GitBucket
Treat a directory of SPARQL queries as a test suite.
|
|
|||
|---|---|---|---|
| project | 5 years ago | ||
| src/ main | 4 years ago | ||
| .dockerignore | 5 years ago | ||
| .gitignore | 5 years ago | ||
| Dockerfile | 5 years ago | ||
| LICENSE | 6 years ago | ||
| README.md | 6 years ago | ||
| build.sbt | 5 years ago | ||
SPARQL test runner
SPARQL can be used to validate assumptions made when curating data. This project goes through a given 'tests' directory looking for files ending in .sparql, and runs the queries against a given RDF dataset or datasets.
The results of each query should either be empty - for SELECT style queries, or TRUE for ASK style queries. More complex results can be checked by creating a file with the expected results and giving it a .expected suffix. The actual results are checked against the expected results - note that the output is in CSV format, so line endings are as for DOS (carriage return, linefeed).
The result of running the SPARQL test runner will be a testresults.xml file in the current directory, in the style of jUnit XML. Errors are also output to stdout/stderr for humans. The exit code of the test runner will be 0 for success, and 1 if there are any errors.
The project essentially bundles the Apache Jena ARQ libraries along with a simple test runner.
The project can be built as a standalone "fat JAR", containing all dependencies such that it can be used directly without having to install anything other than Java. Java version 7 up is supported.
The resulting standalone file can also be used to simply run any SPARQL 1.1 query on any local or remote data, including using the SERVICE keyword to mix together external SPARQL endpoints with local data, etc.
Building
Build using the Scala Build Tool [1]. To build the fat JAR, run 'sbt assembly'. This will result in a single executable file under target/scala-2.11/sparql, which can be copied wherever needed. The file has a prolog invoking Bash to run Java on the embedded JAR.
Running
Running the 'sparql' executable is the same as running Apache Jena's sparql command, but just includes all dependent JARs and classes.
The test runner can be run as follows:
java -cp sparql uk.org.floop.sparqlTestRunner.Run
and will describe usage:
Usage: sparql-testrunner [options] <file>...
-t <dir> | --testdir <dir>
location of SPARQL queries to run, defaults to tests/sparql
<file>...
data to run the queries against
The SPARQL queries are expected to live under the tests/sparql directory, which will be recursed into looking for files ending with .sparql.
The RDF data can be in any format recognizable by Apache Jena, including the quads formats. To keep things simple, quads are treated as triples so that the queries range over the union of all graphs.