|
| 1 | +# Copyright (c) Microsoft. All rights reserved. |
| 2 | + |
| 3 | +from __future__ import annotations |
| 4 | + |
| 5 | +import logging |
| 6 | +import sys |
| 7 | +from collections.abc import Sequence |
| 8 | +from typing import Any, ClassVar, Generic, TypedDict |
| 9 | + |
| 10 | +from agent_framework import ( |
| 11 | + BaseEmbeddingClient, |
| 12 | + Embedding, |
| 13 | + EmbeddingGenerationOptions, |
| 14 | + GeneratedEmbeddings, |
| 15 | + UsageDetails, |
| 16 | + load_settings, |
| 17 | +) |
| 18 | +from agent_framework._settings import SecretString |
| 19 | +from agent_framework.observability import EmbeddingTelemetryLayer |
| 20 | +from mistralai.client import Mistral |
| 21 | + |
| 22 | +if sys.version_info >= (3, 13): |
| 23 | + from typing import TypeVar # type: ignore # pragma: no cover |
| 24 | +else: |
| 25 | + from typing_extensions import TypeVar # type: ignore # pragma: no cover |
| 26 | + |
| 27 | + |
| 28 | +logger = logging.getLogger("agent_framework.mistral") |
| 29 | + |
| 30 | + |
| 31 | +class MistralEmbeddingOptions(EmbeddingGenerationOptions, total=False): |
| 32 | + """Mistral AI-specific embedding options. |
| 33 | +
|
| 34 | + Extends EmbeddingGenerationOptions with Mistral-specific fields. |
| 35 | +
|
| 36 | + Examples: |
| 37 | + .. code-block:: python |
| 38 | +
|
| 39 | + from agent_framework_mistral import MistralEmbeddingOptions |
| 40 | +
|
| 41 | + options: MistralEmbeddingOptions = { |
| 42 | + "model": "mistral-embed", |
| 43 | + "dimensions": 1024, |
| 44 | + } |
| 45 | + """ |
| 46 | + |
| 47 | + |
| 48 | +MistralEmbeddingOptionsT = TypeVar( |
| 49 | + "MistralEmbeddingOptionsT", |
| 50 | + bound=TypedDict, # type: ignore[valid-type] |
| 51 | + default="MistralEmbeddingOptions", |
| 52 | + covariant=True, |
| 53 | +) |
| 54 | + |
| 55 | + |
| 56 | +class MistralEmbeddingSettings(TypedDict, total=False): |
| 57 | + """Mistral AI embedding settings. |
| 58 | +
|
| 59 | + Fields: |
| 60 | + api_key: Mistral API key. Resolved from ``MISTRAL_API_KEY``. |
| 61 | + embedding_model: Embedding model name. Resolved from ``MISTRAL_EMBEDDING_MODEL``. |
| 62 | + server_url: Optional server URL override. Resolved from ``MISTRAL_SERVER_URL``. |
| 63 | + """ |
| 64 | + |
| 65 | + api_key: str | None |
| 66 | + embedding_model: str | None |
| 67 | + server_url: str | None |
| 68 | + |
| 69 | + |
| 70 | +class RawMistralEmbeddingClient( |
| 71 | + BaseEmbeddingClient[str, list[float], MistralEmbeddingOptionsT], |
| 72 | + Generic[MistralEmbeddingOptionsT], |
| 73 | +): |
| 74 | + """Raw Mistral AI embedding client without telemetry. |
| 75 | +
|
| 76 | + Keyword Args: |
| 77 | + model: The Mistral embedding model (e.g. "mistral-embed"). |
| 78 | + Can also be set via environment variable ``MISTRAL_EMBEDDING_MODEL``. |
| 79 | + api_key: Mistral API key. Defaults to ``MISTRAL_API_KEY`` environment variable. |
| 80 | + server_url: Optional server URL override. Defaults to ``MISTRAL_SERVER_URL`` |
| 81 | + environment variable, or the Mistral default. |
| 82 | + client: Optional pre-configured ``Mistral`` client instance. |
| 83 | + additional_properties: Additional properties stored on the client instance. |
| 84 | + env_file_path: Path to ``.env`` file for settings. |
| 85 | + env_file_encoding: Encoding for ``.env`` file. |
| 86 | + """ |
| 87 | + |
| 88 | + INJECTABLE: ClassVar[set[str]] = {"client"} |
| 89 | + |
| 90 | + def __init__( |
| 91 | + self, |
| 92 | + *, |
| 93 | + model: str | None = None, |
| 94 | + api_key: str | SecretString | None = None, |
| 95 | + server_url: str | None = None, |
| 96 | + client: Mistral | None = None, |
| 97 | + additional_properties: dict[str, Any] | None = None, |
| 98 | + env_file_path: str | None = None, |
| 99 | + env_file_encoding: str | None = None, |
| 100 | + ) -> None: |
| 101 | + """Initialize a raw Mistral AI embedding client.""" |
| 102 | + mistral_settings = load_settings( |
| 103 | + MistralEmbeddingSettings, |
| 104 | + env_prefix="MISTRAL_", |
| 105 | + required_fields=["embedding_model", "api_key"], |
| 106 | + api_key=str(api_key) if isinstance(api_key, SecretString) else api_key, |
| 107 | + embedding_model=model, |
| 108 | + server_url=server_url, |
| 109 | + env_file_path=env_file_path, |
| 110 | + env_file_encoding=env_file_encoding, |
| 111 | + ) |
| 112 | + |
| 113 | + self.model: str = mistral_settings["embedding_model"] # type: ignore[assignment] |
| 114 | + resolved_api_key: str = mistral_settings["api_key"] # type: ignore[assignment] |
| 115 | + resolved_server_url = mistral_settings.get("server_url") |
| 116 | + |
| 117 | + if client is not None: |
| 118 | + self.client = client |
| 119 | + else: |
| 120 | + client_kwargs: dict[str, Any] = {"api_key": resolved_api_key} |
| 121 | + if resolved_server_url: |
| 122 | + client_kwargs["server_url"] = resolved_server_url |
| 123 | + self.client = Mistral(**client_kwargs) |
| 124 | + |
| 125 | + self.server_url = resolved_server_url |
| 126 | + super().__init__(additional_properties=additional_properties) |
| 127 | + |
| 128 | + def service_url(self) -> str: |
| 129 | + """Get the URL of the service.""" |
| 130 | + return self.server_url or "https://api.mistral.ai" |
| 131 | + |
| 132 | + async def get_embeddings( |
| 133 | + self, |
| 134 | + values: Sequence[str], |
| 135 | + *, |
| 136 | + options: MistralEmbeddingOptionsT | None = None, |
| 137 | + ) -> GeneratedEmbeddings[list[float], MistralEmbeddingOptionsT]: |
| 138 | + """Call the Mistral AI embeddings API. |
| 139 | +
|
| 140 | + Args: |
| 141 | + values: The text values to generate embeddings for. |
| 142 | + options: Optional embedding generation options. |
| 143 | +
|
| 144 | + Returns: |
| 145 | + Generated embeddings with usage metadata. |
| 146 | +
|
| 147 | + Raises: |
| 148 | + ValueError: If model is not provided or values is empty. |
| 149 | + """ |
| 150 | + if not values: |
| 151 | + return GeneratedEmbeddings([], options=options) |
| 152 | + |
| 153 | + opts: dict[str, Any] = options or {} # type: ignore |
| 154 | + model = opts.get("model") or self.model |
| 155 | + if not model: |
| 156 | + raise ValueError("model is required") |
| 157 | + |
| 158 | + kwargs: dict[str, Any] = {"model": model, "inputs": list(values)} |
| 159 | + if "dimensions" in opts: |
| 160 | + kwargs["output_dimension"] = opts["dimensions"] |
| 161 | + |
| 162 | + response = await self.client.embeddings.create_async(**kwargs) |
| 163 | + |
| 164 | + embeddings: list[Embedding[list[float]]] = [] |
| 165 | + if response and response.data: |
| 166 | + items = sorted(response.data, key=lambda d: d.index if d.index is not None else 0) |
| 167 | + for item in items: |
| 168 | + vector = list(item.embedding) if item.embedding else [] |
| 169 | + embeddings.append( |
| 170 | + Embedding( |
| 171 | + vector=vector, |
| 172 | + dimensions=len(vector), |
| 173 | + model=response.model or model, |
| 174 | + ) |
| 175 | + ) |
| 176 | + |
| 177 | + usage_dict: UsageDetails | None = None |
| 178 | + if response and response.usage: |
| 179 | + usage_dict = { |
| 180 | + "input_token_count": response.usage.prompt_tokens, |
| 181 | + "total_token_count": response.usage.total_tokens, |
| 182 | + } |
| 183 | + |
| 184 | + return GeneratedEmbeddings(embeddings, options=options, usage=usage_dict) |
| 185 | + |
| 186 | + |
| 187 | +class MistralEmbeddingClient( |
| 188 | + EmbeddingTelemetryLayer[str, list[float], MistralEmbeddingOptionsT], |
| 189 | + RawMistralEmbeddingClient[MistralEmbeddingOptionsT], |
| 190 | + Generic[MistralEmbeddingOptionsT], |
| 191 | +): |
| 192 | + """Mistral AI embedding client with telemetry support. |
| 193 | +
|
| 194 | + Keyword Args: |
| 195 | + model: The Mistral embedding model (e.g. "mistral-embed"). |
| 196 | + Can also be set via environment variable ``MISTRAL_EMBEDDING_MODEL``. |
| 197 | + api_key: Mistral API key. Defaults to ``MISTRAL_API_KEY`` environment variable. |
| 198 | + server_url: Optional server URL override. Defaults to ``MISTRAL_SERVER_URL`` |
| 199 | + environment variable, or the Mistral default. |
| 200 | + client: Optional pre-configured ``Mistral`` client instance. |
| 201 | + otel_provider_name: Optional telemetry provider name override. |
| 202 | + env_file_path: Path to ``.env`` file for settings. |
| 203 | + env_file_encoding: Encoding for ``.env`` file. |
| 204 | +
|
| 205 | + Examples: |
| 206 | + .. code-block:: python |
| 207 | +
|
| 208 | + from agent_framework_mistral import MistralEmbeddingClient |
| 209 | +
|
| 210 | + # Using environment variables |
| 211 | + # Set MISTRAL_API_KEY=your-key |
| 212 | + # Set MISTRAL_EMBEDDING_MODEL=mistral-embed |
| 213 | + client = MistralEmbeddingClient() |
| 214 | +
|
| 215 | + # Or passing parameters directly |
| 216 | + client = MistralEmbeddingClient( |
| 217 | + model="mistral-embed", |
| 218 | + api_key="your-api-key", |
| 219 | + ) |
| 220 | +
|
| 221 | + # Generate embeddings |
| 222 | + result = await client.get_embeddings(["Hello, world!"]) |
| 223 | + print(result[0].vector) |
| 224 | + """ |
| 225 | + |
| 226 | + OTEL_PROVIDER_NAME: ClassVar[str] = "mistralai" |
| 227 | + |
| 228 | + def __init__( |
| 229 | + self, |
| 230 | + *, |
| 231 | + model: str | None = None, |
| 232 | + api_key: str | SecretString | None = None, |
| 233 | + server_url: str | None = None, |
| 234 | + client: Mistral | None = None, |
| 235 | + otel_provider_name: str | None = None, |
| 236 | + additional_properties: dict[str, Any] | None = None, |
| 237 | + env_file_path: str | None = None, |
| 238 | + env_file_encoding: str | None = None, |
| 239 | + ) -> None: |
| 240 | + """Initialize a Mistral AI embedding client.""" |
| 241 | + super().__init__( |
| 242 | + model=model, |
| 243 | + api_key=api_key, |
| 244 | + server_url=server_url, |
| 245 | + client=client, |
| 246 | + additional_properties=additional_properties, |
| 247 | + otel_provider_name=otel_provider_name, |
| 248 | + env_file_path=env_file_path, |
| 249 | + env_file_encoding=env_file_encoding, |
| 250 | + ) |
0 commit comments