eRAG-PDPA-v1 / app.py
amornpan's picture
Update app.py
96c3305 verified
raw
history blame
6.41 kB
import streamlit as st
import requests
from datetime import datetime
import base64
import os
import re
if "dialog_done" not in st.session_state:
st.session_state.dialog_done = False
if not st.session_state.dialog_done:
st.title("เริ่มต้นคำถาม")
st.write("คำถาม: ใครคือคนที่หน้าตาดีที่สุด?")
answer = st.text_input("กรุณาตอบคำถามนี้:")
if st.button("ส่งคำตอบ"):
if answer == "พี่ก้องคนหล่อ":
st.session_state.dialog_done = True
st.success("คำตอบถูกต้อง! กดปุ่ม 'เข้าสู่หน้าหลัก' เพื่อดำเนินการต่อ")
else:
st.error("คำตอบไม่ถูกต้อง กรุณาลองใหม่อีกครั้ง.")
if st.session_state.dialog_done:
if st.button("เข้าสู่หน้าหลัก"):
st.experimental_rerun()
else:
st.title("AI สนับสนุนความรู้ด้าน PDPA")
st.write("เราสอบถาม AI สืบค้น และสรุป")
system_prompt = "คุณเป็นผู้ช่วยที่มีความรู้ด้านกฎหมาย PDPA และสามารถให้คำตอบที่เกี่ยวข้องเฉพาะตาม context ที่ได้รับ"
def clean_text_for_search(text):
text = re.sub(r'P-\d+\s*$', '', text, flags=re.MULTILINE)
text = re.sub(r'Confidential.*$', '', text, flags=re.MULTILINE)
text = ' '.join(text.split())
return text
def create_highlighted_pdf(pdf_path, search_text, page_number):
try:
search_text = clean_text_for_search(search_text)
doc = fitz.open(pdf_path)
page = doc[int(page_number) - 1]
words = [word for word in search_text.split() if word]
for word in words:
if len(word) > 3:
text_instances = page.search_for(word)
for inst in text_instances:
highlight = page.add_highlight_annot(inst)
highlight.set_colors(stroke=(1, 1, 0))
highlight.update()
new_doc = fitz.open()
new_doc.insert_pdf(doc, from_page=int(page_number) - 1, to_page=int(page_number) - 1)
pdf_bytes = new_doc.write()
doc.close()
new_doc.close()
return pdf_bytes
except Exception as e:
st.error(f"Error in create_highlighted_pdf: {str(e)}")
return None
def format_file_size(size_in_bytes):
for unit in ['B', 'KB', 'MB', 'GB']:
if size_in_bytes < 1024:
return f"{size_in_bytes:.2f} {unit}"
size_in_bytes /= 1024
return f"{size_in_bytes:.2f} GB"
def display_search_result(result, index):
with st.expander(f"🔍 Search Result #{index + 1} (Score: {result['score']:.4f})"):
st.markdown("#### 📄 Document Information")
col1, col2 = st.columns(2)
with col1:
st.markdown("**File Details:**")
st.write(f"• File Name: {result['metadata']['file_name']}")
st.write(f"• Page: {result['metadata']['page_label']}")
st.write(f"• Type: {result['metadata']['file_type']}")
st.write(f"• Size: {format_file_size(result['metadata']['file_size'])}")
with col2:
st.markdown("**Dates:**")
st.write(f"• Created: {result['metadata']['creation_date']}")
st.write(f"• Modified: {result['metadata']['last_modified_date']}")
st.markdown("#### 📝 Content")
st.markdown(f"```\n{result['text']}\n```")
try:
pdf_path = result['file_path']
if os.path.exists(pdf_path):
st.markdown("#### 📄 PDF Preview (with highlighted text)")
highlighted_pdf = create_highlighted_pdf(
pdf_path,
result['text'],
result['metadata']['page_label']
)
if highlighted_pdf:
base64_pdf = base64.b64encode(highlighted_pdf).decode('utf-8')
pdf_display = f'''
<iframe
src="data:application/pdf;base64,{base64_pdf}"
width="100%"
height="800px"
type="application/pdf"
style="border: 1px solid #ccc; border-radius: 5px;"
></iframe>
'''
st.markdown(pdf_display, unsafe_allow_html=True)
else:
st.error("Failed to create highlighted PDF")
except Exception as e:
st.error(f"Error displaying PDF: {str(e)}")
if "chat_history" not in st.session_state:
st.session_state.chat_history = []
with st.form(key="input_form"):
user_input = st.text_input("You:", key="input")
submit_button = st.form_submit_button("Send")
if submit_button:
if user_input:
if st.session_state.chat_history:
st.session_state.chat_history.insert(0, ("###", "###"))
st.session_state.chat_history.insert(0, ("You", user_input))
try:
response = requests.post("http://113.53.253.50:8002/search", json={"query": user_input})
response.raise_for_status()
data = response.json()
search_results = data["results"]
st.markdown("### 🔎 Search Results")
for idx, result in enumerate(search_results):
display_search_result(result, idx)
response_text = "\n\n---\n\n".join([f"Text: {result['text']}" for result in search_results])
except requests.RequestException as e:
st.error(f"Error: {str(e)}")