personal-data/personal_data/data.py

import dataclasses
import requests
from enum import Enum
import abc

class DeduplicateMode(Enum):
    NONE = 0
    BY_FIRST_COLUMN = 1
    BY_ALL_COLUMNS = 2

@dataclasses.dataclass(frozen = True)
class Scraper(abc.ABC):
    session: requests.Session

    @staticmethod
    @property
    @abc.abstractmethod
    def dataset_name(self) -> str:
        pass

    @staticmethod
    @property
    @abc.abstractmethod
    def deduplicate_mode(self) -> DeduplicateMode:
        pass

    @staticmethod
    @property
    def dataset_format(self) -> str:
        return 'list-of-dicts'

    @abc.abstractmethod
    def scrape(self):
        pass