Today we launch the OSMNames open-source project which provides a basic place search / geocoding functionality.
It is a lightweight search system powered by full-text engine with world-wide coverage. At the moment it is limited to place names like streets, towns, cities, administrative units, countries, etc.. It can be extended in future with other place types like lake or mountain names and even with POIs. The enhancement with postal addresses (house numbers and zip codes) is also possible. Try it at:
http://osmnames.org/ Install the software easily together with the provided global data on your own server with the docker container osmnames-sphinxsearch. The process is described in the documentation.
The OSMNames project consists of two parts:
OSMNames data available for download
The core of the OSMNames project is the pre-processed data. It comes in easy to handle text format, the TSV (tab separated value). This allows filtering with UNIX text processing tools (grep, sed, awk, less, cut, etc.) and direct indexing via a full-text search engine.
You can download the data at http://osmnames.org/download/
It contains over 21 million gazetteer records with lat/lon location, bounding box, names in several languages, identifiers for OSM and Wikidata, etc. All the records are sorted by precalculated numerical importance (derived from Wikipedia and existing Nominatim ranking). Our website recommends how to filter the data by country, geographical location or type to create smaller extracts.
Search engine software
The project comes with an open-source server software providing a place search API suitable for a basic geocoder with autocomplete functionality. The server is written in Python and is powered by the high-performance C++ open-source full-text engine called Sphinx Search.
The OSMNames project is a result of the cooperation of Klokan Technologies GmbH and HSR Rapperswil.
Klokan Technologies GmbH is using the service in products such as MapTiler or Georeferencer. We are excited to release it now and hope more users can help us to make it even better! Hope to see OSMNames adopted by more individuals and companies soon.er is written in Python and is powered by the high-performance C++ open-source full-text engine called Sphinx Search.