Check and filter out NaN in dcids for input arguments (#122)

* Remove NaN values from API input dcids

* Add unittest

Author: shifucun

datacommons/core.py

@@ -96,6 +96,7 @@ def get_property_labels(dcids, out=True):
     }
   """
   # Generate the GetProperty query and send the request
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   url = utils._API_ROOT + utils._API_ENDPOINTS['get_property_labels']
   payload = utils._send_request(url, req_json={'dcids': dcids})
@@ -152,6 +153,7 @@ def get_property_values(dcids,
     }
   """
   # Convert the dcids field and format the request to GetPropertyValue
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   if out:
     direction = 'out'
@@ -232,6 +234,7 @@ def get_triples(dcids, limit=utils._MAX_LIMIT):
     }
   """
   # Generate the GetTriple query and send the request.
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   url = utils._API_ROOT + utils._API_ENDPOINTS['get_triples']
   payload = utils._send_request(url, req_json={'dcids': dcids, 'limit': limit})

datacommons/places.py

@@ -60,6 +60,7 @@ def get_places_in(dcids, place_type):
     ]
     }
   """
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   url = utils._API_ROOT + utils._API_ENDPOINTS['get_places_in']
   payload = utils._send_request(url, req_json={
@@ -119,6 +120,7 @@ def get_related_places(dcids, population_type, measured_property,
     ]
     }
   """
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   url = utils._API_ROOT + utils._API_ENDPOINTS['get_related_places']
   pvs = []

datacommons/populations.py

@@ -97,6 +97,7 @@ def get_populations(dcids, population_type, constraining_properties={}):
     }
   """
   # Convert the dcids field and format the request to GetPopulations
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   pv = [{'property': k, 'value': v} for k, v in constraining_properties.items()]
   url = utils._API_ROOT + utils._API_ENDPOINTS['get_populations']
@@ -176,6 +177,7 @@ def get_observations(dcids,
       "dc/p/lr52m1yr46r44": 3075662.0
     }
   """
+  dcids = filter(lambda v: v==v, dcids)  # Filter out NaN values
   dcids = list(dcids)
   req_json = {
     'dcids': dcids,

datacommons/test/BUILD.bazel

@@ -7,6 +7,7 @@ py_test(
     deps = [
         "//datacommons:datacommons",
         requirement("mock"),
+        requirement("six"),
     ],
     python_version = "PY3"
 )
@@ -17,6 +18,7 @@ py_test(
     deps = [
         "//datacommons:datacommons",
         requirement("mock"),
+        requirement("six"),
     ],
     python_version = "PY3"
 )
@@ -27,6 +29,7 @@ py_test(
     deps = [
         "//datacommons:datacommons",
         requirement("mock"),
+        requirement("six"),
     ],
     python_version = "PY3"
 )
@@ -37,6 +40,7 @@ py_test(
     deps = [
         "//datacommons:datacommons",
         requirement("mock"),
+        requirement("six"),
     ],
     python_version = "PY3"
 )

datacommons/test/core_test.py

@@ -411,6 +411,15 @@ def test_multiple_dcids(self, urlopen_mock):
       'geoId/24031': ['geoId/2462850']
     })
 
+    dcids = ['geoId/06085', 'geoId/24031', float('nan')]
+    # Handle NaN values
+    towns = dc.get_property_values(
+      dcids, 'containedInPlace', out=False, value_type='Town')
+    self.assertDictEqual(towns, {
+      'geoId/06085': ['geoId/0643294', 'geoId/0644112'],
+      'geoId/24031': ['geoId/2462850']
+    })
+
     # Get the name of Santa Clara and Montgomery County.
     names = dc.get_property_values(dcids, 'name')
     self.assertDictEqual(names, {

requirements.bazel.txt

@@ -1,3 +1,2 @@
 six
-mock
-httplib2
+mock

setup.py

@@ -29,7 +29,6 @@
 
 REQUIRED = [
     'six',
-    'httplib2',
 ]
 
 PACKAGES = ['datacommons']