Skip to content

:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.

2.1 KB
Raw Blame
from collections.abc import Generator
from typing import cast

from django.conf import settings

from azure.ai.inference import EmbeddingsClient
from azure.ai.inference.aio import EmbeddingsClient as EmbeddingsClientAsync
from azure.core.credentials import AzureKeyCredential


def _validate_azure_config() -> tuple[str, str]:
    endpoint = str(settings.AZURE_INFERENCE_ENDPOINT).strip()
    credential = str(settings.AZURE_INFERENCE_CREDENTIAL).strip()
    if not endpoint:
        raise ValueError(
            "AZURE_INFERENCE_ENDPOINT is not configured. The Azure embeddings client requires a valid endpoint URL."
        )
    if not credential:
        raise ValueError(
            "AZURE_INFERENCE_CREDENTIAL is not configured. The Azure embeddings client requires a valid API key."
        )
    return endpoint, credential


def get_azure_embeddings_client() -> EmbeddingsClient:
    endpoint, credential = _validate_azure_config()
    return EmbeddingsClient(
        endpoint=endpoint,
        credential=AzureKeyCredential(credential),
    )


def get_async_azure_embeddings_client() -> EmbeddingsClientAsync:
    endpoint, credential = _validate_azure_config()
    return EmbeddingsClientAsync(
        endpoint=endpoint,
        credential=AzureKeyCredential(credential),
    )


async def aembed_documents(client: EmbeddingsClientAsync, texts: list[str]) -> Generator[list[float]]:
    """Embed documents for storing in a vector database."""
    response = await client.embed(
        input=texts,
        encoding_format="float",
        model="embed-v-4-0",
        input_type="document",
    )
    if not response.data:
        raise ValueError("No embeddings returned")
    return (cast(list[float], res.embedding) for res in response.data)


def embed_search_query(client: EmbeddingsClient, text: str) -> list[float]:
    """Embed a search query for semantic search by stored documents."""
    response = client.embed(
        input=[text],
        encoding_format="float",
        model="embed-v-4-0",
        input_type="query",
    )
    if not response.data:
        raise ValueError("No embeddings returned")
    return cast(list[float], response.data[0].embedding)