story-summary: facts migration + recall enhancements

modules/story-summary/vector/recall.js

@@ -11,7 +11,7 @@ import { getAllEventVectors, getAllChunkVectors, getChunksByFloors, getMeta } fr
 import { embed, getEngineFingerprint } from './embedder.js';
 import { xbLog } from '../../../core/debug-core.js';
 import { getContext } from '../../../../../../extensions.js';
-import { getSummaryStore } from '../data/store.js';
+import { getSummaryStore, getFacts, getNewCharacters, isRelationFact } from '../data/store.js';
 import { filterText } from './text-filter.js';
 import {
     searchStateAtoms,
@@ -258,6 +258,23 @@ function buildEntityLexicon(store, allEvents) {
     const userName = normalize(name1);
     const set = new Set();
 
+    const facts = getFacts(store);
+    for (const f of facts) {
+        if (f?.retracted) continue;
+        const s = normalize(f?.s);
+        if (s) set.add(s);
+        if (isRelationFact(f)) {
+            const o = normalize(f?.o);
+            if (o) set.add(o);
+        }
+    }
+
+    const chars = getNewCharacters(store);
+    for (const m of chars || []) {
+        const s = normalize(typeof m === 'string' ? m : m?.name);
+        if (s) set.add(s);
+    }
+
     for (const e of allEvents || []) {
         for (const p of e.participants || []) {
             const s = normalize(p);
@@ -265,30 +282,11 @@ function buildEntityLexicon(store, allEvents) {
         }
     }
 
-    const json = store?.json || {};
-
-    for (const m of json.characters?.main || []) {
-        const s = normalize(typeof m === 'string' ? m : m?.name);
-        if (s) set.add(s);
-    }
-
-    for (const a of json.arcs || []) {
+    for (const a of store?.json?.arcs || []) {
         const s = normalize(a?.name);
         if (s) set.add(s);
     }
 
-    for (const w of json.world || []) {
-        const t = normalize(w?.topic);
-        if (t && !t.includes('::')) set.add(t);
-    }
-
-    for (const r of json.characters?.relationships || []) {
-        const from = normalize(r?.from);
-        const to = normalize(r?.to);
-        if (from) set.add(from);
-        if (to) set.add(to);
-    }
-
     const stop = new Set([userName, '我', '你', '他', '她', '它', '用户', '角色', 'assistant'].map(normalize).filter(Boolean));
 
     return Array.from(set)
@@ -296,6 +294,79 @@ function buildEntityLexicon(store, allEvents) {
         .slice(0, 5000);
 }
 
+function buildFactGraph(facts) {
+    const graph = new Map();
+
+    for (const f of facts || []) {
+        if (f?.retracted) continue;
+        if (!isRelationFact(f)) continue;
+
+        const s = normalize(f?.s);
+        const o = normalize(f?.o);
+        if (!s || !o) continue;
+
+        if (!graph.has(s)) graph.set(s, new Set());
+        if (!graph.has(o)) graph.set(o, new Set());
+        graph.get(s).add(o);
+        graph.get(o).add(s);
+    }
+
+    return graph;
+}
+
+function expandByFacts(presentEntities, facts, maxDepth = 2) {
+    const graph = buildFactGraph(facts);
+    const expanded = new Map();
+    const seeds = Array.from(presentEntities || []).map(normalize).filter(Boolean);
+
+    seeds.forEach(e => expanded.set(e, 1.0));
+
+    let frontier = [...seeds];
+    for (let d = 1; d <= maxDepth && frontier.length; d++) {
+        const next = [];
+        const decay = Math.pow(0.5, d);
+
+        for (const e of frontier) {
+            const neighbors = graph.get(e);
+            if (!neighbors) continue;
+
+            for (const neighbor of neighbors) {
+                if (!expanded.has(neighbor)) {
+                    expanded.set(neighbor, decay);
+                    next.push(neighbor);
+                }
+            }
+        }
+
+        frontier = next.slice(0, 20);
+    }
+
+    return Array.from(expanded.entries())
+        .sort((a, b) => b[1] - a[1])
+        .slice(0, 15)
+        .map(([term]) => term);
+}
+
+function stripFloorTag(s) {
+    return String(s || '').replace(/\s*\(#\d+(?:-\d+)?\)\s*$/, '').trim();
+}
+
+export function buildEventEmbeddingText(ev) {
+    const parts = [];
+
+    if (ev?.title) parts.push(ev.title);
+
+    const people = (ev?.participants || []).join(' ');
+    if (people) parts.push(people);
+
+    if (ev?.type) parts.push(ev.type);
+
+    const summary = stripFloorTag(ev?.summary);
+    if (summary) parts.push(summary);
+
+    return parts.filter(Boolean).join(' ');
+}
+
 /**
  * 从分段消息中提取实体，继承消息权重
  * @param {string[]} segments
@@ -621,6 +692,7 @@ function formatRecallLog({
     chunkPreFilterStats = null,
     l0Results = [],
     textGapInfo = null,
+    expandedTerms = [],
 }) {
     const lines = [
         '\u2554' + '\u2550'.repeat(62) + '\u2557',
@@ -663,6 +735,9 @@ function formatRecallLog({
     } else {
         lines.push('  (无)');
     }
+    if (expandedTerms?.length) {
+        lines.push(`  扩散: ${expandedTerms.join('、')}`);
+    }
 
     lines.push('');
     lines.push('\u250c' + '\u2500'.repeat(61) + '\u2510');
@@ -757,12 +832,15 @@ export async function recallMemory(queryText, allEvents, vectorConfig, options =
     const lexicon = buildEntityLexicon(store, allEvents);
     const queryEntityWeights = extractEntitiesWithWeights(segments, weights, lexicon);
     const queryEntities = Array.from(queryEntityWeights.keys());
+    const facts = getFacts(store);
+    const expandedTerms = expandByFacts(queryEntities, facts, 2);
 
     // 构建文本查询串：最后一条消息 + 实体 + 关键词
     const lastSeg = segments[segments.length - 1] || '';
     const queryTextForSearch = [
         lastSeg,
         ...queryEntities,
+        ...expandedTerms,
         ...(store?.json?.keywords || []).slice(0, 5).map(k => k.text),
     ].join(' ');
 
@@ -824,6 +902,7 @@ export async function recallMemory(queryText, allEvents, vectorConfig, options =
         chunkPreFilterStats,
         l0Results,
         textGapInfo,
+        expandedTerms,
     });
 
     console.group('%c[Recall]', 'color: #7c3aed; font-weight: bold');