idvpackage 3.0.11__py3-none-any.whl → 3.0.12__py3-none-any.whl

idvpackage/common.py

@@ -1,18 +1,11 @@
-import re
-from datetime import datetime
-from itertools import permutations
+
 import cv2
 import numpy as np
 from PIL import Image
-import io
-from google.cloud import vision_v1
-import tempfile
-import os
-from io import BytesIO
 import logging
 import base64
 from concurrent.futures import ThreadPoolExecutor, as_completed
-from scipy.spatial.distance import cosine
+
 
 # Global variables to store lazily loaded modules
 _deepface = None
@@ -34,553 +27,8 @@ def get_face_recognition():
         _face_recognition = face_recognition
     return _face_recognition
 
-def func_common_dates(  extract_no_space):
-    dob = ''
-    expiry_date = ''
-    try:
-        matches = re.findall(r'\d{2}/\d{2}/\d{4}', extract_no_space)
-        y1 = matches[0][-4:]
-        y2 = matches[1][-4:]
-        if int(y1) < int(y2):
-            dob = matches[0]
-            expiry_date = matches[1]
-        else:
-            dob = matches[1]
-            expiry_date = matches[0]
-    except:
-        dob = ''
-        expiry_date = ''
-
-    return dob, expiry_date
-
-def convert_dob(input_date):
-    day = input_date[4:6]
-    month = input_date[2:4]
-    year = input_date[0:2]
-
-    current_year = datetime.now().year
-    current_century = current_year // 100
-    current_year_last_two_digits = current_year % 100
-
-    century = current_century
-    # If the given year is greater than the last two digits of the current year, assume last century
-    if int(year) > current_year_last_two_digits:
-        century = current_century - 1
-
-    final_date = f"{day}/{month}/{century}{year}"
-
-    return final_date
-
-def func_dob(  extract):
-    extract_no_space = extract.replace(' ','')
-    dob, expiry_date = func_common_dates(extract_no_space)
-    if dob == '':  
-        match_dob = re.findall(r'\d{7}(?:M|F)\d', extract_no_space)
-        for i in match_dob:
-            # print(i)
-            raw_dob = i[0:6]
-            # print(raw_dob)
-            year = str(datetime.today().year)[2:4]
-            temp = '19'
-            if int(raw_dob[0:2]) > int(year):
-                temp = '19'
-            else:
-                temp = '20'      
-            dob = raw_dob[4:6]+'/'+raw_dob[2:4]+'/'+temp+raw_dob[0:2]
-            try:
-                dt_obj = datetime.strptime(dob, '%d/%m/%Y')
-                break
-            except:
-                # print(f'invalid date {dob}')
-                dob = ''
-        else:
-            pattern = r"\b(\d{14}).*?\b"
-
-            new_dob_match = re.search(pattern, extract_no_space)
-
-            if new_dob_match:
-                new_dob = new_dob_match.group(1)
-                new_dob = new_dob[:7]
-                dob = convert_dob(new_dob)
-
-    return dob
-
-def func_expiry_date(  extract):
-    extract_no_space = extract.replace(' ','')
-    dob, expiry_date = func_common_dates(extract_no_space)
-    if expiry_date == '':
-        match_doe = re.findall(r'\d{7}[A-Z]{2,3}', extract_no_space) 
-        for i in match_doe:
-         
-            raw_doe = i[0:6]
-            # print(raw_doe)
-            expiry_date = raw_doe[4:6]+'/'+raw_doe[2:4]+'/20'+raw_doe[0:2]
-            try:
-                dt_obj = datetime.strptime(expiry_date, '%d/%m/%Y')
-                break
-            except:
-   
-                expiry_date = ''
-
-    return expiry_date
-
-def convert_expiry_date(input_date):
-    day = input_date[4:6]
-    month = input_date[2:4]
-    year = input_date[0:2]
-
-    current_year = datetime.now().year
-    current_century = current_year // 100
-    current_year_last_two_digits = current_year % 100
-    century = current_century
-
-    if int(year) <= current_year_last_two_digits:
-        century = current_century
-    else:
-        century = current_century
-    final_date = f"{day}/{month}/{century}{year}"
-
-    return final_date
-
-def extract_first_9_digits(string_input):
-    match = re.search(r'\b\d{9}\b', string_input)
-    if match:
-        sequence = match.group(0)
-        return sequence
-    else:
-        return ""
-
-def func_card_number(  extract):
-    extract_no_space = extract.replace(' ','')
-    try:
-        card_number = re.search(r'\d{9}', extract_no_space).group()
-    except:
-        card_number=  extract_first_9_digits(extract_no_space)
-
-    return card_number
-
-
-def count_digits_after_pattern(s):
-    """
-    Counts the number of digits that come after a specified pattern in a string.
-
-    Parameters:
-    s (str): The input string.
-    pattern (str): The pattern to search for.
-
-    Returns:
-    int: The count of digits that come after the pattern.
-    """
-    # Construct the regex pattern to find the specified pattern followed by digits
-    pattern = "<<<<"
-    regex_pattern = re.compile(f"{re.escape(pattern)}(\d+)")
-    
-    # Search for the pattern in the string
-    match = regex_pattern.search(s)
-    
-    # If a match is found, count the digits
-    if match:
-        digits_after_pattern = match.group(1)
-        return len(digits_after_pattern)
-    else:
-        return 0  # Pattern not found
-
-def remove_special_characters1(string):
-    # This pattern matches any character that is not a letter, digit, or space
-    #pattern = r'[^a-zA-Z0-9<\s]'
-    pattern = r'[^a-zA-Z0-9<>]'
-    return re.sub(pattern, '', string)
-
-def remove_special_characters_mrz2(string):
-    # This pattern matches any character that is not a letter, digit, or space
-    pattern = r'[^a-zA-Z0-9\s]'
-    return re.sub(pattern, '', string)
-
-def validate_string(s):
-    """
-    Validates if the string follows the specific structure.
-
-    Structure: 7 digits, followed by 'M' or 'F', then 7 digits again,
-    then 3 uppercase letters, and ending with 1 digit.
-
-    Parameters:
-    s (str): The string to be validated.
-
-    Returns:
-    bool: True if the string follows the structure, False otherwise.
-    """
-    pattern = r'^\d{7}[MF]\d{7}[A-Z]{3}\d$'
-    return bool(re.match(pattern, s))
-
-
-def remove_special_characters2(string):
-    # This pattern matches any character that is not a letter, digit, or space
-    pattern = r'[^a-zA-Z0-9\s]'
-    return re.sub(pattern, ' ', string)
-
-def func_name(extract):
-    bio_data = extract[-40:]
-    breakup = bio_data.split('\n')
-    if len(breakup) == 2:
-        name_extract = breakup.pop(0)
-    else:
-        country_extract = breakup.pop(0).replace(" ","")
-        name_extract = breakup.pop(0)
-
-# Check the alphanumeric nature of name_extract
-    if not name_extract.isupper():
-        name_extract = breakup.pop(0)
-    
-    try:
-        name = name_extract.replace("<", " ").replace(">", " ").replace(".", " ").replace(":", " ").replace('«','').strip()
-        name = ' '.join(name.split())
-        name = name.replace("0", "O") # special case fix
-    except:
-        name = ""
-
-    return name
-
-def func_nationality(  extract):
-    extract_no_space = extract.replace(' ','')
-    try:
-        pattern = r'\d{5}[A-Z]{3}|\d{5}[A-Z]{2}'
-
-        m = re.findall(pattern, extract_no_space)
-        country = m[len(m)-1].replace("<", "")[5:]
-    except:
-        country = ""
-
-    if country == '':
-        try:
-            pattern = r'\d{2}[a-z][A-Z]{2}'
-
-            m = re.findall(pattern, extract_no_space)
-            country = m[len(m)-1].replace("<", "")[2:].upper()
-        except:
-            country = ""
-
-    return country
-
-def clean_string(input_string):
-        cleaned_string = re.sub(r'[^\w\s]', ' ', input_string)
-        return cleaned_string.strip()
-
-def count_digits(element):
-    digits = [char for char in element if char.isdigit()]
-    return len(digits)
-    
-def find_and_slice_number(input_number, digits):
-    # Generate all possible permutations of the digits
-    perms = [''.join(p) for p in permutations(digits)]
-    
-    # Initialize variables to keep track of the found pattern and its index
-    found_pattern = None
-    found_index = -1
-
-    # Search for any permutation of the digits in the input_number
-    for perm in perms:
-        found_index = input_number.find(perm)
-        if found_index != -1:
-            found_pattern = perm
-            break
-
-    # If a pattern is found, slice the number accordingly
-    if found_pattern:
-        if found_index > len(input_number) - found_index - len(found_pattern):
-            # Slice to the left
-            sliced_number = input_number[:found_index + len(found_pattern)]
-        else:
-            # Slice to the right
-            sliced_number = input_number[found_index:]
-        
-        return sliced_number
-    else:
-        return ''
-    
-def func_id_number(extract,dob):
-        
-    try:
-        p = "784" + "\d{12}"
-        id_re = re.search(p, clean_string(extract).replace(' ',''))
-        id_number = id_re.group()
-    except:
-        
-        try:
-            pattern = r'\d{15,}'
-            digits = '784'
-            matches = re.findall(pattern, clean_string(extract).replace(' ',''))
-            input_number = matches[0]
-            dob=dob[-4:]
-            id_number='784'+dob+find_and_slice_number(input_number, digits)[:8]
-            
-        except:
-            id_number = ''
-
-    return id_number
-
-
-# #year = dob[-4:]
-# p = "784" + "\d{12}"
-# id_re = re.search(p, clean_string(data).replace(' ',''))
-# id_number = id_re.group()
 
 
-
-def convert_to_date(date_str):
-    year = '19' + date_str[:2] if int(date_str[:2]) >= 50 else '20' + date_str[:2]
-    month = date_str[2:4]
-    day = date_str[4:6]
-    return f"{day}/{month}/{year}"
-
-def check_valid_date(date_str, format="%d/%m/%Y"):
-    try:
-        datetime.strptime(date_str, format)
-        return True
-    except ValueError:
-        return False
-    
-
-def find_expiry_date(original_text,mrz2):
-    
-    dates = re.findall(r'\b\d{2}/\d{2}/\d{4}\b', original_text)
-    expiry_date = ''
-
-    if len(dates) == 2:
-        
-        date1 = datetime.strptime(dates[0], '%d/%m/%Y')
-        date2 = datetime.strptime(dates[1], '%d/%m/%Y')
-        
-        if date2 < date1:
-            expiry_date = dates[0]
-        elif date2 > date1:
-            expiry_date = dates[1]
-            
-    elif mrz2:
-        match_expiry_date = re.search(r'[A-Za-z](\d+)', mrz2)
-        if match_expiry_date:
-            expiry_date = match_expiry_date.group(1)[:6]
-            expiry_date = convert_to_date(expiry_date)
-    
-            
-    if not check_valid_date(expiry_date):
-            expiry_date=''
-    return expiry_date
-
-def find_dob(original_text,mrz2):
-    
-     dates = re.findall(r'\b\d{2}/\d{2}/\d{4}\b', original_text)
-     dob = ''
-  
-     if len(dates) == 2:
-            date1 = datetime.strptime(dates[0], '%d/%m/%Y')
-            date2 = datetime.strptime(dates[1], '%d/%m/%Y')
-
-            if date2 < date1:
-                dob = dates[1]
-            elif date2 > date1:
-                dob = dates[0] 
-                
-     elif mrz2:
-        match_dob = re.search(r'(\d+)[A-Za-z]', mrz2)
-        if match_dob:
-            dob = match_dob.group(1)[:6] 
-            dob=convert_to_date(dob)
-    
-     if not check_valid_date(dob):
-            dob=''
-     return dob
-
-
-def convert_date_format(date_str):
-    # Parse the date from DD/MM/YYYY format
-    date_obj = datetime.strptime(date_str, '%d/%m/%Y')
-    # Convert it to YYYY-MM-DD format
-    formatted_date = date_obj.strftime('%Y-%m-%d')
-    return formatted_date
-
-
-def convert_gender(gender_char):
-    if gender_char.lower() == 'm':
-        return 'Male'
-    elif gender_char.lower() == 'f': 
-        return 'Female'
-    else:  
-        return ''
-
-
-def compute_ela_cv(orig_img, quality):
-    SCALE = 15
-    orig_img = cv2.cvtColor(orig_img, cv2.COLOR_BGR2RGB)
-    
-    with tempfile.NamedTemporaryFile(delete=False, suffix='.jpg') as temp_file:
-        temp_filename = temp_file.name
-
-        cv2.imwrite(temp_filename, orig_img, [cv2.IMWRITE_JPEG_QUALITY, quality])
-        # read compressed image
-        compressed_img = cv2.imread(temp_filename)
-
-        # get absolute difference between img1 and img2 and multiply by scale
-        diff = SCALE * cv2.absdiff(orig_img, compressed_img)
-    
-    # delete the temporary file
-    if os.path.exists(temp_filename):
-        os.remove(temp_filename)
-
-    return diff
-
-
-def calculate_error_difference(orig_img, country=None):
-    if isinstance(orig_img, Image.Image):
-        orig_img = np.array(orig_img)
-        
-    if np.any(orig_img):
-        ela_val = compute_ela_cv(orig_img, quality=94)
-        diff_avg = ela_val.mean()
-
-        print(f"DIFFERENCE: {diff_avg}")
-        if country == 'UAE':
-            if diff_avg <= 25:
-                label = 'Genuine'
-            else:
-                label = 'Tampered'
-        else:
-            if diff_avg <= 10.5:
-                label = 'Genuine'
-            else:
-                label = 'Tampered'
-                
-        return label
-    else:
-        print(f"ISSUE")
-        return 'Genuine'
-
-
-def eastern_arabic_to_english(eastern_numeral):
-    try:
-        arabic_to_english_map = {
-            '٠': '0', '۰': '0',
-            '١': '1', '۱': '1',
-            '٢': '2', '۲': '2',
-            '٣': '3', '۳': '3',
-            '٤': '4', '۴': '4',
-            '٥': '5', '۵': '5',
-            '٦': '6', '۶': '6',
-            '٧': '7', '۷': '7',
-            '٨': '8', '۸': '8',
-            '٩': '9', '۹': '9',
-            '/': '/'
-        }
-
-        english_numeral = ''.join([arabic_to_english_map[char] if char in arabic_to_english_map else char for char in eastern_numeral])
-        
-        return english_numeral
-
-    except:
-        return eastern_numeral
-
-def english_to_eastern_arabic(english_numeral):
-    try:
-        english_to_arabic_map = {
-            '0': '٠', 
-            '1': '١', 
-            '2': '٢', 
-            '3': '٣', 
-            '4': '٤', 
-            '5': '٥', 
-            '6': '٦', 
-            '7': '٧', 
-            '8': '٨', 
-            '9': '٩'
-        }
-
-        eastern_arabic_numeral = ''.join([english_to_arabic_map[char] if char in english_to_arabic_map else char for char in english_numeral])
-        
-        return eastern_arabic_numeral
-
-    except Exception as e:
-        return str(e)  
-
-def crop_third_part(img):
-    width, height = img.size
-    part_height = height // 3
-    third_part = img.crop((0, 2 * part_height, width, height))
-    # third_part.save("/Users/fahadpatel/Pictures/thirdpart.jpg")
-    return third_part
-
-def extract_text_from_image_data(client, image):
-    """Detects text in the file."""
-
-    # with io.BytesIO() as output:
-    #     image.save(output, format="PNG")
-    #     content = output.getvalue()
-
-    compressed_image = BytesIO()
-    image.save(compressed_image, format="JPEG", quality=100, optimize=True)
-    content = compressed_image.getvalue()
-
-    image = vision_v1.types.Image(content=content)
-
-    response = client.text_detection(image=image)
-    texts = response.text_annotations
-    
-    return texts[0].description
-
-def detect_id_card_uae(client, image_data, id_text, part=None):
-    if id_text:
-        vertices = id_text[0].bounding_poly.vertices
-        left = vertices[0].x
-        top = vertices[0].y
-        right = vertices[2].x
-        bottom = vertices[2].y
-
-        padding = 30
-        left -= padding
-        top -= padding
-        right += padding
-        bottom += padding
-
-        # img = image_data
-
-        with Image.open(io.BytesIO(image_data)) as img:
-            id_card = img.crop((max(0, left), max(0, top), right, bottom))
-            width, height = id_card.size
-            if width < height:
-                id_card = id_card.rotate(90, expand=True)
-            
-            tampered_result = calculate_error_difference(id_card, country = 'UAE')
-
-            part_text = id_text[0].description
-            if part == 'third':
-                part_img = crop_third_part(id_card)
-                part_text = extract_text_from_image_data(client, part_img)
-
-            return  tampered_result, part_text
-
-def rotate_image(img):
-        from skimage.transform import radon
-        
-        img_array = np.array(img)
-
-        if len(img_array.shape) == 2:
-            gray = img_array
-        else:
-            gray = cv2.cvtColor(img_array, cv2.COLOR_RGB2GRAY)
-
-        h, w = gray.shape
-        if w > 640:
-            gray = cv2.resize(gray, (640, int((h / w) * 640)))
-        gray = gray - np.mean(gray)
-        sinogram = radon(gray)
-        r = np.array([np.sqrt(np.mean(np.abs(line) ** 2)) for line in sinogram.transpose()])
-        rotation = np.argmax(r)
-        angle = round(abs(90 - rotation) + 0.5)
-
-        if abs(angle) > 5:
-            rotated_img = img.rotate(angle, expand=True)
-            return rotated_img
-
-        return img
-
 def deepface_to_dlib_rgb(face):
     """
     Convert DeepFace face output to uint8 RGB for face_recognition
@@ -632,7 +80,8 @@ def load_and_process_image_deepface_topup(image_input):
                 return face_objs, img_to_process, confidence
 
             # Clear memory if no face found
-            del img_to_process
+            if 'img_to_process' in locals():
+                del img_to_process
             return None, None, 0
         except Exception as e:
             logging.info(f"Error processing angle {angle}: {e}")
@@ -775,10 +224,6 @@ def load_and_process_image_deepface_topup(image_input):
             del processed_image
 
 
-
-
-
-
 def load_and_process_image_deepface(image_input, country=None):
     DeepFace = get_deepface()
     face_recognition = get_face_recognition()
@@ -884,7 +329,6 @@ def load_and_process_image_deepface(image_input, country=None):
             print("Empty image input")
             return [], []
 
-
         # -------------------- ANGLE LOOP (NO THREADS) --------------------
 
         best_face_objs = None
@@ -907,8 +351,6 @@ def load_and_process_image_deepface(image_input, country=None):
                 break  # Exit loop on first valid detection
 
             # Keep best fallback (just in case)
-            
-
 
         if best_face_objs is None or best_confidence < CONFIDENCE_THRESHOLD:
             print(f"No valid face found (threshold={CONFIDENCE_THRESHOLD})")
@@ -965,403 +407,3 @@ def extract_face_and_compute_similarity(front_face_locations, front_face_encodin
 
     return min(1, similarity_score)
 
-def load_and_process_image_deepface_all_orientations(image_input):
-    """Similar to load_and_process_image_deepface but processes all orientations to find best confidence"""
-    DeepFace = get_deepface()  # Only load when needed
-    face_recognition = get_face_recognition()  # Only load when needed
-    def process_angle(img, angle):
-        try:
-            # Create a view instead of copy when possible
-            if angle != 0:
-                # Minimize memory usage during rotation
-                with np.errstate(all='ignore'):
-                    img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
-                    img_pil = Image.fromarray(img_rgb)
-                    # Use existing buffer when possible
-                    rotated = np.ascontiguousarray(img_pil.rotate(angle, expand=True))
-                    img_to_process = cv2.cvtColor(rotated, cv2.COLOR_RGB2BGR)
-                    # Clear references to intermediate arrays
-                    del img_rgb, img_pil, rotated
-            else:
-                img_to_process = img
-
-            # Extract faces with memory optimization
-            face_objs = DeepFace.extract_faces(
-                img_to_process,
-                detector_backend='fastmtcnn',
-                enforce_detection=False,
-                align=True
-            )
-
-            if face_objs and len(face_objs) > 0:
-                confidence = face_objs[0].get('confidence', 0)
-                # print(f"Face detected at {angle} degrees with confidence {confidence}")
-
-                return face_objs, img_to_process, confidence
-
-            # Clear memory if no face found
-            del img_to_process
-            return None, None, 0
-        except Exception as e:
-            print(f"Error processing angle {angle}: {e}")
-            return None, None, 0
-        finally:
-            # Ensure memory is cleared
-            if 'img_to_process' in locals():
-                del img_to_process
-
-    try:
-        # Process input image efficiently
-        if isinstance(image_input, np.ndarray):
-            # Use view when possible
-            image = np.ascontiguousarray(image_input)
-            if image.dtype != np.uint8:
-                image = image.astype(np.uint8, copy=False)
-        elif isinstance(image_input, str):
-            # Decode base64 directly to numpy array
-            image_data = base64.b64decode(image_input)
-            image = cv2.imdecode(np.frombuffer(image_data, np.uint8), cv2.IMREAD_COLOR)
-            del image_data  # Clear decoded data
-        else:
-            print(f"Unexpected input type: {type(image_input)}")
-            return [], []
-
-        if image is None or image.size == 0:
-            print("Empty image")
-            return [], []
-
-        # Process all angles in parallel
-        angles = [0, 90, 180, 270]
-        best_confidence = 0
-        best_face_objs = None
-        best_image = None
-
-        # Use context manager to ensure proper cleanup
-        with ThreadPoolExecutor(max_workers=4) as executor:
-            # Submit tasks
-            futures = {
-                executor.submit(process_angle, image, angle): angle
-                for angle in angles
-            }
-
-            try:
-                for future in as_completed(futures):
-                    face_objs, processed_image, confidence = future.result()
-                    if face_objs is not None and confidence > best_confidence:
-                        best_confidence = confidence
-                        best_face_objs = face_objs
-                        best_image = processed_image
-            finally:
-                # Ensure all futures are cancelled
-                for future in futures:
-                    if not future.done():
-                        future.cancel()
-
-        if best_face_objs is None:
-            print("No faces detected with fastmtcnn at any angle")
-            return [], []
-
-        # print(f"Using best detected face with confidence {best_confidence}")
-        try:
-            biggest_face = max(best_face_objs, key=lambda face: face['facial_area']['w'] * face['facial_area']['h'])
-            facial_area = biggest_face['facial_area']
-            x, y, w, h = facial_area['x'], facial_area['y'], facial_area['w'], facial_area['h']
-
-            # Minimize memory during final processing
-            image_rgb = cv2.cvtColor(best_image, cv2.COLOR_BGR2RGB)
-            face_locations = [(y, x + w, y + h, x)]
-            face_encodings = face_recognition.face_encodings(image_rgb, face_locations)
-
-            if face_encodings:
-                return face_locations, face_encodings
-
-            print("Failed to extract face encodings")
-            return [], []
-        finally:
-            # Clear final processing memory
-            del image_rgb, best_image, best_face_objs
-
-    except Exception as e:
-        print(f"Error in face detection: {e}")
-        return [], []
-    finally:
-        # Ensure main image is cleared
-        if 'image' in locals():
-            del image
-
-
-def load_and_process_image_deepface_all_orientations(image_input):
-    """Similar to load_and_process_image_deepface but processes all orientations to find best confidence"""
-    DeepFace = get_deepface()  # Only load when needed
-    face_recognition = get_face_recognition()  # Only load when needed
-    def process_angle(img, angle):
-        try:
-            # Create a view instead of copy when possible
-            if angle != 0:
-                # Minimize memory usage during rotation
-                with np.errstate(all='ignore'):
-                    img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
-                    img_pil = Image.fromarray(img_rgb)
-                    # Use existing buffer when possible
-                    rotated = np.ascontiguousarray(img_pil.rotate(angle, expand=True))
-                    img_to_process = cv2.cvtColor(rotated, cv2.COLOR_RGB2BGR)
-                    # Clear references to intermediate arrays
-                    del img_rgb, img_pil, rotated
-            else:
-                img_to_process = img
-
-            # Extract faces with memory optimization
-            face_objs = DeepFace.extract_faces(
-                img_to_process,
-                detector_backend='fastmtcnn',
-                enforce_detection=False,
-                align=True
-            )
-
-            if face_objs and len(face_objs) > 0:
-                confidence = face_objs[0].get('confidence', 0)
-                # print(f"Face detected at {angle} degrees with confidence {confidence}")
-
-                return face_objs, img_to_process, confidence
-
-            # Clear memory if no face found
-            del img_to_process
-            return None, None, 0
-        except Exception as e:
-            print(f"Error processing angle {angle}: {e}")
-            return None, None, 0
-        finally:
-            # Ensure memory is cleared
-            if 'img_to_process' in locals():
-                del img_to_process
-
-    try:
-        # Process input image efficiently
-        if isinstance(image_input, np.ndarray):
-            # Use view when possible
-            image = np.ascontiguousarray(image_input)
-            if image.dtype != np.uint8:
-                image = image.astype(np.uint8, copy=False)
-        elif isinstance(image_input, str):
-            # Decode base64 directly to numpy array
-            image_data = base64.b64decode(image_input)
-            image = cv2.imdecode(np.frombuffer(image_data, np.uint8), cv2.IMREAD_COLOR)
-            del image_data  # Clear decoded data
-        else:
-            print(f"Unexpected input type: {type(image_input)}")
-            return [], []
-
-        if image is None or image.size == 0:
-            print("Empty image")
-            return [], []
-
-        # Process all angles in parallel
-        angles = [0, 90, 180, 270]
-        best_confidence = 0
-        best_face_objs = None
-        best_image = None
-
-        # Use context manager to ensure proper cleanup
-        with ThreadPoolExecutor(max_workers=4) as executor:
-            # Submit tasks
-            futures = {
-                executor.submit(process_angle, image, angle): angle
-                for angle in angles
-            }
-
-            try:
-                for future in as_completed(futures):
-                    face_objs, processed_image, confidence = future.result()
-                    if face_objs is not None and confidence > best_confidence:
-                        best_confidence = confidence
-                        best_face_objs = face_objs
-                        best_image = processed_image
-            finally:
-                # Ensure all futures are cancelled
-                for future in futures:
-                    if not future.done():
-                        future.cancel()
-
-        if best_face_objs is None:
-            print("No faces detected with fastmtcnn at any angle")
-            return [], []
-
-        # print(f"Using best detected face with confidence {best_confidence}")
-        try:
-            biggest_face = max(best_face_objs, key=lambda face: face['facial_area']['w'] * face['facial_area']['h'])
-            facial_area = biggest_face['facial_area']
-            x, y, w, h = facial_area['x'], facial_area['y'], facial_area['w'], facial_area['h']
-
-            # Minimize memory during final processing
-            image_rgb = cv2.cvtColor(best_image, cv2.COLOR_BGR2RGB)
-            face_locations = [(y, x + w, y + h, x)]
-            face_encodings = face_recognition.face_encodings(image_rgb, face_locations)
-
-            if face_encodings:
-                return face_locations, face_encodings
-
-            print("Failed to extract face encodings")
-            return [], []
-        finally:
-            # Clear final processing memory
-            del image_rgb, best_image, best_face_objs
-
-    except Exception as e:
-        print(f"Error in face detection: {e}")
-        return [], []
-    finally:
-        # Ensure main image is cleared
-        if 'image' in locals():
-            del image
-
-def get_facial_encodings_deepface_irq(image, country: str, model_name="ArcFace", detector_backend="retinaface"):
-    logging.info(f"Type: {type(image)} | Shape: {image.shape} | Dtype: {image.dtype}")
-
-    from deepface import DeepFace
-    import numpy as np
-
-    if image.ndim == 3 and image.shape[2] == 4:
-        # Convert RGBA to BGR (DeepFace expects BGR or RGB)
-        # OpenCV uses BGR by default
-        image = cv2.cvtColor(image, cv2.COLOR_BGRA2BGR)
-    elif image.ndim == 2:
-        # grayscale to BGR
-        image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
-
-    if image.dtype != np.uint8:
-        image = image.astype(np.uint8)
-
-
-    # DeepFace doesnt accepts 4 channel images (RGBA), so we convert to RGB or BGR
-    try:
-        logging.info(f"[DeepFace] Starting face extraction for {country} with model {model_name} and detector {detector_backend}")
-
-        face_objs = DeepFace.represent(
-            img_path=image,
-            model_name=model_name,
-            detector_backend=detector_backend,
-            enforce_detection=False,
-            align=True
-        )
-
-        logging.info(f"[DeepFace] Face extraction successful: {len(face_objs)} faces detected.")
-        # logging.info(f"[DeepFace] Face details: {face_objs}")
-    except Exception as e:
-        logging.error(f"[DeepFace] Face extraction failed: {e}")
-        return [],[]
-
-    if not face_objs:
-        logging.warning("No faces detected.")
-        return [], []
-
-    # Normalize to list
-    if isinstance(face_objs, dict):
-        face_objs = [face_objs]
-
-    # Select face with largest area
-    def get_area(face):
-        fa = face.get("facial_area", {})
-        if "w" in fa and "h" in fa:
-            return fa["w"] * fa["h"]
-        elif {"x", "x2", "y", "y2"}.issubset(fa):
-            return (fa["x2"] - fa["x"]) * (fa["y2"] - fa["y"])
-        return 0
-
-    largest_face = max(face_objs, key=get_area)
-    facial_area = largest_face.get("facial_area", {})
-    confidence = largest_face.get("face_confidence", 0.0)
-    embedding = np.array(largest_face["embedding"], dtype=np.float32)
-
-    # Compute w, h from bounding box
-    w, h = 0, 0
-    if "w" in facial_area and "h" in facial_area:
-        w, h = facial_area["w"], facial_area["h"]
-    elif {"x", "x2", "y", "y2"}.issubset(facial_area): #other libraries, models returns x2,y2 instead of w,h
-        w = facial_area["x2"] - facial_area["x"]
-        h = facial_area["y2"] - facial_area["y"]
-
-    # Country-specific rejection
-    if country.upper() == "SDN":
-        if w < 40 or h < 50:
-            logging.warning(f"Rejected: SDN face too small ({w}x{h}), minimum 40x50")
-            return [], []
-
-    if confidence < 0.95:
-        logging.warning(f"Rejected: low face detection confidence ({confidence:.3f})")
-        return [], []
-
-    facial_area_filtered = {}
-    x = facial_area.get("x", 0)
-    y = facial_area.get("y", 0)
-    w = facial_area.get("w", 0)
-    h = facial_area.get("h", 0)
-
-    facial_area_filtered = [(x,y,w,h)]
-    return facial_area_filtered, [embedding]
-
-
-def cosine_similarity(vec1, vec2):
-    # vec1 = np.array(vec1)
-    # vec2 = np.array(vec2)
-    #
-    # if vec1.shape != vec2.shape:
-    #     raise ValueError(f"Shape mismatch: {vec1.shape} vs {vec2.shape}")
-    #
-    # similarity = dot(vec1, vec2) / (norm(vec1) * norm(vec2))
-    # return round(float(similarity), 4)
-    similarity = 1 - cosine(vec1, vec2)
-    similarity = round(similarity, 5)
-    similarity = min(1, similarity)
-    return similarity
-
-def extract_face_and_compute_cosine_similarity(selfie, front_face_locations, front_face_encodings):
-        try:
-            if selfie is None:
-                print("Error: Selfie image is None")
-                return 0
-
-            # Ensure the input array is contiguous and in the correct format
-            if not selfie.flags['C_CONTIGUOUS']:
-                selfie = np.ascontiguousarray(selfie)
-
-            # Convert array to uint8 if needed
-            if selfie.dtype != np.uint8:
-                if selfie.max() > 255:
-                    selfie = (selfie / 256).astype(np.uint8)
-                else:
-                    selfie = selfie.astype(np.uint8)
-
-            # Try DeepFace first as it's generally more reliable
-            # start_time = time.time()
-            face_locations1, face_encodings1 = get_facial_encodings_deepface_irq(selfie, country='IRQ')
-            # end_time = time.time()
-
-            if not face_locations1 or not face_encodings1:
-                print("No face detected in Selfie Video by DeepFace")
-                return 0
-
-            # print(f"Face detection took {end_time - start_time:.3f} seconds")
-            face_locations2, face_encodings2 = front_face_locations, front_face_encodings
-
-            if not face_encodings2.any():
-                print('No face detected in front ID')
-                return 0
-
-            largest_face_index1 = face_locations1.index(
-                max(face_locations1, key=lambda loc: (loc[2] - loc[0]) * (loc[3] - loc[1])))
-            largest_face_index2 = face_locations2.index(
-                max(face_locations2, key=lambda loc: (loc[2] - loc[0]) * (loc[3] - loc[1])))
-
-            face_encoding1 = face_encodings1[largest_face_index1]
-            face_encoding2 = face_encodings2[largest_face_index2]
-
-            similarity_score = cosine_similarity(face_encoding1, face_encoding2)
-            # print(f"Calculated similarity score: {similarity_score}")
-
-            return similarity_score
-
-        except Exception as e:
-            print(f"Error in extract_face_and_compute_similarity: {e}")
-            import traceback
-            traceback.print_exc()
-            return 0