@konneal/engine 0.1.0 → 0.1.1

package/workers/worker_public/src/research.ts

@@ -13,6 +13,8 @@ import { graphExpand, editionNote } from "./graph";
 import researchPromptText from "../prompts/research.md";
 import type { Env } from "./env";
 import type { Hit } from "../../shared/chunk.ts";
+import { P } from "./profile.ts";
+import { fill, promptVars } from "./pipeline.ts";
 
 /** Deep-research mode (G10 v1): bounded agentic loop for members —
  *  retrieve → sufficiency judge → re-retrieve targeting the gap → answer
@@ -22,7 +24,7 @@ import type { Hit } from "../../shared/chunk.ts";
  *  path when runs outgrow a single request. */
 export async function handleResearch(env: Env, ctx: ExecutionContext, req: Request, session: any): Promise<Response> {
   if (!session) {
-    return err(403, "forbidden", "Deep research is a member feature — sign in with your OIML SMART account.");
+    return err(403, "forbidden", `Deep research is a member feature — sign in with your ${P().publisher.product_name} account.`);
   }
   const body = await readJson(req);
   const q = validateQuery(body);
@@ -71,7 +73,7 @@ export async function handleResearch(env: Env, ctx: ExecutionContext, req: Reque
       try {
         const res: any = await env.AI.run(MODELS.grader, {
           messages: [
-            { role: "system", content: researchPromptText.trimEnd() },
+            { role: "system", content: fill(researchPromptText, promptVars()).trimEnd() },
             { role: "user", content: `Research question: ${q.query}\n\n${digest}Collected passages (${recent.length}):\n${recent.map((h, n) => `[${n + 1}] ${h.metadata.docidentifier ?? ""} §${h.metadata.clause_anchor ?? ""}: ${h.text.slice(0, 700)}`).join("\n")}` },
           ],
           max_tokens: 3072,

package/workers/worker_public/src/stages/conceptGraph.ts

@@ -19,6 +19,7 @@
 // family's content reaches it).
 import { THRESHOLDS } from "../config.ts";
 import type { Stage } from "./types.ts";
+import { refCodec } from "../codecs.ts";
 
 export const conceptGraph: Stage = {
   name: "concept-graph",
@@ -43,8 +44,8 @@ export const conceptGraph: Stage = {
     );
     for (const rows of termRows) {
       for (const r of rows.results ?? []) {
-        const m = String(r.doc ?? "").match(/^doc:OIML-[A-Z]-(\d+)-/);
-        if (m) numbers.add(m[1]!);
+        const mNum = refCodec().graphDocNumber(String(r.doc ?? ""));
+        if (mNum) numbers.add(mNum);
       }
     }
     if (numbers.size) {

package/workers/worker_public/src/stages/corpusScope.ts

@@ -7,8 +7,14 @@
 // the toggle model doesn't name (serving lanes, unknown values) pass
 // untouched: the toggles name DATASETS, not lanes.
 import type { Stage } from "./types.ts";
+import { P } from "../profile.ts";
 
-const DATASET_CORPORA = new Set(["oiml", "dirty", "clean", "synthetic", "smart-model", "iso-internal"]);
+/** The corpora a dataset toggle can name: the union of the datasets'
+ *  declared corpora. Corpora outside it (serving lanes, unknown values)
+ *  pass untouched — the toggles name datasets, not lanes. */
+function datasetCorpora(): Set<string> {
+  return new Set(P().datasets.flatMap((d: { corpora?: string[] }) => d.corpora ?? []));
+}
 
 export const corpusScope: Stage = {
   name: "corpus-scope",
@@ -17,7 +23,7 @@ export const corpusScope: Stage = {
     const before = c.hits.length;
     c.hits = c.hits.filter((h) => {
       const corpus = h.metadata.corpus;
-      if (!corpus || !DATASET_CORPORA.has(corpus)) return true;
+      if (!corpus || !datasetCorpora().has(corpus)) return true;
       return c.opts.datasetScope!.has(corpus);
     });
     if (c.hits.length !== before) console.log("corpus scope:", before, "→", c.hits.length, "candidates");

package/workers/worker_public/src/stages/editionCover.ts

@@ -11,10 +11,8 @@ import { toHits, type Stage } from "./types.ts";
 
 const maxDocs = 2;
 
-const familyOf = (di: string): string | null => {
-  const m = /^(?:OIML\s+)?([A-Z])\s?(\d{1,3})(?:[-–]([0-9A-Za-z]+))?/.exec(di);
-  return m ? `${m[1]}-${m[2]}` : null;
-};
+import { refCodec } from "../codecs.ts";
+const familyOf = (di: string): string | null => refCodec().familyOf(di);
 
 export const editionCover: Stage = {
   name: "edition-cover",

package/workers/worker_public/src/understand.ts

@@ -10,6 +10,7 @@ export type { QueryUnderstanding };
 
 // The prompt is data (prompts/understanding.md), bundled as text.
 import SYSTEM from "../prompts/understanding.md";
+import { fill, promptVars } from "./pipeline.ts";
 
 /** Understand the query with the cheap model. Null = use the regex fallback. */
 export async function understandQuery(
@@ -29,7 +30,7 @@ export async function understandQuery(
   const user = `${convo ? "Conversation so far:\n" + convo + "\n\n" : ""}${entityLine}Question: ${query}`;
   const body = {
     messages: [
-      { role: "system", content: SYSTEM },
+      { role: "system", content: fill(SYSTEM, promptVars()) },
       { role: "user", content: user },
     ],
     // the model always reasons; reasoning tokens share this budget — too

package/dist/chunk-LLWPT2XV.js

@@ -1,49 +0,0 @@
-import {
-  P
-} from "./chunk-MB74PTRM.js";
-
-// workers/worker_public/src/refusal.ts
-function refusalAnswer() {
-  return P().prompts.vars.refusal_sentence;
-}
-var REFUSAL_VARIANT = /^\s*I don[’']?t have information on .{1,120}? in the indexed OIML(?: \w+){0,2} (?:publications|passages|documents|corpus)\.?/i;
-var REFUSAL_DRIFT = [
-  /\b(can'?t|cannot|couldn'?t|unable)\b[^.]{0,120}?\b(indexed )?OIML(?: \w+){0,2} publications\b/i,
-  /\bno real answer to give\b[^.]{0,120}?\bOIML\b/i,
-  /\b(?:falls|well) outside\b[^.]{0,120}?\b(?:what I can answer|my scope|the scope of)\b/i,
-  /\boutside (?:of )?what (?:I|this service) can answer\b/i,
-  // "I can't answer that — weather forecasting is outside my scope":
-  // requires the refusal verb, so a scope DISCUSSION inside a real answer
-  // ("this exemption is outside the scope of R 60") never matches
-  /\bI can[’']?t answer\b[^.]{0,100}?\bscope\b/i,
-  /^\s*I don[’']?t have any indexed OIML \w+(?:s)? (?:covering|about|on)\b/im
-];
-function sentenceStart(answer, i) {
-  let s = 0;
-  for (const sep of [". ", "! ", "? ", "\n"]) {
-    const j = answer.lastIndexOf(sep, i);
-    if (j >= 0) s = Math.max(s, j + sep.length);
-  }
-  return s;
-}
-function sentenceEnd(answer, i) {
-  const m = /[.!?\n]/.exec(answer.slice(i));
-  return m ? i + m.index + 1 : answer.length;
-}
-function canonicalRefusal(answer) {
-  const CANON = refusalAnswer();
-  if (answer.includes(CANON)) return answer;
-  const variant = answer.match(REFUSAL_VARIANT);
-  if (variant) return answer.replace(variant[0], CANON);
-  for (const drift of REFUSAL_DRIFT) {
-    const m = drift.exec(answer);
-    if (!m) continue;
-    return answer.slice(0, sentenceStart(answer, m.index)) + CANON + answer.slice(sentenceEnd(answer, m.index));
-  }
-  return answer;
-}
-
-export {
-  refusalAnswer,
-  canonicalRefusal
-};

package/dist/requestScope.js

@@ -1,10 +0,0 @@
-import {
-  requestSalt,
-  resolveRequestScope
-} from "./chunk-WWNCWKKC.js";
-import "./chunk-WOGQM7DJ.js";
-import "./chunk-MB74PTRM.js";
-export {
-  requestSalt,
-  resolveRequestScope
-};