Skip to main content

generate_all_processes.py

What it is​

A script that generates a set of Turtle (.ttl) ontology files describing NEXUS “processes” in a BFO 7 Buckets-style structure. Each generated file is based on a static in-code catalog (ALL_PROCESSES) and written into a processes/ directory grouped by folder.

Public API​

  • ttl_string(s: str) -> str
    • Escapes a Python string for use as a Turtle literal (escapes backslashes, quotes, newlines).
  • write_process(folder, filename, label, definition, impl, who, where, when, how_know, how_is, why_roles, why_disps, steps, data_props=None, precedes=None, preceded_by=None) -> pathlib.Path
    • Writes one .ttl file describing a process ontology:
      • Adds common RDF/BFO/NEXUS prefixes.
      • Creates an owl:Ontology header with dc:title, dc:description, dc:created, and owl:imports.
      • Creates an owl:Class for the process as a subclass of bfo:BFO_0000015 with label/definition/comment/example steps.
      • Optionally emits bucket classes for: WHO/WHERE/WHEN/HOW WE KNOW/HOW IT IS/WHY roles/WHY dispositions.
      • Optionally emits a small set of object properties linking the process class to the first item in each bucket (if present), plus preceded_by if provided.
      • Optionally emits datatype properties from data_props.
  • main() -> None
    • Iterates over ALL_PROCESSES, calls write_process(**proc) for each, prints a per-file log and a folder summary.

Configuration/Dependencies​

  • Filesystem output location
    • OUTPUT_DIR = Path(__file__).parent.parent / "processes"
    • Files are written to: OUTPUT_DIR / folder / f"{filename}.ttl".
  • Runtime dependencies
    • Python standard library only (pathlib.Path).
  • Input data
    • ALL_PROCESSES: populated at import time via helper p(**kwargs) calls in this file.

Usage​

Run as a script:

# from a shell
python libs/naas-abi/naas_abi/apps/nexus/ontology/scripts/generate_all_processes.py

Or call programmatically:

from naas_abi.apps.nexus.ontology.scripts.generate_all_processes import main

main()

Caveats​

  • Relationship generation is minimal:
    • For each bucket (who/where/when/why_roles), only the first key in the dict is used to create an object property.
    • precedes is accepted by write_process(...) but is not used when emitting relationships.
  • Identifiers:
    • The process class name is derived from label by removing spaces (cls = label.replace(" ", "")), not from filename.
  • Output overwrites:
    • Re-running will overwrite existing .ttl files with the same folder/filename.