img_dnl_2_download

img_dnl_2_download.py

# STEP 1
BASE_DIR = 'images/'

# STEP 2
SUB_DIRS = [topic + '/' for topic in df['Topic'].unique()]

# Print a message to the user
print('Image Download Started...')
start_time = datetime.datetime.now()

# STEP 3
for sub_dir in SUB_DIRS:
    if not os.path.exists(BASE_DIR + sub_dir):
        os.makedirs(BASE_DIR + sub_dir)
    
# STEP 4
for topic in SUB_DIRS:
    # filtering based on topic values
    for ind, row in df[df['Topic'] == topic.split('/')[0]].iterrows():
        
        # STEP 5
        urllib.request.urlretrieve(
            row['Thumbnail'],
            '{}{}.jpg'.format(
                BASE_DIR + topic,
                (row['Title']
                     # removing unwanted characters
                    .lower()
                    .replace(' ', '_')
                    .replace('.', '')
                    .replace(',', '')
                    .replace(':', '')
                    .replace('\'', '')
                    .replace('’', '')
                    .replace('#', '')
                    .replace('*', ''))
            )
        )
       
# Tell the user that download has finished
end_time = datetime.datetime.now()
print('\tDownload Finished! It took {} seconds.'.format(int((end_time - start_time).total_seconds())))