lumbung-calendar-prototype/event_feed.py

#!/bin/python3

#lumbung.space calendar feed generator
#© 2021 roel roscam abbing gplv3 etc

from ics import Calendar  
import requests
import jinja2
import os
import shutil
from slugify import slugify
from natural import date
from event_feed_config import calendar_url, output_dir
from urllib.parse import urlparse
import arrow
import re

cal = Calendar(requests.get(calendar_url).text)

env = jinja2.Environment(
    loader=jinja2.FileSystemLoader(os.path.curdir) 
    )

if not os.path.exists(output_dir):
    os.mkdir(output_dir)

template = env.get_template('event_template.md')

existing_posts = os.listdir(output_dir)

def findURLs(string):
    """
    return all URLs in a given string
    """
    regex = r"(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'\".,<>?«»“”‘’]))"
    url = re.findall(regex,string)      
    return [x[0] for x in url]     

def find_imageURLS(string):
    """
    return all image URLS in a given string
    """
    regex = r"(?:http\:|https\:)?\/\/.*?\.(?:png|jpg|jpeg|gif|svg)" 

    img_urls = re.findall(regex, string, flags=re.IGNORECASE)
    return img_urls 

def create_metadata(event):
    """
    construct a formatted dict of event metadata for use as frontmatter for HUGO post
    """

    if event.location:
        location_urls = findURLs(event.location)

        if location_urls:
            location_url = location_urls[0]
            event.location = '[{}]({})'.format(urlparse(location_url).netloc, location_url)


    event_metadata = {
    'name':event.name,
    'created':event.created.format(),
    'description': event.description,
    'localized_begin': '           '.join(localize_time(event.begin)), #non-breaking space characters to defeat markdown
    'begin': event.begin.format(),
    'end': event.end.format(),
    'duration': date.compress(event.duration),
    'location': event.location,
    'uid': event.uid,
    'images' : find_imageURLS(event.description) # currently not used in template
    }

    return event_metadata

def localize_time(date):
    """
    Turn a given date into various timezones
    """
    # Dates need to be displayed for the various TZs
    # takes arrow objects

    # 3 PM Kassel, Germany, 4 PM Ramallah/Jerusalem, Palestina (QoF),
    # 8 AM Bogota, Colombia (MaMa), 8 PM Jakarta, Indonesia (Gudskul),
    # 1 PM (+1day) Wellington, New Zealand (Fafswag), 9 AM Havana, Cuba (Instar).

    tzs = [
          ('Kassel','Europe/Berlin'),
          ('Bamako', 'Europe/London'),
          ('Palestine','Asia/Jerusalem'),
          ('Bogota','America/Bogota'),
          ('Jakarta','Asia/Jakarta'),
          ('Makassar','Asia/Makassar'),
          ('Wellington', 'Pacific/Auckland')
          ]

    localized_begins =[]
    for location, tz in tzs:
        localized_begins.append( #javascript formatting because of string creation from hell
            '__{}__ {}'.format(
                str(location),
                str(date.to(tz).format("YYYY-MM-DD __HH:mm__"))
                )
            )
    return localized_begins

def create_event_post(post_dir, event):
    
    if not os.path.exists(post_dir):
        os.mkdir(post_dir)

    event_metadata = create_metadata(event)

    for img in event_metadata['images']:
        img_name = img.split('/')[-1]
        local_image = os.path.join(post_dir, img_name)
        #TODO: handle the fact that someone might update a post
        #and delete / replace an image
        if not os.path.exists(local_image):
            #download preview image
            response = requests.get(img, stream=True)
            with open(local_image, 'wb') as img_file:
                shutil.copyfileobj(response.raw, img_file)
                print('Downloaded event image for event "{}"'.format(event.name))
            event_metadata['description'] = event_metadata['description'].replace(img, '![]({})'.format(img_name))

    with open(os.path.join(post_dir,'index.md'),'w') as f:
        post = template.render(event = event_metadata)
        f.write(post)
        print('created post for', event.name, '({})'.format(event.uid))

    with open(os.path.join(post_dir,'.timestamp'),'w') as f:
        f.write(event_metadata['created'])


def update_event_post(post_dir, event):
    """
    Update a post based on the VCARD event 'created' field which changes when updated
    """ 
    if os.path.exists(post_dir):
        old_timestamp = open(os.path.join(post_dir,'.timestamp')).read()
        if event.created > arrow.get(old_timestamp):
            print('Updating', event.name, '({})'.format(event.uid))
            create_event_post(post_dir, event)
        else:
            print('Event current: ', event.name, '({})'.format(event.uid))

for event in list(cal.events):

    post_dir = os.path.join(output_dir, event.uid)

    if event.uid not in existing_posts: 
        #if there is an event we dont already have, make it
        create_event_post(post_dir, event)

    elif event.uid in existing_posts: 
        #if we already have it, update
        update_event_post(post_dir, event)
        existing_posts.remove(event.uid) # create list of posts which have not been returned by the calendar
        

for post in existing_posts: 
    #remove events not returned by the calendar (deletion) 
    print('deleted', post) 
    shutil.rmtree(os.path.join(output_dir,post))