GeoDesk is a fast and storage-efficient geospatial database for OpenStreetMap data. Also available for Python and for Java.

Why GeoDesk?

• Small storage footprint — GeoDesk's GOL files are only 20% to 50% larger than the original OSM data in PBF format — that's less than a tenth of the storage consumed by a traditional SQL-based database.

• Fast queries — typically 50 times faster than SQL.

• Fast to get started — Converting .osm.pbf data to a GOL is 20 times faster than an import into an SQL database. Alternatively, download pre-made data tiles for just the regions you need and automatically assemble them into a GOL.

• Intuitive API — No need for object-relational mapping; GeoDesk queries return lightweight C++ objects. Quickly discover tags, way-nodes and relation members. Get a feature's geometry, measure its length/area.

• Proper handling of relations — (Traditional geospatial databases deal with geometric shapes and require workarounds to support this unique and powerful aspect of OSM data.)

• Optional integration with GEOS for advanced geometric operations, such as buffer, union, simplify, convex and concave hulls, Voronoi diagrams, and much more.

• Modest hardware requirements — any 64-bit Windows, Linux or MacOS system will run GeoDesk.

• Lightweight — the full query engine adds less than 250 KB and has no link-time dependencies. It runs entirely in-process, no database server required.

Get Started

Requirements

• CMake 3.14 or later
• C++20 compiler with a Standard Library for Windows, Linux or MacOS
• Java 16 or above (to run the GOL Tool)

Build & Link

If your project uses CMake, add this to your CMakeLists.txt:

include(FetchContent)
FetchContent_Declare(geodesk GIT_REPOSITORY 
    https://github.com/clarisma/libgeodesk.git)
FetchContent_MakeAvailable(geodesk)

target_link_libraries(my_program geodesk)

Alternatively, build GeoDesk explicitly:

git clone https://github.com/clarisma/libgeodesk.git
cd libgeodesk
mkdir build
cd build
cmake ..
cmake --build .

Create a GOL

Create a Geographic Object Library based on any .osm.pbf file, using the GOL Tool (Requires Java 16+).

For example:

gol build switzerland switzerland-latest.osm.pbf

Example Application

Find all the pubs in Zurich (Switzerland) and print their names:

#include <geodesk/geodesk.h>

using namespace geodesk;

int main(int argc, char* argv[])
{
    // Open switzerland.gol
    Features features("switzerland");      

    // Get the feature that represents the area of the city of Zurich
    Feature zurich = features(
        "a[boundary=administrative][admin_level=8]"
        "[name:en=Zurich]").one();

    // Define a set that contains nodes and areas that are pubs
    Features pubs = features("na[amenity=pub]");

    // Iterate through the pubs that are contained 
    // in the area of Zurich and print their names
    for (Feature pub: pubs.within(zurich)
    {
        std::cout << pub["name"] << std::endl;
    }
}

More Examples

Find all movie theaters within 500 meters from a given point:

Features movieTheaters = features("na[amenity=cinema]")
    .maxMetersFromLonLat(500, myLon, myLat);

Remember, OSM uses British English for its terminology.

Discover the bus routes that traverse a given street:

for (Relation route: street.parents("[route=bus]"))
{
    std::cout << route["ref"] 
        << " from " << route["from"] 
        << " to " << route["to"] << std::endl;
}

Count the number of entrances of a building:

int numberOfEntrances = building.nodes("[entrance]").count();

Documentation

• GeoDesk Developer's Guide
• API Reference

Related Repositories

• geodesk — GeoDesk for Java
• geodesk-py — GeoDesk for Python
• gol-tool — command-line utility for building, maintaining and querying GOL files

---

OpenStreetMap is a trademark of the OpenStreetMap Foundation, and is used with their permission. GeoDesk is not endorsed by or affiliated with the OpenStreetMap Foundation.