Files
upstream-aw-core/aw_transform/split_url_events.py
upstream-mirror 8757f2fd5e
Some checks failed
Build / Test on ubuntu-latest, py-3.8 (push) Successful in 4m21s
Lint / lint (push) Successful in 25s
Build / Test on macOS-latest, py-3.8 (push) Has been cancelled
Build / Test on windows-latest, py-3.8 (push) Has been cancelled
CodeQL / Analyze (python) (push) Failing after 1s
snapshot of https://github.com/ActivityWatch/aw-core.git (master, 2026-09-16; blobs > 131072B omitted)
2026-09-16 00:17:07 +05:00

33 lines
1.1 KiB
Python

import logging
from typing import List
from urllib.parse import urlparse
from aw_core.models import Event
logger = logging.getLogger(__name__)
def split_url_events(events: List[Event]) -> List[Event]:
for event in events:
if "url" in event.data:
url = event.data["url"]
parsed_url = urlparse(url)
event.data["$protocol"] = parsed_url.scheme
netloc = parsed_url.netloc
if netloc:
domain = netloc[4:] if netloc[:4] == "www." else netloc
elif parsed_url.scheme:
# For URLs without a domain (e.g. file://, about:),
# use the scheme as domain so they don't all cluster as empty.
domain = parsed_url.scheme
else:
domain = ""
event.data["$domain"] = domain
event.data["$path"] = parsed_url.path
event.data["$params"] = parsed_url.params
event.data["$options"] = parsed_url.query
event.data["$identifier"] = parsed_url.fragment
# TODO: Parse user, port etc aswell
return events