Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,89 +1,46 @@
|
|
1 |
import os
|
2 |
-
from langchain_core.prompts import PromptTemplate
|
3 |
-
from langchain_community.output_parsers.rail_parser import GuardrailsOutputParser
|
4 |
-
from langchain_community.document_loaders import PyPDFLoader
|
5 |
-
import google.generativeai as genai
|
6 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
|
8 |
-
# Function for initialization
|
9 |
-
def initialize(pdf_file, question):
|
10 |
-
try:
|
11 |
-
# Access the uploaded file information from Gradio
|
12 |
-
file_info = pdf_file
|
13 |
-
|
14 |
-
# Check if a file was uploaded
|
15 |
-
if file_info is not None:
|
16 |
-
# Construct potential file path based on temporary directory and filename
|
17 |
-
file_path = os.path.join("/tmp", file_info.name) # Adjust temporary directory if needed
|
18 |
-
if os.path.exists(file_path):
|
19 |
-
# Process the PDF
|
20 |
-
pdf_loader = PyPDFLoader(file_path)
|
21 |
-
pages = pdf_loader.load_and_split()
|
22 |
-
processed_context = "\n".join(str(page.page_content) for page in pages[:30]) # Limit to first 30 pages
|
23 |
-
|
24 |
-
# Configure Google Generative AI (replace with your API key)
|
25 |
-
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
|
26 |
-
|
27 |
-
# Prompt template for formatting context and question
|
28 |
-
prompt_template = """Answer the question as precise as possible using the provided context. If the answer is not contained in the context, say "answer not available in context"
|
29 |
-
|
30 |
-
Context:
|
31 |
-
{context}
|
32 |
-
|
33 |
-
Question:
|
34 |
-
{question}
|
35 |
-
|
36 |
-
Answer:
|
37 |
-
"""
|
38 |
-
|
39 |
-
prompt = PromptTemplate(template=prompt_template, input_variables=["context", "question"])
|
40 |
-
|
41 |
-
# Load the GeminiPro model
|
42 |
-
model = genai.GenerativeModel('gemini-pro')
|
43 |
-
|
44 |
-
# Option 1: Using GeminiPro's Text Generation (if applicable)
|
45 |
-
|
46 |
-
# Check if the model has a 'generate' method (or similar) - adjust based on actual method
|
47 |
-
if hasattr(model, 'generate'):
|
48 |
-
# Process context and question (already done)
|
49 |
-
|
50 |
-
# Generate answer using GeminiPro's generate method
|
51 |
-
generated_answer = model.generate(prompt=prompt) # Replace with the appropriate method
|
52 |
-
|
53 |
-
# Extract the answer (parse the output from 'generate')
|
54 |
-
# ... (implementation depends on the model's output format)
|
55 |
-
|
56 |
-
return generated_answer
|
57 |
-
|
58 |
-
# Option 2: Alternative LLM Integration (if GeminiPro methods not suitable)
|
59 |
-
|
60 |
-
# Replace this section with code using an alternative library/framework
|
61 |
-
# for question answering (e.g., transformers, haystack)
|
62 |
-
# Ensure the code integrates with your chosen LLM and handles context processing,
|
63 |
-
# question answering, and answer extraction.
|
64 |
-
|
65 |
-
# Example placeholder (replace with your actual implementation):
|
66 |
-
# return "Alternative LLM integration not yet implemented."
|
67 |
-
|
68 |
-
else:
|
69 |
-
return "Error: The uploaded file could not be found."
|
70 |
-
else:
|
71 |
-
return "Error: No PDF file was uploaded."
|
72 |
-
|
73 |
-
except Exception as e:
|
74 |
-
return f"An error occurred: {e}" # Generic error handling
|
75 |
-
|
76 |
-
# Create a Gradio interface
|
77 |
-
interface = gr.Interface(
|
78 |
-
fn=initialize,
|
79 |
-
inputs=[
|
80 |
-
gr.File(label="Upload PDF"), # No need for 'type' argument
|
81 |
-
gr.Textbox(label="Question")
|
82 |
-
],
|
83 |
-
outputs="text",
|
84 |
-
title="GeminiPro Q&A Bot",
|
85 |
-
description="Ask questions about the uploaded PDF document.",
|
86 |
-
)
|
87 |
-
|
88 |
-
# Launch the interface
|
89 |
-
interface.launch()
|
|
|
1 |
import os
|
|
|
|
|
|
|
|
|
2 |
import gradio as gr
|
3 |
+
from langchain import PromptTemplate
|
4 |
+
from langchain.chains.question_answering import load_qa_chain
|
5 |
+
from langchain.document_loaders import PyPDFLoader
|
6 |
+
from langchain_google_genai import ChatGoogleGenerativeAI
|
7 |
+
import google.generativeai as genai
|
8 |
+
from dotenv import load_dotenv
|
9 |
+
|
10 |
+
# Load environment variables from .env file
|
11 |
+
load_dotenv()
|
12 |
+
|
13 |
+
# Fungsi untuk inisialisasi
|
14 |
+
def initialize(file_path, question):
|
15 |
+
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
|
16 |
+
model = genai.GenerativeModel('gemini-pro')
|
17 |
+
model = ChatGoogleGenerativeAI(model="gemini-pro", temperature=0.3)
|
18 |
+
prompt_template = """Answer the question as precise as possible using the provided context. If the answer is
|
19 |
+
not contained in the context, say "answer not available in context" \n\n
|
20 |
+
Context: \n {context}?\n
|
21 |
+
Question: \n {question} \n
|
22 |
+
Answer:
|
23 |
+
"""
|
24 |
+
prompt = PromptTemplate(template=prompt_template, input_variables=["context", "question"])
|
25 |
+
if os.path.exists(file_path):
|
26 |
+
pdf_loader = PyPDFLoader(file_path)
|
27 |
+
pages = pdf_loader.load_and_split()
|
28 |
+
context = "\n".join(str(page.page_content) for page in pages[:30])
|
29 |
+
stuff_chain = load_qa_chain(model, chain_type="stuff", prompt=prompt)
|
30 |
+
stuff_answer = stuff_chain({"input_documents": pages, "question": question, "context": context}, return_only_outputs=True)
|
31 |
+
return stuff_answer['output_text']
|
32 |
+
else:
|
33 |
+
return "Error: Unable to process the document. Please ensure the PDF file is valid."
|
34 |
+
|
35 |
+
# Define Gradio Interface
|
36 |
+
input_file = gr.File(label="Upload PDF File")
|
37 |
+
input_question = gr.Textbox(label="Ask about the document")
|
38 |
+
output_text = gr.Textbox(label="Answer - GeminiPro")
|
39 |
+
|
40 |
+
def pdf_qa(file, question):
|
41 |
+
answer = initialize(file.name, question)
|
42 |
+
return answer
|
43 |
+
|
44 |
+
# Create Gradio Interface
|
45 |
+
gr.Interface(fn=pdf_qa, inputs=[input_file, input_question], outputs=output_text, title="PDF Question Answering System", description="Upload a PDF file and ask questions about the content.").launch()
|
46 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|