Retrieve all rewrite rules in goto.tm .htaccess

goto_rewrite_rules.py

import fileinput
import datetime
import pymongo
import time
import re

from pymongo import MongoClient

content = []
new_rules = []
search_string = ''
ts = time.time()
isodate = datetime.datetime.fromtimestamp(ts, None)

for line in fileinput.input():
    # RewriteRule ^FreedomSteam?$ http://steamcommunity.com/groups/MCNFDM/ [NC,L,R=302]
    if "RewriteRule" in line:
        search_string = re.search('(\^(.+)(?=\$))(\$ )(.+(?= \[))', line.strip())
        # FreedomSteam?$ http://steamcommunity.com/groups/MCNFDM/
        if search_string:
            short_url = re.sub('\W(?=\?|\n)', '', search_string.group(2) + "\n")
            content.append([short_url.replace('', '')[:-1], search_string.group(4)])
        else:
            print(line.strip())

fileinput.close()

for rule in content:
    new_rules.append({
                "long_url": rule[1],
                "short_url": rule[0],
                "updated_at": isodate,
                "created_at": isodate
            })

print('Records to insert: ' + str(len(new_rules)))
client = MongoClient('localhost', 27017)

db = client.gototm

result = db.redirect_rules.insert_many(new_rules)

print('Records inserted: ' + str(db.redirect_rules.count()))