Apache Jena CLI tools
Jelly-JVM fully supports Apache Jena's command-line interface (CLI) utilities.
Parsing
Jena will automatically detect Jelly files based on their extension (.jelly, .jelly.gz) and parse them. You can also manually set the --syntax option to jelly.
Writing
You can use Jelly as an output format for Jena's CLI utilities by specifying the --output or --stream options with the jelly format. We recommend using the --stream option for better performance.
By default Jena will use the "big, strict" Jelly preset (name table: 4000 entries, prefix table: 150, datatype table: 32). You may want to change it to better fit your data, for example to use less memory with small files.
The following presets are available:
SMALL_STRICT– 128 name table entries, 16 prefix table entries, 16 datatype table entriesBIG_STRICT(default) – 4000 name table entries, 150 prefix table entries, 32 datatype table entries (recommended for larger files)
The presets SMALL_GENERALIZED, SMALL_RDF_STAR, SMALL_ALL_FEATURES, BIG_GENERALIZED, BIG_RDF_STAR and BIG_ALL_FEATURES still work, but are deprecated and will be removed in Jelly-JVM 5.0.0. Their RDF-star and generalized RDF flags only apply to Jelly-RDF 1.0 and 1.1 output. Jelly-RDF 1.2, which Jena writes, never has generalized statements and always allows triple terms as objects, so for it they are the same as SMALL_STRICT and BIG_STRICT.
To use one of these presets, use the --set CLI option with the https://neverblink.eu/jelly/riot/symbols#preset symbol:
Example: converting a Turtle file to Jelly with a big preset (strict)
Example: dumping a TDB2 database to Jelly with a small preset