Remcon 2024 - Timeline to XLSX

timeline.py

import re
import json
from datetime import datetime, timedelta

import requests
import xlsxwriter

DEBUG_DUMP_JSON = False


def clean_text(text: str):
    text = re.sub(r'\\"', '"', text)
    text = re.sub(r'(\\r|\\n)+', '\n', text)
    return text.strip()


class Event:
    start_date: datetime
    duration: int  # always 30...
    room: int

    name: str
    author: str
    description: str

    h: int
    x: int
    y: int

    def __init__(self, data: dict):
        self.start_date = datetime.strptime(data["startDate"], '%a, %d %b %Y %H:%M:%S %Z')
        #self.duration = int(data["time"])
        self.room = data["room"]

        self.name = data["name"]
        self.author = data["author"]
        self.description = data["description"]

        self.h = int(data["h"])
        self.x = int(data["x"])
        self.y = int(data["y"])


def get_current_build():
    r = requests.get("https://remcon.pl")
    r.raise_for_status()

    return re.search(r"\"buildId\":\"(?P<id>[A-Za-z0-9_-]+)\"", r.text).group('id')


def get_timeline_data_from_website():
    r = requests.get(f"https://remcon.pl/_next/data/{get_current_build()}/timeline.json")
    data = r.json()

    if DEBUG_DUMP_JSON:
        with open("data.json", "w") as f:
            json.dump(data, f, indent=4)

    return data


def get_timeline_data_from_file(filename="timeline.json"):
    with open("timeline.json") as f:
        return json.load(f)


#data = get_timeline_data_from_file()
data = get_timeline_data_from_website()

event_data = data["pageProps"]["timeline"]["events"]
events = [Event(event) for event in event_data]

earliest = events[0].start_date
latest = events[0].start_date
for event in events:
    earliest = min(earliest, event.start_date)
    latest = max(latest, event.start_date)

room_data = data["pageProps"]["timeline"]["rooms"]
room_to_col = {room["id"]: index + 1 for index, room in enumerate(room_data)}
room_to_label = {room["id"]: room["label"] for room in room_data}

workbook = xlsxwriter.Workbook("export.xlsx")
ws = workbook.add_worksheet()
# ws.write(ROW, COLUMN, CONTENT), rows/cols are 0-indexed

header_format = workbook.add_format({'bold': True})
event_format = workbook.add_format()
event_format.set_align('center')
event_format.set_align('vcenter')
event_format.set_text_wrap()
event_format.set_border()

ws.write(0, 0, "Remcon 2024")
for room_id, col in room_to_col.items():
    ws.write_string(0, col, room_to_label[room_id], cell_format=header_format)

date_row = 1
dt = earliest + timedelta(hours=1)  # forced TZ
while dt < latest + timedelta(hours=4):
    ws.write_string(date_row, 0, dt.strftime("%H:%M"))
    dt += timedelta(minutes=30)
    date_row += 1

for event in events:
    title = clean_text(f"{event.name.strip()} [{event.author.strip()}]")
    desc = clean_text(event.description)

    row = event.y + 1
    col = event.x + 1
    span = event.h

    try:
        if span == 1:
            ws.write_string(row, col, title, cell_format=event_format)
        else:
            ws.merge_range(row, col, row + span - 1, col, title, cell_format=event_format)

        if desc:
            ws.write_comment(row, col, desc)
    except Exception as e:
        print("[!] FAILURE: ", title)
        print(e)


workbook.close()