gen_entities.py

#!/usr/bin/env python3

"""
Converts entity WHATWG HTML5 entity summary into
Text.HTML.Parser.Entities module.

Usage:

    curl https://html.spec.whatwg.org/entities.json | python3 gen_entities.py > src/Text/HTML/Parser/Entities.hs

"""

import sys
import json

entities = json.load(sys.stdin)

def escape(details):
    return ''.join(f'\\x{cp:04x}' for cp in details['codepoints'])

print('''
{-# LANGUAGE OverloadedStrings #-}

-- | N.B. This file is generated by @gen_entities.py@. Do not edit.
module Text.HTML.Parser.Entities (entities) where

import Data.Text (Text)

entities :: [(Text, Text)]
entities = [
'''.strip())
print(',\n'.join(
    f'''  ("{name[1:-1]}", "{escape(details)}")'''
    for name, details in entities.items()
))
print('  ]')