58 lines
2.3 KiB
Raw Normal View History

2021-08-16 18:23:04 -04:00
#!/usr/bin/env python3
This mirror status checker determines whether CSC mirror is up-to-date with upstream
import requests
from arch import Arch
from debian import Debian
from kernel import Kernel
from openbsd import OpenBSD
from dateparser.search import search_dates # this library seems to be super slow but the other library: dateutil.parser gets some errors
# http://theautomatic.net/2018/12/18/2-packages-for-extracting-dates-from-a-string-of-text-in-python/
import re # import regular expressions to remove stray numbers in string that might interfere with date finding
import json # import json to read distro info stored in json file
import datefinder # another date finding library
def checker(directory_URL, file_name):
page = requests.get(directory_URL).text
indexOfFile = page.find(file_name)
# print(page)
# remove stray numbers (file size numbers in particular) that might interfere with date finding
segment_clean = re.sub(r'\s\d+\s', ' ', page[indexOfFile:]) # removes numbers for size
segment_clean = re.sub(r'\s\d+\w*\s', ' ', page[indexOfFile:]) # removes numbers + size unit. e.x. 50kb
# print(segment_clean)
# implementation using dateparser.search.search_dates
# notes: some dates don't parse correctly with this tool
# print(search_dates(page[indexOfFile:], languages=['en']))
# print(search_dates(page[indexOfFile:])[0])
# finds the dates in the segment after the file name
# notes: a generator will be returned by the datefinder module. I'm typecasting it to a list. Please read the note of caution provided at the bottom.
matches = list(datefinder.find_dates(segment_clean))
# print(matches)
if len(matches) > 0:
date = matches[0] # date is of type datetime.datetime
return(date.strftime("%m/%d/%Y, %H:%M:%S"))
return('No dates found')
2021-08-19 17:00:55 -04:00
2021-08-16 18:23:04 -04:00
if __name__ == "__main__":
"""for distro in [Arch, Debian, Kernel, OpenBSD]:
except requests.exceptions.RequestException as err:
print(f"Error: {distro.name()}\n{err}")"""
distros = json.load(open('distros.json',))
for distro in distros:
print(distro[0] + ":")
print("CSC mirror: " + checker(distro[1], distro[3]))
print("Official distro: " + checker(distro[2], distro[3]))