Open Street Map Public Transport Parser

A tool to parse broken/disconnected relations from openstreetmap, reconnect them with some tolerance, sort the internal points, and output them as GeoJSON linestrings.

It can understand and reconstruct

• relations that represent public transport v2
• ways and relations that represent areas

Blazing fast by design: every step in the process pipeline is fully parallelized for maximum speed using all available cpus by default.

How it works

1. Reads data from a .pbf file, extracting ways and relations filtered by attributes specified via the --filter argument.
2. Processes each public transport relation as follows:
   1. Sorts all the ways within the relation based on proximity.
   2. Joins ways that share identical lat/lng coordinates at their first or last nodes, combining them into a single LineString.
   3. If multiple LineStrings remain after this step:
      • It checks for gaps.
      • If the gaps are smaller than the specified gap threshold (in meters, default: 150), they are joined into one LineString.
3. Outputs a JSON array, with one GeoJSON feature per way or relation found, including:
   • For areas: a MultiPolygon feature. Ways with role inner are holes, every other way is part of an outer ring; rings are chained by shared nodes in any member order, a gap smaller than the threshold is bridged, and an area with any ring that cannot be closed is reported as Broken with no geometry.
   • For public transport: a LineString representing the full path and an array of points representing the stops.

See the blogpost for a very detailed description

Status

Try it

git clone git@github.com:cualbondi/osmptparser.git
wget http://download.geofabrik.de/south-america/ecuador-latest.osm.pbf
cargo run --example main ecuador-latest.osm.pbf

Time it

cargo build --release --example main && /usr/bin/time -v target/release/examples/main ecuador-latest.osm.pbf

CLI

cargo run --release ./ecuador-latest.osm.pbf --filter "boundary=national_park"

you should get a json list with one geojson per area that matches with the filter

cargo run --release ./ecuador-latest.osm.pbf --filter-ptv2

you should get a json list with one geojson per ptv2 containing a linestring and each stop

Library

use osmptparser::{ParseStatus, Parser};

let parser = Parser::new_aa("portugal-latest.osm.pbf", 0)?; // name&admin_level&boundary=administrative
for area in parser.get_areas(150.0) {
    // area.id_type is 'r' (relation) or 'w' (standalone closed way)
    // area.geometry: Vec<polygon>, polygon = Vec<ring>, ring = Vec<(lon, lat)>, first ring is the outer one
    if area.parse_status != ParseStatus::Broken {
        println!("{}{} {} {} polygon(s)", area.id_type, area.id, area.tags["name"], area.geometry.len());
    }
}

let parser = Parser::new_ptv2("portugal-latest.osm.pbf", 0)?;
for pt in parser.get_public_transports(150.0) {
    // pt.geometry: Vec<(lon, lat)>; pt.stops[i].role is the member role ("stop", "platform", ...)
}

// raw relations with their members: Way.role / Node.role keep the osm member role
let relation = parser.get_relation_from_id(85965);

Run CI linter + recommendations + tests

cargo fmt -- --check && cargo clippy --all-targets -- -D warnings -A clippy::ptr-arg && cargo test

Build pbf test files

tests/areas.pbf is built from the hand-written tests/areas.osm (administrative areas with outer/inner rings, an exclave, a standalone closed way and a broken ring):

osmium cat -f pbf --overwrite -o tests/areas.pbf tests/areas.osm

tests/test.pbf (two ptv2 routes from Ecuador):

wget http://download.geofabrik.de/south-america/ecuador-latest.osm.pbf
osmconvert ecuador-latest.osm.pbf -o=ecuador.o5m
osmfilter ecuador.o5m --keep= --keep-relations="@id=85965 =2030162" > test.o5m
osmconvert test.o5m -o=test.pbf