This repository contains the source code for a modified version of Apache Marmotta
Modifications are minor and prototypical and enable storage of polyhedra WKT and making use of PostGIS/SFCGAL 3d operators.
The provided dataset in the ./example/ folder is a subset of the 3d office building model provided by US ACE and buildingSMART alliance [1] converted to RDF using [2]. An example query for spaces without air inlets is given below:
PREFIX ifcowl: <http://www.buildingsmart-tech.org/ifcOWL/IFC2X3_TC1#>
PREFIX geo: <http://www.opengis.net/ont/geosparql#>
PREFIX geof: <http://www.opengis.net/def/function/geosparql/>
PREFIX express: <http://purl.org/voc/express#>
SELECT ?name WHERE {
?s a ifcowl:IfcSpace .
?s geo:hasGeometry/geo:asWKT ?sg .
?s ifcowl:name_IfcRoot/express:hasString ?name .
FILTER NOT EXISTS {
?t a ifcowl:IfcFlowTerminal .
?t geo:hasGeometry/geo:asWKT ?tg .
?rel ifcowl:relatedObjects_IfcRelDefines ?t .
?rel ifcowl:relatingType_IfcRelDefinesByType ?type .
?type a ifcowl:IfcAirTerminalType .
FILTER(geof:distance(?tg, ?sg) < 0.01)
}
}
Which translates to the following SQL query. Take note of the the CASE statement to switch between a 2d and 3d distance implementation.
SELECT P5.object AS V5
FROM triples P1
CROSS JOIN triples P2
CROSS JOIN triples P3
INNER JOIN nodes AS P3_object_V3 ON P3.object = P3_object_V3.id
CROSS JOIN triples P4
CROSS JOIN triples P5
WHERE P1.deleted = false
AND P1.predicate = 982184394049863680
AND P1.object = 982184507212185602
AND P2.deleted = false
AND P1.subject = P2.subject
AND P2.predicate = 982184530184388611
AND P3.deleted = false
AND P2.object = P3.subject
AND P3.predicate = 982184530196971522
AND P4.deleted = false
AND P1.subject = P4.subject
AND P4.predicate = 982184434302599170
AND P5.deleted = false
AND P4.object = P5.subject
AND P5.predicate = 982184433149165571
AND NOT ( EXISTS (
SELECT *
FROM triples _P1
CROSS JOIN triples _P2
CROSS JOIN triples _P3
INNER JOIN nodes AS _P3_object_V3 ON _P3.object = _P3_object_V3.id
CROSS JOIN triples _P4
CROSS JOIN triples _P5
CROSS JOIN triples _P6
WHERE _P1.deleted = false
AND _P1.predicate = 982184394049863680
AND _P1.object = 982184434369708034
AND _P2.deleted = false
AND _P1.subject = _P2.subject
AND _P2.predicate = 982184530184388611
AND _P3.deleted = false
AND _P2.object = _P3.subject
AND _P3.predicate = 982184530196971522
AND _P4.deleted = false
AND _P1.subject = _P4.object
AND _P4.predicate = 982184468532314120
AND _P5.deleted = false
AND _P4.subject = _P5.subject
AND _P5.predicate = 982184468964327424
AND _P6.deleted = false
AND _P5.object = _P6.subject
AND _P6.predicate = 982184394049863680
AND _P6.object = 982184468976910338
AND _P3_object_V3.gvalue IS NOT NULL
AND P3_object_V3.gvalue IS NOT NULL
AND ( CASE
WHEN (ST_CoordDim(_P3_object_V3.gvalue) = 3 and ST_CoordDim(P3_object_V3.gvalue) = 3)
THEN (ST_3DDistance(_P3_object_V3.gvalue, P3_object_V3.gvalue))
ELSE (ST_Distance(_P3_object_V3.gvalue, P3_object_V3.gvalue))
END ) < 0.01
) )
A graphical depiction of the query resulted
A similar approach has been implemented for geof:sfIntersects. A more elaborate (but not performant) variant that appeared most reliable for (volumetric!) solid intersections.
CASE
WHEN (
ST_CoordDim({0}) = 3 and
ST_CoordDim({1}) = 3 and
ST_NumPatches({0}) is not null and
ST_NumPatches({1}) is not null
)
THEN (
CASE
WHEN {0} &&& {1}
THEN (
( ST_Volume(ST_3Dunion(
st_makesolid({0}),
st_makesolid({1})
)) + 0.000001 ) <
ST_Volume(st_makesolid({0})) +
ST_Volume(st_makesolid({1}))
)
ELSE FALSE
END
)
ELSE ST_Intersects({0}, {1})
END
Users are advised to use the Dockerfile.
Further reading:
Pauwels P., Krijnen T., Terkaj W., Beetz J. (2017) Enhancing the ifcOWL ontology with an alternative representation for geometric data, Automation in Construction, Volume 80, Pages 77-94, https://doi.org/10.1016/j.autcon.2017.03.001.
[1] https://www.nibs.org/?page=bsa_commonbimfiles#project2 [2] https://github.com/IDLabResearch/IFC-to-RDF-converter
- ... official readme continues below ... *
Apache Marmotta uses Maven to build, test, and install the software. A basic build requires downloading and installing Maven and then running:
mvn clean install
This will compile, package and test all Apache Marmotta modules and install it in your local Maven repository. In case you want to build your own projects based on some of the libraries provided by Apache Marmotta, this usually suffices.
The default loglevel for most unit and integration tests executed during the build is INFO. To change the loglevel for either more or less output, you can pass the loglevel as system property:
mvn clean install -Droot-level=TRACE|DEBUG|INFO|WARN|ERROR
Note that some of the integration tests start up parts of the Marmotta platform during execution. The log level for these tests cannot be changed, as Marmotta is taking over the log configuration in these cases.
Apache Marmotta also includes a default configuration for building a Java Web Application that can be deployed in any Java Application Server. To build the web application, first run
mvn clean install
in the project root. Then change to the launchers/marmotta-webapp directory and run
mvn package
This will create a marmotta.war file in the target/ directory. You can deploy this archive to any Java Application Server by copying it into its deployment directory; more details.
Alternatively, you can directly startup the Apache Marmotta Web Application from Maven with a default configuration suitable for development. To try this out, run
mvn tomcat7:run
wait until the system is started up and point your browser to http://localhost:8080
When developing it is sometimes useful to always start with a clean confi- guration of the system. Therefore, you can also start up the web application as follows:
mvn clean tomcat7:run -Pcleanall
This command will remove any existing configuration directory before startup.
The build environment also offers to automatically build an installer package that guides users through the installation with an easy-to-use installation wizard. The installer is based on izPack and dynamically assembled when building the package. To build the installer, first run
mvn clean install
in the project root. Then change to the launchers/marmotta-installer directory and run
mvn package -Pinstaller
The build process will automatically create an appropriate installer confi- guration from the Maven dependencies through the Apache Marmotta refpack build plugin.
The installer can then be tried out by running
java -jar target/marmotta-installer-x.x.x.jar
Marmotta also comes with support for creating a Docker images that you can use for development or testing:
- Locate at the root of the source repository
- Build image:
docker build -t marmotta . - Run the container:
docker run -p 8080:8080 marmotta - Access Marmotta at localhost:8080/marmotta (IP address may be different, see information bellow).
An official images is available from Docker Hub as an automated build, so you just need to pull it from there to replace the second step above:
docker pull apache/marmotta
Sometimes it is useful to check if the build runs properly on a clean local repository, i.e. simulate what happens if a user downloads the source and runs the build. This can be achieved by running Maven as follows:
mvn clean install -Dmaven.repo.local=/tmp/testrepo
The command changes the local repository location from ~/.m2 to the directory passed as argument
To test the release build without actually deploying the software, we have created a profile that will deploy to the local file system. You can simulate the release by running
mvn clean deploy -Pdist-local,marmotta-release,installer
Please keep in mind that building a release involves creating digital signatures, so you will need a GPG key and a proper GPG configuration to run this task.
Read more about our release process.