togs: a simple feature based geocoder

The name togs stands for trivial OGR-based geocoding server, and togs is just that. It takes a number of OGR readable feature files (ESRI shapefiles, GeoJSONs, etc.) and turns them into Nominatim -like searchable intefaces.

Togs has been built to provide search functionality for planetary maps built with VTS 3D geospatial software stack atop of IAU planetary nomenclature files but it may be readily used with other libraries and tools.

Installation and configuration

To install locally, do:

	$ npm install togs

Now edit the togs configuration file at $(npm root)/togs/conf/togs.conf to include definition of your geocoding interfaces.

Togs comes with an example dataset, a GeoJSON containing a list of power plants in the state of Washington. The corresponding interface is configured as follows:

	[interface.powerplants-wa]
	; path to the dataset
	dataset = ./datasets/GIS_Layer__PowerPlants_WA.geojson

	; property uniquely identifying each feature 
	idProperty = UTILITY_ID

	; properties matching this regexp shall not be indexed  
	searchExclude = ^Latitude|Longitude$

	; templates for some of the response entry values
	class=landuse
	type=industrial
	displayName = {PLANT_NAME}, {UTILITY_NA} ({primary_fu})
	addressRegion = {PLANT_NAME}
	addressState = {UTILITY_NA}

The most important parts of this configuration are the path to the dataset (relative to $(npm root)/togs), the property considered to be feature ID (idProperty) and the template for display name (displayName). In all templates, property names enclosed in braces are expanded to their corresponding values when geocoder response is formed.

Usage

You can start the server as follows:

	$ $(npm root)/.bin/togs

Pointing your browser to

	http://<yourserver:8100>/powerplants-wa?q=wind&format=json&limit=1

will give you the following output:

    [
    	{
            "lon": -120.75305600043436,
            "lat": 47.13916700003981,
            "boundingbox": [
                47.13916700003981,
                47.13916700003981,
                -120.75305600043436,
                -120.75305600043436
            ],
            "display_name": "Swauk Wind LLC, Gamesa Wind US LLC (WND)",
            "class": "landuse",
            "type": "industrial",
            "importance": 1,
            "icon": null,
            "address": {
                "region": "Swauk Wind LLC",
                "state": "Gamesa Wind US LLC"
            },
            "place_id": null,
            "osm_type": null,
            "licence": null,
            "osm_id": null
        }
    ]

Limitations

Nominatim geocoding interface is OSM specific. Togs input files are not OSM, thus togs can only mimic Nominatim output format: place_id, osm_type and osm_id are all set to null in togs output.

Togs relies on elasticlunr for in-memory indexing and stores every dataset in memory in its entirety. It won't scale to very large datasets.

Currently, JSON is the only supported output format, though adding support for XML is probably trivial.

Development

    $ git clone https://github.com/melown/togs.git
    $ npm install
    $ node bin/togs