PyWorldAtlas: offline world geography for Python

Offline world atlas for Python: 248 countries, 11 million places, statistics since 1960, and 3D maps, with zero dependencies. Built for low-cost computers with 512 MB of memory, such as a Raspberry Pi Zero 2 W; the core installs in under a minute on slow 3G.

248 countries · 200,000 places built in · 11 million with packs · statistics since 1960 · airports and World Heritage sites · 248 3D maps · 0 dependencies

Try it in 10 seconds

No install: open the browser playground, type a country or place, take a quiz, or run a ready-made program.

Or install it (Python 3.10 to 3.14, nothing else needed):

python -m pip install pyworldatlas
pyworldatlas --version
pyworldatlas find Everest

A taste

from datetime import date
from pyworldatlas import Atlas

with Atlas() as atlas:
    print(atlas.country("Japan").travel.summary)      # Plugs A, B · 100 V, 50/60 Hz · drives on the left
    print(atlas.search_places("Everest", limit=1)[0])  # Mount Everest, mountain (NP)
    print(atlas.sun("Oslo", date(2026, 12, 21)).daylight)  # 5 h 54 min
    print(atlas.antipode("Madrid").nearest[0].place)  # Ossa, mountain (NZ): dig through the Earth
    atlas.map("Iceland").show(auto_rotate=True)        # 3D map (pyworldatlas get maps)

What's inside

Everything works offline after installation, needs no API key, and is built for low-cost computers and slow connections. Every value is traced to a named source; missing data stays missing.

Start here

• 🧒 Students: the playground and learning activities
• 🍎 Teachers and classrooms: teaching with PyWorldAtlas, with a 15-minute lesson and printable quizzes
• 💻 Developers: the quickstart, example programs, and API reference

Trust and neutrality

PyWorldAtlas presents factual geography from public sources, with no political commentary. Country names and codes follow published international standards, and the project takes no position on disputes. See the educational and neutrality policy, data sources, and data quality.

Contributing

Questions, corrections, and focused contributions are welcome: see CONTRIBUTING.md, the code of conduct, and SECURITY.md. python maintain.py check runs the full test, build, and documentation gate.

License

PyWorldAtlas code is available under the MIT License. The data keeps its own terms, and every source is credited:

• Places, cities, and country metadata: GeoNames, CC BY 4.0
• Local names, currencies, and languages: Unicode CLDR, Unicode License v3
• Statistics pack: World Bank World Development Indicators, CC BY 4.0
• Also: CIA World Factbook and Natural Earth (public domain), Wikidata, IANA, NOAA ETOPO 2022, and Beck et al. Köppen-Geiger maps (CC0), OurAirports (public domain), and United Nations M49 and naming documents

Full credits, the changes made, and license texts are in THIRD_PARTY_NOTICES.md and licenses/. The package metadata license is MIT AND CC-BY-4.0 AND Unicode-3.0 AND CC0-1.0.