draft.py

import cv2
import pytesseract
import re
import numpy as np
from PIL import Image
import streamlit as st

st.markdown(
  """
  <style>
  .stApp{
    background-color: #1f2b2b;
    color: #E5E5E5;
  }
  
    html, body, [class*="css"] {
    color: #E5E5E5 !important;
    }
    
    header[data-testid="stHeader"] {
      background-color: #1f2b2b;
      display: flex;
      justify-content: space-between;
      align-items: center;
      padding: 0.5rem 1rem;
    }
    
    .nav-buttons {
      display: flex;
      gap: 1rem;
     } 
     
    .nav-buttons a {
        text-decoration: none;
        color: #1f2b2b;
        background-color: #E5E5E5;
        padding: 0.5rem 1rem;
        border-radius: 5px;
        font-weigth: bold;
        transition: background-color 0.3s,color 0.3s;
      }
      
      .nav-buttons a:hover {
          background-color: #1f2b2b;
          color: #e5E5E5;
        }
        
       </style>
       <header data-testid="stHeader">
         <div class="nav-buttons">
           <a href="/?nav=home">Home</a>
           <a href="/?nav=about">About Us</a>
          </div>
        </header>
  """
)
st.title("PII shield")
# query_params = st.experimental_get_query_params()
# nav = query+params.get("nav" , ["home"])[0]

# if nav == "home":
st.title("Home")
st.write("Welcome to the Home Page!")
st.write("Our solution for the PII shield involves a robust software application designed to scan and detect government issued PII in documents or data.Uses Regular expression (Regex) patterns to identify PII in Aadhaar cards, PAN, Driving Lincense, etc. Allows user to redact or mask PII to ensure privacy compliance.")
# elif nav == "about":
st.title("About Us")
st.title("Learn more about us on this page")
st.title("Who are we?")
st.write("We are an enthusiastic team of learners and innovators, starting our journey to help individuals and businesses safeguard their sensitive information in the digital age. With a growing passion for data security,and privacy protection, we are committed to building simple, effective tools that make privacy management accessible to everyone. ")
st.write("Our Mission")
st.write("Our mission is to create effective and easy-to-use tools that help safeguard Personally Identifiable Information (PII). As beginners, we are dedicated to developing solutions for detecting, masking, and alerting users about sensitive information, making privacy protection simple, accessible, and reliable for everyone.")

pytesseract.pytesseract.tesseract_cmd= r"C:\Program Files\Tesseract-OCR\tesseract.exe"

face_cascade=cv2.CascadeClassifier(cv2.data.haarcascades+ "haarcasade_frontalface_default.xml")

def extract_aadhaar_number (text):
  aadhaar_pattern =r"\b\d{4}[- ]?\d{4}[- ]?\d{4}\b"
  match = re.search(aadhaar_pattern,text)
  return match.group (0) if match else None
 
def extract_pan_number(text):
  pan_pattern= r"\b[A-Z]{5}\d{4}[A-Z]\b"
  match= re.search(pan_pattern, text)
  return match.group (0) if match else None 

def extract_voter_id(text):
  voter_id_pattern = r"\b[A-Z]{3}\d{7}\b"
  match = re.search(voter_id_pattern, text) 
  return match.group (0) if match else None

def extract_driving_license(text):
  dl_pattern = r"\b[A-Z]{2}[- ]?\d{2}[- ]?\d{7,13}\b"
  match = re.search(dl_pattern, text) 
  return match.group (0) if match else None

def extract_dob(text):
  dob_pattern = r"\b\d{2}/\d{2}/\d{4}\b"
  match = re.search(dob_pattern, text) 
  return match.group (0) if match else None

def extract_address(text):
  lines = text.split('\n')
  address_block=[]
  address_started = False
  
  for line in lines:
    if "Address" in line or "पता" in line:
      address_started = True
      if address_started:
        address_block.append(line.strip()) 
        if len(address_block) >=4:
          break

  address = " ".join(address_block).replace("Address","").replace("पता","").strip()
  return address if address else None
    
def mask_aadhaar_number (img,aadhaar_number):
  text_data = pytesseract.image_to_data(img, output_type =pytesseract.output.DICT)
  for i,word in enumerate (text_data["text"]):
    if word in aadhaar_number and len(word)==4:
      x,y,w,h= (
        text_data["left"][i],
        text_data["top"][i],
        text_data["width"][i],
        text_data["height"][i],
      )
    img =cv2.rectangle(img, (x, y) , (x + w, y + h ), (0, 0, 0), -1)
  return img

def mask_pan_number (img,pan_number):
  text_data = pytesseract.image_to_data(img, output_type =pytesseract.output.DICT)
  for i,word in enumerate (text_data["text"]):
    if word ==pan_number:
      x,y,w,h= (
        text_data["left"][i],
        text_data["top"][i],
        text_data["width"][i],
        text_data["height"][i],
      )
    img =cv2.rectangle(img , (x, y) , (x + w, y + h) , (0, 0, 0) , -1)
  return img

def mask_voter_id (img,voter_id):
  text_data = pytesseract.image_to_data(img, output_type =pytesseract.output.DICT)
  for i,word in enumerate (text_data["text"]):
    if word ==voter_id:
      x,y,w,h= (
        text_data["left"][i],
        text_data["top"][i],
        text_data["width"][i],
        text_data["height"][i],
      )
    img =cv2.rectangle(img , (x,y) , (x+w,y+h) , (0,0,0) , -1)
  return img

def mask_driving_license (img,driving_license):
  text_data = pytesseract.image_to_data(img, output_type =pytesseract.output.DICT)
  for i , word in enumerate (text_data["text"]):
    if word ==driving_license:
      x,y,w,h = (
        text_data["left"][i],
        text_data["top"][i],
        text_data["width"][i],
        text_data["height"][i],
    )
    img =cv2.rectangle(img , (x,y) , (x+w,y+h) , (0,0,0) , -1)
  return img

def mask_text (img,text_to_mask):
  text_data = pytesseract.image_to_data( img, output_type =pytesseract.output.DICT)
  for i, word in enumerate (text_data["text"]):
    if text_to_mask in word:
      x,y,w,h = (
        text_data["left"][i],
        text_data["top"][i],
        text_data["width"][i],
        text_data["height"][i],
      )
      img =cv2.rectangle(img , (x,y) , (x+w,y+h) , (0,0,0) , -1)
  return img
   
def mask_address (img,address):
  text_data = pytesseract.image_to_data(img, output_type =pytesseract.output.DICT)
  coords=[]
  for i , word in enumerate (text_data["text"]):
      if word.strip() in address and len(word.strip()) > 1:
        x, y, w, h = (
          text_data["left"][i],
          text_data["top"][i],
          text_data["width"][i],
          text_data["height"][i],
        )
  coords.append((x , y , x + w , y + h))
  if not coords:
    return img
  coords=np.array(coords)
  img=cv2.rectangle(img, (coords[:, 0].min(), coords[:, 1].min()),
                     (coords[:, 2].max(), coords[:, 3].max()), (0, 0, 0), -1)
  return img

def blur_faces(img):
  gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
  faces = face_cascade.detectMultiScale(gray_img, scalefactor=1.1, minNeighbors=5, minSize=(30, 30))
  for (x, y, w, h) in faces:
    face=img[y:y+h, x:x+w]
    blurred_face=cv2.GaussianBlur(face, (51,51), 30)
    img[y:y+h,x:x+w]=blurred_face
  return img
 
st.title("Document Details Extractor & Masker")
st.write("Upload an image to extract Aadhaar, PAN, Voter ID, Driving License numbers, DOB details, and optionally mask them along with face blurring.")

uploaded_file = st.file_uploader("Choose an image", type=["png", "jpg", "jpeg"])

if uploaded_file is not None:
    image = Image.open(uploaded_file)
    img_array = np.array(image)
    #img_array = cv2.erode(img_array, kernel=np.ones((2,1)), iterations=1)
    img_rgb = img_array.copy()

    extracted_text = pytesseract.image_to_string(img_array)
  
    aadhaar_number = extract_aadhaar_number(extracted_text)
    pan_number = extract_pan_number(extracted_text)
    voter_id = extract_voter_id(extracted_text)
    driving_license = extract_driving_license(extracted_text)
    dob = extract_dob(extracted_text)
    address=extract_address(extracted_text)
  
    st.write("Extracted Information: ")

    extracted_info = {}
    if aadhaar_number:
        st.success(f"Aadhaar Number: {aadhaar_number}")
        extracted_info["Aadhaar Number"] = aadhaar_number
    if pan_number:
        st.success(f"PAN Number: {pan_number}")
        extracted_info["PAN Number"] = pan_number
    if voter_id:
        st.success(f"Voter ID: {voter_id}")
        extracted_info["Voter ID"] = voter_id
    if driving_license:
        st.success(f"Driving License: {driving_license}")
        extracted_info["Driving License"] = driving_license
    if dob:
        st.success(f"Date of Birth: {dob}")
        extracted_info["Date of Birth"] = dob
    if address:
        st.success(f"Address: {address}")
        extracted_info["Address"] = address
      
    extracted_info["Blur Faces"] = "Yes"

    if extracted_info:
        
        selected_pii = st.multiselect(
            "Select the PII to Mask or Blur",
            options=extracted_info.keys(),
            default=list(extracted_info.keys())
        )
      
         if st.button("Apply Masking"):
           for pii_type in selected_pii:
                    if pii_type == "Aadhaar Number":
                        img_rgb = mask_aadhaar_number(img_rgb, extracted_info[pii_type])
                    elif pii_type == "PAN Number":
                        img_rgb = mask_pan_number(img_rgb, extracted_info[pii_type])
                    elif pii_type == "Voter ID":
                        img_rgb = mask_voter_id(img_rgb, extracted_info[pii_type])
                    elif pii_type == "Driving License":
                        img_rgb = mask_driving_license(img_rgb, extracted_info[pii_type])
                    elif pii_type == "Date of Birth":
                        img_rgb = mask_text(img_rgb, extracted_info[pii_type])
                    elif pii_type == "Address":
                        img_rgb = mask_address(img_rgb, extracted_info[pii_type])
                    elif pii_type=="Blur Faces":
                        img_rgb = blur_faces(img_rgb)

                masked_pil = Image.fromarray(img_rgb)
                st.image(masked_pil, caption="Masked and Blurred Image", use_container_width = True)

                masked_pil.save("masked_image.png")
                with open("masked_image.png", "rb") as file:
                    st.download_button(
                        label="Download Masked and Blurred Image",
                        data=file,
                        file_name="masked_image.png",
                        mime="image/png",
                    )