A CodeSights flagship and the platform's largest geographic dataset: 82,996 rivers from Wikidata's half-million-strong river tree, kept by the measured-or-notable rule (a stated length, discharge or basin area, or an English Wikipedia article). The flagship trick is that the file is its own network: Mouth_ID joins River_ID within the same CSV, 60,072 edges resolve internally, and the test suite proves it by walking upstream from the Mississippi and reconstructing its basin as a directed tree, with the Ohio and Missouri asserting their famous confluences and a Czech stream that claimed to flow into itself edited out as the graph error it is. Quantities are best-rank with units converted from the statement values (the graph states lengths in kilometres, metres and miles): 56,233 lengths with the Nile at 6,650 km and the Amazon at 6,400, both shipped, neither crowned, because the rivalry is a measurement dispute and the dataset refuses to referee it; the Amazon settles for its two uncontested crowns, 209,000 cubic metres per second of discharge and a 7.05 million square kilometre basin. A corroboration guard protects the giants: any length over 4,000 km must carry a stated basin or discharge, which caught an Indonesian river claiming 5,806 km on a unit error. Coordinates on 75,014 rows (QLever writes POINT where WDQS writes Point, the parser now reads both), countries on 68,000-plus with Russia's 27,989 rivers leading, Romania's 8,983 a surprise of thorough cataloguing, and the Danube crossing at least eight countries by test. Names resolve en-then-mul, and namesakes are everywhere: Canada has a Mississippi, so the acceptance suite selects famous rivers by the longest bearer of the name. Sixteen tests across scale, anchors, network integrity, physics guards and geography.
82,996 rows in one CSV file, with 12 columns in total: River, River_ID, Countries, Length_km, Discharge_m3_s, Basin_km2, Source, Mouth, Mouth_ID, Latitude, Longitude, River_URL.
It is built by programmatically scraping wikidata.org, last pulled on 2026-10-13. Datasets are versioned; older versions stay downloadable.
Yes. Every CodeSights dataset is completely free as a CSV download; a free account is all it takes. Anyone can preview the data without signing in.
Yes. The exact Python scraper that built it is viewable on the dataset page by any signed-in member, so every number is reproducible.
Mouth_ID joins River_ID within this same file: rows form the global tributary network, so drainage basins reconstruct by walking the edges upstream.
Automated scraping leaves room for error and the underlying sources change over time, so no version is guaranteed accurate or complete. If a number matters, verify it against the original source.