README.md

Horned-Bin

This is a set of command line tools built using the horned crate and provide core services for the manipulation of OWL ontologies.

Building

The tool suite is available as a single binary horned, collecting all tools as subcommands:

cargo build --release --bin horned

or as a collection of standalone binaries, each can be built using:

cargo build --release --bin horned-[TOOLNAME]

using the tools listed below.

Running

To use one of the utilities described below, run

horned [TOOLNAME] <ARGUMENT_LIST>

or

horned-[TOOLNAME] <ARGUMENT_LIST>

using the requested arguments.

Description

TOOLNAME	Arguments	Summary	Source file
big	n: unsigned integer	Generates an OWL file containing n class declarations.	source
compare	ont-1, ont-2: paths	Compares the statistics of ontologies specified in ont-1 and ont-2.	source
convert	ont: path, --to: format	Converts ont to another OWL serialisation format.	source
dump	ont: path	Parses ont and returns the content of the data structures created by the parser.	source
materialize	ont: path	Parses ont, downloading and resolving all of the ontologies imported by ont.	source
parse	ont: path	Parses ont and exits.	source
profile	ont: path	Reports which OWL 2 profile(s) (EL/QL/RL/DL) ont conforms to.	source
round	ont: path	Parses ont and renders the obtained ontology.	source
summary	ont: path	Parses ont and returns statistics related to the obtained ontology.	source
triples	ont: path	Parses ont as an ontology written using the OWL/RDF format and returns the obtained triples.	source
unparsed	ont: path	Parses ont as an ontology written using the OWL/RDF format and shows what has not been successfully parsed.	source
validate	ont: path	Parses ont and fails if any part of it could not be parsed.	source

Exit codes

Every tool exits 0 on success and 1 on failure (there are no other exit codes -- this is the default behaviour Rust gives a fn main() -> Result<(), E> when it returns Err, printing the error to stderr). Argument errors, I/O errors, and syntax errors during parsing always cause a non-zero exit, in every tool.

What differs between tools is whether a semantically incomplete parse also counts as failure. Some OWL/RDF/Turtle input can parse without a syntax error yet still leave part of the model unrepresented -- e.g. leftover triples an AnnotatedComponent couldn't be built from (see IncompleteParse). Whether that counts as success or failure depends on the tool:

• validate treats an incomplete parse as failure: it prints the unparsed remainder and exits 1. This is the tool to use in a script or CI check that needs to know an ontology parsed completely, not just that it parsed at all.
• parse, unparsed, dump, round, convert, summary, compare, materialize, and profile all exit 0 as long as parsing didn't throw a syntax/IO error, even if the resulting model is incomplete -- they discard the IncompleteParse the parser returns rather than checking it. unparsed (always) and dump (for RDF/XML/Turtle input, unless run with --incomplete/-l to suppress the full ontology dump) still print the unparsed remainder to stdout -- they just don't fail the process over it. This is deliberate, not an oversight: these tools' job is to show you what parsed, not to gate on completeness (see #135/#138, where validate was added specifically to fill that gap rather than changing parse's existing contract, which scripts may already depend on).
• big and triples don't parse a pre-existing ontology into a MutableOntology at all (big only generates one; triples reads raw RDF quads directly, bypassing the axiom-building parser entirely), so the concept of an incomplete parse doesn't apply to them.

If you need a hard failure on incomplete parses, use validate rather than relying on parse's exit code -- or inspect unparsed's/dump's printed output yourself if you need the incomplete remainder itself, not just a pass/fail signal.