Most “open data” about places is heavier than you need. Opening hours change, shop directories go stale, and a wide schema makes a simple demo harder than the problem you actually have. Sometimes you just want a short, stable list of names — for a typeahead, a test fixture, or a collation example.
danske-storcentre is that kind of list: 65 names of shopping centres currently operating in Denmark, in two files that contain the same data. storcentre.csv has a single column, navn. storcentre.json is a JSON array of strings. Both are dedicated to the public domain under CC0 1.0. The same files are also on Hugging Face.
The list is not exhaustive, and it is names only on purpose. Names change less often than hours or tenant lists, and they are enough when the job is “load some real-world Danish strings and do something useful with them.”
Load the CSV in pandas
import pandas as pd url = "https://codeberg.org/storcentredk/danske-storcentre/raw/branch/main/storcentre.csv" df = pd.read_csv(url) print(len(df), "names") # 65 print(df.columns.tolist()) # ['navn'] print(df["navn"].head(3).tolist()) # ['Aalborg Storcenter', 'Albertslund Centrum', 'Amager Centret'] If you cloned the repo, point read_csv at the local path instead.
The published order is alphabetical with str.casefold(), so Ballerup Centret comes before BIG Shopping Herlev. That is a sensible default, but it is not a Danish dictionary order — which is the interesting part of this dataset.
Load it with 🤗 Datasets
from datasets import load_dataset ds = load_dataset("storcentredk/danske-storcentre") print(ds["train"][0]) # {'navn': 'Aalborg Storcenter'} print(len(ds["train"])) # 65 Hugging Face detects the CSV and exposes a train split with the same navn column.
Danish sorting: æ, ø, å — and Aa
In the Danish alphabet, æ, ø, and å come after z, in that order: … x, y, z, æ, ø, å. Two other traps show up as soon as you sort these names:
- Default Unicode order puts å before æ (
Åis U+00C5,Æis U+00C6). A name starting with Å would be misplaced, even thoughØlbycenterhappens to land at the end of the published file. - A traditional Danish collator treats the letter sequence Aa as Å. So
Aalborg Storcentermoves to the end, next to any Å-name. Python’s default sort andstr.casefold()keep it at the front, which is also how the files are published.
locale.strxfrm is the first tool people reach for, but it only works if the locale is actually installed on the machine:
import locale locale.setlocale(locale.LC_COLLATE, "da_DK.UTF-8") # locale.Error if missing sorted(df["navn"], key=locale.strxfrm) On many servers, CI images, and containers, locale -a will not list da_DK.UTF-8, and the call raises locale.Error. That is not a theoretical caveat: a minimal Linux environment with only the C and POSIX locales fails exactly like this.
PyICU is the robust Python option, because it ships the collation data itself — at the cost of a native ICU dependency:
import icu collator = icu.Collator.createInstance(icu.Locale("da_DK")) sorted(df["navn"], key=collator.getSortKey) If you cannot install ICU, a small sort key is enough to get æ, ø, å, and the optional Aa rule right. It is not a full collator: punctuation such as the apostrophe in RO's Torv needs ICU or Intl if you care about dictionary order (RO's Torv should sort next to other “Ro…” names, not after a stray ').
def danish_key(s: str) -> list[int]: s = s.casefold().replace("aa", "å") # traditional: Aa = Å order = {ch: i for i, ch in enumerate("abcdefghijklmnopqrstuvwxyzæøå")} return [order.get(ch, 1000 + ord(ch)) for ch in s] demo = [ "Ølbycenter", "Aalborg Storcenter", "WAVES", "Søndercentret", "Field's", "Århus Storcenter", # not in the dataset; useful as a check "Sædding Centret", "Shoppen", ] sorted(demo, key=danish_key) # ["Field's", "Shoppen", "Sædding Centret", "Søndercentret", # "WAVES", "Ølbycenter", "Aalborg Storcenter", "Århus Storcenter"] Without the Aa → Å replacement, Aalborg Storcenter stays at the front. Both behaviours are defensible; just be explicit about which one you want.
JavaScript: fetch the JSON and sort with Intl.Collator
Modern JavaScript does this without extra libraries. Intl.Collator('da') includes the Aa/Å rule and handles punctuation better than a hand-rolled key. The raw file on Codeberg is a plain JSON array, so fetch is enough:
const url = "https://codeberg.org/storcentredk/danske-storcentre/raw/branch/main/storcentre.json"; const names = await (await fetch(url)).json(); console.log(names.length); // 65 const da = new Intl.Collator("da"); const sorted = [...names].sort(da.compare); const demo = [ "Ølbycenter", "Aalborg Storcenter", "WAVES", "Søndercentret", "Field's", "Århus Storcenter", "Sædding Centret", "Shoppen", ]; console.log([...demo].sort(da.compare)); // ["Field's", "Shoppen", "Sædding Centret", "Søndercentret", // "WAVES", "Ølbycenter", "Aalborg Storcenter", "Århus Storcenter"] A names-only list will not replace a full directory. It will, however, give you real Danish strings, a clear licence, and a concrete way to test whether your app sorts ø after z and treats Å as the last letter of the alphabet. If a centre is missing, renamed, or closed, the Codeberg repo accepts issues and pull requests.
The list is maintained by storcentre.dk.


