feat: add passing http_client to components using OpenAI clients

[alexengrig]

Related Issues

• fixes Add passing http_client to OpenAI clients #9169

Proposed Changes:

When using components with an OpenAI client, we can configure the HTTP client by passing http_client.

How did you test it?

I don’t have any ideas for suitable tests, could you suggest some tests that could be added?

I added unit tests for the new parameter http_client.

Notes for the reviewer

I would like to add documentation on using proxies, authentication, and HTTP_PROXY support, could you tell me where I should add this documentation?

I added HTTP_PROXY as an example to the docstring.

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added unit tests
• I updated the docstrings
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test: and added ! in case the PR includes breaking changes.
• I documented my code
• I ran pre-commit hooks and fixed any issue

[CLAassistant]

All committers have signed the CLA.

[alexengrig]

I would like to add tests of using HTTP_PROXY/HTTPS_PROXY and what do you think – should I write it as a unit test (see below) or as an integration test (with a running mock server)?:

    def test_init_with_proxy_in_http_client_by_env(self, monkeypatch):
        monkeypatch.setenv('HTTP_PROXY', 'http://user:password@my-proxy.com:8080')
        monkeypatch.setenv('HTTPS_PROXY', 'https://user:password@my-proxy.com:4433')
        embedder = OpenAITextEmbedder(
            api_key=Secret.from_token('fake-api-key'),
        )
        proxies_with_auth_headers = {
            (str(mount._pool._proxy_url.origin), mount._pool._proxy_headers[0][1])
            for mount in embedder.client._client._mounts.values()
        }
        assert proxies_with_auth_headers == {
            ('http://my-proxy.com:8080', b'Basic dXNlcjpwYXNzd29yZA=='),
            ('https://my-proxy.com:4433', b'Basic dXNlcjpwYXNzd29yZA=='),
        }

[vblagoje]

Hey @alexengrig thanks for this effort. Let's split it in a few PRs. Perhaps we can start with OpenAIChatGenerator only as a test pilot PR. Note that you seem to be using 3.10 syntax, we still need to support 3.9 so let's use that syntax still. Perhaps open a new PR for OpenAIChatGenerator only?

[alexengrig]

Hi @vblagoje!

Perhaps open a new PR for OpenAIChatGenerator only?

I agree, and I'll open it.

Note that you seem to be using 3.10 syntax, we still need to support 3.9 so let's use that syntax still.

OK, I'll fix it.

[anakin87]

Hello!

For serialization/deserialization purposes, I think we should allow passing http_client_kwargs instead of the client itself.
You can find an example implementation in the AzureOpenAITextEmbedder.

[Amnah199]

Hi @alexengrig,
Thank you for picking this up! I’ve opened a PR #9215 that adds client config support to both OpenAIChatGenerator and AzureChatGenerator. Feel free to use it as a reference for adding similar support to other components. Additionally, we’ve introduced a new utility method for initializing the HTTP client, which might be helpful to integrate as well.

[alexengrig]

Perhaps open a new PR for OpenAIChatGenerator only?

@vblagoje
This PR (#9215) enables OpenAI client configuration in OpenAIChatGenerator and all Azure* components. The only remaining task is to add the configuration to the following components:

• OpenAIDocumentEmbedder and OpenAITextEmbedder
• OpenAIGenerator
• RemoteWhisperTranscriber
• DALLEImageGenerator

What should I do with the current PR, and how should I split the changes for the components into additional PRs?
Embedders, generators and transcriber (3 PRs)?