lukas
/
shovel
1
0
Fork 0
Code Issues 1 Pull Requests Releases Wiki Activity

Update formatting

This commit is contained in:
Lukáš Kucharczyk 2021-03-18 17:47:37 +01:00
parent 1ffe83e493
commit bf65cc41e7
No known key found for this signature in database
GPG Key ID: 65524498C0196B64
1 changed files with 23 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
main.py
View File

@ -6,12 +6,16 @@ from collections import Counter
from bs4 import BeautifulSoup
if __name__ == '__main__':
parser = argparse.ArgumentParser(description="Dig out links from a website.")
parser.add_argument('site', type=str, help="Website that you want to scrape for links.")
parser = argparse.ArgumentParser(
description="Dig out links from a website.")
parser.add_argument(
'site', type=str, help="Website that you want to scrape for links.")
parser.add_argument('-f', '--filter', type=str, default=r'..*$',
help="Only return file names matching this regular expression.")
parser.add_argument('-s', '--summary', action='store_true', help="Print summary and quit.")
parser.add_argument('-d', '--debug', action='store_true', help="Print out debugging information.")
parser.add_argument('-s', '--summary', action='store_true',
help="Print summary and quit.")
parser.add_argument('-d', '--debug', action='store_true',
help="Print out debugging information.")
parser.add_argument('-r', '--save-report', action='store_true',
help='Save report of all the original links to a file.')
@ -40,7 +44,8 @@ if __name__ == '__main__':
has_jpeg = True
extensions.append(extension)
if args.debug:
print(f'Original: {link}, splitext[0]: {filename}, splitext[1]: {extension}')
print(
f'Original: {link}, splitext[0]: {filename}, splitext[1]: {extension}')
# todo: merge jpeg and jpg into one extension (display as: xxx jpg (yyy as jpeg))
print(f'URL: {args.site}')
print(f'Total file names: {len(links)}')