refactor: rename app.py to model_runner.py

bin/oversightml-mr-entry-point.py

@@ -11,7 +11,7 @@
 from codeguru_profiler_agent import Profiler
 from pythonjsonlogger import jsonlogger
 
-from aws.osml.model_runner.app import ModelRunner
+from aws.osml.model_runner import ModelRunner
 from aws.osml.model_runner.common import ThreadingLocalContextFilter
 
 

src/aws/osml/model_runner/__init__.py

@@ -3,3 +3,5 @@
 # Telling flake8 to not flag errors in this file. It is normal that these classes are imported but not used in an
 # __init__.py file.
 # flake8: noqa
+
+from .model_runner import ModelRunner

src/aws/osml/model_runner/api/__init__.py

@@ -8,5 +8,5 @@
 from .image_request import ImageRequest
 from .inference import VALID_MODEL_HOSTING_OPTIONS, ModelInvokeMode
 from .region_request import RegionRequest
-from .request_utils import shared_properties_are_valid
+from .request_utils import get_image_path, shared_properties_are_valid
 from .sink import SinkMode, SinkType

src/aws/osml/model_runner/api/image_request.py

@@ -4,12 +4,10 @@
 from json import dumps, loads
 from typing import Any, Dict, List, Optional
 
-import boto3
 import shapely.geometry
 import shapely.wkt
 from shapely.geometry.base import BaseGeometry
 
-from aws.osml.model_runner.app_config import BotoConfig
 from aws.osml.model_runner.common import (
     FeatureDistillationAlgorithm,
     FeatureDistillationNMS,
@@ -19,10 +17,8 @@
     MRPostProcessing,
     MRPostprocessingStep,
     deserialize_post_processing_list,
-    get_credentials_for_assumed_role,
 )
 
-from .exceptions import InvalidS3ObjectException
 from .inference import ModelInvokeMode
 from .request_utils import shared_properties_are_valid
 from .sink import SinkType
@@ -193,37 +189,3 @@ def get_feature_distillation_option(self) -> List[FeatureDistillationAlgorithm]:
             if op.step == MRPostprocessingStep.FEATURE_DISTILLATION
             and isinstance(op.algorithm, FeatureDistillationAlgorithm)
         ]
-
-    @staticmethod
-    def validate_image_path(image_url: str, assumed_role: str) -> bool:
-        """
-        Validate if an image exists in S3 bucket
-
-        :param image_url: str = formatted image path to S3 bucket
-        :param assumed_role: str = containing a formatted arn role
-
-        :return: bool
-        """
-        bucket, key = image_url.replace("s3://", "").split("/", 1)
-        if assumed_role:
-            assumed_credentials = get_credentials_for_assumed_role(assumed_role)
-            # Here we will be writing to S3 using an IAM role other than the one for this process.
-            s3_client = boto3.client(
-                "s3",
-                aws_access_key_id=assumed_credentials["AccessKeyId"],
-                aws_secret_access_key=assumed_credentials["SecretAccessKey"],
-                aws_session_token=assumed_credentials["SessionToken"],
-                config=BotoConfig.default,
-            )
-        else:
-            # If no invocation role is provided the assumption is that the default role for this
-            # container will be sufficient to read/write to the S3 bucket.
-            s3_client = boto3.client("s3", config=BotoConfig.default)
-
-        try:
-            # head_object is the fastest approach to determine if it exists in S3
-            # also its less expensive to do the head_object approach
-            s3_client.head_object(Bucket=bucket, Key=key)
-            return True
-        except Exception as err:
-            raise InvalidS3ObjectException("This image does not exist!") from err

src/aws/osml/model_runner/api/request_utils.py

@@ -1,7 +1,11 @@
 #  Copyright 2023-2024 Amazon.com, Inc. or its affiliates.
 
-from aws.osml.model_runner.common import VALID_IMAGE_COMPRESSION, VALID_IMAGE_FORMATS
+import boto3
 
+from aws.osml.model_runner.app_config import BotoConfig
+from aws.osml.model_runner.common import VALID_IMAGE_COMPRESSION, VALID_IMAGE_FORMATS, get_credentials_for_assumed_role
+
+from .exceptions import InvalidS3ObjectException
 from .inference import VALID_MODEL_HOSTING_OPTIONS
 
 
@@ -55,3 +59,56 @@ def shared_properties_are_valid(request) -> bool:
         return False
 
     return True
+
+
+def get_image_path(image_url: str, assumed_role: str) -> str:
+    """
+    Returns the formatted image path for GDAL to read the image, either from S3 or a local file.
+
+    If the image URL points to an S3 path, this method validates the image's existence in S3
+    and reformats the path to use GDAL's /vsis3/ driver. Otherwise, it returns the local or
+    network image path.
+
+    :param image_url: str = formatted image path to S3 bucket
+    :param assumed_role: str = containing a formatted arn role
+
+    :return: The formatted image path.
+    """
+    if "s3:/" in image_url:
+        validate_image_path(image_url, assumed_role)
+        return image_url.replace("s3:/", "/vsis3", 1)
+    return image_url
+
+
+def validate_image_path(image_url: str, assumed_role: str) -> bool:
+    """
+    Validate if an image exists in S3 bucket
+
+    :param image_url: str = formatted image path to S3 bucket
+    :param assumed_role: str = containing a formatted arn role
+
+    :return: bool
+    """
+    bucket, key = image_url.replace("s3://", "").split("/", 1)
+    if assumed_role:
+        assumed_credentials = get_credentials_for_assumed_role(assumed_role)
+        # Here we will be writing to S3 using an IAM role other than the one for this process.
+        s3_client = boto3.client(
+            "s3",
+            aws_access_key_id=assumed_credentials["AccessKeyId"],
+            aws_secret_access_key=assumed_credentials["SecretAccessKey"],
+            aws_session_token=assumed_credentials["SessionToken"],
+            config=BotoConfig.default,
+        )
+    else:
+        # If no invocation role is provided the assumption is that the default role for this
+        # container will be sufficient to read/write to the S3 bucket.
+        s3_client = boto3.client("s3", config=BotoConfig.default)
+
+    try:
+        # head_object is the fastest approach to determine if it exists in S3
+        # also its less expensive to do the head_object approach
+        s3_client.head_object(Bucket=bucket, Key=key)
+        return True
+    except Exception as err:
+        raise InvalidS3ObjectException("This image does not exist!") from err