Resolved comments

Author: howieleung

sdk/ai/azure-ai-projects/.env.template

@@ -41,41 +41,18 @@ SHAREPOINT_PROJECT_CONNECTION_ID=
 A2A_PROJECT_CONNECTION_ID=
 BROWSER_AUTOMATION_PROJECT_CONNECTION_ID=
 OPENAPI_PROJECT_CONNECTION_ID=
+AI_SEARCH_USER_INPUT=
+SHAREPOINT_USER_INPUT=
+
 
 #######################################################################
 #
 # Used in tests
+#######################################################################
 
-# Default Foundry endpoint for tests (specific test groups may read other variables below)
-AZURE_AI_PROJECTS_TESTS_PROJECT_ENDPOINT=
-
-# Used in Agent tests
-AZURE_AI_PROJECTS_TESTS_AGENTS_PROJECT_ENDPOINT=
-AZURE_AI_PROJECTS_TESTS_MODEL_DEPLOYMENT_NAME=
-
-# Used in tracing tests
-AZURE_AI_PROJECTS_TESTS_TRACING_PROJECT_ENDPOINT=
-
-# Used in Container App Agents tests
-AZURE_AI_PROJECTS_TESTS_CONTAINER_PROJECT_ENDPOINT=
-AZURE_AI_PROJECTS_TESTS_CONTAINER_APP_RESOURCE_ID=
-AZURE_AI_PROJECTS_TESTS_CONTAINER_INGRESS_SUBDOMAIN_SUFFIX=
-
-# Used in Agent Image Generation tests
-AZURE_AI_PROJECTS_TESTS_IMAGE_GENERATION_MODEL_DEPLOYMENT_NAME=
-
-# Used in Memory Store tests
-AZURE_AI_PROJECTS_TESTS_MEMORY_STORE_CHAT_MODEL_DEPLOYMENT_NAME=
-AZURE_AI_PROJECTS_TESTS_MEMORY_STORE_EMBEDDING_MODEL_DEPLOYMENT_NAME=
-
-# Connection IDs and settings used in agent tool tests
-AZURE_AI_PROJECTS_TESTS_BING_PROJECT_CONNECTION_ID=
-AZURE_AI_PROJECTS_TESTS_AI_SEARCH_PROJECT_CONNECTION_ID=
-AZURE_AI_PROJECTS_TESTS_AI_SEARCH_INDEX_NAME=
-AZURE_AI_PROJECTS_TESTS_AI_SEARCH_USER_INPUT=
-AZURE_AI_PROJECTS_TESTS_MCP_PROJECT_CONNECTION_ID=
-AZURE_AI_PROJECTS_TESTS_SHAREPOINT_PROJECT_CONNECTION_ID=
-AZURE_AI_PROJECTS_TESTS_SHAREPOINT_USER_INPUT=
+# Used for recording or playback
+AZURE_TEST_RUN_LIVE=false
+AZURE_SKIP_LIVE_RECORDING=true
 
 # Used in Fine-tuning tests
 AZURE_AI_PROJECTS_TESTS_COMPLETED_OAI_MODEL_SFT_FINE_TUNING_JOB_ID=

sdk/ai/azure-ai-projects/.gitignore

@@ -22,9 +22,8 @@
     add_body_regex_sanitizer,
 )
 
-if not load_dotenv(find_dotenv(filename=".test.env"), override=True):
-    print("Failed to apply environment variables for azure-ai-projects tests.")
-
+if not load_dotenv(find_dotenv(), override=True):
+    print("Did not find a .env file. Using default environment variable values for tests.")
 
 def pytest_collection_modifyitems(items):
     if os.environ.get("AZURE_TEST_RUN_LIVE") == "true":

sdk/ai/azure-ai-projects/tests/conftest.py

@@ -1,14 +1,16 @@
-## Recorded sample tests
+# Recorded sample tests
 
 Use recorded tests to validate samples with `SyncSampleExecutor` and `AsyncSampleExecutor`. Tests run the sample code, fail on exceptions, and can optionally validate captured `print` output through LLM instructions.
 
-### Prerequisites
+## Prerequisites
+
 - In `.env`, set the variables required by your samples plus `AZURE_TEST_RUN_LIVE=true` and `AZURE_SKIP_LIVE_RECORDING=false` when capturing recordings. CI playback typically uses `AZURE_TEST_RUN_LIVE=false` and `AZURE_SKIP_LIVE_RECORDING=true`.
 - Provide sanitized defaults via `servicePreparer` so recordings do not leak secrets. All mandatory environment variables used by the tests/samples must be specified in `servicePreparer` (as sanitized values) so playback can run without access to real secrets.
 
 **VS Code tip (record a single sample):** Open the **Testing** tab, expand the pytest tree to find the specific sample test case (for example, one parameterized case for a particular `sample_path`), then right-click it and choose **Run Test** (or **Debug Test**). Make sure your `.env` (or your test run environment) includes `AZURE_TEST_RUN_LIVE=true` and `AZURE_SKIP_LIVE_RECORDING=false` so that run captures a new recording for just that sample.
 
-### Sync example
+## Sync example
+
 ```python
 import pytest
 import os
@@ -49,7 +51,8 @@ class TestSamples(AzureRecordedTestCase):
         )        
 ```
 
-### Async example
+## Async example
+
 ```python
 import pytest
 from devtools_testutils.aio import recorded_by_proxy_async
@@ -86,9 +89,11 @@ class TestSamplesAsync(AzureRecordedTestCase):
         )
 ```
 
-### Key pieces
+## Key pieces
+
 - `@servicePreparer()`: A custom helper you create for your test suite. It supplies sanitized environment variables for playback (often via `EnvironmentVariableLoader`). The name is up to you; these examples call it `servicePreparer` by convention.
 - Example:
+
 ```python
 import functools
 from devtools_testutils import EnvironmentVariableLoader
@@ -101,8 +106,10 @@ servicePreparer = functools.partial(
     # add other sanitized vars here
 )
 ```
+
 - If all your test environment variables share a prefix (for example `azure_ai_projects_tests`), pass that prefix as the second positional
     argument to `EnvironmentVariableLoader` and include the prefix on the keys you provide.  For example:
+
 ```python
 servicePreparer = functools.partial(
     EnvironmentVariableLoader,
@@ -112,14 +119,16 @@ servicePreparer = functools.partial(
     # add other sanitized vars here
 )
 ```
+
 - `@pytest.mark.parametrize`: Drives one test per sample file. Use `samples_to_test` or `samples_to_skip` with `get_sample_paths` / `get_async_sample_paths`.
 - `@SamplePathPasser`: Forwards the sample path to the recorder decorators.
 - `recorded_by_proxy` / `recorded_by_proxy_async`: Wrap tests for recording/playback. Include `RecordedTransport.HTTPX` when samples use httpx in addition to the default `RecordedTransport.AZURE_CORE`.
 - `execute` / `execute_async`: Run the sample; any exception fails the test.
 - `validate_print_calls_by_llm` / `validate_print_calls_by_llm_async`: Optionally validate captured print output with LLM instructions and an explicit `project_endpoint` (and optional `model`).
 - `kwargs` in the test function: A dictionary with environment variables in key and value pairs.
 
-### Optional test environment variables mapping
+## Optional test environment variables mapping
+
 If you need to remap the environment variable names provided by your fixtures to the names the sample expects, pass a dictionary via the `env_var_mapping` kwarg on the sample executors. When your fixtures already supply the sample-ready names, you can omit `env_var_mapping`.
 
 ```python
@@ -130,7 +139,8 @@ env_var_mapping = {
 executor = SyncSampleExecutor(self, sample_path, env_var_mapping=env_var_mapping, **kwargs)
 ```
 
-### Optional environment variables
+## Optional environment variables
+
 To run the “hero path” of a sample, your `@servicePreparer` should provide all mandatory environment variables.
 If a sample supports optional environment variables (for optional features/paths), use `@additionalSampleTests` to generate additional recorded test cases with those optional variables set.
 
@@ -169,12 +179,12 @@ def test_agent_tools_samples(self, sample_path: str, **kwargs) -> None:
 ```
 
 Notes:
+
 - `AdditionalSampleTestDetail.env_vars` is a mapping of **sample env-var name** -> **sanitized playback value**.
-    - Live/recording (`AZURE_TEST_RUN_LIVE=true`): reads the real value from your environment (for example, from `.env`) and sanitizes it to the provided playback value.
-    - Playback (`AZURE_TEST_RUN_LIVE=false`): sets each key to the provided playback value.
+  - Live/recording (`AZURE_TEST_RUN_LIVE=true`): reads the real value from your environment (for example, from `.env`) and sanitizes it to the provided playback value.
+  - Playback (`AZURE_TEST_RUN_LIVE=false`): sets each key to the provided playback value.
 - `AdditionalSampleTestDetail.test_id` customizes the parameter id used for that variant.
-    - If omitted, the id is auto-generated from the sample filename and env-var keys (for example, `sample_agent_computer_use-[COMPUTER_USE_MODEL_DEPLOYMENT_NAME]`).
-    - If provided, it becomes the full parameter id for that variant (no filename prefix).
-    - If the auto-generated id makes the recording file path too long (most commonly on Windows), the recording file may fail to generate; set a short `test_id` (and keep it unique across all parametrized cases).
+  - If omitted, the id is auto-generated from the sample filename and env-var keys (for example, `sample_agent_computer_use-[COMPUTER_USE_MODEL_DEPLOYMENT_NAME]`).
+  - If provided, it becomes the full parameter id for that variant (no filename prefix).
+  - If the auto-generated id makes the recording file path too long (most commonly on Windows), the recording file may fail to generate; set a short `test_id` (and keep it unique across all parametrized cases).
 - In VS Code’s **Testing** tab, these show up as additional parameterized cases for the same test, using the parameter id.
-