docs(help): jsonwrite and jsonparse, and rewire JsonKB to the builtin

[ddehilster]

Docs and library side of VisualText/nlp-engine#712 (jsonwrite / jsonparse builtins).

⚠️Do not merge before nlp-engine#712 is released.KBFuncs.nlp here calls the jsonwrite builtin; on an older engine that name doesn't exist and JsonKB would fail.

spec/KBFuncs.nlp — JsonKB becomes a wrapper

JsonKB(L("file"),L("con")) {
jsonwrite(L("file"),L("con"));
}

Byte-identical output, ~3× faster on the pass. Serializing the KB is where a knowledge-base analyzer spends most of its time — on date-time, the thirteen-line output.nlp that calls JsonKB and SaveKB was 43.9% of total runtime.

Scope, stated plainly: analyzers carry their own copy of KBFuncs.nlp (13 in the analyzers repo, 21 bundled in the extension), and those copies already differ from this template. So this gives newly created analyzers the speedup automatically; an existing analyzer gets it when its copy is refreshed. It is not a silent global upgrade.

JsonKBRecurse, JsonAttributes and JsonStr are left in place with a note explaining why — an analyzer may call them directly (JsonStr is independently useful for escaping), and they document exactly what the builtin reproduces.

New pages

jsonwrite.md and jsonparse.md, including the mapping table and the two quirks the builtin reproduces deliberately for byte compatibility:

• indentation for level n is max(1,n) steps, not n
• a concept name with interior digits is split at the wrong place when deciding whether it's a counted sibling

Both are documented rather than silently fixed, because changing either would alter existing analyzers' output files.

jsonparse also documents why it exists beyond convenience: the json2kbb Python pass needs Python on the machine — which the npm and pypi distributions do not provide — and it won't overwrite an existing .kbb, so editing the .json appears to do nothing until you delete it.

Version notes also record the 3.8.1 python-pass fix.

🤖 Generated with Claude Code