fix: Ensure variant JSONB output from tap-postgres is mapped to JSONB in this target (#395)

fixes: #394

---------

Co-authored-by: Edgar Ramírez-Mondragón <[email protected]>
Co-authored-by: Edgar Ramírez Mondragón <[email protected]>

Author: hudovisk

target_postgres/connector.py

@@ -291,14 +291,34 @@ def pick_individual_type(self, jsonschema_type: dict):
         if "object" in jsonschema_type["type"]:
             return JSONB()
         if "array" in jsonschema_type["type"]:
-            items_type = jsonschema_type.get("items")
-            if "string" == items_type:
-                return ARRAY(TEXT())
-            if "integer" == items_type:
-                return ARRAY(BIGINT())
-            else:
+            items = jsonschema_type.get("items")
+            # Case 1: items is a string
+            if isinstance(items, str):
+                return ARRAY(self.to_sql_type({"type": items}))
+
+            # Case 2: items are more complex
+            if isinstance(items, dict):
+                # Case 2.1: items are variants
+                if "type" not in items:
+                    return ARRAY(JSONB())
+
+                items_type = items["type"]
+
+                # Case 2.2: items are a single type
+                if isinstance(items_type, str):
+                    return ARRAY(self.to_sql_type({"type": items_type}))
+
+                # Case 2.3: items are a list of types
+                if isinstance(items_type, list):
+                    return ARRAY(self.to_sql_type({"type": items_type}))
+
+            # Case 3: tuples
+            if isinstance(items, list):
                 return ARRAY(JSONB())
 
+            # All other cases, return JSONB
+            return JSONB()
+
         # string formats
         if jsonschema_type.get("format") == "date-time":
             return TIMESTAMP()

target_postgres/tests/data_files/jsonb_data.singer

@@ -0,0 +1,3 @@
+{"type":"SCHEMA", "stream":"test_jsonb_data", "key_properties":["id"], "schema":{"required":["id"], "type":"object", "properties":{"id":{"type":["integer"]},"event_data":{"type":["string","number","integer","array","object","boolean","null"]}}}}
+{"type":"RECORD","stream":"test_jsonb_data","record":{"id":1,"event_data":null,"time_extracted":"2024-07-27T12:24:43.774995+00:00"}}
+{"type":"RECORD","stream":"test_jsonb_data","record":{"id":2,"event_data":{"test":{"test_name":"test_value"}},"time_extracted":"2024-07-27T12:24:43.774995+00:00"}}

target_postgres/tests/test_target_postgres.py

@@ -423,6 +423,16 @@ def test_array_data(postgres_target):
     verify_data(postgres_target, "test_carts", 4, "id", row)
 
 
+def test_jsonb_data(postgres_target):
+    file_name = "jsonb_data.singer"
+    singer_file_to_target(file_name, postgres_target)
+    row = [
+        {"id": 1, "event_data": None},
+        {"id": 2, "event_data": {"test": {"test_name": "test_value"}}},
+    ]
+    verify_data(postgres_target, "test_jsonb_data", 2, "id", row)
+
+
 def test_encoded_string_data(postgres_target):
     """
     We removed NUL characters from the original encoded_strings.singer as postgres doesn't allow them.