }
+        }
+        """
+        corp = self.api.search('Slovenia', max_documents=1)
+        self.assertEqual(len(corp), 1)

orangecontrib/text/widgets/owguardian.py

@@ -11,7 +11,7 @@
 from Orange.widgets.widget import Output
 
 from orangecontrib.text.corpus import Corpus
-from orangecontrib.text.guardian import TheGuardianCredentials, TheGuardianAPI
+from orangecontrib.text.guardian import TheGuardianCredentials, TheGuardianAPI, APILimitError
 from orangecontrib.text.widgets.utils import CheckListLayout, QueryBox, DatePickerInterval, gui_require, asynchronous
 
 
@@ -88,6 +88,7 @@ class Warning(OWWidget.Warning):
     class Error(OWWidget.Error):
         no_api = Msg('Please provide a valid API key.')
         no_query = Msg('Please provide a query.')
+        limit_exceeded = Msg('Requests limit reached.')
 
     def __init__(self):
         super().__init__()
@@ -149,7 +150,11 @@ def start_stop(self):
     @gui_require('api', 'no_api')
     @gui_require('recent_queries', 'no_query')
     def run_search(self):
-        self.search()
+        self.Error.limit_exceeded.clear()
+        try:
+            self.search()
+        except APILimitError:
+            self.Error.limit_exceeded()
 
     @asynchronous
     def search(self):