Image Download using Python3

image_download.py

import os
import shutil 

import pandas as pd 
import requests


datadir = '/your/path/to/url_csv/file/ulfile.csv'  # Change this variable and point 2 sitable csv file path

downloadDir='/your/path/to/download/directory'     # Change this variable and point 2 image download directory

if not os.path.exists(datadir):
    raise Exception(' data dir {} not found'.format(datadir))

if not os.path.exists(downloadDir):
    raise Exception(' data dir {} not found'.format(downloadDir))

target_col = ['Item #', 'Image url']
target_df = pd.read_csv(datadir, sep='^', encoding='ISO-8859-1', usecols = target_col, index_col=None)
target_df.rename(columns={'Item #': 'id', 'Image url': 'url'}, inplace=True)
target_df.drop_duplicates()

print('Dtatframe Shape : {}'.format(target_df.shape))

print('Unique Image Id Count {}'.format(target_df['id'].nunique()))

print('Unique Image URL Count {}'.format(target_df['url'].nunique()))

target_df.head()

imageCounter = 0

image_count = len(os.listdir(downloadDir))
print('Total Number Image present in Directory before downLoad {}'.format(image_count))

for index, row in target_df.iterrows():
    #print(row['url'])
    newFilefileName=str(row['id'])+'.PNG'
    download_path=os.path.join(downloadDir,newFilefileName)
    if os.path.exists(download_path):
        print('Image {} is present in Download directory')
        continue
    try:    
        r = requests.get(str(row['url']), stream = True)
        r.raw.decode_content = True    
        with open(download_path,'wb') as f:
            shutil.copyfileobj(r.raw, f)
        imageCounter = imageCounter + 1    
    except Exception as e:
        print(e)
    finally:    
        print('{} : {}'.format(index, row['id']))
    
print('Total Number Image DownLoaded {}'.format(imageCounter))
image_count = len(os.listdir(downloadDir))
print('Total Number Image present in Directory after downLoad {}'.format(image_count))