unit tests for inferring semantic domain from Domain object

great-expectations · Apr 29, 2022 · 59387ad · 59387ad
1 parent e68b640
commit 59387ad
Show file tree

Hide file tree

Showing 3 changed files with 123 additions and 12 deletions.
diff --git a/great_expectations/rule_based_profiler/helpers/util.py b/great_expectations/rule_based_profiler/helpers/util.py
@@ -433,9 +433,9 @@ def integer_semantic_domain_type(domain: Domain) -> bool:
     """
     This method examines "INFERRED_SEMANTIC_TYPE_KEY" attribute of "Domain" argument to check whether or not underlying
     "SemanticDomainTypes" enum value is an "integer".  Because explicitly designated "SemanticDomainTypes.INTEGER" type
-    is unavaiable, "SemanticDomainTypes.LOGIC",  "SemanticDomainTypes.BINARY",  and "SemanticDomainTypes.IDENTIFIER",
-    are intepreted as taking on "integer" values.  Note: In certain settings, this method should be used as pre-filter
-    to "NumericMetricRangeMultiBatchParameterBuilder._get_round_decimals_using_heuristics()".
+    is unavaiable, "SemanticDomainTypes.LOGIC" and "SemanticDomainTypes.IDENTIFIER" are intepreted as "integer" values.
+
+    This method can be used "NumericMetricRangeMultiBatchParameterBuilder._get_round_decimals_using_heuristics()".
 
     Note: Inability to assess underlying "SemanticDomainTypes" details of "Domain" object produces "False" return value.
 
@@ -457,7 +457,6 @@ def integer_semantic_domain_type(domain: Domain) -> bool:
             semantic_domain_type
             in [
                 SemanticDomainTypes.LOGIC,
-                SemanticDomainTypes.BINARY,
                 SemanticDomainTypes.IDENTIFIER,
             ]
             for semantic_domain_type in (inferred_semantic_domain_type.values())

diff --git a/...le_based_profiler/parameter_builder/numeric_metric_range_multi_batch_parameter_builder.py b/...le_based_profiler/parameter_builder/numeric_metric_range_multi_batch_parameter_builder.py
@@ -12,6 +12,7 @@
     compute_bootstrap_quantiles_point_estimate,
     compute_quantiles,
     get_parameter_value_and_validate_return_type,
+    integer_semantic_domain_type,
 )
 from great_expectations.rule_based_profiler.parameter_builder import (
     AttributedResolvedMetrics,
@@ -351,12 +352,17 @@ def _estimate_metric_value_range(
         lower_bound: Optional[float] = truncate_values.get("lower_bound")
         upper_bound: Optional[float] = truncate_values.get("upper_bound")
 
-        round_decimals: int = self._get_round_decimals_using_heuristics(
-            metric_values=metric_values,
-            domain=domain,
-            variables=variables,
-            parameters=parameters,
-        )
+        round_decimals: int
+
+        if integer_semantic_domain_type(domain=domain):
+            round_decimals = 0
+        else:
+            round_decimals = self._get_round_decimals_using_heuristics(
+                metric_values=metric_values,
+                domain=domain,
+                variables=variables,
+                parameters=parameters,
+            )
 
         min_value: Number
         max_value: Number

diff --git a/tests/rule_based_profiler/domain_builder/test_domain.py b/tests/rule_based_profiler/domain_builder/test_domain.py
@@ -212,7 +212,9 @@ def test_integer_semantic_domain_type():
     domain = Domain(
         rule_name="my_rule",
         domain_type="column",
-        domain_kwargs={"column": "passenger_count"},
+        domain_kwargs={
+            "column": "passenger_count",
+        },
         details={
             INFERRED_SEMANTIC_TYPE_KEY: {
                 "VendorID": SemanticDomainTypes.NUMERIC,
@@ -224,10 +226,114 @@ def test_integer_semantic_domain_type():
     domain = Domain(
         rule_name="my_rule",
         domain_type="column",
-        domain_kwargs={"column": "VendorID"},
+        domain_kwargs={
+            "column": "VendorID",
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "VendorID": SemanticDomainTypes.IDENTIFIER,
+            },
+        },
+    )
+    assert integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column": "is_night_time",
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "is_night_time": SemanticDomainTypes.LOGIC,
+            },
+        },
+    )
+    assert integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column_A": "passenger_count",
+            "column_B": "fare_amount",
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "passenger_count": SemanticDomainTypes.NUMERIC,
+                "fare_amount": SemanticDomainTypes.NUMERIC,
+            },
+        },
+    )
+    assert not integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column_A": "passenger_count",
+            "column_B": "VendorID",
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "passenger_count": SemanticDomainTypes.NUMERIC,
+                "VendorID": SemanticDomainTypes.IDENTIFIER,
+            },
+        },
+    )
+    assert not integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column_A": "is_night_time",
+            "column_B": "VendorID",
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "is_night_time": SemanticDomainTypes.LOGIC,
+                "VendorID": SemanticDomainTypes.IDENTIFIER,
+            },
+        },
+    )
+    assert integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column_list": [
+                "passenger_count",
+                "fare_amount",
+                "is_night_time",
+            ],
+        },
+        details={
+            INFERRED_SEMANTIC_TYPE_KEY: {
+                "passenger_count": SemanticDomainTypes.NUMERIC,
+                "fare_amount": SemanticDomainTypes.NUMERIC,
+                "is_night_time": SemanticDomainTypes.LOGIC,
+            },
+        },
+    )
+    assert not integer_semantic_domain_type(domain=domain)
+
+    domain = Domain(
+        rule_name="my_rule",
+        domain_type="column",
+        domain_kwargs={
+            "column_list": [
+                "VendorID",
+                "fare_amount",
+                "is_night_time",
+            ],
+        },
         details={
             INFERRED_SEMANTIC_TYPE_KEY: {
                 "VendorID": SemanticDomainTypes.IDENTIFIER,
+                "RatecodeID": SemanticDomainTypes.IDENTIFIER,
+                "is_night_time": SemanticDomainTypes.LOGIC,
             },
         },
     )