osmfeatures
7.1.0indexedFast dictionary of OpenStreetMap features, enabling term and tag-based searches. Utilizes indices for speed and supports translations and brand preset data integration.
26
Stars
—
Used by
dependents
—
Health
/ 100
Fast dictionary of OpenStreetMap features, enabling term and tag-based searches. Utilizes indices for speed and supports translations and brand preset data integration.
A Kotlin multiplatform dictionary of OSM map features, accessible by terms and by tags. Supported platforms are Android, JVM and iOS.
Due to heavy use of indices, it is very fast.
It is currently used in StreetComplete.
© 2019-2025 Tobias Zwick. This library is released under the terms of the Apache License Version 2.0.
Add de.westnordost:osmfeatures:7.1.0 as a Maven dependency or download the jar from there.
The data for the dictionary is not maintained in this repository. It actually uses the preset data from iD, its translations and optionally additionally the brand preset data from the name suggestion index. Each are © iD contributors, licensed under the ISC license.
So, just dump all the translations and the presets.json into the same directory. To be always up-to-date, it is advisable to have an automatic build task that fetches the current version of the presets from the repository.
The app for which this library was developed (StreetComplete), uses the following tasks:
Point the dictionary to the directory where the data is located (see above). Use FeatureDictionary as a singleton, as initialization takes a moment (loading files, building indices).
val dictionary = FeatureDictionary.create(fileSystem, "path/to/data")
For Android, use
val dictionary = FeatureDictionary.create(assetManager, "path/within/assets/folder/to/data")
If brand features from the name suggestion index should be included in the dictionary, you can specify the path to these presets as a third parameter. These will be loaded on-demand depending on for which countries you search for.
Translations will also be loaded on demand when first querying features using a certain language.
val matches = dictionary.getByTags(
tags = mapOf("amenity" to "bench"), // look for features that have the given tags
languages = listOf("de"), // show results in German only, don't fall back to English or unlocalized results
geometry = GeometryType.POINT, // limit the search to features that may be points
)
// prints "Parkbank" (or something like this)
// or null if no preset for amenity=bench exists that is localized to German
println(matches[0]?.getName())
val matches = dictionary.getByTerm(
term = "Bank", // look for features matching "Bank"
languages = listOf("de", null), // show results in German or fall back to unlocalized results
// (brand features are usually not localized)
country = "DE", // also include things (brands) that only exist in Germany
geometry = GeometryType.AREA, // limit the search to features that may be areas
)
// result sequence will have matches with at least amenity=bank, but not amenity=bench because it is a point-feature
val match = dictionary.getById(
id = "amenity/bank",
languages = listOf("de", "en-US", null), // show results in German, otherwise fall back to American
// English or otherwise unlocalized results
country = "DE", // also include things (brands) that only exist in Germany
)
For a more convenient interface on Java, the above functions continue to be available as builders, e.g.
List<Feature> matches = dictionary
.byTags(Map.of("amenity", "bench"))
.forGeometry(GeometryType.POINT)
.inLanguage("de")
.find();
Surfaced from shared tags and platforms — no rankings paid for.