parse_access_apache_log_txt

access.log

127.0.0.1 - - [01/Apr/2022:18:19:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"
127.0.0.1 - - [01/Apr/2022:18:19:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"
127.0.0.1 - - [01/Apr/2022:18:20:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"
127.0.0.1 - - [01/Apr/2022:18:20:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"
::1 - - [01/Apr/2022:18:22:10 +1000] "GET /phpmyadmin/ HTTP/1.1" 200 18361 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:13 +1000] "GET /phpmyadmin/themes/pmahomme/css/theme.css?v=5.1.1&nocache=2250110580ltr&server=1 HTTP/1.1" 200 214799 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:14 +1000] "GET /phpmyadmin/js/messages.php?l=ru&v=5.1.1 HTTP/1.1" 200 12908 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:15 +1000] "POST /phpmyadmin/index.php?route=/navigation&ajax_request=1 HTTP/1.1" 200 2898 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:15 +1000] "POST /phpmyadmin/index.php?route=/config/get HTTP/1.1" 200 1855 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:15 +1000] "POST /phpmyadmin/index.php?route=/config/get HTTP/1.1" 200 1764 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:15 +1000] "POST /phpmyadmin/index.php?route=/version-check HTTP/1.1" 200 64 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:15 +1000] "POST /phpmyadmin/index.php?route=/database/structure/favorite-table&ajax_request=1&favorite_table=1&sync_favorite_tables=1 HTTP/1.1" 200 192 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
::1 - - [01/Apr/2022:18:22:16 +1000] "POST /phpmyadmin/index.php?route=/config/set HTTP/1.1" 200 1756 "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/99.0.4844.74 Safari/537.36 Edg/99.0.1150.55"
127.0.0.1 - - [01/Apr/2022:18:25:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"
127.0.0.1 - - [01/Apr/2022:18:25:15 +1000] "\x16\x03\x01\x02" 400 326 "-" "-"

main.py

import json
from datetime import datetime
import re
from file_read_backwards import FileReadBackwards
from threading import Thread
from rapidfuzz import fuzz


class ParserAccessLog(Thread):

    def __init__(self, path_file):
        super().__init__()
        self.path_file = path_file
        self.new_lines = []
        self.parse_logs = []
        self.lines = []

    def _read_file(self):
        self.parse_logs = []
        try:
            with FileReadBackwards(self.path_file, encoding="utf-8") as frb:
                self.new_lines = []
                for idx, l in enumerate(frb):
                    if len(self.lines) == 0:
                        self.new_lines.append(l)
                    else:
                        try:
                            if fuzz.ratio(l, self.lines[0]) == 100:
                                break
                            else:
                                self.new_lines.append(l)
                        except IndexError:
                            break
                if len(self.new_lines) == 0:
                    return
                else:
                    return self._parse_file()

        except PermissionError or IndexError:
            return

    def _parse_file(self):
        for el in self.new_lines:
            ip = re.findall(r"^\d+[.]\d+[.]\d+", el)
            if len(ip) == 0:
                ip = re.findall(r"^\d+[:]\d+[:]\d+", el)
                if len(ip) == 0:
                    ip = re.findall(r"::\d+", el)
            date = str(datetime.strptime(re.findall(r"\d+[/]\D+[/]\d+[:]\d+[:]\d+[:]\d+", el)[-1], '%d/%b/%Y:%H:%M:%S'))
            data = re.split(r"] ", el)[-1]
            if len(ip) == 1:
                data = data.replace('\"', '')
                self.parse_logs.append({"ip": ip[-1], "date": date, "data": data})
        for idx, el in enumerate(self.new_lines):
            self.lines.insert(idx, el)

        return json.dumps({"data": self.parse_logs})

    def _update_json(self):
        while True:
            json = self._read_file()
            if json is not None:
                print(json)

    def run(self):
        self._update_json()


if __name__ == '__main__':
    thread1 = ParserAccessLog(r"access.log")
    thread1.start()

print('основной поток')