File size: 1,935 Bytes
ed4d993
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
from typing import List, Optional

import aiohttp
import requests
from langchain_core.callbacks import (
    AsyncCallbackManagerForRetrieverRun,
    CallbackManagerForRetrieverRun,
)
from langchain_core.documents import Document
from langchain_core.retrievers import BaseRetriever


class RemoteLangChainRetriever(BaseRetriever):
    """`LangChain API` retriever."""

    url: str
    """URL of the remote LangChain API."""
    headers: Optional[dict] = None
    """Headers to use for the request."""
    input_key: str = "message"
    """Key to use for the input in the request."""
    response_key: str = "response"
    """Key to use for the response in the request."""
    page_content_key: str = "page_content"
    """Key to use for the page content in the response."""
    metadata_key: str = "metadata"
    """Key to use for the metadata in the response."""

    def _get_relevant_documents(
        self, query: str, *, run_manager: CallbackManagerForRetrieverRun
    ) -> List[Document]:
        response = requests.post(
            self.url, json={self.input_key: query}, headers=self.headers
        )
        result = response.json()
        return [
            Document(
                page_content=r[self.page_content_key], metadata=r[self.metadata_key]
            )
            for r in result[self.response_key]
        ]

    async def _aget_relevant_documents(
        self, query: str, *, run_manager: AsyncCallbackManagerForRetrieverRun
    ) -> List[Document]:
        async with aiohttp.ClientSession() as session:
            async with session.request(
                "POST", self.url, headers=self.headers, json={self.input_key: query}
            ) as response:
                result = await response.json()
        return [
            Document(
                page_content=r[self.page_content_key], metadata=r[self.metadata_key]
            )
            for r in result[self.response_key]
        ]