Merge branch 'master' into local-artifacts-cleanup

Author: pintaoz-aws

CHANGELOG.md

@@ -1,5 +1,20 @@
 # Changelog
 
+## v2.237.3 (2025-01-09)
+
+### Bug Fixes and Other Changes
+
+ * pin metadata-version to 2.3
+ * model server might have already done a serialization. honor that by not decoding the request again if it is not already bytes or bytestream
+ * Disable jumpstart tests missing clean up logic
+ * Jumpstart ap southeast 5
+ * add autogluon 1.2
+ * updated inference script to cover context
+ * security update -> use sha256 instead of md5 for file hashing
+ * Fix Flake8 Violations
+ * Added parsing string support for situations where custom code might be used (ie. mlflow)
+ * Updating Inference Optimization Validations
+
 ## v2.237.2 (2024-12-17)
 
 ### Bug Fixes and Other Changes

VERSION

@@ -1 +1 @@
-2.237.3.dev0
+2.237.4.dev0

pyproject.toml

@@ -73,13 +73,15 @@ pattern = "(?P<version>.+)"
 [tool.hatch.metadata.hooks.custom]
 
 [tool.hatch.build.targets.wheel]
+core-metadata-version = "2.3"
 packages = ["src/sagemaker"]
 exclude = ["src/sagemaker/serve/model_server/triton/pack_conda_env.sh"]
 
 [tool.hatch.build.targets.wheel.shared-scripts]
 "src/sagemaker/serve/model_server/triton/pack_conda_env.sh" = "pack_conda_env.sh"
 
 [tool.hatch.build.targets.sdist]
+core-metadata-version = "2.3"
 only-include = [
   "/requirements/extras",
   "/src",

src/sagemaker/serve/model_server/multi_model_server/inference.py

@@ -46,18 +46,28 @@ def input_fn(input_data, content_type, context=None):
         if hasattr(schema_builder, "custom_input_translator"):
             deserialized_data = schema_builder.custom_input_translator.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type,
             )
         else:
             deserialized_data = schema_builder.input_deserializer.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type[0],
             )

src/sagemaker/serve/model_server/torchserve/inference.py

@@ -68,18 +68,28 @@ def input_fn(input_data, content_type):
         if hasattr(schema_builder, "custom_input_translator"):
             deserialized_data = schema_builder.custom_input_translator.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type,
             )
         else:
             deserialized_data = schema_builder.input_deserializer.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type[0],
             )

src/sagemaker/serve/model_server/torchserve/xgboost_inference.py

@@ -71,18 +71,28 @@ def input_fn(input_data, content_type):
         if hasattr(schema_builder, "custom_input_translator"):
             return schema_builder.custom_input_translator.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type,
             )
         else:
             return schema_builder.input_deserializer.deserialize(
                 (
-                    io.BytesIO(input_data)
-                    if type(input_data) == bytes
-                    else io.BytesIO(input_data.encode("utf-8"))
+                    io.BytesIO(input_data.encode("utf-8"))
+                    if not any(
+                        [
+                            isinstance(input_data, bytes),
+                            isinstance(input_data, bytearray),
+                        ]
+                    )
+                    else io.BytesIO(input_data)
                 ),
                 content_type[0],
             )