Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
import os
|
2 |
import time
|
3 |
-
from fastapi import FastAPI,Request
|
4 |
from fastapi.responses import HTMLResponse
|
5 |
from fastapi.staticfiles import StaticFiles
|
6 |
from llama_index.core import StorageContext, load_index_from_storage, VectorStoreIndex, SimpleDirectoryReader, ChatPromptTemplate, Settings
|
@@ -15,12 +15,12 @@ from fastapi.templating import Jinja2Templates
|
|
15 |
from huggingface_hub import InferenceClient
|
16 |
import json
|
17 |
import re
|
18 |
-
|
19 |
-
|
20 |
|
21 |
# Define Pydantic model for incoming request body
|
22 |
class MessageRequest(BaseModel):
|
23 |
message: str
|
|
|
24 |
repo_id = "meta-llama/Meta-Llama-3-8B-Instruct"
|
25 |
llm_client = InferenceClient(
|
26 |
model=repo_id,
|
@@ -29,10 +29,8 @@ llm_client = InferenceClient(
|
|
29 |
|
30 |
os.environ["HF_TOKEN"] = os.getenv("HF_TOKEN")
|
31 |
|
32 |
-
|
33 |
app = FastAPI()
|
34 |
|
35 |
-
|
36 |
@app.middleware("http")
|
37 |
async def add_security_headers(request: Request, call_next):
|
38 |
response = await call_next(request)
|
@@ -40,7 +38,6 @@ async def add_security_headers(request: Request, call_next):
|
|
40 |
response.headers["X-Frame-Options"] = "ALLOWALL"
|
41 |
return response
|
42 |
|
43 |
-
|
44 |
# Allow CORS requests from any domain
|
45 |
app.add_middleware(
|
46 |
CORSMiddleware,
|
@@ -50,17 +47,14 @@ app.add_middleware(
|
|
50 |
allow_headers=["*"],
|
51 |
)
|
52 |
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
@app.get("/favicon.ico")
|
57 |
async def favicon():
|
58 |
return HTMLResponse("") # or serve a real favicon if you have one
|
59 |
|
60 |
-
|
61 |
app.mount("/static", StaticFiles(directory="static"), name="static")
|
62 |
|
63 |
templates = Jinja2Templates(directory="static")
|
|
|
64 |
# Configure Llama index settings
|
65 |
Settings.llm = HuggingFaceInferenceAPI(
|
66 |
model_name="meta-llama/Meta-Llama-3-8B-Instruct",
|
@@ -82,6 +76,7 @@ os.makedirs(PDF_DIRECTORY, exist_ok=True)
|
|
82 |
os.makedirs(PERSIST_DIR, exist_ok=True)
|
83 |
chat_history = []
|
84 |
current_chat_history = []
|
|
|
85 |
def data_ingestion_from_directory():
|
86 |
documents = SimpleDirectoryReader(PDF_DIRECTORY).load_data()
|
87 |
storage_context = StorageContext.from_defaults()
|
@@ -92,6 +87,7 @@ def initialize():
|
|
92 |
start_time = time.time()
|
93 |
data_ingestion_from_directory() # Process PDF ingestion at startup
|
94 |
print(f"Data ingestion time: {time.time() - start_time} seconds")
|
|
|
95 |
def split_name(full_name):
|
96 |
# Split the name by spaces
|
97 |
words = full_name.strip().split()
|
@@ -111,7 +107,6 @@ def split_name(full_name):
|
|
111 |
|
112 |
initialize() # Run initialization tasks
|
113 |
|
114 |
-
|
115 |
def handle_query(query):
|
116 |
chat_text_qa_msgs = [
|
117 |
(
|
@@ -133,19 +128,23 @@ def handle_query(query):
|
|
133 |
if past_query.strip():
|
134 |
context_str += f"User asked: '{past_query}'\nBot answered: '{response}'\n"
|
135 |
|
136 |
-
|
137 |
query_engine = index.as_query_engine(text_qa_template=text_qa_template, context_str=context_str)
|
138 |
answer = query_engine.query(query)
|
139 |
|
140 |
if hasattr(answer, 'response'):
|
141 |
-
response=answer.response
|
142 |
elif isinstance(answer, dict) and 'response' in answer:
|
143 |
-
response =answer['response']
|
144 |
else:
|
145 |
-
response ="Sorry, I couldn't find an answer."
|
146 |
current_chat_history.append((query, response))
|
147 |
return response
|
148 |
|
|
|
|
|
|
|
|
|
|
|
149 |
@app.post("/hist/")
|
150 |
async def save_chat_history(history: dict):
|
151 |
# Check if 'userId' is present in the incoming dictionary
|
@@ -160,10 +159,23 @@ async def save_chat_history(history: dict):
|
|
160 |
hist = ''.join([f"'{entry['sender']}: {entry['message']}'\n" for entry in history['history']])
|
161 |
hist = "You are a Redfernstech summarize model. Your aim is to use this conversation to identify user interests solely based on that conversation: " + hist
|
162 |
print(hist)
|
|
|
163 |
# Get the summarized result from the client model
|
164 |
-
result = hist
|
|
|
165 |
return {"summary": result, "message": "Chat history saved"}
|
166 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
167 |
|
168 |
@app.post("/chat/")
|
169 |
async def chat(request: MessageRequest):
|
@@ -177,7 +189,7 @@ async def chat(request: MessageRequest):
|
|
177 |
}
|
178 |
chat_history.append(message_data)
|
179 |
return {"response": response}
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
|
|
1 |
import os
|
2 |
import time
|
3 |
+
from fastapi import FastAPI, Request
|
4 |
from fastapi.responses import HTMLResponse
|
5 |
from fastapi.staticfiles import StaticFiles
|
6 |
from llama_index.core import StorageContext, load_index_from_storage, VectorStoreIndex, SimpleDirectoryReader, ChatPromptTemplate, Settings
|
|
|
15 |
from huggingface_hub import InferenceClient
|
16 |
import json
|
17 |
import re
|
18 |
+
from gradio_client import Client
|
|
|
19 |
|
20 |
# Define Pydantic model for incoming request body
|
21 |
class MessageRequest(BaseModel):
|
22 |
message: str
|
23 |
+
|
24 |
repo_id = "meta-llama/Meta-Llama-3-8B-Instruct"
|
25 |
llm_client = InferenceClient(
|
26 |
model=repo_id,
|
|
|
29 |
|
30 |
os.environ["HF_TOKEN"] = os.getenv("HF_TOKEN")
|
31 |
|
|
|
32 |
app = FastAPI()
|
33 |
|
|
|
34 |
@app.middleware("http")
|
35 |
async def add_security_headers(request: Request, call_next):
|
36 |
response = await call_next(request)
|
|
|
38 |
response.headers["X-Frame-Options"] = "ALLOWALL"
|
39 |
return response
|
40 |
|
|
|
41 |
# Allow CORS requests from any domain
|
42 |
app.add_middleware(
|
43 |
CORSMiddleware,
|
|
|
47 |
allow_headers=["*"],
|
48 |
)
|
49 |
|
|
|
|
|
|
|
50 |
@app.get("/favicon.ico")
|
51 |
async def favicon():
|
52 |
return HTMLResponse("") # or serve a real favicon if you have one
|
53 |
|
|
|
54 |
app.mount("/static", StaticFiles(directory="static"), name="static")
|
55 |
|
56 |
templates = Jinja2Templates(directory="static")
|
57 |
+
|
58 |
# Configure Llama index settings
|
59 |
Settings.llm = HuggingFaceInferenceAPI(
|
60 |
model_name="meta-llama/Meta-Llama-3-8B-Instruct",
|
|
|
76 |
os.makedirs(PERSIST_DIR, exist_ok=True)
|
77 |
chat_history = []
|
78 |
current_chat_history = []
|
79 |
+
|
80 |
def data_ingestion_from_directory():
|
81 |
documents = SimpleDirectoryReader(PDF_DIRECTORY).load_data()
|
82 |
storage_context = StorageContext.from_defaults()
|
|
|
87 |
start_time = time.time()
|
88 |
data_ingestion_from_directory() # Process PDF ingestion at startup
|
89 |
print(f"Data ingestion time: {time.time() - start_time} seconds")
|
90 |
+
|
91 |
def split_name(full_name):
|
92 |
# Split the name by spaces
|
93 |
words = full_name.strip().split()
|
|
|
107 |
|
108 |
initialize() # Run initialization tasks
|
109 |
|
|
|
110 |
def handle_query(query):
|
111 |
chat_text_qa_msgs = [
|
112 |
(
|
|
|
128 |
if past_query.strip():
|
129 |
context_str += f"User asked: '{past_query}'\nBot answered: '{response}'\n"
|
130 |
|
|
|
131 |
query_engine = index.as_query_engine(text_qa_template=text_qa_template, context_str=context_str)
|
132 |
answer = query_engine.query(query)
|
133 |
|
134 |
if hasattr(answer, 'response'):
|
135 |
+
response = answer.response
|
136 |
elif isinstance(answer, dict) and 'response' in answer:
|
137 |
+
response = answer['response']
|
138 |
else:
|
139 |
+
response = "Sorry, I couldn't find an answer."
|
140 |
current_chat_history.append((query, response))
|
141 |
return response
|
142 |
|
143 |
+
@app.get("/ch/{id}", response_class=HTMLResponse)
|
144 |
+
async def load_chat(request: Request, id: str):
|
145 |
+
return templates.TemplateResponse("index.html", {"request": request, "user_id": id})
|
146 |
+
|
147 |
+
# Route to save chat history
|
148 |
@app.post("/hist/")
|
149 |
async def save_chat_history(history: dict):
|
150 |
# Check if 'userId' is present in the incoming dictionary
|
|
|
159 |
hist = ''.join([f"'{entry['sender']}: {entry['message']}'\n" for entry in history['history']])
|
160 |
hist = "You are a Redfernstech summarize model. Your aim is to use this conversation to identify user interests solely based on that conversation: " + hist
|
161 |
print(hist)
|
162 |
+
|
163 |
# Get the summarized result from the client model
|
164 |
+
result = hist
|
165 |
+
|
166 |
return {"summary": result, "message": "Chat history saved"}
|
167 |
|
168 |
+
@app.post("/webhook")
|
169 |
+
async def receive_form_data(request: Request):
|
170 |
+
form_data = await request.json()
|
171 |
+
# Generate a unique ID (for tracking user)
|
172 |
+
unique_id = str(uuid.uuid4())
|
173 |
+
|
174 |
+
# Here you can do something with form_data like saving it to a database
|
175 |
+
print("Received form data:", form_data)
|
176 |
+
|
177 |
+
# Send back the unique id to the frontend
|
178 |
+
return JSONResponse({"id": unique_id})
|
179 |
|
180 |
@app.post("/chat/")
|
181 |
async def chat(request: MessageRequest):
|
|
|
189 |
}
|
190 |
chat_history.append(message_data)
|
191 |
return {"response": response}
|
192 |
+
|
193 |
+
@app.get("/")
|
194 |
+
def read_root():
|
195 |
+
return {"message": "Welcome to the API"}
|