jev22: extend + optimize config

Author: saerdnaer

html2meta.py

@@ -23,7 +23,6 @@
 
 acronym = 'jev22'  # args.acronym.lower()
 
-
 @dataclass
 class Location:
     id: str
@@ -72,37 +71,39 @@ def fetch_post(source_url):
         cid = chapter['id'].split('-')[3]
         print("\n==", cid, chapter.string)
 
-        region = None
-        location = None
-        subsection = None
-
-        # TODO: next_siblings vs next_elements
-        for region in chapter.find_next_siblings('h3'):
-            print(region.string)
-            
-            for element in region.find_next_siblings('h4'):
-                location = Location(
-                    id=element['id'],
-                    name=element.string,
-                    region=region,
-                    country=cid
-                )
-                print("-", location.name)
-                conferences.append(location)
-
-                subsection = None
-                for p in element.find_all('p'):
-                    if p.get('class') == ['subsection']:
-                        subsection = p.string.strip()
-                        continue
-
-                    if subsection == 'Info':
-                        location.infos = [list(li.children) for li in p.select('li')]
-                        continue
-
-                    if subsection and location:
-                        location.description[subsection] += p.text.prettify()
-                        continue
+        elements = chapter.find_next_siblings('h4')
+        element = next(elements)
+
+        while True:
+            location = Location(
+                id=element['id'],
+                name=element.string,
+                region=element.find_previous_sibling('h3'),
+                country=cid
+            )
+            print("-", location.name)
+            conferences.append(location)
+
+            subsection = None
+            for p in element.next_siblings:
+                if p.name != 'p':
+                    break
+
+                if p.get('class') == ['subsection']:
+                    subsection = p.string.strip()
+                    continue
+
+                if subsection == 'Info':
+                    location.infos = [list(li.children) for li in p.select('li')]
+                    continue
+
+                if subsection and location:
+                    location.description[subsection] += p.text.prettify()
+                    continue
+
+            element = next(elements)
+            if not element or element.previous_sibling.name == 'h2':
+                break
 
     return conferences
 

schedule_jev22.py

@@ -1,7 +1,6 @@
 #!/usr/bin/env python3
 # -*- coding: UTF-8 -*-
 
-import os
 from typing import List
 import requests
 import json
@@ -83,6 +82,21 @@
         },
         options={"track": "Hellarious"},
     ),
+    GenericConference(
+        url="https://data.jtbx.de/jev22_ccl/schedule.json",
+        data={
+            "name": "Curious Community Labs e.V.",  # (27.–29.)
+            "location": "Hamburg",
+            "links": [
+                "https://curious.bio/2022/11/remote-chaos-experience/",
+            ],
+            "osm_url": "https://www.openstreetmap.org/node/9615054340#v1"
+        },
+        options={
+            "track": "Curious Community Labs",
+            "id_offsets": -200
+        },
+    ),
     PretalxConference(
         url="https://pretalx.c3voc.de/xrelog-2022",
         data={
@@ -94,6 +108,16 @@
             ],
         },
     ),
+    PretalxConference(
+        url="https://talks.w.icmp.camp/wicmp1",
+        data={
+            "slug": "wicmp1",
+            "name": "Wintergalaktische Club Mate Party",  # (27.–30.)
+            "location": "Erlangen, Bits'n'Bugs + ZAM",
+            "links": ["https://w.icmp.camp"],
+        },
+        options={"track": "Erlangen"},
+    ),
     WebcalConference(
         url="webcals://ramac.mudbyte.de/remote.php/dav/public-calendars/YTtwyZcXsmZDfSoo/?export",
         data={
@@ -105,16 +129,6 @@
             "track": "Gießen/Wetzlar",
         },
     ),
-    PretalxConference(
-        url="https://talks.w.icmp.camp/wicmp1",
-        data={
-            "slug": "wicmp1",
-            "name": "Wintergalaktische Club Mate Party",  # (27.–30.)
-            "location": "Erlangen, Bits'n'Bugs + ZAM",
-            "links": ["https://w.icmp.camp"],
-        },
-        options={"track": "Erlangen"},
-    ),
     GenericConference(
         url="https://laborluxeria.github.io/winterchaos2022/schedule.json",
         # https://github.com/laborluxeria/winterchaos2022/tree/main/_sessions – michi
@@ -162,7 +176,8 @@
         },
     ),
     GenericConference(
-        url="TBD",
+        # TODO: add XML import from https://dezentrale.space/2022/12/localverse2022-fahrplan.xml
+        url="TBD",  # https://dezentrale.space/2022/12/localverse2022-fahrplan.xml",
         data={
             "name": "localverse2022",  # (27.–30.)
             "location": "Leipzig",
@@ -173,16 +188,18 @@
         },
     ),
     GenericConference(
-        url="https://data.jtbx.de/jev22_ccl/schedule.json",
+        url="https://woodbine.nyc/2022/12/dweb-jev/schedule.json",
         data={
-            "name": "jev22_ccl",  # (27.–29.)
-            "location": "Hamburg",
+            "name": "DWeb NY Pre-kickoff + RemoteCCC @ Woodbine",
+            "location": "New York, USA",
+            "address": "Basement, Woodbine, 585 Woodward Ave, Ridgewood, Queens, NY 11385",
+            "description": "A new DWeb Node is being formed! What does this mean? Come and find out! \n\n Relevant topics include: decentralized web, digital commons, co-ops, governance, privacy, anti-surveillance, software freedom.",
             "links": [
-                "https://curious.bio/2022/11/remote-chaos-experience/",
+                "https://dweb.events/stuff/ny/2022/woodbine-remoteccc/",
             ],
-        },
-        options={"track": "Curious Community Labs"},
-    )
+            "osm_url": "https://www.openstreetmap.org/node/5365063316#v13"
+        }
+    ),
 ]
 
 targets = [
@@ -314,6 +331,7 @@ def main():
         git("rm events/*")
 
     # write seperate file for each event, to get better git diffs
+    # TODO: use Event.export()
     def export_event(event: Event):
         origin_system = None
         if isinstance(event, Event):

voc/schedule.py

@@ -437,7 +437,7 @@ def add_events_from(self, other_schedule, id_offset=None, options={}, context: E
                         event["do_not_record"] = options['do_not_record'](event) if callable(options["do_not_record"]) else options["do_not_record"]
 
                     if options.get("remove_title_additions"):
-                        event["title"], subtitle, event_type = re.match(r"^(.+?)(?:( ?[:–] .+?))?(?: \((.+?)\))?$", event["title"]).groups()
+                        event["title"], subtitle, event_type = re.match(r"^(.{5,}?)(?:( ?[:–] .+?))?(?: \((.+?)\))?$", event["title"]).groups()
                         if not event.get("subtitle") and subtitle:
                             event["subtitle"] = subtitle
 
@@ -457,7 +457,7 @@ def add_events_from(self, other_schedule, id_offset=None, options={}, context: E
                         # TODO? offset for person IDs?
 
                     # workaround for fresh pretalx instances
-                    elif options.get("randomize_small_ids") and int(event["id"]) < 1000:
+                    elif options.get("randomize_small_ids") and int(event["id"]) < 1500:
                         event["id"] = int(re.sub("[^0-9]+", "", event["guid"])[0:4])
 
                     # overwrite slug for pretalx schedule.json input