theater.py

import logging
import os
import sys

import lxml.etree as etree
from openpyxl import load_workbook

logging.basicConfig(stream=sys.stdout, level=logging.DEBUG)


def append_something(filename, what):
    return "{0}_{2}{1}".format(*os.path.splitext(filename) + (what,))


def paranid_story_lines(xml, row):
    d = row[3]
    f = row[5]

    d_text = d.value

    if type(d_text) is int:
        nodes = xml.xpath(f'//t[@id="{d_text}"]')

        if len(nodes) > 0:
            f.value = nodes[0].text


def paranid_sink_lines(xml, row):
    d = row[3]
    f = row[5]

    d_text = d.value

    if type(d_text) is int:
        nodes = xml.xpath(f'//t[@id="{d_text}"]')

        if len(nodes) > 0:
            f.value = nodes[0].text


def split_story_lines(xml, row):
    c = row[2]
    d = row[3]
    f = row[5]

    c_text = c.value
    d_text = d.value

    if not c_text:
        return

    if type(d_text) is int and '"' in c_text:
        intro = c_text.split('"')[1]
        nodes = xml.xpath(f'//t[@id="{intro}{d_text:03}"]')

        if len(nodes) > 0:
            f.value = nodes[0].text


def main(args):
    path_xlsx_in = args[1]
    path_xlsx_out = append_something(path_xlsx_in, 'it')
    path_xml = args[2]

    logging.info(f'XML file as input:   {path_xml}')
    logging.info(f'XLSX file as input:  {path_xlsx_in}')
    logging.info(f'XLSX file as output: {path_xlsx_out}')

    # if not os.path.exists(path_xlsx_out):
    #     copyfile(path_xlsx_in, path_xlsx_out)

    xml = etree.parse(path_xml)

    wb = load_workbook(path_xlsx_in)
    sheet_names = wb.sheetnames

    function_sheet = [paranid_story_lines, paranid_sink_lines, split_story_lines]

    for s in range(len(sheet_names)):
        sheet_name = sheet_names[s]
        sheet = wb[sheet_name]
        f = function_sheet[s]

        logging.info(sheet_name)

        for r in sheet.rows:
            f(xml, r)

    wb.save(path_xlsx_out)


if __name__ == '__main__':
    main(sys.argv)