Merge pull request #217 from JVickery-TBS/fix/type-guessing-numeric

Add Mixed Integers Type Guessing to NUMERIC Tests

ckanext/xloader/tests/samples/mixed_numeric_string_sample.csv

@@ -1,3 +1,7 @@
-Funding agency,Program title,Maximum (indicative) grant amount
-DTIS,Accessible Tourism Infrastructure Grants,Five hundred thousand dollars
-DTIS,Boosting Accessible Tourism Experiences Grants,5000
+Funding agency,Program title,Maximum (indicative) grant amount,Some mixed integers
+DTIS,Accessible Tourism Infrastructure Grants,Five hundred thousand dollars,1
+DTIS,Boosting Accessible Tourism Experiences Grants,5000,-1
+DTIS,Some Other Grants,5 hundred thousand,0
+DTIS,Some Other Grants,$5000,6
+DTIS,Some Other Grants,$5 hundred thousand,
+DTIS,Some Other Grants,"$5,000.00","5"

ckanext/xloader/tests/test_loader.py

@@ -678,7 +678,7 @@ def test_with_mixed_types(self, Session):
             mimetype="text/csv",
             logger=logger,
         )
-        assert len(self._get_records(Session, resource_id)) == 2
+        assert len(self._get_records(Session, resource_id)) == 6
 
     def test_reload(self, Session):
         csv_filepath = get_sample_filepath("simple.csv")
@@ -979,6 +979,27 @@ def test_simple_large_file(self, Session):
             u"text",
         ]
 
+    def test_with_mixed_types(self, Session):
+        csv_filepath = get_sample_filepath("mixed_numeric_string_sample.csv")
+        resource = factories.Resource()
+        resource_id = resource['id']
+        loader.load_table(
+            csv_filepath,
+            resource_id=resource_id,
+            mimetype="text/csv",
+            logger=logger,
+        )
+        assert len(self._get_records(Session, resource_id)) == 6
+
+        assert self._get_column_types(Session, resource_id) == [
+            u'int4',
+            u'tsvector',
+            u'text',
+            u'text',
+            u'text',
+            u'numeric'
+        ]
+
     # test disabled by default to avoid adding large file to repo and slow test
     @pytest.mark.skip
     def test_boston_311_complete(self):