Prague integrated transport (PID)

Use it from code

Every day is a Parquet file at a fixed address. Read it straight from DuckDB, Polars or anything that reads Parquet over HTTP: only the columns and rows your query needs are downloaded. You need a free API token from an account.

Try it in your browser Developer guide

Source id

cz-pid

Services

TripUpdate live delays: daily files from 28 Sep 2026 to 3 Nov 2026 (37 days)

VehiclePosition train positions: daily files from 28 Sep 2026 to 3 Nov 2026 (37 days)

Daily file

https://data.train-archive.eu/daily/provider=cz-pid/service=TripUpdate/date=YYYY-MM-DD.parquet final delays per train and stop

Every update

https://data.train-archive.eu/provider=cz-pid/service=SERVICE/date=YYYY-MM-DD/day.parquet

Today

Still being collected in 15-minute files. The day’s file is built after 02:15 UTC. The developer guide shows how to read the 15-minute files.

Timetables

Service static: 14 versions, listed in cz-pid/static/latest.json

INSTALL httpfs; LOAD httpfs;

-- The final delay of each train at each stop, in minutes: the 20 largest
SELECT route, stop_name, scheduled_arrival, arrival_delay_min, delay_source
FROM read_parquet('https://data.train-archive.eu/daily/provider=cz-pid/service=TripUpdate/date=2026-11-01.parquet?token=YOUR_TOKEN')
WHERE trip_status = 'scheduled'
ORDER BY arrival_delay_min DESC NULLS LAST
LIMIT 20;

import polars as pl

token = "YOUR_TOKEN"
day = pl.read_parquet(f"https://data.train-archive.eu/daily/provider=cz-pid/service=TripUpdate/date=2026-11-01.parquet?token={token}")

# The final delay of each train at each stop, in minutes: the 20 largest
print(
    day.filter(pl.col("trip_status") == "scheduled")
    .select("route", "stop_name", "scheduled_arrival", "arrival_delay_min", "delay_source")
    .sort("arrival_delay_min", descending=True, nulls_last=True)
    .head(20)
)

curl -L -o cz-pid-TripUpdate-2026-11-01.parquet \
  'https://data.train-archive.eu/daily/provider=cz-pid/service=TripUpdate/date=2026-11-01.parquet?token=YOUR_TOKEN'

Replace YOUR_TOKEN with one of your API tokens. Delays are in seconds in the Parquet files; the examples turn them into minutes.

import json, urllib.request
import polars as pl

base, token = "https://data.train-archive.eu", "YOUR_TOKEN"
latest = json.load(urllib.request.urlopen(f"{base}/cz-pid/static/latest.json?token={token}"))
stops = pl.read_parquet(f"{base}/{latest['tables']['stops']}?token={token}")
print(stops.select("stop_id", "stop_name"))

Also: every file of this source, listed, and asking an AI assistant to query it for you.