src/rule_based_system/HTMLRule.py

from bs4 import BeautifulSoup

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


class HTMLRule(Rule):

    def get_verdict(self, comment_text: str) -> Verdict:
        html = self.find_html(comment_text)

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

    @staticmethod
    def find_html(text: str) -> list:
        html = BeautifulSoup(text, "html.parser").find_all()

        return [str(tag) for tag in html]

    def is_strict(self) -> bool:
        """
        This rule occasionally removes appropriate comments when names are enclosed in triangular brackets e.g. <name>
        """
        return False

    def get_rule_description(self) -> str:
        return 'HTML used in comment text'