@konneal/engine 0.2.103 → 0.2.109

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

@@ -17,6 +17,7 @@ import { hyde } from "./hyde.ts";
 import { glossary } from "./glossary.ts";
 import { conceptGraph } from "./conceptGraph.ts";
 import { graphLane } from "./graphLane.ts";
+import { graphPpr } from "./graphPpr.ts";
 import { licensedLane } from "./licensedLane.ts";
 import { licensedCover } from "./licensedCover.ts";
 import { multiQuery } from "./multiQuery.ts";
@@ -48,6 +49,7 @@ export const STAGES: Stage[] = [
   glossary,
   conceptGraph,
   graphLane,
+  graphPpr,
   licensedLane,
   multiQuery,
   subQuery,
@@ -75,3 +77,20 @@ export const STAGES: Stage[] = [
   dedup,
   windowFloor,
 ];
+
+export const STAGE_NAMES: string[] = STAGES.map((s) => s.name);
+
+/** The ablation switch (TODO.sota/09): a configuration is a stage list.
+ *  The projection keeps REGISTRY order — a subset can narrow the
+ *  pipeline, never reorder it (the ordering invariants are
+ *  load-bearing). Unknown names throw so a typo'd configuration can
+ *  never measure the wrong pipeline silently. */
+export function projectStages(names: readonly string[]): Stage[] {
+  const wanted = new Set(names);
+  const projected = STAGES.filter((s) => wanted.has(s.name));
+  if (projected.length !== wanted.size) {
+    const known = new Set(STAGE_NAMES);
+    for (const n of wanted) if (!known.has(n)) throw new Error(`unknown stage: ${n}`);
+  }
+  return projected;
+}

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

@@ -1,15 +1,60 @@
-// Section-summary units → leaf evidence (FABLE multi-granularity): a
-// depth-1 summary vector that ranked is a navigation node, not quotable
-// evidence: fetch its top child clauses (metadata-filtered, same query
-// vector) so the model gets source text to cite, then retire the
-// synthetic summary — the answer contract grounds claims in source
-// clauses, never in our own summaries. The summary stays only when no
-// child answered (it is then the doc's sole representative).
+// Section-summary units → leaf evidence (FABLE multi-granularity, the
+// RAPTOR extension of TODO.sota/06's catalog row 3): a ranked summary
+// is a navigation node, not quotable evidence — the stage descends it
+// to source clauses and retires the synthetic node. Two levels:
+//   level 1 (section summary): child_anchors are CLAUSE anchors — the
+//     same-query vector search finds the clauses within the document.
+//   level 2 (document summary): child_anchors are LEVEL-1 UNIT ids —
+//     they come back by id (the ≤20 getByIds law), and each consumed
+//     section's clauses come from the D1 corpus DIRECTLY (the l6a law:
+//     the corpus is the source of truth; the index metadata filters
+//     are dead on this index).
+// The summary stays only when no child answered (it is then the doc's
+// sole representative).
 import { THRESHOLDS } from "../config.ts";
 import type { ChunkMeta } from "../../../shared/chunk";
 import type { Hit } from "../../../shared/chunk";
 import type { Stage } from "./types.ts";
 
+const CHUNK_COLS =
+  "id, doc_id, docidentifier, doctype, doc_number, edition, language, clause_anchor, clause_title, status, superseded_by, corpus, tier, text, unit_id, block";
+
+/** A level-2 document summary's section children, by id (the ≤20
+ *  getByIds law caps each read). */
+async function sectionChildren(env: any, unitIds: string[]): Promise<Hit[]> {
+  const kids: Hit[] = [];
+  for (let i = 0; i < unitIds.length; i += 20) {
+    const got = (await env.VECTORIZE.getByIds(unitIds.slice(i, i + 20))) ?? [];
+    for (const v of got) {
+      if (!v?.metadata?.child_anchors) continue;
+      kids.push({
+        id: v.id,
+        score: 0,
+        metadata: v.metadata as ChunkMeta,
+        text: String((v.metadata as any).chunk_text ?? ""),
+      });
+    }
+  }
+  return kids;
+}
+
+/** A section's child clauses, from the D1 corpus directly. */
+async function clausesOf(env: any, docId: string, anchors: string[], limit: number): Promise<Hit[]> {
+  const marks = anchors.map(() => "?").join(",");
+  const res = await env.DB.prepare(
+    `SELECT ${CHUNK_COLS} FROM chunks WHERE doc_id = ? AND clause_anchor IN (${marks}) LIMIT ${limit}`,
+  )
+    .bind(docId, ...anchors)
+    .all()
+    .catch(() => ({ results: [] }));
+  return (res.results ?? []).map((r: any) => ({
+    id: String(r.id),
+    score: 0,
+    metadata: r as unknown as ChunkMeta,
+    text: String(r.text ?? ""),
+  }));
+}
+
 export const sectionDescent: Stage = {
   name: "section-descent",
   failure: "additive",
@@ -20,6 +65,45 @@ export const sectionDescent: Stage = {
     const sectionHit = c.finalHits.find(
       (h) => h.metadata.section_summary === "1" && h.metadata.child_anchors && h.score > 0,
     )!;
+
+    if ((sectionHit.metadata as any).summary_level === "2") {
+      // document summary → its leading section nodes → their clauses
+      const unitIds = sectionHit
+        .metadata.child_anchors!.split(",")
+        .map((s) => s.trim())
+        .filter(Boolean)
+        .slice(0, 12);
+      const kids = await sectionChildren(c.env, unitIds);
+      // document reading order (child_anchors order) is the doc-scope
+      // prior; the summaries were written in it
+      const order = new Map(unitIds.map((id, i) => [id, i]));
+      kids.sort((a, b) => (order.get(a.id) ?? 99) - (order.get(b.id) ?? 99));
+      const collected: Hit[] = [];
+      const consumed: string[] = [];
+      for (const kid of kids.slice(0, 2)) {
+        const anchors = kid.metadata.child_anchors!.split(",").map((s) => s.trim()).filter(Boolean).slice(0, 25);
+        if (!anchors.length) continue;
+        const clauses = (await clausesOf(c.env, String(kid.metadata.doc_id), anchors, 6))
+          .filter((x) => !c.finalHits.some((h) => h.id === x.id) && !collected.some((h) => h.id === x.id))
+          .slice(0, 2)
+          .map((x) => ({ ...x, score: sectionHit.score * THRESHOLDS.sectionDescentDiscount }));
+        if (clauses.length) {
+          collected.push(...clauses);
+          consumed.push(kid.id);
+        }
+      }
+      if (collected.length) {
+        c.finalHits = [...c.finalHits.filter((h) => h !== sectionHit), ...collected];
+        console.log(
+          "section descent (doc):",
+          sectionHit.metadata.docidentifier,
+          "→",
+          collected.map((x) => "§" + x.metadata.clause_anchor).join(", "),
+        );
+      }
+      return;
+    }
+
     const kids = sectionHit
       .metadata.child_anchors!.split(",")
       .map((s) => s.trim())

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

@@ -67,6 +67,12 @@ export interface RetrieveOptions {
    *  metadata stays). Null = the deployment declares no licensed
    *  content, the scope is inert. */
   standardKeys?: Set<string> | null;
+  /** The ablation switch (TODO.sota/09): run the registry PROJECTED to
+   *  these stage names (registry order preserved, unknown names throw —
+   *  projectStages owns the semantics). Absent = the full registry.
+   *  Admin-gated at the ask boundary (src/ablate.ts); a stage-subset
+   *  ask never reads or writes the answer caches. */
+  stageSubset?: string[];
 }
 
 export interface GlossaryEntry {

package/dist/ask-M3VRAIJB.js

@@ -1,15 +0,0 @@
-import {
-  handleAsk
-} from "./chunk-NCZHN6QW.js";
-import "./chunk-ANUHRJ4Z.js";
-import "./chunk-25ECO5M7.js";
-import "./chunk-KNCRYAX3.js";
-import "./chunk-DTW4UW2E.js";
-import "./chunk-A3QHHUN5.js";
-import "./chunk-U4ERFUOK.js";
-import "./chunk-BOAITSVV.js";
-import "./chunk-JZFXSTYX.js";
-import "./chunk-3FYJM7LH.js";
-export {
-  handleAsk
-};

package/dist/search-5SNP5CHL.js

@@ -1,13 +0,0 @@
-import {
-  handleSearch
-} from "./chunk-TFGJ6NYS.js";
-import "./chunk-25ECO5M7.js";
-import "./chunk-KNCRYAX3.js";
-import "./chunk-DTW4UW2E.js";
-import "./chunk-A3QHHUN5.js";
-import "./chunk-U4ERFUOK.js";
-import "./chunk-BOAITSVV.js";
-import "./chunk-3FYJM7LH.js";
-export {
-  handleSearch
-};