grouping.py

import pandas as pd
import os
from PIL import Image
import glob

df = pd.read_excel('result.xlsx')

out = {
    'user_id': [],
    'type': [],
    'validity': [],
    'height': [],
    'width': [],
    'size(kb)': []
}

for idx, row in df.iterrows():
    user_id = str(int(row[0])) # user_id
    files = [f for f in [glob.glob(s) for s in ('document/'+user_id+'/jpg/*.jpg', 'document/'+user_id+'/jpg/*.jpeg')] for f in f]
    # print(files)
    for f in files:
        if 'KTP' in f:
            ktp = Image.open(f)
            ktp_size = os.path.getsize(f)
        elif 'TRANSKRIP' in f:
            transkrip = Image.open(f)
            transkrip_size = os.path.getsize(f)
        elif 'IJAZAH' in f:
            ijazah = Image.open(f)
            ijazah_size = os.path.getsize(f)

    for i in range(1,len(row)//4+1):
        out['user_id'].append(user_id)

        if i*4 == 4:
            out['type'].append('name')
            try:
                width, height = ktp.size
                size = ktp_size/1000
            except:
                print('file not found')
                width, height = (0,0)
                size = 0.0
        elif i*4 == 8:
            out['type'].append('dob')
            try:
                width, height = ktp.size
                size = ktp_size/1000
            except:
                print('file not found')
                width, height = (0,0)
                size = 0.0
        elif i*4 == 12:
            out['type'].append('nik')
            try:
                width, height = ktp.size
                size = ktp_size/1000
            except:
                print('file not found')
                width, height = (0,0)
                size = 0.0
        elif i*4 ==16:
            out['type'].append('institution')
            try:
                width, height = ijazah.size
                size = ijazah_size/1000
            except:
                print('file not found')
                width, height = (0,0)
                size = 0.0
        elif i*4 == 20:
            out['type'].append('ipk')
            try:
                width, height = transkrip.size
                size = transkrip_size/1000
            except:
                print('file not found')
                width, height = (0,0)
                size = 0.0
        
        out['validity'].append(row[i*4])
        out['height'].append(height)
        out['width'].append(width)
        out['size(kb)'].append(size)

        print('processing file')
    ktp = None
    ktp_size = None
    transkrip = None
    transkrip_size = None
    ijazah = None
    ijazah_size = None
        

output = pd.DataFrame(out)
output.to_excel('image-details.xlsx', index=False)