How to import .trig files with named graphs to fuseki-geosparql

[navarral]

Version

4.7.0

Question

Hi, I am trying to load .trig file in fuseki-geosparql that has a named graph, but the .trig file does not load correctly resulting into an empty dataset. However, when Ioad the file in the fuseki-server it works.

Do you have any advice on what to include in the command to make it work for fuseki-geosparql?

Please find below the necessary information to reproduce the error,

I have downloaded jena-fuseki-geosparql-4.7.0.jar from https://repo1.maven.org/maven2/org/apache/jena/jena-fuseki-geosparql/4.7.0/ .

I am using the following command: java -jar jena-fuseki-geosparql-4.7.0.jar -rf "test-geosparql.trig" -i

Error I get is:

[2023-02-27 16:57:47] DatasetOperations INFO Reading RDF - Started - File: test-geosparql.trig, Graph Name: , RDF Format: Turtle/pretty [2023-02-27 16:57:47] riot WARN Only triples or default graph data expected : named graph data ignored [2023-02-27 16:57:47] DatasetOperations INFO Reading RDF - Completed - File: test-geosparql.trig, Graph Name: , RDF Format: Turtle/pretty [2023-02-27 16:57:48] GeoSPARQLOperations INFO Applying GeoSPARQL Schema - Started [2023-02-27 16:57:48] GeoSPARQLOperations INFO GeoSPARQL schema not applied to empty graph: default [2023-02-27 16:57:48] GeoSPARQLOperations INFO Applying GeoSPARQL Schema - Completed [2023-02-27 16:57:48] DatasetOperations WARN Dataset empty. Spatial Index not constructed. Server will require restarting after adding data and any updates to build Spatial Index.

The test-geosparql.trig is the following (I have removed most of the triples just to leave a minimal example):

@prefix rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns# . @prefix rdfs: http://www.w3.org/2000/01/rdf-schema# . @prefix xsd: http://www.w3.org/2001/XMLSchema# . @prefix owl: http://www.w3.org/2002/07/owl# . @prefix dct: http://purl.org/dc/terms/ . @prefix qb: http://purl.org/linked-data/cube# .

https://example.org/graph-A { https://example.org/dataset-1 a qb:DataSet ; dct:title "Test Dataset"@en ; . }

[rvesse]

This looks like a limitation of the code, hopefully @galbiston can confirm this.

The code looks to assume that each RDF file is read into a specific model and doesn't properly support reading an entire dataset from a TriG file AFAICT - https://github.com/apache/jena/blob/64f96d76324c615a7a98ec96d6f5d97168141359/jena-fuseki2/jena-fuseki-geosparql/src/main/java/org/apache/jena/fuseki/geosparql/DatasetOperations.java#L147-L162

A workaround would be to create a TDB dataset with other Jena commands first and then use the --tdb option to point to that instead

[rvesse]

Also seems like the code assumes a default format of Turtle regardless of the file format unless you explicitly declare a format (but even then the above noted code would still cause this use case not to work)

https://github.com/apache/jena/blob/64f96d76324c615a7a98ec96d6f5d97168141359/jena-fuseki2/jena-fuseki-geosparql/src/main/java/org/apache/jena/fuseki/geosparql/cli/RDFFileParameter.java#L53-L59

[galbiston]

As @rvesse has noted the file reading relies on graph name being specified rather than read from the file.

It should be possible to set the graph name using the arg:

-rf test-geosparql.trig#https://example.org/graph-A&trig

There is also this warning from RIOT though:

[2023-02-27 16:57:47] riot WARN Only triples or default graph data expected : named graph data ignored

Setting up the data in TDB first is another option and then launching with -t arg to use the TDB dataset: -t <my_tdb> for TDB or -t <my_tdb> -t2 for TDB2. There are several options that may be wanted on first use to add additional GeoSPARQL triples.

[navarral]

@rvesse @galbiston thanks for looking into it!

I tried the following but it does not load correctly the test-geosparql.trig file:

1. I loaded a sample RDF file, that I know it works, to set up a dataset in a "TestTDB"

java -jar jena-fuseki-geosparql-4.7.0.jar -rf "EU-nuts-rdf-geosparql.ttl" -t "TestTDB" -i

2. I tested a GeoSPARQL query and it works just fine but then, I try to upload the .trig file following your guidance:

java -jar jena-fuseki-geosparql-4.7.0.jar -rf "test-geosparql.trig#https://example.org/graph-A&trig" -t "TestTDB" -i

but the data does not load. I ran a query and it only finds the first file from 1. . I have also pasted http://localhost:3030/ds to export all the data and it is not there.

[2023-03-09 10:32:26] Main INFO Arguments Received: [-rf, test-geosparql.trig#https://example.org/graph-A&trig, -t, TestTDB] [2023-03-09 10:32:26] DatasetOperations INFO Server Configuration: port=3030, datsetName=ds, loopbackOnly=true, updateAllowed=false, inference=false, applyDefaultGeometry=false, validateGeometryLiteral=false, convertGeoPredicates=false, removeGeoPredicates=false, queryRewrite=true, tdbFile=TestTDB, fileGraphFormats=[FileGraphFormat{rdfFile=test-geosparql.trig, graphName=https://example.org/graph-A&trig, rdfFormat=Turtle/pretty}], fileGraphDelimiters=[], indexEnabled=true, indexSizes=[-1, -1, -1], indexExpiries=[5000, 5000, 5000], spatialIndexFile=null, tdb2=false, transformGeometry=true, help=false [2023-03-09 10:32:26] DatasetOperations INFO TDB Dataset: TestTDB, TDB2: false [2023-03-09 10:32:26] system WARN The “SIS_DATA” environment variable is not set. [2023-03-09 10:32:27] DatasetOperations INFO Reading RDF - Started - File: test-geosparql.trig, Graph Name: https://example.org/graph-A&trig, RDF Format: Turtle/pretty [2023-03-09 10:32:27] riot WARN Only triples or default graph data expected : named graph data ignored [2023-03-09 10:32:27] DatasetOperations INFO Reading RDF - Completed - File: test-geosparql.trig, Graph Name: https://example.org/graph-A&trig, RDF Format: Turtle/pretty

3. I tried with specifying the conversion to .trig with ">trig", but I got the same results

java -jar jena-fuseki-geosparql-4.7.0.jar -rf "test-geosparql.trig#https://example.org/graph-A&trig>trig" -t "TestTDB" -i

[2023-03-09 10:31:31] Main INFO Arguments Received: [-rf, test-geosparql.trig#https://example.org/graph-A&trig>trig, -t, TestTDB] [2023-03-09 10:31:31] DatasetOperations INFO Server Configuration: port=3030, datsetName=ds, loopbackOnly=true, updateAllowed=false, inference=false, applyDefaultGeometry=false, validateGeometryLiteral=false, convertGeoPredicates=false, removeGeoPredicates=false, queryRewrite=true, tdbFile=TestTDB, fileGraphFormats=[FileGraphFormat{rdfFile=test-geosparql.trig, graphName=https://example.org/graph-A&trig, rdfFormat=TriG/pretty}], fileGraphDelimiters=[], indexEnabled=true, indexSizes=[-1, -1, -1], indexExpiries=[5000, 5000, 5000], spatialIndexFile=null, tdb2=false, transformGeometry=true, help=false [2023-03-09 10:31:31] DatasetOperations INFO TDB Dataset: TestTDB, TDB2: false [2023-03-09 10:31:31] system WARN The “SIS_DATA” environment variable is not set. [2023-03-09 10:31:31] DatasetOperations INFO Reading RDF - Started - File: test-geosparql.trig, Graph Name: https://example.org/graph-A&trig, RDF Format: TriG/pretty [2023-03-09 10:31:31] riot WARN Only triples or default graph data expected : named graph data ignored [2023-03-09 10:31:31] DatasetOperations INFO Reading RDF - Completed - File: test-geosparql.trig, Graph Name: https://example.org/graph-A&trig, RDF Format: TriG/pretty

Do you have any further workarounds or command options that I could try?

Many thanks,

[galbiston]

Sorry, by suggesting setting up the TDB first I was meaning using the tdbloader to import the *.trig file and create the TDB dataset. Then run GeoSPARQL Fuseki pointing at that new dataset. This should give better support for file loading.

[navarral]

Apologies but I am new to Jena and I might have missed something when following the tdbloader documentation. I have tried the following options:

1. Set up a dataset with geosparql-fuseki and then load with tdbloader:

bin/tdbloader --loc /path/for/database/TestTDB test-geosparql.trig

11:13:13 INFO loader :: -- Start triples data phase 11:13:13 INFO loader :: Load into triples table with existing data 11:13:13 INFO loader :: -- Start quads data phase 11:13:13 INFO loader :: Load into quads table with existing data 11:13:13 INFO loader :: Load: test-geosparql.trig -- 2023/03/09 11:13:13 GMT 11:13:13 INFO loader :: -- Finish triples data phase 11:13:13 INFO loader :: -- Finish quads data phase 11:13:13 INFO loader :: Data: 2 quads loaded in 0.25 seconds [Rate: 7.97 per second] 11:13:13 INFO loader :: -- Start quads index phase 11:13:13 INFO loader :: -- Finish quads index phase 11:13:13 INFO loader :: -- Finish triples load 11:13:13 INFO loader :: -- Finish quads load 11:13:13 INFO loader :: Completed: 2 quads loaded in 0.26 seconds [Rate: 7.78 per second]

But it does not load correctly as I cannot see it there with java -jar jena-fuseki-geosparql-4.7.0.jar -t "TestTDB"

2. Set up a TDB dataset with bin/tdb1.xloader --loc /path/for/database/TestTDB2 test-geosparql.trig

11:18:49 INFO -- TDB1 Bulk Loader Start 11:18:49 INFO Data Load Phase 11:18:49 INFO Got 1 data files to load 11:18:49 INFO Data file 1: /path/for/file/test-geosparql.trig 11:18:50 INFO loader :: Total: 2 tuples : 0.18 seconds : 10.87 tuples/sec [2023/03/09 11:18:50 GMT] 11:18:50 INFO Data Load Phase Completed 11:18:50 INFO Index Building Phase 11:18:50 INFO Creating Index GSPO 11:18:50 INFO Sort GSPO 11:18:50 INFO Sort GSPO Completed 11:18:50 INFO Build GSPO 11:18:51 INFO Build GSPO Completed 11:18:51 INFO Creating Index GPOS 11:18:51 INFO Sort GPOS 11:18:51 INFO Sort GPOS Completed 11:18:51 INFO Build GPOS 11:18:52 INFO Build GPOS Completed 11:18:52 INFO Creating Index GOSP 11:18:52 INFO Sort GOSP 11:18:52 INFO Sort GOSP Completed 11:18:52 INFO Build GOSP 11:18:53 INFO Build GOSP Completed 11:18:53 INFO Creating Index SPOG 11:18:53 INFO Sort SPOG 11:18:53 INFO Sort SPOG Completed 11:18:53 INFO Build SPOG 11:18:54 INFO Build SPOG Completed 11:18:54 INFO Creating Index POSG 11:18:54 INFO Sort POSG 11:18:54 INFO Sort POSG Completed 11:18:54 INFO Build POSG 11:18:54 INFO Build POSG Completed 11:18:54 INFO Creating Index OSPG 11:18:54 INFO Sort OSPG 11:18:54 INFO Sort OSPG Completed 11:18:54 INFO Build OSPG 11:18:55 INFO Build OSPG Completed 11:18:55 INFO Index Building Phase Completed 11:18:55 INFO -- TDB1 Bulk Loader Finish 11:18:55 INFO -- 6 seconds

But then, when I run java -jar jena-fuseki-geosparql-4.7.0.jar -t "TestTDB2" I get the following error

WARNING: sun.reflect.Reflection.getCallerClass is not supported. This will impact performance. [2023-03-09 11:26:55] Main INFO Arguments Received: [-t, TestTDB2] [2023-03-09 11:26:55] DatasetOperations INFO Server Configuration: port=3030, datsetName=ds, loopbackOnly=true, updateAllowed=false, inference=false, applyDefaultGeometry=false, validateGeometryLiteral=false, convertGeoPredicates=false, removeGeoPredicates=false, queryRewrite=true, tdbFile=TestTDB2, fileGraphFormats=[], fileGraphDelimiters=[], indexEnabled=true, indexSizes=[-1, -1, -1], indexExpiries=[5000, 5000, 5000], spatialIndexFile=null, tdb2=false, transformGeometry=true, help=false [2023-03-09 11:26:55] DatasetOperations INFO TDB Dataset: TestTDB2, TDB2: false [2023-03-09 11:26:56] system WARN The “SIS_DATA” environment variable is not set. [2023-03-09 11:26:56] GeoSPARQLOperations INFO Find Mode SRS - Started [2023-03-09 11:26:56] GeoSPARQLOperations INFO Find Mode SRS - Completed [2023-03-09 11:26:56] Main ERROR GeoSPARQL Server: Exiting - No SRS found. Check 'http://www.opengis.net/ont/geosparql#hasSerialization' or 'http://www.w3.org/2003/01/geo/wgs84_pos#lat'/'http://www.w3.org/2003/01/geo/wgs84_pos#lon' predicates are present in the source data. Hint: Inferencing with GeoSPARQL schema may be required.: ds

3. I have tried to import the sample RDF file that works together with the .trig file but I get the same as in 1.

Am I missing something? Could you help me with the list of commands I should be using to create a TDB dataset and to be used in geosparql-fuseki?

Thanks again for your time

[galbiston]

Just to confirmat that the working directory when running the jena-fuseki-geosparql-4.7.0.jar command is the /path/for/database folder? Otherwise you would need to include the path as part of the arg -t "TestTDB2".

Just looking at your minimal sample and there are no GeoSPARQL prefixes. The error Main ERROR GeoSPARQL Server: Exiting - No SRS found. is saying that the server cannot find any recognised geospatial information to complete setup and stopping. You will need to expand your minimal sample to include geospatial information.

[SimonBin]

Everything said here is right. But if you just want to quickly play with geosparql, you can also try @Aklakan 's RdfProcessingToolkit powered by Apache Jena

java -jar rpt-1.9.3-rc3.jar integrate NGraph_GeohiveData.trig  --geoindex --server

then run a geo query:

PREFIX uom: <http://www.opengis.net/def/uom/OGC/1.0/>
PREFIX geo: <http://www.opengis.net/ont/geosparql#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX spatial: <http://jena.apache.org/spatial#>
select distinct ?g ?feature ?geo ?geoLabel ?myPoint ("red" as ?myPointColor) {
  values ?myPoint {
    "Point(-6.22675717146472 53.3681724340328)"^^geo:wktLiteral
  }
  graph ?g {
    ?feature spatial:nearbyGeom(?myPoint 3 uom:kilometre) .
    ?feature rdfs:label ?geoLabel ;
             geo:hasGeometry/geo:asWKT ?geo .
  }
}

you can use https://yasgui.triply.cc/ to visualise the shapes

[navarral]

I expanded the minimal example to include geosparql information but it still did not work.

@prefix rdf:            <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs:           <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xsd:            <http://www.w3.org/2001/XMLSchema#> .
@prefix owl:            <http://www.w3.org/2002/07/owl#> .
@prefix geo:            <http://www.opengis.net/ont/geosparql#> .

@prefix dct:            <http://purl.org/dc/terms/> .
@prefix qb:             <http://purl.org/linked-data/cube#> .
@prefix locn:           <http://www.w3.org/ns/locn#> .

# -- Named Graph --------------
<https://example.org/graph-A> {
# -- Data Set -------------------
<https://example.org/dataset-1>
    a           qb:DataSet , geo:Feature ;
    dct:title       "Test Dataset"@en ;
    locn:geometry   <https://example.org/dataset-1-geo> ;
.

<https://example.org/dataset-1-geo> 
    a       geo:Geometry ;
    geo:asWKT   "POINT(10.5341 42.9391)"^^geo:wktLiteral ;
.
}

@galbiston I confirm that the /path/for/database/ is the same in both of the following commands

bin/tdbloader --loc /path/for/database/TestTDB test-geosparql.trig

07:54:51 INFO loader :: -- Start triples data phase 07:54:51 INFO loader :: Load empty triples table 07:54:51 INFO loader :: -- Start quads data phase 07:54:51 INFO loader :: Load into quads table with existing data 07:54:51 INFO loader :: Load: test-geosparql.trig -- 2023/03/13 07:54:51 GMT 07:54:52 INFO loader :: -- Finish triples data phase 07:54:52 INFO loader :: -- Finish quads data phase 07:54:52 INFO loader :: Data: 6 quads loaded in 0.27 seconds [Rate: 22.39 per second] 07:54:52 INFO loader :: -- Start quads index phase 07:54:52 INFO loader :: -- Finish quads index phase 07:54:52 INFO loader :: -- Finish triples load 07:54:52 INFO loader :: -- Finish quads load 07:54:52 INFO loader :: Completed: 6 quads loaded in 0.28 seconds [Rate: 21.43 per second]

java -jar jena-fuseki-geosparql-4.7.0.jar -t "/path/for/database/TestTDB"

WARNING: sun.reflect.Reflection.getCallerClass is not supported. This will impact performance. [2023-03-13 08:05:08] Main INFO Arguments Received: [-t, TestTDB] [2023-03-13 08:05:08] DatasetOperations INFO Server Configuration: port=3030, datsetName=ds, loopbackOnly=true, updateAllowed=false, inference=false, applyDefaultGeometry=false, validateGeometryLiteral=false, convertGeoPredicates=false, removeGeoPredicates=false, queryRewrite=true, tdbFile=TestTDB, fileGraphFormats=[], fileGraphDelimiters=[], indexEnabled=true, indexSizes=[-1, -1, -1], indexExpiries=[5000, 5000, 5000], spatialIndexFile=null, tdb2=false, transformGeometry=true, help=false [2023-03-13 08:05:08] DatasetOperations INFO TDB Dataset: TestTDB, TDB2: false [2023-03-13 08:05:08] system WARN The “SIS_DATA” environment variable is not set. [2023-03-13 08:05:08] GeoSPARQLOperations INFO Find Mode SRS - Started [2023-03-13 08:05:08] GeoSPARQLOperations INFO Find Mode SRS - Completed [2023-03-13 08:05:08] SpatialIndex INFO Saving Spatial Index - Started: /path/for/database/TestTDB/spatial.index [2023-03-13 08:05:08] SpatialIndex INFO Saving Spatial Index - Completed: /path/for/database/TestTDB/spatial.index [2023-03-13 08:05:08] GeosparqlServer INFO GeoSPARQL Server: Running - Port: 3030, Dataset: /ds, Loopback Only: true, Allow Update: false [2023-03-13 08:05:08] Server INFO Start Fuseki (http=3030)

The .trig files seems to be uploaded but not accessible as it returns an empty result:

[2023-03-13 08:05:41] Fuseki INFO [1] POST http://localhost:3030/ds [2023-03-13 08:05:41] Fuseki INFO [1] Query = PREFIX rdfs: http://www.w3.org/2000/01/rdf-schema# PREFIX xsd: http://www.w3.org/2001/XMLSchema# PREFIX qb: http://purl.org/linked-data/cube# SELECT * { ?s ?p ?o } LIMIT 5 [2023-03-13 08:05:41] Fuseki INFO [1] 200 OK (136 ms)

Am I missing something?

@SimonBin thanks for your answer! I will keep it in mind for certain use cases. However, in this particular one, I am looking for an open-source triplestore with GeoSPARQL support (like Jena) to import ~50Milion triples to run GeoSPARQL queries offline as there is some health data involved. Any advice would be welcome.

I would have preferred to use geosparql-fuseki for the full GeoSPARQL support but if it does not support named graphs I might try to use jena-fuseki with GeoSPARQL enabled. Is there a ready to use fuseki server with GeoSPARQL enabled?

[SimonBin]

@navarral if you are working with named graphs then of course select ?s ?p ?o will turn up empty! try ```select { graph https://example.org/graph-A { ?s ?p ?o } }```

[navarral]

You are right! I totally missed that, I was using an old script to test the query. Thanks for pointing it out, it is working now.