Get lucky

App/Embedding/utils/Initialize.py

@@ -1,74 +1,24 @@
 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
+from langchain.vectorstores import Pinecone
+import os
+import pinecone
 
+index_name = "movie-recommender-fast"
+model_name = "thenlper/gte-base"
+embeddings = HuggingFaceEmbeddings(model_name=model_name)
 
-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}
- """
+# get api key from app.pinecone.io
+PINECONE_API_KEY = (
+    os.environ.get("PINECONE_API_KEY") or "0712a5e4-bcf3-4152-a726-27ee3a2676bb"
+)
+# find your environment next to the api key in pinecone console
+PINECONE_ENV = os.environ.get("PINECONE_ENVIRONMENT") or "us-west4-gcp-free"
 
+pinecone.init(api_key=PINECONE_API_KEY, environment=PINECONE_ENV)
 
-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)
+docsearch = Pinecone.from_existing_index(index_name, embeddings)
 
 
 def generateChunks(chunks, task_id, n=100):
@@ -103,7 +53,6 @@ def search(query: str, task_id: str):
         embedding=embeddings,
         index_name="test_embedding",
     )
-
     data = vectorstore.similarity_search(
         query=query,
         pre_filter={"text": {"path": "task_id", "query": task_id}},
@@ -116,7 +65,10 @@ def search(query: str, task_id: str):
     # 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]
+    return [
+        {"text": d.page_content, "start": d.metadata["start"], "end": d.metadata["end"]}
+        for d in data
+    ]
     # agent =vectorstore.as_retriever(
 
     # )

App/app.py

@@ -47,7 +47,7 @@ def authjwt_exception_handler(request: Request, exc: AuthJWTException):
 
 @app.on_event("startup")
 async def startup_event():
-    # await bot.start(bot_token="6183919505:AAEhHFt4mI18bQeAf2Lj7AePXFRPVLrOFM8")
+    await bot.start(bot_token="6183919505:AAEhHFt4mI18bQeAf2Lj7AePXFRPVLrOFM8")
     # await upload_bot.start()
     # await models.create_all()
     # models.metadata.create_all()