Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,8 +3,8 @@
|
|
| 3 |
# =========================
|
| 4 |
import os
|
| 5 |
import tempfile
|
| 6 |
-
|
| 7 |
import gradio as gr
|
|
|
|
| 8 |
from groq import Groq
|
| 9 |
from duckduckgo_search import DDGS
|
| 10 |
|
|
@@ -17,7 +17,7 @@ from langchain_community.vectorstores import Chroma
|
|
| 17 |
# =========================
|
| 18 |
# CONFIG
|
| 19 |
# =========================
|
| 20 |
-
GROQ_API_KEY = os.getenv("GROQ_API_KEY") #
|
| 21 |
|
| 22 |
client = Groq(api_key=GROQ_API_KEY)
|
| 23 |
|
|
@@ -57,18 +57,16 @@ def web_search(query):
|
|
| 57 |
|
| 58 |
|
| 59 |
# =========================
|
| 60 |
-
# PROCESS PDF (
|
| 61 |
# =========================
|
| 62 |
def process_pdf(file):
|
| 63 |
|
| 64 |
global vectorstore, retriever
|
| 65 |
|
| 66 |
-
#
|
| 67 |
-
|
| 68 |
-
tmp.write(file.read())
|
| 69 |
-
tmp_path = tmp.name
|
| 70 |
|
| 71 |
-
loader = PyPDFLoader(
|
| 72 |
documents = loader.load()
|
| 73 |
|
| 74 |
splitter = RecursiveCharacterTextSplitter(
|
|
@@ -89,11 +87,11 @@ def process_pdf(file):
|
|
| 89 |
|
| 90 |
retriever = vectorstore.as_retriever(search_kwargs={"k": 4})
|
| 91 |
|
| 92 |
-
return "✅ PDF
|
| 93 |
|
| 94 |
|
| 95 |
# =========================
|
| 96 |
-
#
|
| 97 |
# =========================
|
| 98 |
def ask_rag(query):
|
| 99 |
|
|
@@ -105,10 +103,10 @@ def ask_rag(query):
|
|
| 105 |
docs = retriever.invoke(query)
|
| 106 |
pdf_context = "\n\n".join([d.page_content for d in docs])
|
| 107 |
|
| 108 |
-
#
|
| 109 |
if len(pdf_context.strip()) < 50:
|
| 110 |
web_context = web_search(query)
|
| 111 |
-
context = pdf_context + "\n\nWEB
|
| 112 |
else:
|
| 113 |
context = pdf_context
|
| 114 |
|
|
@@ -123,7 +121,7 @@ def ask_rag(query):
|
|
| 123 |
|
| 124 |
|
| 125 |
# =========================
|
| 126 |
-
# CHAT FUNCTION (
|
| 127 |
# =========================
|
| 128 |
def chat(user_message, history):
|
| 129 |
|
|
@@ -132,22 +130,23 @@ def chat(user_message, history):
|
|
| 132 |
if history is None:
|
| 133 |
history = []
|
| 134 |
|
| 135 |
-
history.append(
|
|
|
|
| 136 |
|
| 137 |
return history, history
|
| 138 |
|
| 139 |
|
| 140 |
# =========================
|
| 141 |
-
#
|
| 142 |
# =========================
|
| 143 |
with gr.Blocks() as app:
|
| 144 |
|
| 145 |
-
gr.Markdown("# 🧠 Hybrid RAG Chatbot (PDF + Web
|
| 146 |
|
| 147 |
file = gr.File(label="Upload PDF")
|
| 148 |
status = gr.Textbox(label="Status")
|
| 149 |
|
| 150 |
-
chatbot = gr.Chatbot()
|
| 151 |
msg = gr.Textbox(placeholder="Ask your question...")
|
| 152 |
state = gr.State([])
|
| 153 |
|
|
|
|
| 3 |
# =========================
|
| 4 |
import os
|
| 5 |
import tempfile
|
|
|
|
| 6 |
import gradio as gr
|
| 7 |
+
|
| 8 |
from groq import Groq
|
| 9 |
from duckduckgo_search import DDGS
|
| 10 |
|
|
|
|
| 17 |
# =========================
|
| 18 |
# CONFIG
|
| 19 |
# =========================
|
| 20 |
+
GROQ_API_KEY = os.getenv("GROQ_API_KEY") # HF SECRET KEY
|
| 21 |
|
| 22 |
client = Groq(api_key=GROQ_API_KEY)
|
| 23 |
|
|
|
|
| 57 |
|
| 58 |
|
| 59 |
# =========================
|
| 60 |
+
# PROCESS PDF (HF SAFE)
|
| 61 |
# =========================
|
| 62 |
def process_pdf(file):
|
| 63 |
|
| 64 |
global vectorstore, retriever
|
| 65 |
|
| 66 |
+
# safe file handling
|
| 67 |
+
file_path = file.name
|
|
|
|
|
|
|
| 68 |
|
| 69 |
+
loader = PyPDFLoader(file_path)
|
| 70 |
documents = loader.load()
|
| 71 |
|
| 72 |
splitter = RecursiveCharacterTextSplitter(
|
|
|
|
| 87 |
|
| 88 |
retriever = vectorstore.as_retriever(search_kwargs={"k": 4})
|
| 89 |
|
| 90 |
+
return "✅ PDF successfully processed"
|
| 91 |
|
| 92 |
|
| 93 |
# =========================
|
| 94 |
+
# RAG ENGINE (HYBRID)
|
| 95 |
# =========================
|
| 96 |
def ask_rag(query):
|
| 97 |
|
|
|
|
| 103 |
docs = retriever.invoke(query)
|
| 104 |
pdf_context = "\n\n".join([d.page_content for d in docs])
|
| 105 |
|
| 106 |
+
# hybrid fallback
|
| 107 |
if len(pdf_context.strip()) < 50:
|
| 108 |
web_context = web_search(query)
|
| 109 |
+
context = pdf_context + "\n\nWEB:\n" + web_context
|
| 110 |
else:
|
| 111 |
context = pdf_context
|
| 112 |
|
|
|
|
| 121 |
|
| 122 |
|
| 123 |
# =========================
|
| 124 |
+
# CHAT FUNCTION (FIXED FORMAT)
|
| 125 |
# =========================
|
| 126 |
def chat(user_message, history):
|
| 127 |
|
|
|
|
| 130 |
if history is None:
|
| 131 |
history = []
|
| 132 |
|
| 133 |
+
history.append({"role": "user", "content": user_message})
|
| 134 |
+
history.append({"role": "assistant", "content": response})
|
| 135 |
|
| 136 |
return history, history
|
| 137 |
|
| 138 |
|
| 139 |
# =========================
|
| 140 |
+
# UI (HUGGING FACE SAFE)
|
| 141 |
# =========================
|
| 142 |
with gr.Blocks() as app:
|
| 143 |
|
| 144 |
+
gr.Markdown("# 🧠 Hybrid RAG Chatbot (PDF + Web)")
|
| 145 |
|
| 146 |
file = gr.File(label="Upload PDF")
|
| 147 |
status = gr.Textbox(label="Status")
|
| 148 |
|
| 149 |
+
chatbot = gr.Chatbot(type="messages") # IMPORTANT FIX
|
| 150 |
msg = gr.Textbox(placeholder="Ask your question...")
|
| 151 |
state = gr.State([])
|
| 152 |
|