__author__ = 'Ray'

import io
import re
from mediawiki import get_wikipage_text

def get_descendants(title):
    """Get and parse the descendant section from en.wiktionary.org for a specific title (usually word)"""
    wikipage_text_string = get_wikipage_text(title)
    # ET.dump(root)
    # print(wikipage_text_string)
    match = re.search(r"\=+Descendants\=+", wikipage_text_string)
    descendants_string = wikipage_text_string[match.end():-1]
    descendant_words = []
    # {{l|ang|hungor}} start_tag = "{{" end_tag = "}}" parameter_tag = "|"
    descendants_buffer = io.StringIO(descendants_string)
    for line in descendants_buffer:
        matches = re.findall(r"(?<={{).*?(?=}})",
                             line)  # find all matches for every string between {{ and }} in a non-greedy manner
        for match in matches:
            match_split = match.split("|")
            if (match_split[0] == "l"):
                descendant_words.append({"language": match_split[1], "lemma": match_split[2]})
    return descendant_words


def get_PG_decendants(word):
    """Get and parse the descendant section from en.wiktionary.org for a specific gem-pro word"""
    pg_prefix = "Reconstruction:Proto-Germanic/"
    return get_descendants(pg_prefix + word)