Merge branch 'main' into hod/unicode

.devcontainer/Dockerfile

@@ -0,0 +1,23 @@
+FROM python:3.9-slim-bullseye AS base
+
+RUN set -x
+
+RUN apt-get update \
+    && apt-get -y install curl \
+    && apt-get -y install net-tools \
+    && apt-get -y install procps \
+    && apt-get -y install build-essential \
+    && apt-get -y install docker.io
+
+# specific pip version to workaround: https://github.com/conda/conda/issues/10178
+RUN pip install --upgrade pip==20.1.1
+RUN pip install ipython ipykernel
+RUN ipython kernel install --user --name aml
+
+# FROM base AS promptflow
+COPY default_requirements.txt .
+RUN pip install -r default_requirements.txt
+
+RUN set +x
+
+CMD bash

.devcontainer/README.md

@@ -0,0 +1,17 @@
+# Devcontainer for promptflow
+To facilitate your promptflow project development and empower you to work on LLM projects using promptflow more effectively, 
+we've configured the necessary environment for developing promptflow projects and utilizing flows through the dev container feature. 
+You can seamlessly initiate your promptflow project development and start leveraging flows by simply using the dev container feature via VS Code or Codespaces.
+
+# Use devcontainer
+1. Use vscode to open promptflow repo, and install vscode extension: Dev Containers and then open promptflow with dev containers.
+   ![devcontainer](./devcontainers.png)  
+   **About dev containers please refer to: [dev containers](https://code.visualstudio.com/docs/devcontainers/containers)**
+2. Use codespaces to open promptflow repo, it will automatically build the dev containers environment and open promptflow with dev containers.
+   ![codespaces](./codespaces.png)
+
+### Notes
+1. If you only want to try out promptflow without developing promptflow, you can simply install Docker and use promptflow within Docker without the need for using DevContainers functionality.
+   1. `docker build -t promptflow_container`
+   2. `docker run -it promptflow_container`
+2. When using the dev containers function, the promptflow and promptflow-tools installed in the container are the code of the current repo.

.devcontainer/default_requirements.txt

@@ -0,0 +1,5 @@
+azure-cli
+azure-identity
+opencensus-ext-azure
+promptflow[azure]
+promptflow-tools

.devcontainer/devcontainer.json

@@ -0,0 +1,25 @@
+{
+	"name": "Promptflow-Python39",
+	// "context" is the path that the Codespaces docker build command should be run from, relative to devcontainer.json
+	"context": ".",
+	"dockerFile": "Dockerfile",
+
+	// Set *default* container specific settings.json values on container create.
+	"settings": {
+		"terminal.integrated.shell.linux": "/bin/bash"
+	},
+
+	// Add the IDs of extensions you want installed when the container is created.
+	"extensions": [
+		"ms-python.python",
+		"ms-toolsai.vscode-ai",
+		"ms-toolsai.jupyter",
+		"redhat.vscode-yaml",
+		"prompt-flow.prompt-flow"
+	],
+
+	// Use 'postCreateCommand' to run commands after the container is created.
+	"postCreateCommand": "pip install -r ${containerWorkspaceFolder}/src/promptflow-tools/requirements.txt --force-reinstall;pip install -r ${containerWorkspaceFolder}/src/promptflow/dev_requirements.txt --force-reinstall;pip install -e ${containerWorkspaceFolder}/src/promptflow-tools;pip install -e ${containerWorkspaceFolder}/src/promptflow",
+
+	"runArgs": ["-v", "/var/run/docker.sock:/var/run/docker.sock"]
+}

docs/how-to-guides/deploy-a-flow/deploy-using-dev-server.md

@@ -14,6 +14,13 @@ an example to show how to deploy a flow.
 Please ensure you have [create the connection](../manage-connections.md#create-a-connection) required by flow, if not, you could
 refer to [Setup connection for web-classification](https://github.com/microsoft/promptflow/tree/main/examples/flows/standard/web-classification).
 
+Note: We will use relevant environment variable ({connection_name}_{key_name}) to override connection configurations in 
+serving mode, white space in connection name will be removed directly from environment variable name. For instance, 
+if there is a custom connection named 'custom_connection' with a configuration key called 'chat_deployment_name,' the 
+function will attempt to retrieve 'chat_deployment_name' from the environment variable 
+'CUSTOM_CONNECTION_CHAT_DEPLOYMENT_NAME' by default. If the environment variable is not set, it will use the original 
+value as a fallback.
+
 
 The following CLI commands allows you serve a flow folder as an endpoint. By running this command, a [flask](https://flask.palletsprojects.com/en/) app will start in the environment where command is executed, please ensure all prerequisites required by flow have been installed.
 ```bash

examples/flows/standard/gen-docstring/main.py

@@ -3,14 +3,16 @@
 from diff import show_diff
 from load_code_tool import load_code
 from promptflow import PFClient
+from pathlib import Path
 
 
 if __name__ == "__main__":
+    current_folder = Path(__file__).absolute().parent
     parser = argparse.ArgumentParser(description="The code path of code that need to generate docstring.")
-    parser.add_argument("--source", help="Path for the code file", default='./azure_open_ai.py')
+    parser.add_argument("--source", help="Path for the code file", default=str(current_folder / 'azure_open_ai.py'))
     args = parser.parse_args()
 
     pf = PFClient()
     source = args.source
-    flow_result = pf.test(flow="./", inputs={"source": source})
+    flow_result = pf.test(flow=str(current_folder), inputs={"source": source})
     show_diff(load_code(source), flow_result['code'], File(source).filename)

src/promptflow/promptflow/_sdk/_serving/flow_invoker.py

@@ -12,6 +12,7 @@
     get_local_connections_from_executable,
     resolve_connections_environment_variable_reference,
     update_environment_variables_with_connections,
+    override_connection_config_with_environment_variable,
 )
 from promptflow._sdk.entities._connection import _Connection
 from promptflow.executor import FlowExecutor
@@ -61,6 +62,7 @@ def _init_connections(self, connection_provider):
         else:
             raise UnsupportedConnectionProvider(connection_provider)
 
+        override_connection_config_with_environment_variable(self.connections)
         resolve_connections_environment_variable_reference(self.connections)
         update_environment_variables_with_connections(self.connections)
         logger.info(f"Promptflow get connections successfully. keys: {self.connections.keys()}")

src/promptflow/promptflow/_sdk/_utils.py

@@ -317,6 +317,29 @@ def _match_env_reference(val: str):
     return name
 
 
+def override_connection_config_with_environment_variable(connections: Dict[str, dict]):
+    """
+    The function will use relevant environment variable to override connection configurations. For instance, if there
+    is a custom connection named 'custom_connection' with a configuration key called 'chat_deployment_name,' the
+    function will attempt to retrieve 'chat_deployment_name' from the environment variable
+    'CUSTOM_CONNECTION_CHAT_DEPLOYMENT_NAME' by default. If the environment variable is not set, it will use the
+    original value as a fallback.
+    """
+    logger = logging.getLogger(LOGGER_NAME)
+    for connection_name, connection in connections.items():
+        values = connection.get("value", {})
+        for key, val in values.items():
+            connection_name = connection_name.replace(" ", "_")
+            env_name = f"{connection_name}_{key}".upper()
+            if env_name not in os.environ:
+                continue
+            values[key] = os.environ[env_name]
+            logger.info(
+                f"Connection {connection_name}'s {key} is overridden with environment variable {env_name}"
+            )
+    return connections
+
+
 def resolve_connections_environment_variable_reference(connections: Dict[str, dict]):
     """The function will resolve connection secrets env var reference like api_key: ${env:KEY}"""
     for connection in connections.values():

src/promptflow/tests/sdk_cli_test/unittests/test_utils.py

@@ -30,6 +30,7 @@
     generate_flow_tools_json,
     refresh_connections_dir,
     resolve_connections_environment_variable_reference,
+    override_connection_config_with_environment_variable,
     snake_to_camel,
 )
 
@@ -87,6 +88,29 @@ def test_resolve_connections_environment_variable_reference(self):
         assert connections["test_connection"]["value"]["api_base"] == "BASE"
         assert connections["test_custom_connection"]["value"]["key"] == "CUSTOM_VALUE"
 
+    def test_override_connection_config_with_environment_variable(self):
+        connections = {
+            "test_connection": {
+                "type": "AzureOpenAIConnection",
+                "value": {
+                    "api_key": "KEY",
+                    "api_base": "https://gpt-test-eus.openai.azure.com/",
+                },
+            },
+            "test_custom_connection": {
+                "type": "CustomConnection",
+                "value": {"key": "value1", "key2": "value2"},
+            },
+        }
+        with mock.patch.dict(
+            os.environ, {"TEST_CONNECTION_API_BASE": "BASE", "TEST_CUSTOM_CONNECTION_KEY": "CUSTOM_VALUE"}
+        ):
+            override_connection_config_with_environment_variable(connections)
+        assert connections["test_connection"]["value"]["api_key"] == "KEY"
+        assert connections["test_connection"]["value"]["api_base"] == "BASE"
+        assert connections["test_custom_connection"]["value"]["key"] == "CUSTOM_VALUE"
+        assert connections["test_custom_connection"]["value"]["key2"] == "value2"
+
     def test_generate_flow_tools_json(self) -> None:
         # call twice to ensure system path won't be affected during generation
         for _ in range(2):