src/rule_based_system/UrlRule.py

from src.rule_based_system.Rule import Rule
from src.rule_based_system.Verdict import Verdict


class UrlRule(Rule):

    url_extractor = None

    def __init__(self, url_extractor):
        self.url_extractor = url_extractor

    def get_verdict(self, comment_text: str) -> Verdict:
        urls = self.find_urls(comment_text)

        return Verdict(len(urls) == 0, urls)

    def find_urls(self, text: str) -> list:
        urls = self.url_extractor.find_urls(text)

        # url_extractor does not find url with spaces. Add extra check for urls like http:// goatse info
        if len(urls) == 0 and 'http' in text:
            urls = ['http']

        return urls

    def is_strict(self) -> bool:
        return True

    @staticmethod
    def get_rule_description() -> str:
        return 'Url was mentioned in text'