add "model" and "status" fields

Author: vvkh

data_diff/dbt.py

@@ -52,6 +52,7 @@ class TDiffVars(pydantic.BaseModel):
     where_filter: Optional[str] = None
     include_columns: List[str]
     exclude_columns: List[str]
+    dbt_model: Optional[str] = None
 
 
 def dbt_diff(
@@ -166,6 +167,7 @@ def _get_diff_vars(
     datadiff_model_config = dbt_parser.get_datadiff_model_config(model.meta)
 
     return TDiffVars(
+        dbt_model=model.unique_id,
         dev_path=dev_qualified_list,
         prod_path=prod_qualified_list,
         primary_keys=primary_keys,
@@ -280,10 +282,14 @@ def _local_diff(diff_vars: TDiffVars, json_output: bool = False) -> None:
         # drain the iterator to get accumulated stats in diff.info_tree
         list(diff)
 
-        print(json.dumps(jsonify(diff, with_summary=True, with_columns={
-            "added": columns_added,
-            "removed": columns_removed,
-            "changed": columns_type_changed,
+        print(json.dumps(
+            jsonify(
+                diff, 
+                dbt_model=diff_vars.dbt_model,
+                with_summary=True, with_columns={
+                    "added": columns_added,
+                    "removed": columns_removed,
+                    "changed": columns_type_changed,
         })))
         return
 

data_diff/format.py

@@ -2,11 +2,11 @@
 from typing import Any, Optional, List, Dict, Tuple
 
 from runtype import dataclass
-
 from data_diff.diff_tables import DiffResultWrapper
 
 
 def jsonify(diff: DiffResultWrapper,
+            dbt_model: str,
             with_summary: bool = False,
             with_columns: Optional[Dict[str, List[str]]] = None) -> 'JsonDiff':
     """
@@ -57,7 +57,8 @@ def jsonify(diff: DiffResultWrapper,
         )
     )
     return JsonDiff(
-        isDifferent=is_different,
+        status="different" if is_different else "identical",
+        model=dbt_model,
         table1=list(table1.table_path),
         table2=list(table2.table_path),
         rows=RowsDiff(
@@ -150,7 +151,8 @@ class RowsDiff:
 
 @dataclass
 class JsonDiff:
-    isDifferent: bool
+    status: str # Literal ["identical", "different"]
+    model: str
     table1: List[str]
     table2: List[str]
     rows: RowsDiff
@@ -231,23 +233,23 @@ def _jsonify_exclusive(row: Dict[str, Any], key_columns: List[str]) -> Dict[str,
 
 
 def _jsonify_diff_summary(stats_dict: dict) -> JsonDiffSummary:
-    return {
-        'rows': {
-            'total': {
-               'table1': stats_dict["rows_A"],
-               'table2': stats_dict["rows_B"]
-            },
-            'exclusive': {
-                'table1': stats_dict["exclusive_A"],
-                'table2': stats_dict["exclusive_B"],
-            },
-            'updated': stats_dict["updated"],
-            'unchanged': stats_dict["unchanged"]
-        },
-        'stats': {
-            'diffCounts': stats_dict["stats"]['diff_counts']
-        }
-    }
+    return JsonDiffSummary(
+        rows=Rows(
+            total=Total(
+                table1=stats_dict["rows_A"],
+                table2=stats_dict["rows_B"]
+            ),
+            exclusive=ExclusiveRows(
+                table1=stats_dict["exclusive_A"],
+                table2=stats_dict["exclusive_B"],
+            ),
+            updated=stats_dict["updated"],
+            unchanged=stats_dict["unchanged"]
+        ),
+        stats=Stats(
+            diffCounts=stats_dict["stats"]['diff_counts']
+        )
+    )
 
 
 def _jsonify_columns_diff(columns_diff: Dict[str, List[str]]) -> JsonColumnsSummary:

tests/test_dbt.py

@@ -1151,6 +1151,7 @@ def test_get_diff_vars_meta_where(self, mock_prod_path_from_manifest, mock_prod_
         mock_model.config.schema_ = None
         mock_model.config.database = None
         mock_model.alias = "a_model_name"
+        mock_model.unique_id = "unique_id"
         mock_tdatadiffmodelconfig = Mock()
         mock_tdatadiffmodelconfig.where_filter = "where"
         mock_tdatadiffmodelconfig.include_columns = ["include"]
@@ -1187,6 +1188,7 @@ def test_get_diff_vars_meta_unrelated(self, mock_prod_path_from_manifest, mock_p
         mock_model.config.schema_ = None
         mock_model.config.database = None
         mock_model.alias = "a_model_name"
+        mock_model.unique_id = "unique_id"
         mock_tdatadiffmodelconfig = Mock()
         mock_tdatadiffmodelconfig.where_filter = "where"
         mock_tdatadiffmodelconfig.include_columns = ["include"]
@@ -1234,6 +1236,7 @@ def test_get_diff_vars_meta_none(self, mock_prod_path_from_manifest, mock_prod_p
         mock_dbt_parser.get_pk_from_model.return_value = primary_keys
         mock_dbt_parser.requires_upper = False
         mock_model.meta = None
+        mock_model.unique_id = "unique_id"
         mock_dbt_parser.prod_manifest_obj = None
         mock_prod_path_from_config.return_value = ("prod_db", "prod_schema")
 
@@ -1271,6 +1274,7 @@ def test_get_diff_vars_custom_db(self, mock_prod_path_from_manifest, mock_prod_p
         mock_dbt_parser.get_pk_from_model.return_value = primary_keys
         mock_dbt_parser.requires_upper = False
         mock_model.meta = None
+        mock_model.unique_id = "unique_id"
         mock_dbt_parser.prod_manifest_obj = None
         mock_prod_path_from_config.return_value = ("prod_db", "prod_schema")
 
@@ -1309,6 +1313,7 @@ def test_get_diff_vars_upper(self, mock_prod_path_from_manifest, mock_prod_path_
         mock_dbt_parser.get_pk_from_model.return_value = primary_keys
         mock_dbt_parser.requires_upper = True
         mock_model.meta = None
+        mock_model.unique_id = "unique_id"
         mock_dbt_parser.prod_manifest_obj = None
         mock_prod_path_from_config.return_value = ("prod_db", "prod_schema")
 
@@ -1333,6 +1338,7 @@ def test_get_diff_vars_call_get_prod_path_from_manifest(
         mock_model = Mock()
         primary_keys = ["a_primary_key"]
         mock_model.database = "a_dev_db"
+        mock_model.unique_id = "unique_id"
         mock_model.schema_ = "a_schema"
         mock_model.config.schema_ = None
         mock_model.config.database = "custom_database"

tests/test_format.py

@@ -28,10 +28,11 @@ def test_jsonify_diff(self):
             ),
             diff=[], stats={}
         )
-        json_diff = jsonify(diff)
+        json_diff = jsonify(diff, dbt_model='my_model')
         self.assertEqual(json_diff, {
             'version': '1.0.0',
-            'isDifferent': True,
+            'status': 'different',
+            'model': 'my_model',
             'table1': ['db', 'schema', 'table1'],
             'table2': ['db', 'schema', 'table2'],
             'rows': {
@@ -78,10 +79,11 @@ def test_jsonify_diff_no_difeference(self):
             ),
             diff=[], stats={}
         )
-        json_diff = jsonify(diff)
+        json_diff = jsonify(diff, dbt_model='model')
         self.assertEqual(json_diff, {
             'version': '1.0.0',
-            'isDifferent': False,
+            'status': 'identical',
+            'model': 'model',
             'table1': ['db', 'schema', 'table1'],
             'table2': ['db', 'schema', 'table2'],
             'rows': {