\xd1\xa7\xc9\xfa\xb8\xf6\xc8\xcb\xd0\xc5\xcf\xa2\xcf\xb5\xcd\xb3

gen_ics.py

from bs4 import BeautifulSoup
from datetime import datetime, timedelta
import re


def get_start_day():
    return datetime(2016, 2, 29)


def date_to_str(dt):
    return dt.strftime('%Y%m%dT%H%M%S')


bydays = ['SU', 'MO', 'TU', 'WE', 'TH', 'FR', 'SA', 'SU']


def ics_head() -> str:
    return """BEGIN:VCALENDAR
PRODID:-//Ning Humbert//Schedule2ICS//ZH
VERSION:0.1
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Asia/Shanghai
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:CST
DTSTART:19700101T000000
END:STANDARD
END:VTIMEZONE
"""


def ics_tail() -> str:
    return "END:VCALENDAR\n"


def gen_ics(body: str):
    with open('test.ics', 'w') as f:
        f.write(ics_head() + body + ics_tail())


def read_html_to_bs(path: str):
    with open(path, 'rb') as f:
        bin_data = f.read()
    try:
        html_data = bin_data.decode()
    except UnicodeDecodeError:
        html_data = bin_data.decode('gbk')
    return BeautifulSoup(html_data, 'lxml')


def find_table(bs):
    ts = bs.findAll('table')
    for t in reversed(ts):
        if t.attrs.get('bgcolor') == '#F2EDF8':
            return t
    return None


def find_courses(table):
    trs = table.findAll('tr')
    i = 0
    for tr in trs:
        if len(tr.findAll('td')) != 9:
            break
        i += 1
    return trs[:i] + trs[i + 1:]


def vanish_dash(maybe_dash: str) -> list:
    found = maybe_dash.find('-')
    try:
        if found == -1:
            return [int(maybe_dash)]
        start_n = int(maybe_dash[:found])
        end_n = int(maybe_dash[found + 1:])
        ret = []
        for i in range(start_n, end_n + 1):
            ret.append(i)
        return ret
    except ValueError:
        return []


def generate_dash(none_dash: list) -> list:
    if len(none_dash) == 0:
        return []
    ret = []
    none_dash += [1000]
    tmp_start = none_dash[0]
    tmp_end = none_dash[0]
    for i in none_dash[1:]:
        if i - tmp_end == 1:
            tmp_end = i
        else:
            ret.append((tmp_start, tmp_end))
            tmp_start = tmp_end = i
    return ret


def handle_date(date: str):
    maybe_dashes = re.findall('(\d+\-\d+|\d+)', date)
    ret = []
    for i in maybe_dashes:
        ret += vanish_dash(i)
    return generate_dash(ret)


def get_time(start_week: int, course_time: str):
    dash_pos = course_time.find('-')
    assert dash_pos != -1
    course_day = bydays[int(course_time[:dash_pos])]
    course_seq = int(course_time[dash_pos + 1:])
    that_day = get_start_day() + timedelta(weeks=start_week - 1) + timedelta(days=int(course_time[:dash_pos]) - 1)
    assert isinstance(that_day, datetime)
    if course_seq == 1:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=8
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=9,
            minute=45
        )
    elif course_seq == 2:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=10,
            minute=5
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=11,
            minute=50
        )
    elif course_seq == 3:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=14
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=15,
            minute=45
        )
    elif course_seq == 4:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=16,
            minute=5
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=17,
            minute=50
        )
    elif course_seq == 5:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=18,
            minute=40
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=20,
            minute=25
        )
    else:
        that_start_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=20,
            minute=35
        )
        that_end_time = datetime(
            year=that_day.year,
            month=that_day.month,
            day=that_day.day,
            hour=22,
            minute=20
        )
    return {
        'byday': course_day,
        'start_time': that_start_time,
        'end_time': that_end_time
    }


def handle_a_course(_course):
    tds = _course.findAll('td')
    assert len(tds) == 9
    course_name = tds[0].get_text()[:-1]
    course_id = tds[2].get_text()[:-1]
    course_nid = tds[3].get_text()[:-1]
    # course_is_must = tds[4].get_text()[:-1]
    # course_is_exam = tds[5].get_text()[:-1]
    course_location = tds[6].get_text()[:-1]
    course_time = tds[7].get_text()[:-1]
    course_date = tds[8].get_text()[:-1]
    course_uid = course_id + course_nid + course_time
    ret = ""
    date_segments = handle_date(course_date)
    for date_segment in date_segments:
        time_info = get_time(date_segment[0], course_time)
        ret += """BEGIN:VEVENT
DTSTART;TZID=Asia/Shanghai:{start_time}
DTEND;TZID=Asia/Shanghai:{end_time}
RRULE:FREQ=WEEKLY;COUNT={count};BYDAY={byday}
UID:{course_uid}
DESCRIPTION:
LOCATION:{location}
SEQUENCE:1
STATUS:CONFIRMED
SUMMARY:{course_name}
TRANSP:OPAQUE
END:VEVENT
""".format(
            start_time=date_to_str(time_info.get('start_time')),
            end_time=date_to_str(time_info.get('end_time')),
            byday=time_info.get('byday'),
            course_uid=course_uid + str(date_segment[0]),
            location=course_location,
            course_name=course_name,
            count=date_segment[1] - date_segment[0] + 1
        )
    return ret


if __name__ == '__main__':
    all_course = ""
    courses = find_courses(find_table(read_html_to_bs('b.html')))
    for course in courses:
        all_course += handle_a_course(course)
    gen_ics(all_course)