utils.py

import os, re


def header_info(xml_path):
    os.system("head {}".format(xml_path))


def handle_unicode_errors(txt):
    return txt.encode('utf-8', 'replace').decode()


def is_question(elem_attribs):
    if elem_attribs["PostTypeId"] is not None:
        if elem_attribs["PostTypeId"] == "1":
            return True
    return False


def is_answer(elem_attribs):
    if elem_attribs["PostTypeId"] is not None:
        if elem_attribs["PostTypeId"] == "2":
            return True
    return False


def filter_newlines(text):
    return re.sub("\n{3,}", "\n\n", text)


def is_accepted_answer(a_attribs, q_attribs):
    assert is_question(q_attribs), "Must be a question to have an accepted answer"
    assert is_answer(a_attribs), "Must be an answer to be an accepted answer"
    if q_attribs["AcceptedAnswerId"] is not None:
        if q_attribs["AcceptedAnswerId"] == a_attribs["Id"]:
            return True
    else:
        return False


def has_answers(elem_attribs):
    assert is_question(elem_attribs), "Must be a question to have answers"
    if elem_attribs["AnswerCount"] is not None:
        if int(elem_attribs["AnswerCount"]):
            return True
    return False


def trim_attribs(elem_attribs, attrib_type="question"):
    """deletes non-useful data from attribs dict for questions / answers, returns remaining"""
    if attrib_type == "question":
        to_keep = ['Id', 'Body', 'Title', 'Tags', 'AnswerCount', 'AcceptedAnswerId', 'PostTypeId']
        to_delete = [x for x in elem_attribs.keys() if x not in to_keep]
        [elem_attribs.pop(x, None) for x in to_delete]
        elem_attribs["ParsedAnswers"] = 0
        elem_attribs["Answers"] = {}
    elif attrib_type == "answer":
        to_keep = ['Id', 'Body', 'Score']
        new_dict = {}
        for item in to_keep:
            new_dict[item] = elem_attribs[item]
        return new_dict
    else:
        raise Exception('Unrecognized attribute type - please specify either question or answer')