ConvRDF

Converts a file in some RDF formats such as RDF/XML, JSON-LD, Turtle to it in N-Triples.

ConvRDF converts a file in a streaming fashion, so that it can handle huge numbers of triples (> 1 billion) without consuming a huge amount of memory.

Compressed (.gz, .bz2, and .xz) files including .tar.gz can be processed properly.

NOTICE: This tool uses Apache Jena and Apache Commons libraries, which are released under the Apache License Version 2.0. In addition, XZ for Java is used for processing XZ files.

Requirements

• Java 17 or later (LTS recommended)
• Apache Jena 5.x (resolved automatically via Maven)

Build

mvn -B package
# produces target/ConvRDF.jar (self-contained, executable)

Usage

java -jar ConvRDF.jar [options] <file-or-directory>...

Options

Option	Description
-r	Recursively process directories. Default: off.
-c	Enable strict RDF syntax checking by Apache Jena. Default: off.
-i	Ignore lexical-level errors (bad IRIs, malformed literals, invalid Unicode escapes) with a warning, and continue. Structural syntax errors still abort the current file. Implies -c.
-b <URI>	Override the base URI used to resolve relative IRIs. Default: derived from the input file path.
-o <file>	Output file. Extensions .gz and .xz trigger compression. Default: standard output.
-h, --help	Show help and exit.

Exit status

Code	Meaning
0	All inputs were converted successfully.
1	One or more errors occurred (invalid arguments, unreadable files, parse failures, or I/O errors). Errors are reported to stderr.

Examples

# Convert a single Turtle file to N-Triples on stdout
java -jar ConvRDF.jar data.ttl > data.nt

# Convert an xz-compressed RDF/XML file, writing gzip-compressed N-Triples
java -jar ConvRDF.jar -o out.nt.gz ontology.owl.xz

# Recursively convert a tree, tolerating bad IRIs but reporting them
java -jar ConvRDF.jar -r -i -o all.nt.xz /data/rdf/

# Override the base URI used to resolve relative IRIs in a streamed input
java -jar ConvRDF.jar -b http://example.org/ontology -o out.nt input.owl.xz

---

Migration notes (v1 -> v2)

• Apache Jena upgraded to 5.x. Requires Java 17 or later.
• Streaming pipeline rewritten. The old PipedRDFIterator + thread-based producer/consumer pattern has been replaced with a direct StreamRDF pipeline. Memory use is now constant regardless of input size.
• Base URI is set explicitly when reading from compressed streams or tar entries. This fixes the previous {E211} Base URI is null failure on RDF/XML inputs that contain relative IRIs (for example rdf:about="") when the input was a .xz / .gz / .bz2 file.
• New option -i ignores lexical-level errors (bad IRIs, malformed literals, invalid Unicode escapes) with a warning, while still aborting on structural syntax errors.
• New option -b <URI> overrides the auto-derived base URI.
• Exit status is now well-defined: 0 on full success, 1 if any error occurred. Output is flushed and closed cleanly so .gz / .xz trailers are always written.