Open data

What we share back

Some of our datasets build on openly licensed data. OpenStreetMap’s licence, the Open Database License (ODbL) 1.0, asks anyone who publishes something made from it to offer, free, the data they derived or the method of making it. Here they are.

Toll plaza locations

Our toll tariffs come from the Government Gazette, which names each plaza but doesn’t say where it is. The positions and provinces come from OpenStreetMap toll booths. Data © OpenStreetMap contributors, available under the ODbL 1.0.

• locations.json: the plaza positions and provinces we derived (ODbL 1.0)
• plazas.json: each gazetted plaza and the OpenStreetMap names we match it by
• toll-tariffs-locate.py: the script that makes them (MIT licence)
• README.txt: how to rebuild it, and the licences

Every dataset, in bulk (Parquet)

Each live dataset’s full history as Parquet files: every record each release held, and the facts they cite with their documents and pages. Downloads take an API key, like the records; the list, sizes, licences and SHA-256 fingerprints are open to everyone, so a copy can be checked. How to use them: the docs. Every dataset is also described in DCAT at /catalog.json.

Dam outlines

Each dam page shows the water surface measured from space, inside the dam’s full-supply outline. The Department of Water and Sanitation names each dam but gives no outline, so the outlines come from OpenStreetMap water polygons. Data © OpenStreetMap contributors, available under the ODbL 1.0.

• dam-outlines.geojson: one outline per DWS dam station, with how it was matched (ODbL 1.0)
• outline-selection.json: which OpenStreetMap polygons are each dam, and the dams left unmatched with why
• outline-match.sql, outline-match.py and outlines.py: the method (MIT licence)
• README.txt: how to rebuild it, and the licences

Places, streets and house numbers (address search)

Unmapped Search finds streets, suburbs, townships, villages and towns, and a house number on a street where OpenStreetMap has one. They come from OpenStreetMap only: Data © OpenStreetMap contributors, available under the ODbL 1.0. Nothing else is joined in. The ward and municipality a search shows are worked out when you search, from the Municipal Demarcation Board’s ward boundaries, a separate dataset that is not in these files. No personal information: no contributor names, no farms or private roads, and for a house only its number and its point, never a name, an owner or an occupant. Built 11 November 2026 from OpenStreetMap data to 31 October 2026.

How a house number is placed: each point OpenStreetMap tags with a house number goes on the named street its tag gives, the nearest stretch of a street of exactly that name within 400 m. A number with no street in its tag goes on a street only when one named street is within 40 m and no street of another name is near; otherwise it is left out (we don’t guess which street a corner house faces). Numbers that are not a plain number, such as a unit or an erf, are left out. Where a search asks for a number OpenStreetMap doesn’t have, the page places it between the nearest numbers either side, or at the middle of the street, and says so.

• osm-places-2026-10-31.ndjson.gz: 8,175 places, 0.2 MB, sha256 01171f2075bc
• osm-streets-2026-10-31.ndjson.gz: 233,961 street pieces, 10.7 MB, sha256 589dfd3a922a
• osm-houses-2026-10-31.ndjson.gz: 13,813 streets with their house numbers, 0.9 MB, sha256 93fc60e898fb
• address-osm.py: the script that makes the places and streets from the OpenStreetMap extract (MIT licence)
• address-lib.mjs: how they become the search index (MIT licence)
• address-houses.py: the script that puts each house number on its street (MIT licence)
• address-houses-lib.mjs: how the house numbers and street names become the address suggestions (MIT licence)
• houses.json: how many numbers were kept and left out, and why
• open.json: the extract it was built from (sha256), the counts and the files
• README.txt: what is in them, how to rebuild them, and the licences

Where each dam is (dams nearest a spot)

A search for a street or a town lists the dams nearest it, and a dam’s page lists the towns near it. The Department of Water and Sanitation’s weekly report names each dam but doesn’t say where it is. The positions come from OpenStreetMap: for each dam, a point inside its OpenStreetMap water outline, the outline matched to the dam by name and checked against the capacity the Department gives. A dam with no named outline in OpenStreetMap has no point, and the file lists it with the reason. Data © OpenStreetMap contributors, available under the ODbL 1.0.

• dam-points.json: each dam’s point (latitude, longitude and the area of its outline in km², by station), and the dams with none and why (ODbL 1.0)