Python: Bedrock connector (#9100)

### Motivation and Context

<!-- Thank you for your contribution to the semantic-kernel repo!
Please help reviewers and future users, providing the following
information:
  1. Why is this change required?
  2. What problem does it solve?
  3. What scenario does it contribute to?
  4. If it fixes an open issue, please link to the issue here.
-->
Add Amazon Bedrock AI to Semantic Kernel.

Amazon Bedrock is a platform on AWS that offers access to a wide range
to LLMs, including proprietary Amazon Titan models.

We are integrating it into Semantic Kernel as one of our AI connectors
to provide more options to developers.

### Description

<!-- Describe your changes, the overall approach, the underlying design.
These notes will help understanding how your code works. Thanks! -->
1. All interfaces implemented: chat completion (streaming +
non-streaming), text completion (streaming + non-streaming), and text
embedding.
2. Supports for function calling.
3. Supports for image content.
4. Supports for all models available on Bedrock.

### Contribution Checklist

<!-- Before submitting this PR, please make sure: -->

- [x] The code builds clean without any errors or warnings
- [x] The PR follows the [SK Contribution
Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
and the [pre-submission formatting
script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts)
raises no violations
- [x] All unit tests pass, and I have added new tests where possible
- [x] I didn't break anyone 😄

.github/workflows/python-integration-tests.yml

@@ -108,6 +108,12 @@ jobs:
           credentials_json: ${{ secrets.VERTEX_AI_SERVICE_ACCOUNT_KEY }}
       - name: Set up gcloud
         uses: google-github-actions/setup-gcloud@v2
+      - name: Configure AWS Credentials
+        uses: aws-actions/configure-aws-credentials@v4
+        with:
+          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
+          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+          aws-region: ${{ vars.AWS_REGION }}
       - name: Setup Redis Stack Server
         if: matrix.os == 'ubuntu-latest'
         run: docker run -d --name redis-stack-server -p 6379:6379 redis/redis-stack-server:latest
@@ -236,6 +242,12 @@ jobs:
           credentials_json: ${{ secrets.VERTEX_AI_SERVICE_ACCOUNT_KEY }}
       - name: Set up gcloud
         uses: google-github-actions/setup-gcloud@v2
+      - name: Configure AWS Credentials
+        uses: aws-actions/configure-aws-credentials@v4
+        with:
+          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
+          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+          aws-region: ${{ vars.AWS_REGION }}
       - name: Setup Redis Stack Server
         if: matrix.os == 'ubuntu-latest'
         run: docker run -d --name redis-stack-server -p 6379:6379 redis/redis-stack-server:latest

python/.cspell.json

@@ -27,6 +27,7 @@
         "aiplatform",
         "azuredocindex",
         "azuredocs",
+        "boto",
         "contentvector",
         "contoso",
         "datamodel",

python/pyproject.toml

@@ -118,6 +118,9 @@ weaviate = [
 pandas = [
     "pandas ~= 2.2"
 ]
+aws = [
+    "boto3>=1.28.57",
+]
 
 [tool.uv]
 prerelease = "if-necessary-or-explicit"

python/samples/concepts/chat_completion/azure_chat_gpt_api.py

@@ -76,22 +76,25 @@ async def chat() -> bool:
 
     stream = True
     if stream:
-        answer = kernel.invoke_stream(
+        chunks = kernel.invoke_stream(
             chat_function,
             user_input=user_input,
             chat_history=history,
         )
         print("Mosscap:> ", end="")
-        async for message in answer:
+        answer = ""
+        async for message in chunks:
             print(str(message[0]), end="")
+            answer += str(message[0])
         print("\n")
-        return True
-    answer = await kernel.invoke(
-        chat_function,
-        user_input=user_input,
-        chat_history=history,
-    )
-    print(f"Mosscap:> {answer}")
+    else:
+        answer = await kernel.invoke(
+            chat_function,
+            user_input=user_input,
+            chat_history=history,
+        )
+        print(f"Mosscap:> {answer}")
+
     history.add_user_message(user_input)
     history.add_assistant_message(str(answer))
     return True

python/samples/concepts/chat_completion/chat_anthropic_api.py

@@ -55,22 +55,25 @@ async def chat() -> bool:
 
     stream = True
     if stream:
-        answer = kernel.invoke_stream(
+        chunks = kernel.invoke_stream(
             chat_function,
             user_input=user_input,
             chat_history=chat_history,
         )
         print("Mosscap:> ", end="")
-        async for message in answer:
+        answer = ""
+        async for message in chunks:
             print(str(message[0]), end="", flush=True)
+            answer += str(message[0])
         print("\n")
-        return True
-    answer = await kernel.invoke(
-        chat_function,
-        user_input=user_input,
-        chat_history=chat_history,
-    )
-    print(f"Mosscap:> {answer}")
+    else:
+        answer = await kernel.invoke(
+            chat_function,
+            user_input=user_input,
+            chat_history=chat_history,
+        )
+        print(f"Mosscap:> {answer}")
+
     chat_history.add_user_message(user_input)
     chat_history.add_assistant_message(str(answer))
     return True

python/samples/concepts/chat_completion/chat_bedrock_api.py

@@ -0,0 +1,95 @@
+# Copyright (c) Microsoft. All rights reserved.
+
+import asyncio
+
+from semantic_kernel import Kernel
+from semantic_kernel.connectors.ai.bedrock.bedrock_prompt_execution_settings import BedrockChatPromptExecutionSettings
+from semantic_kernel.connectors.ai.bedrock.services.bedrock_chat_completion import BedrockChatCompletion
+from semantic_kernel.contents import ChatHistory
+
+system_message = """
+You are a chat bot. Your name is Mosscap and
+you have one goal: figure out what people need.
+Your full name, should you need to know it, is
+Splendid Speckled Mosscap. You communicate
+effectively, but you tend to answer with long
+flowery prose.
+"""
+
+kernel = Kernel()
+
+service_id = "bedrock-chat"
+kernel.add_service(BedrockChatCompletion(service_id=service_id, model_id="cohere.command-r-v1:0"))
+
+settings = BedrockChatPromptExecutionSettings(
+    max_tokens=2000,
+    temperature=0.7,
+    top_p=0.8,
+    # Cohere Command specific settings: https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-cohere-command-r-plus.html
+    extension_data={
+        "presence_penalty": 0.5,
+        "seed": 5,
+    },
+)
+
+chat_function = kernel.add_function(
+    plugin_name="ChatBot",
+    function_name="Chat",
+    prompt="{{$chat_history}}{{$user_input}}",
+    template_format="semantic-kernel",
+    prompt_execution_settings=settings,
+)
+
+chat_history = ChatHistory()
+chat_history.add_user_message("Hi there, who are you?")
+chat_history.add_assistant_message("I am Mosscap, a chat bot. I'm trying to figure out what people need")
+
+
+async def chat() -> bool:
+    try:
+        user_input = input("User:> ")
+    except KeyboardInterrupt:
+        print("\n\nExiting chat...")
+        return False
+    except EOFError:
+        print("\n\nExiting chat...")
+        return False
+
+    if user_input == "exit":
+        print("\n\nExiting chat...")
+        return False
+
+    stream = True
+    if stream:
+        chunks = kernel.invoke_stream(
+            chat_function,
+            user_input=user_input,
+            chat_history=chat_history,
+        )
+        print("Mosscap:> ", end="")
+        answer = ""
+        async for message in chunks:
+            print(str(message[0]), end="", flush=True)
+            answer += str(message[0])
+        print("\n")
+    else:
+        answer = await kernel.invoke(
+            chat_function,
+            user_input=user_input,
+            chat_history=chat_history,
+        )
+        print(f"Mosscap:> {answer}")
+
+    chat_history.add_user_message(user_input)
+    chat_history.add_assistant_message(str(answer))
+    return True
+
+
+async def main() -> None:
+    chatting = True
+    while chatting:
+        chatting = await chat()
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

python/samples/concepts/chat_completion/chat_mistral_api.py

@@ -55,22 +55,25 @@ async def chat() -> bool:
 
     stream = True
     if stream:
-        answer = kernel.invoke_stream(
+        chunks = kernel.invoke_stream(
             chat_function,
             user_input=user_input,
             chat_history=chat_history,
         )
         print("Mosscap:> ", end="")
-        async for message in answer:
+        answer = ""
+        async for message in chunks:
             print(str(message[0]), end="")
+            answer += str(message[0])
         print("\n")
-        return True
-    answer = await kernel.invoke(
-        chat_function,
-        user_input=user_input,
-        chat_history=chat_history,
-    )
-    print(f"Mosscap:> {answer}")
+    else:
+        answer = await kernel.invoke(
+            chat_function,
+            user_input=user_input,
+            chat_history=chat_history,
+        )
+        print(f"Mosscap:> {answer}")
+
     chat_history.add_user_message(user_input)
     chat_history.add_assistant_message(str(answer))
     return True

python/semantic_kernel/connectors/ai/bedrock/README.md

@@ -0,0 +1,50 @@
+# Amazon - Bedrock
+
+[Amazon Bedrock](https://docs.aws.amazon.com/bedrock/latest/userguide/what-is-bedrock.html) is a service provided by Amazon Web Services (AWS) that allows you to access large language models with a serverless experience. Semantic Kernel provides a connector to access these models from AWS.
+
+## Prerequisites
+
+- An AWS account and [access to the foundation models](https://docs.aws.amazon.com/bedrock/latest/userguide/model-access-permissions.html)
+- [AWS CLI installed](https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html) and [configured](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/quickstart.html#configuration)
+
+### Configuration
+
+Follow this [guide](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/quickstart.html#configuration) to configure your environment to use the Bedrock API.
+
+## Supports
+
+### Region
+
+To find model supports by AWS regions, refer to this [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-regions.html).
+
+### Input & Output Modalities
+
+Foundational models in Bedrock support the multiple modalities, including text, image, and embedding. However, not all models support the same modalities. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html) for more information.
+
+The Bedrock connector supports all modalities except for **image embeddings, and text to image**.
+
+### Text completion vs chat completion
+
+Some models in Bedrock supports only text completion, or only chat completion (aka Converse API), or both. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-features.html) for more information.
+
+### Tool Use
+
+Not all models in Bedrock support tools. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/models-features.html) for more information.
+
+### Streaming vs Non-Streaming
+
+Not all models in Bedrock support streaming. You can use the boto3 client to check if a model supports streaming. Refer to the [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/conversation-inference-supported-models-features.html) and the [Boto3 documentation](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/bedrock/client/get_foundation_model.html) for more information.
+
+You can also directly call the `get_foundation_model_info("model_id")` method from the Bedrock connector to check if a model supports streaming.
+
+> Note: The bedrock connector will check if a model supports streaming before making a streaming request to the model.
+
+## Model specific parameters
+
+Foundation models can have specific parameters that are unique to the model or the model provider. You can refer to this [AWS documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters.html) for more information.
+
+You can pass these parameters via the `extension_data` field in the `PromptExecutionSettings` object.
+
+## Unsupported features
+
+- [Guardrail](https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails.html)

python/semantic_kernel/connectors/ai/bedrock/bedrock_prompt_execution_settings.py

@@ -0,0 +1,44 @@
+# Copyright (c) Microsoft. All rights reserved.
+
+
+from typing import Any
+
+from pydantic import Field
+
+from semantic_kernel.connectors.ai.prompt_execution_settings import PromptExecutionSettings
+
+
+class BedrockPromptExecutionSettings(PromptExecutionSettings):
+    """Bedrock Prompt Execution Settings."""
+
+    temperature: float | None = Field(None, ge=0.0, le=1.0)
+    top_p: float | None = Field(None, ge=0.0, le=1.0)
+    top_k: int | None = Field(None, gt=0)
+    max_tokens: int | None = Field(None, gt=0)
+    stop: list[str] = Field(default_factory=list)
+
+
+class BedrockChatPromptExecutionSettings(BedrockPromptExecutionSettings):
+    """Bedrock Chat Prompt Execution Settings."""
+
+    tools: list[dict[str, Any]] | None = Field(
+        None,
+        max_length=64,
+        description="Do not set this manually. It is set by the service based on the function choice configuration.",
+    )
+    tool_choice: dict[str, Any] | None = Field(
+        None,
+        description="Do not set this manually. It is set by the service based on the function choice configuration.",
+    )
+
+
+class BedrockTextPromptExecutionSettings(BedrockPromptExecutionSettings):
+    """Bedrock Text Prompt Execution Settings."""
+
+    ...
+
+
+class BedrockEmbeddingPromptExecutionSettings(PromptExecutionSettings):
+    """Bedrock Embedding Prompt Execution Settings."""
+
+    ...

python/semantic_kernel/connectors/ai/bedrock/bedrock_settings.py

@@ -0,0 +1,34 @@
+# Copyright (c) Microsoft. All rights reserved.
+
+from typing import ClassVar
+
+from semantic_kernel.kernel_pydantic import KernelBaseSettings
+from semantic_kernel.utils.experimental_decorator import experimental_class
+
+
+@experimental_class
+class BedrockSettings(KernelBaseSettings):
+    """Amazon Bedrock service settings.
+
+    The settings are first loaded from environment variables with
+    the prefix 'BEDROCK_'.
+    If the environment variables are not found, the settings can
+    be loaded from a .env file with the encoding 'utf-8'.
+    If the settings are not found in the .env file, the settings
+    are ignored; however, validation will fail alerting that the
+    settings are missing.
+
+    Optional settings for prefix 'BEDROCK_' are:
+        - chat_model_id: str | None - The Amazon Bedrock chat model ID to use.
+            (Env var BEDROCK_CHAT_MODEL_ID)
+        - text_model_id: str | None - The Amazon Bedrock text model ID to use.
+            (Env var BEDROCK_TEXT_MODEL_ID)
+        - embedding_model_id: str | None - The Amazon Bedrock embedding model ID to use.
+            (Env var BEDROCK_EMBEDDING_MODEL_ID)
+    """
+
+    env_prefix: ClassVar[str] = "BEDROCK_"
+
+    chat_model_id: str | None = None
+    text_model_id: str | None = None
+    embedding_model_id: str | None = None