fix: JsonItemsParser should yield floats, not Decimals

ijson.items() parses non-integer JSON numbers as decimal.Decimal by default, which the
CDK cannot serialize downstream (orjson raises 'Decimal is not JSON serializable'). This
broke any JsonItemsDecoder stream with decimal fields (e.g. Amazon Brand Analytics
clickShare/conversionShare, Sales & Traffic, Vendor reports).

Pass use_float=True so non-integer numbers are parsed as float, matching the
json.loads/orjson behavior of the other JSON parsers. Adds a regression test.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: tolik0

airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py

@@ -149,7 +149,9 @@ def parse(self, data: BufferedIOBase) -> PARSER_OUTPUT_TYPE:
         # ijson auto-selects the best available backend (yajl2_c when present)
         # and reads from `data` lazily — it does not call `.read()` on the
         # whole stream up front.
-        yield from ijson.items(data, f"{self.items_path}.item")
+        # use_float=True yields floats for non-integer numbers instead of Decimal, matching
+        # json.loads/orjson behavior so downstream JSON serialization doesn't choke on Decimal.
+        yield from ijson.items(data, f"{self.items_path}.item", use_float=True)
 
 
 @dataclass

unit_tests/sources/declarative/decoders/test_composite_decoder.py

@@ -451,6 +451,22 @@ def test_json_items_parser_composes_with_gzip(requests_mock) -> None:
     assert list(decoder.decode(response)) == payload["dataByAsin"]
 
 
+def test_json_items_parser_yields_floats_not_decimals(requests_mock) -> None:
+    """Non-integer numbers must be parsed as float (not Decimal) so downstream JSON
+    serialization (orjson) does not fail on Decimal values."""
+    import orjson
+
+    payload = {"data": [{"ratio": 0.5, "rank": 3, "amount": 0.0000}]}
+    requests_mock.register_uri("GET", "https://airbyte.io/", content=json.dumps(payload).encode("utf-8"))
+    response = requests.get("https://airbyte.io/", stream=True)
+
+    records = list(CompositeRawDecoder(parser=JsonItemsParser(items_path="data")).decode(response))
+    assert isinstance(records[0]["ratio"], float)
+    assert isinstance(records[0]["rank"], int)
+    # Must be serializable by orjson (which rejects Decimal).
+    orjson.dumps(records[0])
+
+
 def test_json_items_parser_requires_items_path() -> None:
     parser = JsonItemsParser()
     with pytest.raises(ValueError, match="items_path"):