eulaurarien/feed_dns.py

#!/usr/bin/env python3

import argparse
import database
import logging
import sys
import typing
import multiprocessing
import enum

RecordType = enum.Enum('RecordType', 'A AAAA CNAME PTR')
Record = typing.Tuple[RecordType, int, str, str]

# select, write
FUNCTION_MAP: typing.Any = {
    RecordType.A: (
        database.Database.get_ip4,
        database.Database.set_hostname,
    ),
    RecordType.CNAME: (
        database.Database.get_domain,
        database.Database.set_hostname,
    ),
    RecordType.PTR: (
        database.Database.get_domain,
        database.Database.set_ip4address,
    ),
}


class Writer(multiprocessing.Process):
    def __init__(self,
                 recs_queue: multiprocessing.Queue,
                 index: int = 0):
        super(Writer, self).__init__()
        self.log = logging.getLogger(f'wr')
        self.recs_queue = recs_queue

    def run(self) -> None:
        self.db = database.Database()
        self.db.log = logging.getLogger(f'wr')

        self.db.enter_step('block_wait')
        block: typing.List[Record]
        for block in iter(self.recs_queue.get, None):

            record: Record
            for record in block:

                rtype, updated, name, value = record
                self.db.enter_step('feed_switch')

                select, write = FUNCTION_MAP[rtype]
                for source in select(self.db, value):
                    # write(self.db, name, updated, source=source)
                    write(self.db, name, updated)

            self.db.enter_step('block_wait')

        self.db.enter_step('end')
        self.db.save()


class Parser():
    def __init__(self,
                 buf: typing.Any,
                 recs_queue: multiprocessing.Queue,
                 block_size: int,
                 ):
        super(Parser, self).__init__()
        self.buf = buf
        self.log = logging.getLogger('pr')
        self.recs_queue = recs_queue
        self.block: typing.List[Record] = list()
        self.block_size = block_size
        self.prof = database.Profiler()
        self.prof.log = logging.getLogger('pr')

    def register(self, record: Record) -> None:
        self.prof.enter_step('register')
        self.block.append(record)
        if len(self.block) >= self.block_size:
            self.prof.enter_step('put_block')
            self.recs_queue.put(self.block)
            self.block = list()

    def run(self) -> None:
        self.consume()
        self.recs_queue.put(self.block)
        self.prof.profile()

    def consume(self) -> None:
        raise NotImplementedError


class Rapid7Parser(Parser):
    TYPES = {
        'a': RecordType.A,
        'aaaa': RecordType.AAAA,
        'cname': RecordType.CNAME,
        'ptr': RecordType.PTR,
    }

    def consume(self) -> None:
        data = dict()
        for line in self.buf:
            self.prof.enter_step('parse_rapid7')
            split = line.split('"')

            for k in range(1, 14, 4):
                key = split[k]
                val = split[k+2]
                data[key] = val

            record = (
                Rapid7Parser.TYPES[data['type']],
                int(data['timestamp']),
                data['name'],
                data['value']
            )
            self.register(record)


class DnsMassParser(Parser):
    # dnsmass --output Snrql
    # --retry REFUSED,SERVFAIL --resolvers nameservers-ipv4
    TYPES = {
        'A': (RecordType.A, -1, None),
        'AAAA': (RecordType.AAAA, -1, None),
        'CNAME': (RecordType.CNAME, -1, -1),
    }

    def consume(self) -> None:
        self.prof.enter_step('parse_dnsmass')
        timestamp = 0
        header = True
        for line in self.buf:
            line = line[:-1]
            if not line:
                header = True
                continue

            split = line.split(' ')
            try:
                if header:
                    timestamp = int(split[1])
                    header = False
                else:
                    dtype, name_offset, value_offset = \
                        DnsMassParser.TYPES[split[1]]
                    record = (
                        dtype,
                        timestamp,
                        split[0][:name_offset],
                        split[2][:value_offset],
                    )
                    self.register(record)
                    self.prof.enter_step('parse_dnsmass')
            except KeyError:
                continue


PARSERS = {
    'rapid7': Rapid7Parser,
    'dnsmass': DnsMassParser,
}

if __name__ == '__main__':

    # Parsing arguments
    log = logging.getLogger('feed_dns')
    args_parser = argparse.ArgumentParser(
        description="TODO")
    args_parser.add_argument(
        'parser',
        choices=PARSERS.keys(),
        help="TODO")
    args_parser.add_argument(
        '-i', '--input', type=argparse.FileType('r'), default=sys.stdin,
        help="TODO")
    args_parser.add_argument(
        '-j', '--workers', type=int, default=4,
        help="TODO")
    args_parser.add_argument(
        '-b', '--block-size', type=int, default=100,
        help="TODO")
    args_parser.add_argument(
        '-q', '--queue-size', type=int, default=10,
        help="TODO")
    args = args_parser.parse_args()

    recs_queue: multiprocessing.Queue = multiprocessing.Queue(
            maxsize=args.queue_size)

    writer = Writer(recs_queue)
    writer.start()

    parser = PARSERS[args.parser](args.input, recs_queue, args.block_size)
    parser.run()

    recs_queue.put(None)
    writer.join()