app.py

from flask import Flask, jsonify,request
import pdfplumber
import pandas as pd
import re

app = Flask(__name__)

@app.route("/validate",methods = ["POST"])
def get_details():
    data = []
    try:
        if 'airline' not in request.form and 'Airline' not in request.form:
            return 'Airline name is missing'

        # Get the value of 'airline' from the request
        airlines_name = request.form.get('airline') or request.form.get('Airline')

        # Check if the airlines name is empty
        if not airlines_name:
            return 'Airlines name is missing'

        # Check if the airlines name is 'Vistara'
        if airlines_name.lower()!='vistara':
            return 'This api only for vistara'
        else:
            if 'file' not in request.files:
                return 'No file part'

            file = request.files['file']

            if file.filename == '':
                return 'No selected file'

            if file.filename.split('.')[-1] != 'pdf':
                return "Only read pdf files"

            else:
                text = ""
                with pdfplumber.open(file) as pdf:
                    for page in pdf.pages:
                        text += page.extract_text()

                # Extracting Invoice Number
                invoice_number_match = re.search(r'Invoice No\. (\w+)', text)
                invoice_number = invoice_number_match.group(1) if invoice_number_match else "Not found"

                # Extracting Invoice Date
                invoice_date_match = re.search(r'Invoice Date (\d{2}-[A-Z]{3}-\d{2})', text)
                invoice_date = invoice_date_match.group(1) if invoice_date_match else "Not found"

                # Extracting GSTIN
                gstin_match = re.search(r'\b\d{2}[A-Z]{5}\d{4}[A-Z]\d[Z]\d\b', text)
                gstin = gstin_match.group() if gstin_match else "Not found"

                # Extracting Place of Supply (State)
                place_of_supply_match = re.search(r'Place of supply \(State\) (\w+)', text)
                place_of_supply_state = place_of_supply_match.group(1) if place_of_supply_match else "Not found"

                # Extracting Bill-to Name or Customer Name
                bill_to_name_match = re.search(r'Name\s+(.*?)\s+(?:Credit|Debit) Note Date', text)
                bill_to_name = bill_to_name_match.group(1) if bill_to_name_match else "Not found"

                # Extracting Company Name
                company_name_match = re.search(r"\b\w+\s+\w+\s+\w+\s+(Limited|limited|Pvt|PVT)\b", text)
                company_name = company_name_match.group() if company_name_match else "Not found"

                # Extracting Company Address
                company_add_match = re.search(r'(?<=Limited\n).*',text)
                company_add = company_add_match.group() if company_add_match else "Not found"

                # Extracting HASC Code
                company_sac_match = re.search(r'SAC\s*Code\s*(\d{5,6})', text)
                company_sac = company_sac_match.group(1) if company_sac_match else "Not found"

                # Extracting Description of Service
                description_pattern_match = re.search(r'Description of Service\s*(.+)',text)
                description_of_service  = description_pattern_match.group(1) if description_pattern_match else "Not found"

                # Extracting CGST Rate
                company_cgst_match = re.search(r"CGST Rate\s*(\d+(\.\d+)?%?)",text)
                company_cgst = company_cgst_match.group(1) if company_cgst_match else "Not found"

                # Extracting CGST Amount
                company_cgst_amt_match = re.search(r"CGST\s*Amount\s*\s*(\d+)", text)
                company_cgst_amount = company_cgst_amt_match.group(1) if company_cgst_amt_match else "Not found"

                # Extracting site gstin
                site_gstin_match = re.search(r"GSTN\s+(\w+)",text)
                site_gstin = site_gstin_match.group(1) if site_gstin_match else "Not found"

                # Extracting SGST Rate
                company_sgst_match = re.search(r"SGST Rate\s*(\d+(\.\d+)?%?)",text)
                company_sgst = company_sgst_match.group(1) if company_sgst_match else "Not found"

                # Extracting SGST Amount
                company_sgst_amt_match = re.search(r"SGST\s*Amount\s*\s*(\d+)", text)
                company_sgst_amount = company_sgst_amt_match.group(1) if company_sgst_amt_match else "Not found"

                # Extracting IGST Rate
                company_igst_match = re.search(r"IGST Rate\s*(\d+(\.\d+)?%?)",text)
                company_igst = company_igst_match.group(1) if company_igst_match else "Not found"

                # Extracting IGST Amount
                company_igst_amt_match = re.search(r"IGST\s*Amount\s*\s*(\d+)", text)
                company_igst_amount = company_igst_amt_match.group(1) if company_igst_amt_match else "Not found"

                # Extracting company amount 
                company_amount_match = re.search(r'Total Invoice Value \(in figures\) (\d+(?:,\d+)?)',text)
                company_amount = company_amount_match.group(1) if company_amount_match else "Not found"

                #Extracting Net taxable values
                net_tax_match = re.search(r'Net taxable value\s*(\d+(?:,\d+)?)',text)
                net_tax = net_tax_match.group(1) if net_tax_match else "Not found"
            
                data.append(
                    {
                        "COMPANY_NAME": bill_to_name if bill_to_name != "Not found" else "NIL",
                        "STATE": place_of_supply_state if place_of_supply_state != "Not found" else "NIL",
                        "COMPANY_GSTN":gstin if gstin != "Not found" else "NIL",
                        "INVOICE_SOURCE": None,
                        "SUPPLIER_NAME":company_name if company_name != "Not found" else "NIL",
                        "SUPPLIER_ADDRESS":company_add if company_add != "Not found" else "NIL",
                        "SUPPLIER_CITY": company_add if company_add != "Not found" else "NIL", 
                        "SUPPLIER_CITY_PINCODE": 0,
                        "SITE_NAME":bill_to_name if bill_to_name != "Not found" else "NIL",
                        "SITE_STATE": place_of_supply_state if place_of_supply_state != "Not found" else "NIL",  # Placeholder value
                        "SITE_STATE_CODE":"NIL",
                        "SITE_GSTIN": site_gstin if site_gstin != "Not found" else "NIL",  # Placeholder value
                        "INVOICE_TYPE": "B2B",  # Placeholder value
                        "SUPPLIER_INVOICE_NO": invoice_number if invoice_number != "Not found" else "NIL",
                        "SUPPLIER_INVOICE_DATE": invoice_date if invoice_date != "Not found" else "NIL",
                        "QTY": 0,  # Placeholder value
                        "UOM": 0,
                        "UNIT_PRICE": 0,  # Placeholder value
                        "SUPPLIER_TAXABLE_VALUE_INR": net_tax,
                        "CGST_RATE": company_cgst if company_cgst != "Not found" else 0,
                        "CGST_AMT": company_cgst_amount if company_cgst_amount != "Not found" else 0,
                        "SGST_RATE": company_sgst if company_sgst != "Not found" else 0,
                        "SGST_AMT": company_sgst_amount if company_sgst_amount != "Not found" else 0,
                        "IGST_RATE": company_igst if company_igst != "Not found" else 0,
                        "IGST_AMT": company_igst_amount if company_igst_amount != "Not found" else 0,
                        "IRN_CODE": 0,  # Placeholder value
                        "EINV_STATUS": 0,  # Placeholder value
                        "IRN_CREATION_DATE": 0,  # Placeholder value
                        "INVOICE_AMOUNT_INR": company_amount if company_amount != "Not found" else 0,
                        "INVOICE_CURRENCY_CODE": "INR",  # Placeholder value
                        "INVOICE_AMOUNT": company_amount if company_amount != "Not found" else 0,
                        "HSNSAC_CODE": company_sac if company_sac != "Not found" else 0,
                        "ITEM_DESCRIPTION": description_of_service if description_of_service != "Not found" else "NIL"
                    }
                )
            df = pd.DataFrame(data)
            df.to_csv("output.csv",index=False)
            return jsonify({"message": "Data extracted successfully", "output_csv": "output.csv"})
    
    except Exception as e:
        return str(e)


if __name__ == "__main__":
    app.run(debug=True)