ic

packages/azure-ada/banortegpt/embedding/azure_ada/__init__.py

@@ -0,0 +1,4 @@
+from .blocking import Ada
+from .nonblocking import AsyncAda
+
+__all__ = ["Ada", "AsyncAda"]

packages/azure-ada/banortegpt/embedding/azure_ada/base.py

@@ -0,0 +1,51 @@
+from typing import Protocol
+
+
+class BaseAda:
+    def __init__(
+        self, model: str | None = None, *, endpoint: str, key: str, version: str
+    ) -> None:
+        self.model = model
+
+    class Config(Protocol):
+        embedding_model: str
+        azure_endpoint: str
+        openai_api_key: str
+        openai_api_version: str
+
+    @classmethod
+    def from_config(cls, c: Config):
+        return cls(
+            model=c.embedding_model,
+            endpoint=c.azure_endpoint,
+            key=c.openai_api_key,
+            version=c.openai_api_version,
+        )
+
+    @classmethod
+    def from_vault(
+        cls,
+        vault: str,
+        *,
+        model: str | None = None,
+        url: str | None = None,
+        token: str | None = None,
+        mount_point: str = "secret",
+    ):
+        from hvac import Client
+
+        client = Client(url=url or "https://vault.ia-innovacion.work", token=token)
+
+        if not client.is_authenticated():
+            raise Exception("Vault authentication failed")
+
+        secret_map = client.secrets.kv.v2.read_secret_version(
+            path=vault, mount_point=mount_point
+        )["data"]["data"]
+
+        return cls(
+            model=model,
+            endpoint=secret_map["azure_endpoint"],
+            key=secret_map["openai_api_key"],
+            version=secret_map["openai_api_version"],
+        )

packages/azure-ada/banortegpt/embedding/azure_ada/blocking.py

@@ -0,0 +1,47 @@
+from langfuse.openai import AzureOpenAI
+from openai.types.embedding import Embedding
+
+from .base import BaseAda
+
+
+class Ada(BaseAda):
+    def __init__(
+        self, model: str | None = None, *, endpoint: str, key: str, version: str
+    ) -> None:
+        super().__init__(model, endpoint=endpoint, key=key, version=version)
+        self.client = AzureOpenAI(
+            azure_endpoint=endpoint, api_key=key, api_version=version
+        )
+
+    def embed(
+        self, input: str | list[str], *, model: str | None = None
+    ) -> list[float] | list[list[float]]:
+        if isinstance(input, str):
+            return self.embed_query(input, model)
+        else:
+            return self.batch_embed(input, model)
+
+    def batch_embed(
+        self, texts: list[str], model: str | None = None
+    ) -> list[list[float]]:
+        if model is None:
+            if self.model is None:
+                raise ValueError("No embedding model set")
+            model = self.model
+
+        batches = [texts[i : i + 2048] for i in range(0, len(texts), 2048)]
+        results = [
+            (self.client.embeddings.create(input=batch, model=model)).data
+            for batch in batches
+        ]
+        flattened_results: list[Embedding] = sum(results, [])
+        return [result.embedding for result in flattened_results]
+
+    def embed_query(self, text: str, model: str | None = None) -> list[float]:
+        if model is None:
+            if self.model is None:
+                raise ValueError("No embedding model set")
+            model = self.model
+
+        response = self.client.embeddings.create(input=text, model=model)
+        return response.data[0].embedding

packages/azure-ada/banortegpt/embedding/azure_ada/nonblocking.py

@@ -0,0 +1,47 @@
+from langfuse.openai import AsyncAzureOpenAI
+from openai.types.embedding import Embedding
+
+from .base import BaseAda
+
+
+class AsyncAda(BaseAda):
+    def __init__(
+        self, model: str | None = None, *, endpoint: str, key: str, version: str
+    ) -> None:
+        super().__init__(model, endpoint=endpoint, key=key, version=version)
+        self.client = AsyncAzureOpenAI(
+            azure_endpoint=endpoint, api_key=key, api_version=version
+        )
+
+    async def embed(
+        self, input: str | list[str], *, model: str | None = None
+    ) -> list[float] | list[list[float]]:
+        if isinstance(input, str):
+            return await self.embed_query(input, model)
+        else:
+            return await self.batch_embed(input, model)
+
+    async def batch_embed(
+        self, texts: list[str], model: str | None = None
+    ) -> list[list[float]]:
+        if model is None:
+            if self.model is None:
+                raise ValueError("No embedding model set")
+            model = self.model
+
+        batches = [texts[i : i + 2048] for i in range(0, len(texts), 2048)]
+        results = [
+            (await self.client.embeddings.create(input=batch, model=model)).data
+            for batch in batches
+        ]
+        flattened_results: list[Embedding] = sum(results, [])
+        return [result.embedding for result in flattened_results]
+
+    async def embed_query(self, text: str, model: str | None = None) -> list[float]:
+        if model is None:
+            if self.model is None:
+                raise ValueError("No embedding model set")
+            model = self.model
+
+        response = await self.client.embeddings.create(input=text, model=model)
+        return response.data[0].embedding

packages/azure-ada/pyproject.toml

@@ -0,0 +1,15 @@
+[project]
+name = "azure-ada"
+version = "0.1.0"
+description = "Add your description here"
+readme = "README.md"
+authors = [{ name = "ajac-zero", email = "ajcardoza2000@gmail.com" }]
+requires-python = ">=3.12"
+dependencies = ["hvac","openai>=1.72.0"]
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.hatch.build.targets.wheel]
+packages = ["banortegpt"]