dante-watch 0.1.2 → 0.1.3

package/src/bib/ear.mjs

@@ -1,11 +1,8 @@
-// The two ears, trained once from the sorted bib that ships in the package (plain, readable JSON):
-// bib.json — category → intent → [line, level, language]. Levels: G = first bulk run, E1 general,
-// E2 general but in the person's own way, E3 the person's line itself, washed.
-// The ear for what a human wants hears befehle + fragen + nichtdante; the ear for the
-// "HUMAN, HÖR ZU" moment hears fixgespraech. Routing is flat across intents — measured: going
-// category-first passed every category mix-up on (5 wrong instead of 2).
-// Lines were written by DeepSeek playing real users, seeded by washed real goals; the real
-// goals themselves are not in here.
+// Builds the classifiers ("ears") from the language library shipped as src/bib/bib.json:
+// category -> intent -> [line, level, language]. Levels: G bulk run, E1 general phrasing, E2 personal
+// phrasing, E3 a user line with personal values replaced by placeholders.
+// Intents are ranked across all categories at once; a category-first pass confused more lines in tests.
+// The example lines are generated; no real user input is stored here.
 import { readFileSync } from "node:fs";
 import { train, classify, tokens, sureFor, KNOWN } from "./router.mjs";
 import { wash } from "./wash.mjs";
@@ -19,18 +16,18 @@ export const rowsOf = (cats) => cats.flatMap((c) => Object.entries(load().cats[c
 export const bibDate = () => load().date;
 // Three neurons: the counting ear (bayes), the net, and the combiner (bayes^a · net^b, fitted on
 // lines neither saw). WAY decides who may act:
-//   bayes — the counting ear alone
-//   einig — act only when BOTH ears are sure on their own and say the same; else ask back
-//   kombi — the combiner's joint probability decides
-// The ranked list a human sees ("Meinst du …?") always comes from the joint probability.
+//   bayes - the counting ear alone
+//   einig - act only when both ears are sure on their own and say the same; else ask back
+//   kombi - the combiner's joint probability decides
+// The ranked list a human sees ("Meinst du ...?") always comes from the joint probability.
 // Measured per ear (25.10.): ohr  bayes 131/138 · 1 wrong · einig 130 · 0 wrong · kombi 131 · 3 wrong
 //                            janein bayes 30/33 · einig 30/33 · kombi 33/33 (it can never say yes)
 const WAYS = { ohr: "einig", janein: "kombi", system: "einig", muster: "einig", wissen: "einig" };
 export const setWay = (w) => { WAYS.ohr = WAYS.janein = WAYS.system = WAYS.muster = WAYS.wissen = w; };
 const ears = {};
 let nets = null;
-// Exact memory (like Virgil's value cache): a line that stands 1:1 in the bib — washed, so
-// "server root@1.2.3.4" meets "server {ZIEL}" — is known, no probability in between. This is what
+// Exact memory: a line that stands 1:1 in the bib - washed, so
+// "server root@1.2.3.4" meets "server {ZIEL}" - is known, no probability in between. This is what
 // makes E3 work: the trust rule keeps one line from deciding for OTHER lines, not for itself.
 const exactKey = (t) => tokens(wash(t).text).join(" ");
 // The domain ears (system, muster, wissen) sit behind the main ear's gap. Each also learns the lines of
@@ -69,15 +66,19 @@ export const hearMain = (line) => hear("ohr", ["befehle", "fragen", "smalltalk",
 export const hearConsent = (line) => hear("janein", ["fixgespraech"], [], line);
 // The third ear: only system knowledge (the sixth bib). The main ear decides it is a system job
 // (pruef/server), THEN this ear picks the move: fingerabdruck / finde_php / pruef_php2shell / beleg /
-// fix_php2shell. Kept apart from the five language bibs on purpose — no E1/E2/E3, family is a slot.
+// fix_php2shell. Kept apart from the five language bibs on purpose - no E1/E2/E3, family is a slot.
 export const hearSystem = (line) => hear("system", ["systemwissen"], SYS_PROBE, line);
-// The fourth ear: the Muster-Bib — the watchman's "go and look for what's wrong" moves. Also sits
+// The fourth ear: the Muster-Bib - the watchman's "go and look for what's wrong" moves. Also sits
 // behind the main ear's gap; conservative (einig) so plain talk is never pulled into a patrol.
 export const hearMuster = (line) => hear("muster", ["muster"], MUSTER_PROBE, line);
-// The fifth ear: the Fachwissen-Bib — when a person asks Dante to EXPLAIN (what he watches, one topic,
+// The fifth ear: the Fachwissen-Bib - when a person asks Dante to EXPLAIN (what he watches, one topic,
 // what stands out, what is normal). It learns the way of asking; the topic comes from the cards.
 export const hearWissen = (line) => hear("wissen", ["wissen"], WISSEN_PROBE, line);
 
+// Words that are about Dante's own schedule. A line with one of them is never "end the session", whatever the
+// main ear heard ("schalt dich scharf" sounds like "schalt dich aus" to it): a hard bolt, no probability.
+export const SETUP_WORDS = /\b(scharf\w*|arm yourself|armed|armar|disarm\w*|desarm\w*|patrouill\w*|patrol\w*|patrull\w*|zeitplan\w*|schedul\w*|horario|automatisch\w*|automatic\w*|automátic\w*|von selbst|on your own|por tu cuenta)\b/i;
+
 // The domain ears behind the main ear's gap, in the order they are asked. Each knows what is not its
 // own (the negative class), so the first one that speaks is the right one. null = none of them.
 export function hearDomain(line) {
@@ -94,7 +95,7 @@ const TELL = {
   en: /\b(the|is|my|what|how|why|you|are|did|do|does|this|that|it|i|me|can|someone|safe|hacked|anywhere|even|wont)\b/gi,
   de: /\b(der|die|das|ist|mein|meine|was|wie|warum|du|bin|ich|mir|kann|jemand|sicher|gehackt|nicht|und|heißt|überhaupt)\b/gi,
 };
-// No telling word at all ("fix", "demo") → the fallback, so one bare command does not flip him.
+// No telling word at all ("fix", "demo") -> the fallback, so one bare command does not flip him.
 export const langOf = (line, fallback = "de") => {
   let best = fallback, n = 0;
   for (const [k, re] of Object.entries(TELL)) { const c = (String(line).match(re) || []).length; if (c > n) { n = c; best = k; } }