Spaces:
Sleeping
Sleeping
vhr1007
commited on
Commit
•
500c1ba
1
Parent(s):
b55082a
init
Browse files- .gitignore +2 -0
- Dockerfile +16 -0
- app.py +109 -0
- config.py +7 -0
- requiements.txt +8 -0
- services/__init__py +0 -0
- services/openai_service.py +48 -0
- services/qdrant_searcher.py +44 -0
- utils/__init__.py +0 -0
- utils/auth.py +44 -0
.gitignore
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
.env
|
2 |
+
|
Dockerfile
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Base image
|
2 |
+
FROM python:3.8-slim
|
3 |
+
|
4 |
+
# Set working directory
|
5 |
+
WORKDIR /app
|
6 |
+
|
7 |
+
# Copy requirements.txt and install dependencies
|
8 |
+
COPY requirements.txt .
|
9 |
+
RUN pip install --no-cache-dir -r requirements.txt
|
10 |
+
|
11 |
+
# Copy the rest of the application code
|
12 |
+
COPY . .
|
13 |
+
|
14 |
+
|
15 |
+
# Command to run the FastAPI application
|
16 |
+
CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
|
app.py
ADDED
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from huggingface_hub import login
|
2 |
+
from fastapi import FastAPI, Depends, HTTPException
|
3 |
+
import logging
|
4 |
+
from pydantic import BaseModel
|
5 |
+
from sentence_transformers import SentenceTransformer
|
6 |
+
from services.qdrant_searcher import QdrantSearcher
|
7 |
+
from services.openai_service import generate_rag_response
|
8 |
+
from utils.auth import token_required
|
9 |
+
from dotenv import load_dotenv
|
10 |
+
import os
|
11 |
+
|
12 |
+
load_dotenv() # Load environment variables from .env file
|
13 |
+
|
14 |
+
app = FastAPI()
|
15 |
+
|
16 |
+
os.environ["HF_HOME"] = "/tmp/huggingface_cache"
|
17 |
+
|
18 |
+
# Ensure the cache directory exists
|
19 |
+
cache_dir = os.environ["HF_HOME"]
|
20 |
+
if not os.path.exists(cache_dir):
|
21 |
+
os.makedirs(cache_dir)
|
22 |
+
|
23 |
+
# Setup logging
|
24 |
+
logging.basicConfig(level=logging.INFO)
|
25 |
+
# Load Hugging Face token from environment variable
|
26 |
+
huggingface_token = os.getenv('HUGGINGFACE_HUB_TOKEN')
|
27 |
+
|
28 |
+
|
29 |
+
if huggingface_token:
|
30 |
+
login(token=huggingface_token, add_to_git_credential=True)
|
31 |
+
else:
|
32 |
+
raise ValueError("Hugging Face token is not set. Please set the HUGGINGFACE_HUB_TOKEN environment variable.")
|
33 |
+
|
34 |
+
|
35 |
+
# Initialize the Qdrant searcher
|
36 |
+
qdrant_url = os.getenv('QDRANT_URL')
|
37 |
+
access_token = os.getenv('QDRANT_ACCESS_TOKEN')
|
38 |
+
encoder = SentenceTransformer('paraphrase-MiniLM-L6-v2', trust_remote_code=True) # Replace with your actual encoder
|
39 |
+
searcher = QdrantSearcher(encoder, qdrant_url, access_token)
|
40 |
+
|
41 |
+
# Request body models
|
42 |
+
class SearchDocumentsRequest(BaseModel):
|
43 |
+
query: str
|
44 |
+
limit: int = 3
|
45 |
+
|
46 |
+
class GenerateRAGRequest(BaseModel):
|
47 |
+
search_query: str
|
48 |
+
|
49 |
+
@app.post("/api/search-documents")
|
50 |
+
async def search_documents(
|
51 |
+
body: SearchDocumentsRequest,
|
52 |
+
credentials: tuple = Depends(token_required)
|
53 |
+
):
|
54 |
+
customer_id, user_id = credentials
|
55 |
+
|
56 |
+
# Check if customer_id or user_id is missing
|
57 |
+
if not customer_id or not user_id:
|
58 |
+
logging.error("Failed to extract customer_id or user_id from the JWT token.")
|
59 |
+
raise HTTPException(status_code=401, detail="Invalid token: missing customer_id or user_id")
|
60 |
+
|
61 |
+
logging.info("Received request to search documents")
|
62 |
+
try:
|
63 |
+
collection_name = "my_embeddings"
|
64 |
+
hits, error = searcher.search_documents(collection_name, body.query, user_id, body.limit)
|
65 |
+
|
66 |
+
if error:
|
67 |
+
logging.error(f"Search documents error: {error}")
|
68 |
+
raise HTTPException(status_code=500, detail=error)
|
69 |
+
|
70 |
+
return hits
|
71 |
+
except Exception as e:
|
72 |
+
logging.error(f"Unexpected error: {e}")
|
73 |
+
raise HTTPException(status_code=500, detail=str(e))
|
74 |
+
|
75 |
+
@app.post("/api/generate-rag-response")
|
76 |
+
async def generate_rag_response_api(
|
77 |
+
body: GenerateRAGRequest,
|
78 |
+
credentials: tuple = Depends(token_required)
|
79 |
+
):
|
80 |
+
customer_id, user_id = credentials
|
81 |
+
|
82 |
+
# Check if customer_id or user_id is missing
|
83 |
+
if not customer_id or not user_id:
|
84 |
+
logging.error("Failed to extract customer_id or user_id from the JWT token.")
|
85 |
+
raise HTTPException(status_code=401, detail="Invalid token: missing customer_id or user_id")
|
86 |
+
|
87 |
+
logging.info("Received request to generate RAG response")
|
88 |
+
try:
|
89 |
+
collection_name = "my_embeddings"
|
90 |
+
hits, error = searcher.search_documents(collection_name, body.search_query, user_id)
|
91 |
+
|
92 |
+
if error:
|
93 |
+
logging.error(f"Search documents error: {error}")
|
94 |
+
raise HTTPException(status_code=500, detail=error)
|
95 |
+
|
96 |
+
response, error = generate_rag_response(hits, body.search_query)
|
97 |
+
|
98 |
+
if error:
|
99 |
+
logging.error(f"Generate RAG response error: {error}")
|
100 |
+
raise HTTPException(status_code=500, detail=error)
|
101 |
+
|
102 |
+
return {"response": response}
|
103 |
+
except Exception as e:
|
104 |
+
logging.error(f"Unexpected error: {e}")
|
105 |
+
raise HTTPException(status_code=500, detail=str(e))
|
106 |
+
|
107 |
+
if __name__ == '__main__':
|
108 |
+
import uvicorn
|
109 |
+
uvicorn.run(app, host='0.0.0.0', port=8000)
|
config.py
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from dotenv import load_dotenv
|
2 |
+
import os
|
3 |
+
|
4 |
+
QDRANT_URL = os.getenv('QDRANT_URL')
|
5 |
+
QDRANT_ACCESS_TOKEN = os.getenv('QDRANT_ACCESS_TOKEN')
|
6 |
+
OPENAI_API_KEY = os.getenv('OPENAI_API_KEY')
|
7 |
+
JWKS_URL = os.getenv('JWKS_URL')
|
requiements.txt
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
fastapi==0.78.0
|
2 |
+
uvicorn==0.17.6
|
3 |
+
pandas==1.3.5
|
4 |
+
qdrant-client==0.9.2
|
5 |
+
sentence-transformers==2.2.2
|
6 |
+
openai==0.27.0
|
7 |
+
PyJWT==2.6.0
|
8 |
+
python-dotenv==0.19.2
|
services/__init__py
ADDED
File without changes
|
services/openai_service.py
ADDED
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import logging
|
2 |
+
import os
|
3 |
+
from openai import OpenAI
|
4 |
+
from openai import OpenAIError, RateLimitError
|
5 |
+
from config import OPENAI_API_KEY
|
6 |
+
|
7 |
+
# Initialize the OpenAI client with the API key from the environment variable
|
8 |
+
#api_key = os.getenv('OPENAI_API_KEY')
|
9 |
+
|
10 |
+
client = OpenAI(api_key=OPENAI_API_KEY)
|
11 |
+
|
12 |
+
def generate_rag_response(json_output, user_query):
|
13 |
+
logging.info("Generating RAG response")
|
14 |
+
|
15 |
+
# Extract text from the JSON output
|
16 |
+
context_texts = [hit['chunk_text'] for hit in json_output]
|
17 |
+
|
18 |
+
# Create the context for the prompt
|
19 |
+
context = "\n".join(context_texts)
|
20 |
+
prompt = f"Based on the given context, answer the user query: {user_query}\nContext:\n{context}"
|
21 |
+
|
22 |
+
main_prompt = [
|
23 |
+
{"role": "system", "content": "You are a helpful assistant."},
|
24 |
+
{"role": "user", "content": prompt}
|
25 |
+
]
|
26 |
+
|
27 |
+
try:
|
28 |
+
# Create a chat completion request
|
29 |
+
chat_completion = client.chat.completions.create(
|
30 |
+
messages=main_prompt,
|
31 |
+
model="gpt-4o-mini", # Use the gpt-4o-mini model
|
32 |
+
timeout=10
|
33 |
+
)
|
34 |
+
|
35 |
+
# Log the response from the model
|
36 |
+
logging.info("RAG response generation completed")
|
37 |
+
logging.info(f"RAG response: {chat_completion.choices[0].message.content}")
|
38 |
+
return chat_completion.choices[0].message.content, None
|
39 |
+
|
40 |
+
except RateLimitError as e:
|
41 |
+
logging.error(f"Rate limit exceeded: {e}")
|
42 |
+
return None, "Rate limit exceeded. Please try again later."
|
43 |
+
except OpenAIError as e:
|
44 |
+
logging.error(f"OpenAI API error: {e}")
|
45 |
+
return None, f"An error occurred: {str(e)}"
|
46 |
+
except Exception as e:
|
47 |
+
logging.error(f"Unexpected error: {e}")
|
48 |
+
return None, str(e)
|
services/qdrant_searcher.py
ADDED
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import logging
|
2 |
+
from qdrant_client import QdrantClient
|
3 |
+
from qdrant_client.http.models import Filter, FieldCondition
|
4 |
+
|
5 |
+
class QdrantSearcher:
|
6 |
+
def __init__(self, encoder, qdrant_url, access_token):
|
7 |
+
self.encoder = encoder
|
8 |
+
self.client = QdrantClient(url=qdrant_url, api_key=access_token)
|
9 |
+
|
10 |
+
def search_documents(self, collection_name, query, user_id, limit=3):
|
11 |
+
logging.info("Starting document search")
|
12 |
+
query_vector = self.encoder.encode(query).tolist()
|
13 |
+
query_filter = Filter(must=[FieldCondition(key="user_id", match={"value": user_id})])
|
14 |
+
|
15 |
+
try:
|
16 |
+
hits = self.client.search(
|
17 |
+
collection_name=collection_name,
|
18 |
+
query_vector=query_vector,
|
19 |
+
limit=limit,
|
20 |
+
query_filter=query_filter
|
21 |
+
)
|
22 |
+
except Exception as e:
|
23 |
+
logging.error(f"Error during Qdrant search: {e}")
|
24 |
+
return None, str(e)
|
25 |
+
|
26 |
+
if not hits:
|
27 |
+
logging.info("No documents found for the given query")
|
28 |
+
return None, "No documents found for the given query."
|
29 |
+
|
30 |
+
hits_list = []
|
31 |
+
for hit in hits:
|
32 |
+
hit_info = {
|
33 |
+
"id": hit.id,
|
34 |
+
"score": hit.score,
|
35 |
+
"file_id": hit.payload.get('file_id'),
|
36 |
+
"organization_id": hit.payload.get('organization_id'),
|
37 |
+
"chunk_index": hit.payload.get('chunk_index'),
|
38 |
+
"chunk_text": hit.payload.get('chunk_text'),
|
39 |
+
"s3_bucket_key": hit.payload.get('s3_bucket_key')
|
40 |
+
}
|
41 |
+
hits_list.append(hit_info)
|
42 |
+
|
43 |
+
logging.info(f"Document search completed with {len(hits_list)} hits")
|
44 |
+
return hits_list, None
|
utils/__init__.py
ADDED
File without changes
|
utils/auth.py
ADDED
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import logging
|
2 |
+
from fastapi import Depends, HTTPException
|
3 |
+
from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
|
4 |
+
import jwt
|
5 |
+
from jwt import PyJWKClient
|
6 |
+
from config import JWKS_URL
|
7 |
+
|
8 |
+
security = HTTPBearer()
|
9 |
+
|
10 |
+
def get_public_key(token: str):
|
11 |
+
try:
|
12 |
+
jwks_client = PyJWKClient(JWKS_URL)
|
13 |
+
signing_key = jwks_client.get_signing_key_from_jwt(token)
|
14 |
+
return signing_key.key
|
15 |
+
except Exception as e:
|
16 |
+
logging.error(f"Error fetching public key: {e}")
|
17 |
+
raise
|
18 |
+
|
19 |
+
def token_required(credentials: HTTPAuthorizationCredentials = Depends(security)):
|
20 |
+
token = credentials.credentials
|
21 |
+
try:
|
22 |
+
public_key = get_public_key(token)
|
23 |
+
decoded = jwt.decode(
|
24 |
+
token,
|
25 |
+
public_key,
|
26 |
+
algorithms=['RS256'],
|
27 |
+
issuer="https://assuring-lobster-64.clerk.accounts.dev"
|
28 |
+
)
|
29 |
+
customer_id = decoded.get('org_id')
|
30 |
+
user_id = decoded.get('sub')
|
31 |
+
logging.info(f"Customer/Org ID: {customer_id}, User ID: {user_id}")
|
32 |
+
if not customer_id:
|
33 |
+
logging.error("Customer ID is missing in the token!")
|
34 |
+
raise HTTPException(status_code=401, detail="Customer ID is missing in the token!")
|
35 |
+
return customer_id, user_id
|
36 |
+
except jwt.ExpiredSignatureError:
|
37 |
+
logging.error("Token has expired")
|
38 |
+
raise HTTPException(status_code=401, detail="Token has expired")
|
39 |
+
except jwt.InvalidTokenError as e:
|
40 |
+
logging.error(f"Invalid token: {e}")
|
41 |
+
raise HTTPException(status_code=401, detail="Invalid token")
|
42 |
+
except Exception as e:
|
43 |
+
logging.error(f"Error decoding token: {e}")
|
44 |
+
raise HTTPException(status_code=401, detail=str(e))
|