Script to convert Red's v2 trivia list format to YAML (http://www.yaml.org/).

trivialists_to_yaml.py

"""Script for converting v2 trivia list format to YAML.

Trivia lists must be .txt files placed in a subdirectory named `lists`, and
their corresponding .yaml files will be placed in a subdirectory named `yamls`.
"""
import sys
import pathlib
import yaml
from yaml.events import (NodeEvent, AliasEvent, ScalarEvent,
                         CollectionStartEvent)


# I've overwritten this method from yaml's emitter so we can extend the
# maximum key length (for some reason it is hard-coded at 128).
# The only change is on line 31 (in this file), changed from 128 to 1000
def _check_simple_key(self):
    length = 0
    if isinstance(self.event, NodeEvent) and self.event.anchor is not None:
        if self.prepared_anchor is None:
            self.prepared_anchor = self.prepare_anchor(self.event.anchor)
        length += len(self.prepared_anchor)
    if (isinstance(self.event, (ScalarEvent, CollectionStartEvent))
            and self.event.tag is not None):
        if self.prepared_tag is None:
            self.prepared_tag = self.prepare_tag(self.event.tag)
        length += len(self.prepared_tag)
    if isinstance(self.event, ScalarEvent):
        if self.analysis is None:
            self.analysis = self.analyze_scalar(self.event.value)
        length += len(self.analysis.scalar)
    return (length < 1000
            and (isinstance(self.event, AliasEvent) or
                 (isinstance(self.event, ScalarEvent)
                  and not self.analysis.empty and not self.analysis.multiline)
                 or self.check_empty_sequence() or self.check_empty_mapping()))


yaml.emitter.Emitter.check_simple_key = _check_simple_key


def parse_trivia_list(path: pathlib.Path):
    """Parse a .txt trivia file.

    Returns a list in the following form::
        List[Dict{question: List[answers]}]
    """
    ret = []

    with path.open('r', encoding="ISO-8859-1") as file:
        trivia_list = file.readlines()

    for line in trivia_list:
        if '`' not in line:
            continue
        line = line.replace('\n', '')
        line = line.replace('\'', '"')
        line = line.replace("\xEF\xBB\xBF", '')
        line = line.split('`')
        question = line[0].strip()
        answers = []
        for ans in line[1:]:
            ans = ans.strip()
            if ans.isdigit():
                ans = int(ans)
            answers.append(ans)
        if len(line) >= 2 and question and answers:
            ret.append({question: answers})

    if not ret:
        raise ValueError('Empty trivia list.')

    return ret


def output_yaml(data: object, path: pathlib.Path):
    """Output the python objects in `data` to the path at `path`."""
    # data is in form List[{question: List[answers]}]
    with path.open('w') as stream:
        yaml.dump(
            data,
            stream,
            width=300,
            default_flow_style=False,
            allow_unicode=True)


def main():
    """Run this script on the sub-directories `lists` and `yamls`."""
    list_path = pathlib.Path('lists')
    yaml_path = pathlib.Path('yamls')
    if not list_path.exists():
        print("No 'lists' subdirectory found. Make sure your console is at the"
              " correct directory.")
        sys.exit(1)
    if not yaml_path.exists():
        yaml_path.mkdir()
    for child in list_path.glob('*.txt'):
        print("Reading {}...".format(child))
        data = parse_trivia_list(child)
        filename = child.stem
        newpath = yaml_path.joinpath(filename + '.yaml')
        print("  Outputting to {}...".format(newpath))
        output_yaml(data, newpath)


if __name__ == '__main__':
    main()
    print("Success.")
    sys.exit(0)