File size: 4,487 Bytes
d7d0d8e
 
7fc5d88
 
 
 
d7d0d8e
7d5df78
7fc5d88
 
 
 
 
 
 
 
d7d0d8e
 
7fc5d88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d7d0d8e
7fc5d88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d7d0d8e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
55174eb
d7d0d8e
 
 
 
 
 
 
 
 
 
7fc5d88
d7d0d8e
 
 
 
 
 
 
 
 
 
 
f337592
7fc5d88
d7d0d8e
 
 
 
 
 
 
55174eb
d7d0d8e
 
 
 
 
 
 
 
 
 
7fc5d88
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
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.docstore.document import Document
from langchain.vectorstores import MongoDBAtlasVectorSearch
from pymongo import MongoClient
from motor.motor_asyncio import AsyncIOMotorClient
import os,pprint


completion_base = os.environ.get("completion_base")
openai_api_key = os.environ.get("openai_api_key")
mongoDB = os.environ.get("MONGO_DB")
template = """### Given the following context 
### Context
{context}
### Use it to explain the question: {question}
 """


async def fetch_data(question, context):
    url = completion_base

    payload = json.dumps(
        {
            "messages": [
                {
                    "role": "system",
                    "content": "### You provide explanations based on the provided context",
                },
                {
                    "role": "user",
                    "content": template.format(context=context, question=question),
                },
            ],
            "model": "gpt-3.5-turbo",
            "temperature": 1,
            "presence_penalty": 0,
            "top_p": 0.95,
            "frequency_penalty": 0,
            "stream": False,
        }
    )

    headers = {
        "Content-Type": "application/json",
        "Authorization": f"Bearer {openai_api_key}",
    }

    async with aiohttp.ClientSession() as session:
        async with session.post(url, headers=headers, data=payload) as response:
            response = await response.json()
            return response["choices"][0]["message"]["content"]


async def delete_documents(task_id):
    client = AsyncIOMotorClient(mongoDB)
    db = client["transcriptions"]
    collection = db["videos"]

    result = await collection.delete_many({"task_id": task_id})
    print(f"Deleted {result.deleted_count} document(s)")


# mongo_client = MongoClient(
#    mongoDB
# )
# model_name = "BAAI/bge-base-en"
# collection = mongo_client["transcriptions"]["videos"]
# embeddings = HuggingFaceEmbeddings(model_name=model_name)
# index_name = "test_embeddings"
# vectorstore = MongoDBAtlasVectorSearch(collection, embeddings, index_name=index_name)


def generateChunks(chunks, task_id, n=100):
    combined = [chunks[i : i + n] for i in range(0, len(chunks), n)]
    result = []
    for chunk in combined:
        data = {"text": ""}
        for item in chunk:
            if chunk.index(item) == 0:
                data["start"] = item["start"]
            if chunk.index(item) == len(chunk) - 1:
                data["end"] = item["end"]
            data["text"] += " " + item["text"]

        temp = Document(
            page_content=data["text"],
            metadata={"start": data["start"], "end": data["end"], "task_id": task_id},
        )
        result.append(temp)
    return result


def search(query: str, task_id: str):
    mongo_client = MongoClient(mongoDB)
    model_name = "BAAI/bge-base-en"
    collection = mongo_client["transcriptions"]["videos"]
    embeddings = HuggingFaceEmbeddings(model_name=model_name)
    index_name = "test_embedding"
    k = 5
    vectorstore = MongoDBAtlasVectorSearch(
        collection,
        embedding=embeddings,
        index_name="test_embedding",
    )

    data = vectorstore.similarity_search(
        query=query,
        pre_filter={"text": {"path": "task_id", "query": task_id}},
        search_kwargs={
            "k": k,  # overrequest k during search
            "pre_filter": {"path": "task_id", "equals": task_id},
            "post_filter_pipeline": [{"$limit": k}],  # limit results to top k
        },
    )
    # data =[d.dict() for d in data]
    # print(data[0].metadata.exclude({'_id','embedding'}))
    # pprint.pprint(data[0].metadata)
    return [{"text": d.page_content,'start':d.metadata['start'],"end":d.metadata['end']} for d in data]
    # agent =vectorstore.as_retriever(

    # )
    # return agent.get_relevant_documents


def encode(temp: list[Document]):
    mongo_client = MongoClient(mongoDB)
    model_name = "BAAI/bge-base-en"
    collection = mongo_client["transcriptions"]["videos"]
    embeddings = HuggingFaceEmbeddings(model_name=model_name)
    index_name = "test_embedding"
    vectorstore = MongoDBAtlasVectorSearch(
        collection, embeddings, index_name=index_name
    )
    vectorstore.from_documents(
        temp, embedding=embeddings, collection=collection, index_name=index_name
    )
    # return  embeddings.embed_documents(texts = [d.page_content for d in temp])