DynamoDB: Apply rough type evaluation and dispatching

CHANGES.md

@@ -1,6 +1,8 @@
 # Changelog
 
 ## Unreleased
+- DynamoDB: Apply rough type evaluation and dispatching when computing
+  values for `UPDATE` statements
 
 ## 2024/08/17 v0.0.7
 - DynamoDB: Fixed a syntax issue with `text` data type in `UPDATE` statements

src/commons_codec/transform/dynamodb.py

@@ -9,6 +9,7 @@
 import simplejson as json
 import toolz
 
+from commons_codec.util.data import is_container, is_number
 from commons_codec.vendor.boto3.dynamodb.types import TypeDeserializer
 
 logger = logging.getLogger(__name__)
@@ -145,8 +146,9 @@ def values_to_update(self, keys: t.Dict[str, t.Dict[str, str]]) -> str:
 
         constraints: t.List[str] = []
         for key_name, key_value in values_clause.items():
-            key_value = str(key_value).replace("'", "''")
-            constraint = f"{self.DATA_COLUMN}['{key_name}'] = '{key_value}'"
+            if not is_container(key_value) and not is_number(key_value):
+                key_value = "'" + str(key_value).replace("'", "''") + "'"
+            constraint = f"{self.DATA_COLUMN}['{key_name}'] = {key_value}"
             constraints.append(constraint)
 
         return ", ".join(constraints)

src/commons_codec/util/data.py

@@ -30,3 +30,7 @@ def is_number(s):
         pass
 
     return False
+
+
+def is_container(value):
+    return isinstance(value, (dict, list, set))

tests/transform/test_dynamodb.py

@@ -55,7 +55,7 @@
     },
     "eventSource": "aws:dynamodb",
 }
-MSG_MODIFY = {
+MSG_MODIFY_BASIC = {
     "awsRegion": "us-east-1",
     "eventID": "24757579-ebfd-480a-956d-a1287d2ef707",
     "eventName": "MODIFY",
@@ -84,6 +84,35 @@
     },
     "eventSource": "aws:dynamodb",
 }
+MSG_MODIFY_NESTED = {
+    "awsRegion": "us-east-1",
+    "eventID": "24757579-ebfd-480a-956d-a1287d2ef707",
+    "eventName": "MODIFY",
+    "userIdentity": None,
+    "recordFormat": "application/json",
+    "tableName": "foo",
+    "dynamodb": {
+        "ApproximateCreationDateTime": 1720742302233719,
+        "Keys": {"device": {"S": "foo"}, "timestamp": {"S": "2024-07-12T01:17:42"}},
+        "NewImage": {
+            "device": {"M": {"id": {"S": "bar"}, "serial": {"N": 12345}}},
+            "tags": {"L": [{"S": "foo"}, {"S": "bar"}]},
+            "empty_map": {"M": {}},
+            "empty_list": {"L": []},
+            "timestamp": {"S": "2024-07-12T01:17:42"},
+        },
+        "OldImage": {
+            "humidity": {"N": "84.84"},
+            "temperature": {"N": "42.42"},
+            "device": {"S": "foo"},
+            "location": {"S": "Sydney"},
+            "timestamp": {"S": "2024-07-12T01:17:42"},
+        },
+        "SizeBytes": 161,
+        "ApproximateCreationDateTimePrecision": "MICROSECOND",
+    },
+    "eventSource": "aws:dynamodb",
+}
 MSG_REMOVE = {
     "awsRegion": "us-east-1",
     "eventID": "ff4e68ab-0820-4a0c-80b2-38753e8e00e5",
@@ -145,10 +174,18 @@ def test_decode_cdc_insert_nested():
     )
 
 
-def test_decode_cdc_modify():
+def test_decode_cdc_modify_basic():
+    assert (
+        DynamoCDCTranslatorCrateDB(table_name="foo").to_sql(MSG_MODIFY_BASIC) == 'UPDATE "foo" '
+        "SET data['humidity'] = 84.84, data['temperature'] = 55.66, data['location'] = 'Sydney' "
+        "WHERE data['device'] = 'foo' AND data['timestamp'] = '2024-07-12T01:17:42';"
+    )
+
+
+def test_decode_cdc_modify_nested():
     assert (
-        DynamoCDCTranslatorCrateDB(table_name="foo").to_sql(MSG_MODIFY) == 'UPDATE "foo" '
-        "SET data['humidity'] = '84.84', data['temperature'] = '55.66', data['location'] = 'Sydney' "
+        DynamoCDCTranslatorCrateDB(table_name="foo").to_sql(MSG_MODIFY_NESTED) == 'UPDATE "foo" '
+        "SET data['tags'] = ['foo', 'bar'], data['empty_map'] = {}, data['empty_list'] = [] "
         "WHERE data['device'] = 'foo' AND data['timestamp'] = '2024-07-12T01:17:42';"
     )