Flask Config Documentation to Protobuf Message

flask2proto.py

from io import BytesIO
import re
import functools

import requests
from requests.adapters import HTTPAdapter
import lxml.etree


class HTML:
    """Utilities for interacting with HTML documents"""

    @staticmethod
    def iter(content):
        """Generates each element in the DOM of the HTML one element at a
        time. Will recover on failure. Unparsable documents become a
        single paragraph element containing the text of the content.
        """

        context = lxml.etree.iterparse(BytesIO(content), recover=True, html=True)

        for action, elem in context:
            yield elem


class HTTP:
    _HTTP_TIMEOUT = 2.0 # seconds

    """HTTP constants and verb functions"""
    _session = requests.Session()
    _session.mount('http://', HTTPAdapter(max_retries=0))
    _session.mount('https://', HTTPAdapter(max_retries=0))

    #: The good status code for HTTP responses
    ok = 200

     #: Asyncio friendly wrapper for requests.get with a timeout
    get = functools.partial(_session.get, timeout=_HTTP_TIMEOUT)


class Tag:
    """Utilities for HTML Tags"""

    @staticmethod
    def is_(tag):
        def matcher(element):
            return element.tag == tag

        return matcher

    def attr_is(**kwargs):
        assert len(kwargs) == 1

        key, value = next(iter(kwargs.items()))

        def matcher(element):
            return element.attrib.get(key) == str(value)

        return matcher

is_span = Tag.is_('span')
is_class_pre = Tag.attr_is(**{'class': 'pre'})
is_config_tag = lambda t: is_span(t) and is_class_pre(t)

tags = HTML.iter(HTTP.get('http://flask.pocoo.org/docs/latest/config/').content)
tags = filter(is_config_tag, tags)


keys = sorted(set(re.findall(r'[A-Z]{2,100}(?:(?:_?[A-Z0-9]+)*)', '|'.join(t))))

print('message FlaskConfig {')
print('\n'.join([f'    string {t.lower():<32} = {i:>3} [json_name="{t.upper()}"];' for i, t in enumerate(keys, start=1)]))
print('}')