test(azure openai repro): adds example script to reproduce Azure OpenAI bug

[arthurgousset]

Description of changes

Adds a minimal test case to reproduce the bug described in chroma-core#1770. Can't reproduce the bug with this script, seems to be resolved in the client already unless I'm missing something.

Test plan

Setup environment

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt -r requirements_dev.txt
pip install -e .

Run test:

# From root directory
$ AZURE_OPENAI_API_KEY="azure-openai-api-key" python azure_openai_repro.py

{'ids': [['cat_embedding']], 'embeddings': None, 'documents': [['Cat']], 'uris': None, 'included': ['metadatas', 'documents', 'distances'], 'data': None, 'metadatas': [[None]], 'distances': [[0.8539673089981079]]}

Use OPENAI_LOG=debug to see more verbose debug logs and confirm the embedding request is going to our private Azure OpenAI deployment.

# From root directory
$ OPENAI_LOG=debug AZURE_OPENAI_API_KEY="azure-openai-api-key" python azure_openai_repro.py

[2025-05-17 15:48:10 - openai._base_client:482 - DEBUG] Request options: {'method': 'post', 'url': '/deployments/text-embedding-3-small/embeddings', 'headers': {'api-key': '<redacted>'}, 'files': None, 'idempotency_key': 'stainless-python-retry-74021265-20cd-46a4-9a3b-9eda799efed3', 'post_parser': <function Embeddings.create.<locals>.parser at 0x10508e5c0>, 'json_data': {'input': ['Cat', 'Apple', 'San Francisco'], 'model': 'text-embedding-3-small', 'encoding_format': 'base64'}}
[2025-05-17 15:48:10 - openai._base_client:965 - DEBUG] Sending HTTP Request: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01
[2025-05-17 15:48:10 - httpx:1025 - INFO] HTTP Request: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01 "HTTP/1.1 200 OK"
[2025-05-17 15:48:10 - openai._base_client:1003 - DEBUG] HTTP Response: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01 "200 OK" Headers({'content-length': '24962', 'content-type': 'application/json', 'access-control-allow-origin': '*', 'apim-request-id': 'bdb2632d-73df-4a05-8a05-702544fb05f5', 'strict-transport-security': 'max-age=31536000; includeSubDomains; preload', 'x-content-type-options': 'nosniff', 'x-ms-region': 'East US', 'x-ratelimit-remaining-tokens': '149993', 'x-ratelimit-limit-tokens': '150000', 'x-request-id': '72d69d11-3ae2-4bc9-91e0-2576ddc7d84c', 'azureml-model-session': 'd011-20250512072110', 'x-envoy-upstream-service-time': '30', 'x-ms-client-request-id': 'bdb2632d-73df-4a05-8a05-702544fb05f5', 'x-ms-deployment-name': 'text-embedding-3-small', 'date': 'Sat, 17 May 2025 22:48:10 GMT'})
[2025-05-17 15:48:10 - openai._base_client:1011 - DEBUG] request_id: 72d69d11-3ae2-4bc9-91e0-2576ddc7d84c
[2025-05-17 15:48:11 - openai._base_client:482 - DEBUG] Request options: {'method': 'post', 'url': '/deployments/text-embedding-3-small/embeddings', 'headers': {'api-key': '<redacted>'}, 'files': None, 'idempotency_key': 'stainless-python-retry-797fdf84-a809-4785-842f-35b7745f6f97', 'post_parser': <function Embeddings.create.<locals>.parser at 0x10508e5c0>, 'json_data': {'input': ['Dog'], 'model': 'text-embedding-3-small', 'encoding_format': 'base64'}}
[2025-05-17 15:48:11 - openai._base_client:965 - DEBUG] Sending HTTP Request: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01
[2025-05-17 15:48:11 - httpx:1025 - INFO] HTTP Request: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01 "HTTP/1.1 200 OK"
[2025-05-17 15:48:11 - openai._base_client:1003 - DEBUG] HTTP Response: POST https://chroma-repro-bug.openai.azure.com/openai/deployments/text-embedding-3-small/embeddings?api-version=2024-02-01 "200 OK" Headers({'content-length': '8414', 'content-type': 'application/json', 'access-control-allow-origin': '*', 'apim-request-id': 'cf07bf26-f633-4f5b-bd52-2e80fb4f7251', 'strict-transport-security': 'max-age=31536000; includeSubDomains; preload', 'x-content-type-options': 'nosniff', 'x-ms-region': 'East US', 'x-ratelimit-remaining-tokens': '149992', 'x-ratelimit-limit-tokens': '150000', 'x-request-id': '2521d7bc-b1eb-4820-a57c-d443f8ee9d69', 'azureml-model-session': 'd011-20250512072110', 'x-envoy-upstream-service-time': '24', 'x-ms-client-request-id': 'cf07bf26-f633-4f5b-bd52-2e80fb4f7251', 'x-ms-deployment-name': 'text-embedding-3-small', 'date': 'Sat, 17 May 2025 22:48:10 GMT'})
[2025-05-17 15:48:11 - openai._base_client:1011 - DEBUG] request_id: 2521d7bc-b1eb-4820-a57c-d443f8ee9d69
{'ids': [['cat_embedding']], 'embeddings': None, 'documents': [['Cat']], 'uris': None, 'included': ['metadatas', 'documents', 'distances'], 'data': None, 'metadatas': [[None]], 'distances': [[0.8539673089981079]]}

• Tests pass locally with pytest for python, yarn test for js, cargo test for rust

Documentation Changes

Are all docstrings for user-facing APIs updated if required? Do we need to make documentation changes in the docs section?