import streamlit as st
import requests
from PIL import Image
import io
import json
import pandas as pd
import plotly.express as px
import numpy as np
from typing import Dict, Any
import logging
import pytesseract
import re
from openai import OpenAI, OpenAIError
import boto3
from botocore.exceptions import ClientError
import os
from dotenv import load_dotenv
load_dotenv()
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
# Constants
API_URL = "http://localhost:8000"
SUPPORTED_DOCUMENT_TYPES = ["invoice", "receipt", "form"]
api_key = os.getenv("PERPLEXITY_API_KEY")
client = OpenAI(api_key=api_key, base_url="https://api.perplexity.ai")
REGION = "us-east-1"
dynamodb = boto3.resource('dynamodb', region_name=REGION)
def extract_json_from_llm_output(llm_result):
match = re.search(r'\{.*\}', llm_result, re.DOTALL)
if match:
return match.group(0)
return None
def save_to_dynamodb(data, table_name="Receipts"):
dynamodb = boto3.resource("dynamodb")
table = dynamodb.Table(table_name)
try:
table.put_item(Item=data)
return True
except ClientError as e:
st.error(f"Failed to save to DynamoDB: {e}")
return False
def main():
st.set_page_config(
page_title="FormIQ - Intelligent Document Parser",
page_icon="📄",
layout="wide"
)
st.title("FormIQ: Intelligent Document Parser")
st.markdown("""
Upload your documents to extract and validate information using advanced AI models.
""")
# Sidebar
with st.sidebar:
st.header("Settings")
document_type = st.selectbox(
"Document Type",
options=SUPPORTED_DOCUMENT_TYPES,
index=0
)
confidence_threshold = st.slider(
"Confidence Threshold",
min_value=0.0,
max_value=1.0,
value=0.5,
step=0.05
)
st.markdown("---")
st.markdown("### About")
st.markdown("""
FormIQ uses LayoutLMv3 and GPT-4 to extract and validate information from documents.
""")
# Main content
uploaded_file = st.file_uploader(
"Upload Document",
type=["png", "jpg", "jpeg", "pdf"],
help="Upload a document image to process"
)
if uploaded_file is not None:
# Display uploaded image
image = Image.open(uploaded_file)
st.image(image, caption="Uploaded Document", width=600)
# Process button
if st.button("Process Document"):
with st.spinner("Processing document..."):
try:
# Save the uploaded file to a temporary location
temp_path = "temp_uploaded_image.jpg"
image.save(temp_path)
# Extract fields using OCR + regex
fields = extract_fields(temp_path)
# Extract with Perplexity LLM using the provided API key
with st.spinner("Extracting structured data with Perplexity LLM..."):
try:
llm_result = extract_with_perplexity_llm(pytesseract.image_to_string(Image.open(temp_path)))
st.subheader("Structured Data (Perplexity LLM)")
st.code(llm_result, language="json")
# Extract and save JSON to DynamoDB
raw_json = extract_json_from_llm_output(llm_result)
if raw_json:
try:
llm_data = json.loads(raw_json)
if save_to_dynamodb(llm_data):
st.success("Data saved to DynamoDB!")
except Exception as e:
st.error(f"Failed to parse/save JSON: {e}")
else:
st.error("No valid JSON found in LLM output.")
except Exception as e:
st.error(f"LLM extraction failed: {e}")
except Exception as e:
logger.error(f"Error processing document: {str(e)}")
st.error(f"Error processing document: {str(e)}")
def display_results(results: Dict[str, Any]):
"""Display extraction and validation results."""
# Create tabs for different views
tab1, tab2, tab3 = st.tabs(["Extracted Fields", "Validation", "Visualization"])
with tab1:
st.subheader("Extracted Fields")
if "fields" in results["extraction"]:
fields_df = pd.DataFrame(results["extraction"]["fields"])
st.dataframe(fields_df)
else:
st.info("No fields extracted")
with tab2:
st.subheader("Validation Results")
validation = results["validation"]
# Display validation status
status_color = "green" if validation["is_valid"] else "red"
st.markdown(f"### Status: :{status_color}[{validation['is_valid']}]")
# Display validation errors if any
if validation["validation_errors"]:
st.error("Validation Errors:")
for error in validation["validation_errors"]:
st.markdown(f"- {error}")
# Display confidence score
st.metric(
"Overall Confidence",
f"{validation['confidence_score']:.2%}"
)
with tab3:
st.subheader("Confidence Visualization")
if "confidence_scores" in results["extraction"]["metadata"]:
scores = results["extraction"]["metadata"]["confidence_scores"]
# Create confidence distribution plot
fig = px.histogram(
x=scores,
nbins=20,
title="Confidence Score Distribution",
labels={"x": "Confidence Score", "y": "Count"}
)
st.plotly_chart(fig)
# Display heatmap if available
if "bbox" in results["extraction"]["fields"][0]:
st.subheader("Field Location Heatmap")
# TODO: Implement heatmap visualization
st.info("Heatmap visualization coming soon!")
def group_tokens_by_label(tokens, labels):
structured = {}
current_label = None
current_tokens = []
for token, label in zip(tokens, labels):
if label != current_label:
if current_label is not None:
structured.setdefault(current_label, []).append(' '.join(current_tokens))
current_label = label
current_tokens = [token]
else:
current_tokens.append(token)
if current_label is not None:
structured.setdefault(current_label, []).append(' '.join(current_tokens))
return structured
def extract_fields(image_path):
# OCR
text = pytesseract.image_to_string(Image.open(image_path))
# Display OCR output for debugging
st.subheader("Raw OCR Output (for debugging)")
st.code(text)
# Improved Regex patterns for fields
patterns = {
"name": r"Mrs\s+\w+\s+\w+",
"date": r"Date[:\s]+([\d/]+)",
"product": r"\d+\s+\w+.*Style\s+\d+",
"amount_paid": r"Total Paid\s+\$?([\d.,]+)",
# Improved pattern for receipt number (handles optional dot, colon, spaces)
"receipt_no": r"Receipt No\.?\s*:?\s*(\d+)"
}
results = {}
for field, pattern in patterns.items():
match = re.search(pattern, text, re.IGNORECASE)
if match:
results[field] = match.group(1) if match.groups() else match.group(0)
else:
results[field] = None
return results
def extract_with_perplexity_llm(ocr_text):
prompt = f"""
Extract the following fields from this receipt text:
- name
- date
- product
- amount_paid
- receipt_no
Text:
\"\"\"{ocr_text}\"\"\"
Return the result as a JSON object with those fields.
"""
messages = [
{
"role": "system",
"content": (
"You are an artificial intelligence assistant. "
"Answer user questions as concisely and directly as possible. "
"Limit your responses to 2-3 sentences unless the user asks for more detail."
)
},
{
"role": "user",
"content": prompt
}
]
response = client.chat.completions.create(
model="sonar-pro", # Use a valid model name for your account
messages=messages,
)
return response.choices[0].message.content
def interactive_chatbot_ui():
st.header("🤖 Chatbot")
if "chat_history" not in st.session_state:
st.session_state.chat_history = []
# Display chat history as chat bubbles
for sender, msg in st.session_state.chat_history:
if sender == "You":
st.markdown(f"You: {msg}
", unsafe_allow_html=True)
else:
st.markdown(f"Bot: {msg}
", unsafe_allow_html=True)
# Input at the bottom
with st.form(key="chat_form", clear_on_submit=True):
user_input = st.text_input("Type your message...", key="chat_input_main", placeholder="Ask me anything...")
submitted = st.form_submit_button("Send")
if submitted and user_input:
st.session_state.chat_history.append(("You", user_input))
try:
response = requests.post(
f"{API_URL}/chat",
json={"question": user_input}
)
if response.status_code == 200:
bot_reply = response.json()["answer"]
else:
bot_reply = f"Error: Server returned status code {response.status_code}"
except Exception as e:
bot_reply = f"Error: {e}"
st.session_state.chat_history.append(("Bot", bot_reply))
if __name__ == "__main__":
main()
st.markdown("---")
interactive_chatbot_ui()