rdf-geopackage

Generate RDF out of a GeoPackage (for further processing)

Usage

Install using NPM locally npm install --global @rdmr-eu/rdf-geopackage as a command line tool.

Check if it's installed correctly with rdf-geopackage --help. That should return the following help info:

Generate RDF from an OGC GeoPackage with rdf-geopackage

Options:
      --help                   Show help                               [boolean]
      --version                Show version number                     [boolean]
  -i, --input                  GeoPackage file                          [string]
  -o, --output                 Output quads file                        [string]
      --format                 Override output format (default: nquads)
            [choices: "nq", "nquads", "trig", "nt", "ntriples", "ttl", "turtle"]
      --bbox                   Limit features to bounding box           [string]
      --bbox-crs               Coordinate Reference System code         [string]
      --only-layers            Only output named feature layers and attribute ta
                               bles                                      [array]
      --include-binary-values  Output binary values                    [boolean]
      --base-iri               Base IRI                                 [string]
      --model                  Data meta model             [choices: "facade-x"]
      --geosparql              Output GeoSPARQL
  [array] [choices: "wkt", "geojson", "bbox", "centroid", "length-area"] [defaul
                                                                     t: ["wkt"]]

Options

Basic input and output serializations can be set with the following options:

• --input: the path to the input GeoPackage file (required). With -, it reads the GeoPackage from stdin, e.g., piping a file with curl
• --output: path to the file output. By default, rdf-geopackage outputs nquads to stdout. Its extension sets the serialization format, optionally with .gz to GZip the output. E.g., --output myfile.ttl.gz
• --format: set the output format explicitly. Provide a file extension with .gz to GZip the output.

Work with large GeoPackages by limiting the output features, output tables and binary values:

• --bbox limits the the output features to those in this area (default CRS: WGS84)
• --bbox-crs indicates the CRS for the aforementioned bounding box. Supply a EPSG code (web lookup with EPSG.io) or a projection WKT.
• --only-layers limits which feature layers (or attribute tables!) are output.
• --include-binary-values overrides the default of skipping binary values. These will be base64 encoded string values with a ^^xsd:base64Binary data type. NULL values are never output.

Modify the model and types of the output triples or quads:

• --base-iri: set the relative base for the output RDF data. By default, this value is derived from the present working directory.
• --model: the GeoPackage tables are not natively RDF data, so a module is programmed to generating triples according to a data meta-model. Included modules:
  • default: facade-x
• --geosparql: modify which GeoSPARQL geometries, serializations and properties are output. Only the WKT literal is in layer native CRS, all other are calculated and/or projected. Multiple values (space separated) are allowed. Included feature predicates are listed in the table below.

| Option | GeoSPARQL feature predicates | Note | | ------------------- | ----------------------------------------- | --------------------------------------------------------------- | | wkt (default) | geo:hasGeometry/geo:asWKT | Output the feature's geometry as a WKT string literal | | geojson | geo:hasGeometry/geo:asGeoJSON | Output the feature's geometry as a WGS84 GeoJSON string literal | | bbox | geo:hasBoundingBox | Output a feature's bounding box | | centroid | geo:hasCentroid | Output a feature's centroid point | | length-area | geo:hasMetricLength geo:hasMetricArea | Output the length (in m) and area (in m²) of the feature | | always on | rdf:type geo:Feature | Output the feature class |

RDF output

Model: Facade-X

Facade-X is a data meta-model from the SPARQL-Anything project, that can easily represent tabular data. Facade-X uses RDF containers and blank nodes to represent tables and rows. Column metadata is currently very limited (GH-24) and many values are not typed properly.

Features, geometries and CRS’s

Features and their geometries are represented using GeoSPARQL. Only rows from feature tables are a geo:Feature.

A feature has zero or more geometries predicated with geo:hasDefaultGeometry. There might be no geometry if the underlying library does not support the geometry type. There may be multiple geometries if the feature is from a layer not in EPSG:4326.

That's because a GeoJSON serialization (geo:asGeoJSON) is always (reprojected) in EPSG:4326. A geo:Geometry can be in only one CRS, meaning that when the feature is not originally in EPSG:4326, other serializations should also be reprojected. That is undesirable, so in these cases, rdf-geopackage generates a second geo:Geometry for the WKT serialization (geo:asWKT).

Example RDF output

Example data abridged from NGA: the table media is a feature table, nga_properties is an attribute table.

prefix fx: <http://sparql.xyz/facade-x/ns/>
prefix geo: <http://www.opengis.net/ont/geosparql#>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix xsd: <http://www.w3.org/2001/XMLSchema#>
prefix xyz: <http://sparql.xyz/facade-x/data/>

xyz:nga_properties {  # representing a table
xyz:nga_properties a fx:root ;  # representing a table
  rdf:_1 [  # the first row
    xyz:id 14;
    xyz:property "subject";
    xyz:value "Examples"
  ] .
}

xyz:media {
xyz:media a fx:root ;
  rdf:_1 [
    a geo:Feature ;  # a row from a feature table
    xyz:text "BIT Systems";
    xyz:date "2023-01-23";
    geo:hasDefaultGeometry [  # single geometry as CRS is EPSG:4326
      a geo:Geometry ;
      geo:asWKT "POINT (-104.801918 39.720014)"^^geo:wktLiteral ;
      geo:asGeoJSON "{\"coordinates\":[-104.801918,39.720014],\"type\":\"Point\"}"^^geo:geoJSONLiteral
    ]
  ] .
}

Acknowledgements

This tool was developed for a project funded by the City Deal Openbare ruimte, executed by Stichting Kennisplatform CROW.