dataportalen

Python access to dataportal.se that gives you plain dicts, not RDF.

Sweden's open-data registry describes its 23,500 datasets in RDF, which means every value arrives as a web address: a dataset about roads is filed under http://publications.europa.eu/resource/authority/data-theme/TRAN rather than under "transport". Working out that it means "transport" is, per the registry's own documentation, your problem.

This package makes it not your problem. You search with words and you get dictionaries back.

Install

pip install dataportalen

Distribution and import name are both dataportalen. The unrelated dataportal package on PyPI is a Korean public-data client — not this.

Quick start

from dataportalen import Catalog, text

catalog = Catalog()      # downloads the catalogue the first time: ~7 min, ~95 MB
print(catalog.info())    # {'datasets': 17555, 'data_services': 578, ...}

page = catalog.datasets(theme="transport", format="csv")
print(page.total)                    # 44
print(page.facets["publisher"][:3])  # who publishes them

for dataset in page:
    print(text(dataset["title"]), text(dataset["publisher"]["name"]))
    for dist in dataset["distributions"]:
        print("   ", dist["format"], dist["access_url"])

That first line takes about seven minutes. It downloads the whole catalogue once, and every search after that is local and takes hundredths of a second. When the copy is over a week old, the next Catalog() catches up in under a minute.

By default the catalogue holds the datasets that say public, minus the distributions the registry's nightly link check got an HTTP error for or found no host for. Catalog(access_rights=None, exclude_broken=False) holds everything.

Publishers are there too: catalog.publishers() lists them, and catalog.publisher("scb") is one of them with what it publishes.

No room for a download, or one question to ask? LiveCatalog() has the same methods and asks the registry directly, a page at a time.

Documentation

• docs/cheatsheet.md — every public name once, every argument spelled out, the output shape underneath.
• docs/guide.md — how to use it, in the order you hit it: searching, discovering filters, reading a result, recipes.
• docs/reference.md — every method, filter, record key and error, as tables.
• docs/internals.md — how it is built: the short values, the module layout, development, releasing.

Licence

Code is MIT. The bundled label table is third-party data redistributed under CC BY 4.0 and EU Decision 2011/833/EU; the attributions are in NOTICE, with the details in docs/internals.md.

Metadata you retrieve is published by its respective publishers, each under its own licence — check the license field on the dataset.

Status

Most of the code here was written with an AI coding assistant. The design, features and scope are my decisions, and everything shipped has been reviewed and tested against the live registry.

This package exists to support NordicIntel's main project, so expect it to be maintained rather than extended. Bug reports are welcome; I am not looking for co-maintainers.