@konneal/engine 0.2.40 → 0.2.42

package/dist/{ask-QHODBUU2.js → ask-LFU3M2JA.js}

@@ -1,6 +1,6 @@
 import {
   handleAsk
-} from "./chunk-5SLSRFOG.js";
+} from "./chunk-ZAAKM54K.js";
 import "./chunk-N2UOENDY.js";
 import "./chunk-R2V3X6SQ.js";
 import "./chunk-A3QHHUN5.js";

package/dist/certificates.d.ts

@@ -0,0 +1,23 @@
+export interface RegisterRow {
+    num: string;
+    family: string;
+    holder: string;
+    model: string;
+    year: string;
+    status: string;
+}
+/** A question is register-shaped when it asks about certification
+ *  standing of an identifiable model or holder — the deterministic
+ *  trigger the citation-graph notes already use (query-shaped notes,
+ *  never a router). */
+export declare function isRegisterShaped(query: string): boolean;
+/** Search tokens: words that can identify a holder or a model — words of
+ *  2+ characters that are not register question words, plus bare model
+ *  numbers ("190", "HM14H1"). Upper-cased tokens match case-insensitively
+ *  in SQL LIKE. */
+export declare function registerTokens(query: string): string[];
+export declare function searchRegister(db: any, query: string): Promise<{
+    rows: RegisterRow[];
+    tokens: string[];
+} | null>;
+export declare function registerNote(rows: RegisterRow[]): string;

package/dist/{chunk-5SLSRFOG.js → chunk-ZAAKM54K.js}

@@ -1747,6 +1747,93 @@ function semanticCacheKey(indexVersion, gen, signature) {
   return `sc:${indexVersion}:g${gen}:${signature}`;
 }
 
+// workers/worker_public/src/certificates.ts
+function isRegisterShaped(query) {
+  return /\bcertif(ied|icates?|ication)s?\b/i.test(query) && /\b(is|are|still|currently|valid|status|suspended|revoked|was|were)\b/i.test(query);
+}
+function registerTokens(query) {
+  const stop = /* @__PURE__ */ new Set([
+    "the",
+    "a",
+    "an",
+    "is",
+    "are",
+    "was",
+    "were",
+    "still",
+    "currently",
+    "in",
+    "on",
+    "for",
+    "of",
+    "and",
+    "or",
+    "certificate",
+    "certificates",
+    "certified",
+    "certification",
+    "register",
+    "status",
+    "valid",
+    "suspended",
+    "revoked",
+    "recommendation",
+    "per",
+    "under",
+    "by",
+    "with",
+    "what",
+    "which",
+    "who",
+    "does",
+    "do",
+    "load",
+    "cell",
+    "model",
+    "manufacturer",
+    "holder",
+    "company",
+    "r",
+    "how",
+    "to",
+    "list",
+    "show"
+  ]);
+  const tokens2 = [];
+  const publisherWord = P().publisher.name.toLowerCase();
+  for (const w of query.split(/[^A-Za-z0-9&-]+/)) {
+    if (w.length < 2) continue;
+    if (stop.has(w.toLowerCase())) continue;
+    if (w.toLowerCase() === publisherWord) continue;
+    if (/^\d+$/.test(w) && w.length < 2) continue;
+    if (!tokens2.includes(w)) tokens2.push(w);
+  }
+  return tokens2.slice(0, 6);
+}
+async function searchRegister(db, query) {
+  if (!isRegisterShaped(query)) return null;
+  const tokens2 = registerTokens(query);
+  if (!tokens2.length) return null;
+  const clauses = tokens2.map(() => "(holder LIKE ?1 OR model LIKE ?1 OR num LIKE ?1)").join(" OR ");
+  const params = tokens2.map((t) => `%${t}%`);
+  try {
+    const res = await db.prepare(`SELECT num, family, holder, model, year, status FROM certificates WHERE ${clauses} LIMIT 6`).bind(...params).all();
+    return { rows: res.results ?? [], tokens: tokens2 };
+  } catch {
+    return { rows: [], tokens: tokens2 };
+  }
+}
+function registerNote(rows) {
+  if (!rows.length) {
+    return `Certificate register: NO certificate matching the asked holder or model appears in the ${P().publisher.name}-CS register snapshot. State plainly that no such certificate is in this register, and that the register is a snapshot rather than the live certification status.`;
+  }
+  const lines = rows.map((r) => `- ${r.num}: holder ${r.holder}, model "${r.model}"${r.year ? `, issued ${r.year}` : ""} \u2014 status ${r.status}`);
+  return [
+    "Certificate register \u2014 the following certificates matched the asked holder or model. Quote the certificate number, the holder and the status verbatim; the register is a snapshot, so qualify any statement about current certification accordingly.",
+    ...lines
+  ].join("\n");
+}
+
 // workers/worker_public/src/ask.ts
 function userImageDataUrl(body) {
   const img = body?.image;
@@ -2077,6 +2164,8 @@ async function handleAsk(env, ctx, req, tier, key) {
   console.log("understand:", understanding?.intent ?? "null", understanding?.doc_number ? `doc#${understanding.doc_number}${understanding.edition ? "@" + understanding.edition : ""}` : "nodoc", "|", q.query.slice(0, 50));
   const graphDocNumbers = await graphExpand(env, understanding);
   const eNote = await editionNote(env, understanding);
+  const register = await searchRegister(env.DB, q.query);
+  const regNote = register ? registerNote(register.rows) : void 0;
   if (understanding?.intent !== "conversational" && !nodeScoped && !contextual && !declaredCtx && !draftAct && !apiCallIntent && !userImage && !fresh) {
     const warmVec = await warmEmbed ?? null;
     if (warmVec) {
@@ -2443,14 +2532,15 @@ Answer account questions from these records ONLY: name the record when you use i
     q.lang,
     keptHistory,
     // stage-extracted graph facts (GraphRAG) ride the same note channel
-    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...retrieved.notes ?? []].filter(Boolean).join("\n") || void 0,
+    [processNote, eNote, regNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...retrieved.notes ?? []].filter(Boolean).join("\n") || void 0,
     summary,
     budget
   );
   await attachFigureImages(env, messages, usedHits, q.query);
   if (userImage) {
+    const wantsMarkings = /\b(mark(ing|ings)?|nameplate|label|inscription|engrav|sticker|plate)\b|certificat/i.test(q.query);
     const last = messages[messages.length - 1];
-    const note = "\n\n(The user attached an image with this question; interpret it directly when answering.)";
+    const note = wantsMarkings ? "\n\n(The user attached a photo with this question. First transcribe every inscription you can actually read in the image \u2014 names, model references, accuracy classes, numeric values with their units, certificate or approval numbers \u2014 quoting them verbatim. Then answer the question from the numbered passages, citing the requirement each interpretation rests on. If an inscription is unreadable, say so; never invent a marking.)" : "\n\n(The user attached an image with this question; interpret it directly when answering.)";
     if (Array.isArray(last.content)) {
       const textPart = last.content.find((p) => p.type === "text");
       if (textPart) textPart.text += note;
@@ -2476,7 +2566,7 @@ Answer account questions from these records ONLY: name the record when you use i
 `));
           send({ type: "read", read: readAs() });
           const sourceQuality = answerQuality(cites.map((c) => c.quality));
-          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality), ...sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {} } : {}, ...liveRecords ? { records: liveRecords } : {}, quota });
+          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...sourceQuality ? { source_quality: sourceQuality, confidence_note: qualityNote(sourceQuality), ...sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {} } : {}, ...liveRecords ? { records: liveRecords } : {}, quota });
           let full = "";
           try {
             for await (const tok of sseTokens(stream)) {
@@ -2610,7 +2700,7 @@ Answer account questions from these records ONLY: name the record when you use i
   }
   completionBlocks.push(...await completeFigures(env.DB, answer, [...c2ns.blocks, ...completionBlocks], used));
   const jsonQuality = answerQuality(finalCites.map((c) => c.quality));
-  const out = { answer, citations: finalCites, ...jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality), ...jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {} } : {}, model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...verdictBlock ? [verdictBlock] : [], ...conditionBlock ? [conditionBlock] : [], ...aggregationBlock ? [aggregationBlock] : [], ...completionBlocks], context_applied: ctxApplied, ...liveRecords ? { records: liveRecords } : {} };
+  const out = { answer, citations: finalCites, ...jsonQuality ? { source_quality: jsonQuality, confidence_note: qualityNote(jsonQuality), ...jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {} } : {}, model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...verdictBlock ? [verdictBlock] : [], ...conditionBlock ? [conditionBlock] : [], ...aggregationBlock ? [aggregationBlock] : [], ...completionBlocks], context_applied: ctxApplied, ...liveRecords ? { records: liveRecords } : {} };
   const cacheable = !contextual && !declaredCtx && !answer.includes(refusalAnswer()) && finalAnchors.violations.length === 0;
   if (cacheable) {
     const warmVec = await warmEmbed ?? null;

package/dist/worker_public/src/index.js

@@ -9,7 +9,7 @@ import {
   modelGroundingBlock,
   scoreJudge,
   standardForDocNumber
-} from "../../chunk-5SLSRFOG.js";
+} from "../../chunk-ZAAKM54K.js";
 import {
   buildMessages,
   citations,
@@ -999,7 +999,7 @@ async function handleMcp(env, ctx, req, tier, key) {
       // stream:false forces the JSON lane (anon defaults to SSE)
       body: JSON.stringify({ ...args, stream: false })
     });
-    const res = name === "ask" ? await (await import("../../ask-QHODBUU2.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-QZDQVB57.js")).handleSearch(env, ctx, inner, tier, key);
+    const res = name === "ask" ? await (await import("../../ask-LFU3M2JA.js")).handleAsk(env, ctx, inner, tier, key) : await (await import("../../search-QZDQVB57.js")).handleSearch(env, ctx, inner, tier, key);
     return res.json().catch(() => ({ error: { message: "tool transport failed", status: res.status } }));
   });
   if (out.ok && "accepted" in out) return new Response(null, { status: 202 });

package/package.json

@@ -27,7 +27,7 @@
     "playwright": "^1.62.1",
     "yaml": "^2.9.1"
   },
-  "version": "0.2.40",
+  "version": "0.2.42",
   "description": "The Konneal engine: the publisher-agnostic build pipeline and API plane for standards intelligence (retrieval, answer contract, verdicts, evaluation).",
   "license": "BSD-3-Clause",
   "type": "module",

package/workers/worker_public/src/ask.ts

@@ -37,6 +37,7 @@ export type { Env };
 import { json, err, corsHeaders, readJson, validateQuery, type ApiKey } from "./lib/http";
 import { clientIp, checkQuota, telemetry } from "./quota";
 import { graphExpand, editionNote } from "./graph";
+import { registerNote, searchRegister } from "./certificates";
 import { P } from "./profile.ts";
 
 /** User-uploaded image for multimodal questions: a data URL
@@ -577,6 +578,12 @@ async function handleAsk(
   console.log("understand:", understanding?.intent ?? "null", understanding?.doc_number ? `doc#${understanding.doc_number}${understanding.edition ? "@" + understanding.edition : ""}` : "nodoc", "|", q.query.slice(0, 50));
   const graphDocNumbers = await graphExpand(env, understanding);
   const eNote = await editionNote(env, understanding);
+  // the certificate register as a database (TODO.new-era/6): a
+  // register-shaped question queries the D1 table directly and the rows
+  // join the prompt as an authoritative note — exact matches, never a
+  // similarity guess about certification standing
+  const register = await searchRegister(env.DB, q.query);
+  const regNote = register ? registerNote(register.rows) : undefined;
 
   // semantic cache: near-duplicate of a recently answered question —
   // serves the stored answer with a `similar: true` marker (checked only
@@ -1108,16 +1115,23 @@ async function handleAsk(
     q.lang,
     keptHistory,
     // stage-extracted graph facts (GraphRAG) ride the same note channel
-    [processNote, eNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...(retrieved.notes ?? [])].filter(Boolean).join("\n") || undefined,
+    [processNote, eNote, regNote, contextNote(declaredCtx, docScope), accountNote, modelNote, vocabNote, memNote, machineNote, conditionNote, aggregationNote, boundaryNote, licenseNote, ...(retrieved.notes ?? [])].filter(Boolean).join("\n") || undefined,
     summary,
     budget,
   );
   await attachFigureImages(env, messages, usedHits, q.query);
   if (userImage) {
     // the user's own image rides on the question message — retrieval stays
-    // text-driven; the answer model reads the image as question context
+    // text-driven; the answer model reads the image as question context.
+    // A markings/nameplate/certificate photo gets a TRANSCRIPTION note:
+    // read the inscriptions verbatim first, then interpret from the
+    // passages (the marking requirements are corpus content) — never
+    // invent a marking that cannot be read.
+    const wantsMarkings = /\b(mark(ing|ings)?|nameplate|label|inscription|engrav|sticker|plate)\b|certificat/i.test(q.query);
     const last = messages[messages.length - 1];
-    const note = "\n\n(The user attached an image with this question; interpret it directly when answering.)";
+    const note = wantsMarkings
+      ? "\n\n(The user attached a photo with this question. First transcribe every inscription you can actually read in the image — names, model references, accuracy classes, numeric values with their units, certificate or approval numbers — quoting them verbatim. Then answer the question from the numbered passages, citing the requirement each interpretation rests on. If an inscription is unreadable, say so; never invent a marking.)"
+      : "\n\n(The user attached an image with this question; interpret it directly when answering.)";
     if (Array.isArray(last.content)) {
       const textPart = last.content.find((p: any) => p.type === "text");
       if (textPart) textPart.text += note;
@@ -1147,7 +1161,7 @@ async function handleAsk(
           // retrieval, before a single token of the answer
           send({ type: "read", read: readAs() });
           const sourceQuality = answerQuality(cites.map((c: any) => c.quality));
-          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...(sourceQuality ? { source_quality: sourceQuality, quality_note: qualityNote(sourceQuality), ...(sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {}) } : {}), ...(liveRecords ? { records: liveRecords } : {}), quota, });
+          send({ type: "citations", citations: cites, context_applied: ctxApplied, ...(sourceQuality ? { source_quality: sourceQuality, confidence_note: qualityNote(sourceQuality), ...(sourceQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(cites) } : {}) } : {}), ...(liveRecords ? { records: liveRecords } : {}), quota, });
           let full = "";
           try {
             for await (const tok of sseTokens(stream)) {
@@ -1342,7 +1356,7 @@ async function handleAsk(
   completionBlocks.push(...(await completeFigures(env.DB, answer, [...c2ns.blocks, ...completionBlocks], used)));
 
   const jsonQuality = answerQuality(finalCites.map((c: any) => c.quality));
-  const out = { answer, citations: finalCites, ...(jsonQuality ? { source_quality: jsonQuality, quality_note: qualityNote(jsonQuality), ...(jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {}) } : {}), model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...(verdictBlock ? [verdictBlock] : []), ...(conditionBlock ? [conditionBlock] : []), ...(aggregationBlock ? [aggregationBlock] : []), ...completionBlocks], context_applied: ctxApplied, ...(liveRecords ? { records: liveRecords } : {}) };
+  const out = { answer, citations: finalCites, ...(jsonQuality ? { source_quality: jsonQuality, confidence_note: qualityNote(jsonQuality), ...(jsonQuality === "ocr" ? { experimental_sources: experimentalSourceLabels(finalCites) } : {}) } : {}), model: MODELS.member, query_hash: queryHash, follow_ups: understanding?.follow_ups ?? [], blocks: [...c2ns.blocks, ...(verdictBlock ? [verdictBlock] : []), ...(conditionBlock ? [conditionBlock] : []), ...(aggregationBlock ? [aggregationBlock] : []), ...completionBlocks], context_applied: ctxApplied, ...(liveRecords ? { records: liveRecords } : {}) };
   const cacheable = !contextual && !declaredCtx && !answer.includes(refusalAnswer()) && finalAnchors.violations.length === 0;
   if (cacheable) {
     const warmVec = (await warmEmbed) ?? null;

package/workers/worker_public/src/certificates.ts

@@ -0,0 +1,78 @@
+// The certificate register as a STRUCTURED search surface (TODO.new-era/6):
+// "is model X from holder Y still certified?" is an exact-match question,
+// not a similarity question. The register rows live in D1 (the
+// certificates plane's `certificates` table); this stage decides when a
+// question is register-shaped, queries the table, and returns the rows as
+// an authoritative note for the prompt. A miss is a MISS: "not in this
+// register snapshot" is stated verbatim, never softened into a guess.
+
+import { P } from './profile.ts';
+
+export interface RegisterRow {
+  num: string;
+  family: string;
+  holder: string;
+  model: string;
+  year: string;
+  status: string;
+}
+
+/** A question is register-shaped when it asks about certification
+ *  standing of an identifiable model or holder — the deterministic
+ *  trigger the citation-graph notes already use (query-shaped notes,
+ *  never a router). */
+export function isRegisterShaped(query: string): boolean {
+  return /\bcertif(ied|icates?|ication)s?\b/i.test(query) && /\b(is|are|still|currently|valid|status|suspended|revoked|was|were)\b/i.test(query);
+}
+
+/** Search tokens: words that can identify a holder or a model — words of
+ *  2+ characters that are not register question words, plus bare model
+ *  numbers ("190", "HM14H1"). Upper-cased tokens match case-insensitively
+ *  in SQL LIKE. */
+export function registerTokens(query: string): string[] {
+  const stop = new Set([
+    "the", "a", "an", "is", "are", "was", "were", "still", "currently", "in", "on", "for", "of", "and", "or",
+    "certificate", "certificates", "certified", "certification", "register", "status", "valid", "suspended",
+    "revoked", "recommendation", "per", "under", "by", "with", "what", "which", "who", "does", "do",
+    "load", "cell", "model", "manufacturer", "holder", "company", "r", "how", "to", "list", "show",
+  ]);
+  const tokens: string[] = [];
+  const publisherWord = P().publisher.name.toLowerCase();
+  for (const w of query.split(/[^A-Za-z0-9&-]+/)) {
+    if (w.length < 2) continue;
+    if (stop.has(w.toLowerCase())) continue;
+    if (w.toLowerCase() === publisherWord) continue;
+    if (/^\d+$/.test(w) && w.length < 2) continue;
+    if (!tokens.includes(w)) tokens.push(w);
+  }
+  return tokens.slice(0, 6);
+}
+
+export async function searchRegister(db: any, query: string): Promise<{ rows: RegisterRow[]; tokens: string[] } | null> {
+  if (!isRegisterShaped(query)) return null;
+  const tokens = registerTokens(query);
+  if (!tokens.length) return null;
+  const clauses = tokens.map(() => "(holder LIKE ?1 OR model LIKE ?1 OR num LIKE ?1)").join(" OR ");
+  const params = tokens.map((t) => `%${t}%`);
+  try {
+    const res = await db
+      .prepare(`SELECT num, family, holder, model, year, status FROM certificates WHERE ${clauses} LIMIT 6`)
+      .bind(...params)
+      .all();
+    return { rows: (res.results ?? []) as RegisterRow[], tokens };
+  } catch {
+    // a missing table or a D1 hiccup degrades to no-note honestly
+    return { rows: [], tokens };
+  }
+}
+
+export function registerNote(rows: RegisterRow[]): string {
+  if (!rows.length) {
+    return `Certificate register: NO certificate matching the asked holder or model appears in the ${P().publisher.name}-CS register snapshot. State plainly that no such certificate is in this register, and that the register is a snapshot rather than the live certification status.`;
+  }
+  const lines = rows.map((r) => `- ${r.num}: holder ${r.holder}, model "${r.model}"${r.year ? `, issued ${r.year}` : ""} — status ${r.status}`);
+  return [
+    "Certificate register — the following certificates matched the asked holder or model. Quote the certificate number, the holder and the status verbatim; the register is a snapshot, so qualify any statement about current certification accordingly.",
+    ...lines,
+  ].join("\n");
+}