main.py

#! python3
"""
XLSX-SCHEDULE MAKER

This program is processes schedule text from txt file and puts it into xlsx file

MODULES USED: re, openpyxl, datetime

PLAN:
step1 - read schedule from the file
step2 - analyze schedule with re and datetime, find the first day date
step3 - create data structure Schedule to easily use it later
step4 - import data into xlsx file
"""

from pprint import pprint as pp
import re
import datetime
import Schedule
# import openpyxl


def import_from_file(directory):
    """
    function reads file and assigns its contents to variable schedule
    :param directory: str
    :return schedule: str
    """
    txt = open(directory, "r")
    schedule = txt.read()
    txt.close()
    return schedule


def extract_info(data):
    # splitting schedule into sections (0 - general info, days: 1-5)
    structure = re.split('-{2,}\n', data)

    schedule_info = get_header_info(structure[0])

    create_structure(schedule_info)

    n = split_to_periods(structure[2])

    # list of disciplines
    # d = re.split('\n\* ', n[1])

    # k = re.split(r'\n(?=\d пара)', n[2])
    # print(k)


def get_header_info(data):
    """
    function extracts data from header using regex module
    :param data: text from header
    :return tuple: group, year, semester, start_date
    """
    group = re.search(re.compile(r'(?<=Група )\w{2,3}(-\d{2})*'), data).group()
    year = re.search(re.compile(r'\d{4}-\d{4} н.р.'), data).group()
    semester = 1 if re.search(re.compile(r'\w+(?= семестр)'), data).group() is 'І' else 2
    first_monday = re.search(re.compile(r'\d{2}.\d{2} '), data).group()[:-1]
    start_date = datetime.date(int(year[0:4]) if semester == 1 else int(year[5:9]),
                               int(first_monday[3:]), int(first_monday[:2]))
    print('Extracting header data...\ngroup: {}\nyear: {}\nsemester: {}\nfirst week monday date: {}'
          .format(group, year, semester, start_date))
    return group, year, semester, start_date


def split_to_periods(day):
    # function for splitting weekday into periods
    return re.split('\n(?=\d пара)', day)


def get_disciplines(period):
    """

    :param period: text containing info about all disciplines of this period
    :return: list with information about each discipline
    """


def create_structure(semester_info):
    schedule = Schedule.Schedule  # creating object Schedule
    # adding info
    schedule.group = semester_info[0]
    schedule.year = semester_info[1]
    schedule.semester = semester_info[2]
    schedule.start = semester_info[3]
    cont = schedule.weeks
    # week dates
    week_shift = datetime.timedelta(days=7)
    curr_date = schedule.start
    for i in range(17):
        curr_w = cont[i]
        for j in range(5):
            curr_w

    return schedule





"""
Program
"""

direct = "GІР.txt"
dat = import_from_file(direct)
extract_info(dat)