pokemons.py

from collections import defaultdict

_STARTLINE = r'<div class="infocard-filter-block">'
_EVO_STARTLINE = '<div class="infocard "><span class="infocard-lg-img">'

def _remove_a(s: str):
    while "<a" in s:
        start = s.index("<a")
        end = s.index(">", start) + 1
        sl = list(s)
        sl[start:end] = []
        s = ''.join(sl)
    return s.replace("</a>", "")

def _proc_evo_line(evo_line: str):
    res = defaultdict(list)
    assert evo_line.startswith('<div class="infocard-list-evo">'), evo_line
    evo_line = evo_line.replace('div class="infocard-list-evo">', '')

    chain = evo_line.split(_EVO_STARTLINE)[1:]
    for pokemon in chain:
        if '<i class="icon-arrow icon-arrow-e"></i><small>' not in pokemon:
            continue
        name = pokemon[:pokemon.find("</a><br>")]
        name = name.rsplit(">", 1)[-1]
        evolve_method = pokemon.split('<i class="icon-arrow icon-arrow-e"></i><small>')[-1]
        evolve_method = _remove_a(evolve_method)
        evolve_method = evolve_method.split("<", 1)[0].replace("\n", " ")
        while "  " in evolve_method:
            evolve_method = evolve_method.replace("  ", " ")

        res[name].append(evolve_method)

    return res

def _process_method(method:str):
    assert method.startswith("(") and method.endswith(")"), method
    method = method[1:-1]
    if "Level" in method:
        idx = method.index("Level") + len("Level")
        level_remainder = ''.join([c for c in method[idx+1:idx+3] if not c.isnumeric()])
        lmethod = list(method)
        lmethod[method.index("Level"):idx+3] = f"Evolves with level{level_remainder}"
        method = "".join(lmethod)
    return method

def get_pokemon_evolutions():
    pokemons = defaultdict(list)

    # load html
    with open("pokemondb.net evolution.html", "r", encoding='utf-8') as f:
        html = f.read()

    # all evo line strings
    evo_lines = [i.strip() for i in html.split(_STARTLINE)][1:]

    for evo_line in evo_lines:
        pokemons.update(_proc_evo_line(evo_line))

    proc = {}
    name_map = {}
    for pokemon, methods in pokemons.items():
        proc[pokemon.lower().strip()] = [_process_method(m) for m in methods]
        name_map[pokemon.lower().strip()] = pokemon
    return proc, name_map