stratagate-dsh 0.2.26 → 0.2.29

package/dist/index.js

@@ -364,6 +364,7 @@ function applyGraphProjection(options) {
     const sources2 = validSources(proposal.sourceEventIds);
     if (!ref || !name2 || !NODE_TYPES.has(proposal.type) || sources2.length === 0) continue;
     const aliases = strings(proposal.aliases, 20).filter((alias) => normalizeSearchText(alias) !== normalizeSearchText(name2));
+    const tags = strings(proposal.tags, 12);
     let node = options.nodes.find((candidate) => sameEntity(candidate, { ...proposal, name: name2, aliases }));
     if (!node) {
       node = {
@@ -382,6 +383,7 @@ function applyGraphProjection(options) {
       options.nodes.push(node);
     }
     node.aliases = [.../* @__PURE__ */ new Set([...node.aliases, ...aliases])];
+    if (tags.length > 0) node.tags = [.../* @__PURE__ */ new Set([...node.tags ?? [], ...tags])].slice(0, 12);
     node.status = STATUSES.has(proposal.status ?? "active") ? proposal.status ?? "active" : "active";
     node.confidence = confidence(proposal.confidence);
     node.sourceEventIds = [.../* @__PURE__ */ new Set([...node.sourceEventIds, ...sources2])];
@@ -2626,6 +2628,7 @@ var StrataGate = class _StrataGate {
     const ranked = bm25Rank(candidates, query, (node) => weightedSearchTokens([
       [node.name, 6],
       [node.aliases.join(" "), 5],
+      [(node.tags ?? []).join(" "), 5],
       [node.type, 2],
       [node.currentState, 4],
       [node.facts.map((fact) => `${fact.key} ${Array.isArray(fact.value) ? fact.value.join(" ") : fact.value}`).join(" "), 4],
@@ -3281,6 +3284,7 @@ var GRAPH_NODE = {
     name: { type: "string", required: true },
     type: { type: "string", enum: ["person", "project", "organization", "tool", "place"], required: true },
     aliases: STRING_ARRAY,
+    tags: { ...STRING_ARRAY, required: true },
     state: { type: "string" },
     facts: { type: "array", items: GRAPH_FACT },
     status: { type: "string", enum: ["active", "superseded", "disputed", "archived"] },
@@ -3443,7 +3447,7 @@ var DshModelBridge = class {
     const eventIds = new Set(context.events.map((event) => event.id));
     const raw = object(await this.callStructured(
       "graphProjector",
-      `Project the supplied Events into the current Knowledge Graph, then call ${STRUCTURED_TOOLS.graphProjector.name} exactly once. Events are the sole source of truth; never use legacy Element data. Return stable entity nodes for people, projects, organizations, tools, and places. Use aliases to merge spelling/case/separator variants. Put attributes in node facts and every relationship in a directed edge using fromRef/toRef\u2014never encode a relationship as a fact string. Prefer concise canonical Chinese relation labels such as \u4F7F\u7528\u3001\u5C5E\u4E8E\u3001\u521B\u5EFA\u3001\u53C2\u4E0E\u3001\u8D21\u732E\u3001\u4F9D\u8D56\u3001\u4F4D\u4E8E\u3001\u76F8\u5173. Every node, fact, and edge must cite only supplied Event ids. Do not return text.`,
+      `Project the supplied Events into the current Knowledge Graph, then call ${STRUCTURED_TOOLS.graphProjector.name} exactly once. Events are the sole source of truth; never use legacy Element data. Return stable entity nodes for people, projects, organizations, tools, and places. Use aliases to merge spelling/case/separator variants. Give every returned node 1-6 concise semantic role tags such as benchmark, evaluation, memory-plugin, parser, or development-tool; tags describe the node's specific role and never replace its person/project/organization/tool/place type. Reuse stable tag wording when possible. Put attributes in node facts and every relationship in a directed edge using fromRef/toRef\u2014never encode a relationship as a fact string. Prefer concise canonical Chinese relation labels such as \u4F7F\u7528\u3001\u5C5E\u4E8E\u3001\u521B\u5EFA\u3001\u53C2\u4E0E\u3001\u8D21\u732E\u3001\u4F9D\u8D56\u3001\u4F4D\u4E8E\u3001\u76F8\u5173. Every node, fact, and edge must cite only supplied Event ids. Do not return text.`,
       context
     ));
     const nodes = (Array.isArray(raw.nodes) ? raw.nodes : []).flatMap((candidate) => {
@@ -3464,6 +3468,7 @@ var DshModelBridge = class {
         name: text2(item.name),
         type,
         aliases: strings2(item.aliases),
+        tags: strings2(item.tags).slice(0, 12),
         ...text2(item.state) ? { state: text2(item.state) } : {},
         facts,
         ...typeof item.status === "string" ? { status: item.status } : {},
@@ -4720,7 +4725,119 @@ function registerMemoryTools(ctx, runtime) {
 
 // src/web.ts
 import { createRequire } from "node:module";
-var STRATAGATE_DSH_VERSION = "0.2.22";
+
+// src/graph-clustering.ts
+import createGraph from "ngraph.graph";
+import { detectClusters } from "ngraph.leiden";
+var MAX_CLUSTER_SIZE = 12;
+var LEIDEN_RESOLUTION = 0.55;
+function pairKey(left, right) {
+  return left < right ? `${left}\0${right}` : `${right}\0${left}`;
+}
+function pairIds(key) {
+  const separator = key.indexOf("\0");
+  return [key.slice(0, separator), key.slice(separator + 1)];
+}
+function semanticTags(node) {
+  return Array.isArray(node.tags) ? [...new Set(node.tags.map((tag) => tag.trim()).filter(Boolean))] : [];
+}
+function stableClusterId(nodeIds) {
+  let hash = 2166136261;
+  for (const character of [...nodeIds].sort().join("\0")) {
+    hash ^= character.charCodeAt(0);
+    hash = Math.imul(hash, 16777619);
+  }
+  return `cluster_${(hash >>> 0).toString(36)}`;
+}
+function clusterLabel(members, weightedDegree) {
+  const counts = /* @__PURE__ */ new Map();
+  for (const node of members) for (const tag of semanticTags(node)) {
+    counts.set(tag, (counts.get(tag) ?? 0) + 1);
+  }
+  const orderedTags = [...counts].sort((left, right) => right[1] - left[1] || left[0].localeCompare(right[0])).map(([tag]) => tag);
+  const sharedTags = [...counts].filter(([, count]) => members.length === 1 || count > 1).sort((left, right) => right[1] - left[1] || left[0].localeCompare(right[0])).slice(0, 2).map(([tag]) => tag);
+  if (sharedTags.length > 0) return { label: sharedTags.join(" / "), tags: orderedTags.slice(0, 8) };
+  const names = [...members].sort((left, right) => (weightedDegree.get(right.id) ?? 0) - (weightedDegree.get(left.id) ?? 0) || right.sourceEventIds.length - left.sourceEventIds.length || left.name.localeCompare(right.name)).slice(0, 2).map(({ name: name2 }) => name2);
+  return { label: names.join(" / ") || "\u672A\u547D\u540D\u4E3B\u9898", tags: orderedTags.slice(0, 8) };
+}
+function clusterKnowledgeGraph(rawNodes, rawEdges) {
+  const nodes = rawNodes.filter((node) => node.status !== "archived");
+  if (nodes.length === 0) return [];
+  const nodeMap = new Map(nodes.map((node) => [node.id, node]));
+  const structuralWeights = /* @__PURE__ */ new Map();
+  const addStructuralWeight = (left, right, weight) => {
+    if (left === right || !nodeMap.has(left) || !nodeMap.has(right) || weight <= 0) return;
+    const key = pairKey(left, right);
+    structuralWeights.set(key, (structuralWeights.get(key) ?? 0) + weight);
+  };
+  for (const edge of rawEdges.filter((edge2) => edge2.status === "active")) {
+    addStructuralWeight(edge.fromNodeId, edge.toNodeId, 3 * Math.max(0.25, Number(edge.confidence) || 0.8));
+  }
+  const eventOwners = /* @__PURE__ */ new Map();
+  for (const node of nodes) for (const eventId of new Set(node.sourceEventIds)) {
+    const owners = eventOwners.get(eventId) ?? [];
+    owners.push(node.id);
+    eventOwners.set(eventId, owners);
+  }
+  for (const owners of eventOwners.values()) {
+    const eventWeight = 1.4 / Math.sqrt(Math.max(1, owners.length - 1));
+    for (let left = 0; left < owners.length; left += 1) for (let right = left + 1; right < owners.length; right += 1) {
+      addStructuralWeight(owners[left], owners[right], eventWeight);
+    }
+  }
+  const graph = createGraph({ multigraph: false });
+  for (const node of nodes) graph.addNode(node.id);
+  const weightedDegree = new Map(nodes.map((node) => [node.id, 0]));
+  for (const [key, structuralWeight] of structuralWeights) {
+    const [leftId, rightId] = pairIds(key);
+    const left = nodeMap.get(leftId);
+    const right = nodeMap.get(rightId);
+    const leftTags = new Set(semanticTags(left));
+    const rightTags = new Set(semanticTags(right));
+    const overlap = [...leftTags].filter((tag) => rightTags.has(tag)).length;
+    const union = (/* @__PURE__ */ new Set([...leftTags, ...rightTags])).size;
+    const tagBoost = overlap > 0 && union > 0 ? 0.85 * overlap / union : 0;
+    const typeBoost = left.type === right.type ? 0.15 : 0;
+    const weight = structuralWeight + tagBoost + typeBoost;
+    graph.addLink(leftId, rightId, { weight });
+    weightedDegree.set(leftId, (weightedDegree.get(leftId) ?? 0) + weight);
+    weightedDegree.set(rightId, (weightedDegree.get(rightId) ?? 0) + weight);
+  }
+  const partition = detectClusters(graph, {
+    quality: "cpm",
+    resolution: LEIDEN_RESOLUTION,
+    randomSeed: 42,
+    refine: true,
+    maxCommunitySize: MAX_CLUSTER_SIZE,
+    linkWeight: (link) => link.data.weight
+  });
+  const connectedGroups = [];
+  const isolatedIds = [];
+  for (const community of partition.getCommunities().values()) {
+    const ids = community.map(String).sort();
+    const connected = ids.filter((id) => (weightedDegree.get(id) ?? 0) > 0);
+    const isolated = ids.filter((id) => (weightedDegree.get(id) ?? 0) === 0);
+    if (connected.length > 0) connectedGroups.push(connected);
+    isolatedIds.push(...isolated);
+  }
+  for (let index = 0; index < isolatedIds.length; index += MAX_CLUSTER_SIZE) {
+    connectedGroups.push(isolatedIds.slice(index, index + MAX_CLUSTER_SIZE).sort());
+  }
+  return connectedGroups.map((nodeIds) => {
+    const members = nodeIds.map((id) => nodeMap.get(id));
+    const semantic = clusterLabel(members, weightedDegree);
+    const isolated = members.every((node) => (weightedDegree.get(node.id) ?? 0) === 0);
+    return {
+      id: stableClusterId(nodeIds),
+      label: isolated ? "\u672A\u8FDE\u63A5\u8282\u70B9" : semantic.label,
+      nodeIds,
+      tags: semantic.tags
+    };
+  }).sort((left, right) => right.nodeIds.length - left.nodeIds.length || left.label.localeCompare(right.label));
+}
+
+// src/web.ts
+var STRATAGATE_DSH_VERSION = "0.2.29";
 var LEGACY_THREAD_ID = "__legacy__";
 var nodeRequire = createRequire(import.meta.url);
 function installedPackageVersion(names) {
@@ -5064,6 +5181,7 @@ async function memories(runtime, url) {
         supportingEvents: node.sourceEventIds.flatMap((id) => eventMap.get(id) ?? []).map(eventSummary)
       })),
       edges: snapshot.graphEdges,
+      clusters: clusterKnowledgeGraph(snapshot.graphNodes, snapshot.graphEdges),
       migration: (() => {
         const projected = new Set(snapshot.graphProjectionJobs.filter(({ status, projectorVersion }) => status === "completed" && projectorVersion === KNOWLEDGE_GRAPH_PROJECTOR_VERSION).flatMap(({ sourceEventIds }) => sourceEventIds)).size;
         return {