@chainlesschain/personal-data-hub 0.4.27 → 0.4.28

package/__tests__/adapters/wechat-pc-group-topic.test.js

@@ -0,0 +1,63 @@
+"use strict";
+
+import { describe, it, expect } from "vitest";
+
+const { WeChatPcAdapter } = require("../../lib/adapters/wechat-pc");
+
+// Build the raw envelope shape that WeChatPcAdapter.normalize() consumes for a
+// group message (the sync() generator yields { kind:"message", payload, ... }).
+function groupMessageRaw(payload) {
+  return {
+    adapter: "wechat-pc",
+    kind: "message",
+    originalId: "orig-1",
+    capturedAt: 1780000000000,
+    payload: { kind: "message", ...payload },
+  };
+}
+
+describe("wechat-pc — group topic naming", () => {
+  it("uses the resolved group display name for the topic when groupName is present", () => {
+    const adapter = new WeChatPcAdapter();
+    const out = adapter.normalize(groupMessageRaw({
+      talker: "45498354778@chatroom",
+      isGroup: true,
+      senderWxid: "wxid_friend",
+      groupName: "家庭群",
+      text: "晚饭吃什么",
+      createdTimeMs: 1780000000000,
+    }));
+    expect(out.topics).toHaveLength(1);
+    // Stable id keyed on the chatroom wxid (identity unchanged)...
+    expect(out.topics[0].id).toBe("topic-wechat-group-45498354778@chatroom");
+    // ...but the human-readable display name is used, NOT the numeric id.
+    expect(out.topics[0].name).toBe("家庭群");
+  });
+
+  it("falls back to the raw numeric id when no group name was resolved", () => {
+    const adapter = new WeChatPcAdapter();
+    const out = adapter.normalize(groupMessageRaw({
+      talker: "45498354778@chatroom",
+      isGroup: true,
+      senderWxid: "wxid_friend",
+      groupName: null,
+      text: "hi",
+      createdTimeMs: 1780000000000,
+    }));
+    expect(out.topics).toHaveLength(1);
+    expect(out.topics[0].name).toBe("45498354778");
+  });
+
+  it("blank/whitespace group name falls back to the raw id (no empty topic name)", () => {
+    const adapter = new WeChatPcAdapter();
+    const out = adapter.normalize(groupMessageRaw({
+      talker: "12345@chatroom",
+      isGroup: true,
+      senderWxid: "wxid_x",
+      groupName: "   ",
+      text: "hi",
+      createdTimeMs: 1780000000000,
+    }));
+    expect(out.topics[0].name).toBe("12345");
+  });
+});

package/__tests__/analysis-skills.test.js

@@ -498,6 +498,32 @@ describe("InterestsSkill", () => {
     expect(r.topTopics[0].name).toBe("Travel");
     expect(r.llmInterests).toBeNull();
   });
+
+  it("drops unresolved numeric group-id topics (e.g. WeChat chatroom ids) from the profile", async () => {
+    // Real interest topic
+    rig.vault.putTopic({
+      id: "topic-doubao", type: "topic", name: "豆包",
+      derivedFromEvents: ["e1"],
+      ingestedAt: Date.now(), source: defaultSource("test"),
+    });
+    // Unresolved group-chat topics named by raw numeric chatroom id — noise.
+    rig.vault.putTopic({
+      id: "topic-g1", type: "topic", name: "45498354778",
+      derivedFromEvents: [],
+      ingestedAt: Date.now() + 1, source: defaultSource("test"),
+    });
+    rig.vault.putTopic({
+      id: "topic-g2", type: "topic", name: "54346634535",
+      derivedFromEvents: [],
+      ingestedAt: Date.now() + 2, source: defaultSource("test"),
+    });
+    const skill = new InterestsSkill({ vault: rig.vault });
+    const r = await skill.run({});
+    const names = r.topTopics.map((t) => t.name);
+    expect(names).toContain("豆包");
+    expect(names).not.toContain("45498354778");
+    expect(names).not.toContain("54346634535");
+  });
 });
 
 // ─── TimelineSkill ──────────────────────────────────────────────────────
@@ -534,6 +560,34 @@ describe("TimelineSkill", () => {
     const r = await skill.run({ since: ts(2026, 4, 1) });
     expect(r.llm_narrative).toBe("你这周点了一次外卖。");
   });
+
+  it("excludes inventory-snapshot events (installed-app / contact roster) from the narrative", async () => {
+    // Real activity event (extra has no `kind` → must be kept)
+    makePayment(rig.vault, { id: "act-1", occurredAt: ts(2026, 5, 1), counterpartyName: "美团", amount: 10, adapter: "alipay-bill", title: "外卖" });
+    // Inventory-snapshot events stamped at a LATER (collection) time — these
+    // would dominate a DESC time query but must be filtered out.
+    rig.vault.putEvent({
+      id: "event-android-app-com.x", type: "event", subtype: "other",
+      occurredAt: ts(2026, 6, 1), actor: "person-self",
+      content: { title: "应用：X" },
+      ingestedAt: Date.now(), source: defaultSource("system-data-android"),
+      extra: { kind: "app-snapshot", packageName: "com.x" },
+    });
+    rig.vault.putEvent({
+      id: "event-android-contact-y", type: "event", subtype: "other",
+      occurredAt: ts(2026, 6, 1), actor: "person-self",
+      content: { title: "联系人：Y" },
+      ingestedAt: Date.now(), source: defaultSource("system-data-android"),
+      extra: { kind: "contact-snapshot" },
+    });
+    const skill = new TimelineSkill({ vault: rig.vault });
+    const r = await skill.run({ since: ts(2026, 4, 1) });
+    const ids = r.entries.map((e) => e.id);
+    expect(ids).toContain("act-1");
+    expect(ids).not.toContain("event-android-app-com.x");
+    expect(ids).not.toContain("event-android-contact-y");
+    expect(r.summary.totalEvents).toBe(1);
+  });
 });
 
 // ─── runAnalysisSkill dispatcher ─────────────────────────────────────────

package/lib/adapters/wechat-pc/index.js

@@ -290,6 +290,22 @@ class WeChatPcAdapter {
       null;
     const fallbackCapturedAt = Date.now();
     const messages = (result && Array.isArray(result.messages)) ? result.messages : [];
+    // Harvest group display names from the contact roster: WeChat stores group
+    // chatrooms (wxid ending @chatroom) in contact.db with a nickname/remark.
+    // They are skipped as Person entities below, but their names let us label
+    // group Topics with a human-readable name instead of the raw numeric id.
+    const groupNames = new Map();
+    {
+      const contactsForNames = (result && Array.isArray(result.contacts)) ? result.contacts : [];
+      for (const c of contactsForNames) {
+        if (!c || typeof c.wxid !== "string" || !c.wxid.endsWith("@chatroom")) continue;
+        const nm =
+          (typeof c.remark === "string" && c.remark.trim()) ||
+          (typeof c.nickname === "string" && c.nickname.trim()) ||
+          "";
+        if (nm) groupNames.set(c.wxid, nm);
+      }
+    }
     let emitted = 0;
     // The sidecar already applied `limit` across all sources (chat/biz/sns/
     // favorite). Yield everything it returned — do NOT re-cap here, or the
@@ -311,6 +327,7 @@ class WeChatPcAdapter {
         text: typeof m.text === "string" ? m.text : "",
         senderWxid: isGroup ? (m.sender || null) : null,
         isGroup,
+        groupName: isGroup && conv ? (groupNames.get(conv) || null) : null,
         contentBlob: typeof m.text === "string" ? m.text : null,
         // provenance: chat | biz(公众号) | sns(朋友圈) | favorite(收藏)
         wechatSource: typeof m.source === "string" ? m.source : "chat",
@@ -423,10 +440,16 @@ function normalizeMessage(p, raw, ingestedAt) {
 
   const topics = [];
   if (isGroup && p.talker) {
+    // Prefer the resolved group display name (harvested from contact.db in
+    // sync()); fall back to the raw numeric chatroom id only when unknown.
+    const groupName =
+      (typeof p.groupName === "string" && p.groupName.trim())
+        ? p.groupName.trim()
+        : p.talker.replace("@chatroom", "");
     topics.push({
       id: `topic-wechat-group-${p.talker}`,
       type: ENTITY_TYPES.TOPIC,
-      name: p.talker.replace("@chatroom", ""),
+      name: groupName,
       ingestedAt,
       source,
       extra: { platform: "wechat", source: "pc", wxid: p.talker },

package/lib/analysis-skills/interests.js

@@ -28,6 +28,21 @@
 
 const { AnalysisSkill } = require("./base");
 
+/**
+ * A topic name carries real interest signal only if it is a human-readable
+ * label. Unresolved group-chat IDs (e.g. WeChat group topics named by their
+ * raw numeric chatroom id "45498354778") and empty names are NOT interests —
+ * they would crowd out genuine topics (coffee, photography, 豆包...) in the
+ * profile. Drop them.
+ */
+function isMeaningfulTopicName(name) {
+  if (typeof name !== "string") return false;
+  const s = name.trim();
+  if (s.length === 0) return false;
+  if (/^\d+$/.test(s)) return false; // pure-numeric = unresolved group id
+  return true;
+}
+
 class InterestsSkill extends AnalysisSkill {
   constructor(opts) {
     super({ ...opts, name: "analysis.interests" });
@@ -61,25 +76,30 @@ class InterestsSkill extends AnalysisSkill {
     let topics = [];
     try {
       const db = this.vault._requireOpen();
+      // Over-fetch (×20, capped) before filtering: vaults can hold thousands
+      // of unresolved numeric group-chat topics that would otherwise starve
+      // the few human-readable interest topics out of the top-N budget.
       topics = db.prepare(
         "SELECT id, name, derived_from_events, ingested_at FROM topics ORDER BY ingested_at DESC LIMIT ?"
-      ).all(topN * 3);
+      ).all(Math.min(topN * 20, 2000));
     } catch (_e) {
       // Older vaults may not have topics; non-fatal.
     }
-    const mapped = topics.map((t) => {
-      let eventCount = 0;
-      try {
-        const arr = t.derived_from_events ? JSON.parse(t.derived_from_events) : [];
-        if (Array.isArray(arr)) eventCount = arr.length;
-      } catch (_e) {}
-      return {
-        id: t.id,
-        name: t.name,
-        eventCount,
-        lastSeen: t.ingested_at || null,
-      };
-    });
+    const mapped = topics
+      .filter((t) => isMeaningfulTopicName(t.name))
+      .map((t) => {
+        let eventCount = 0;
+        try {
+          const arr = t.derived_from_events ? JSON.parse(t.derived_from_events) : [];
+          if (Array.isArray(arr)) eventCount = arr.length;
+        } catch (_e) {}
+        return {
+          id: t.id,
+          name: t.name,
+          eventCount,
+          lastSeen: t.ingested_at || null,
+        };
+      });
     return mapped
       .sort((a, b) => (b.eventCount - a.eventCount) || ((b.lastSeen || 0) - (a.lastSeen || 0)))
       .slice(0, topN);

package/lib/analysis-skills/timeline.js

@@ -63,7 +63,12 @@ class TimelineSkill extends AnalysisSkill {
   }
 
   _fetchEvents({ since, until }, limit) {
-    const q = { limit };
+    // Exclude inventory-snapshot events (installed-app roster + contact
+    // roster from system-data-android). They carry a synthetic
+    // collection-time occurredAt — tens of thousands of them cluster at one
+    // recent timestamp and would otherwise crowd out real activity from this
+    // chronological narrative. They remain in the vault for facet counts.
+    const q = { limit, excludeExtraKinds: ["app-snapshot", "contact-snapshot"] };
     if (since != null) q.since = since;
     if (until != null) q.until = until;
     const events = this.vault.queryEvents(q) || [];

package/lib/vault.js

@@ -811,6 +811,18 @@ class LocalVault {
       where.push("source_adapter = @adapter");
       params.adapter = q.adapter;
     }
+    if (Array.isArray(q.excludeExtraKinds) && q.excludeExtraKinds.length > 0) {
+      // Exclude inventory-snapshot events (e.g. installed-app / contact-roster
+      // facet events) whose extra.kind is in the given list. Those carry a
+      // synthetic collection-time occurredAt and would otherwise dominate any
+      // time-ordered (occurred_at DESC) query. Rows with no extra.kind are kept.
+      const placeholders = q.excludeExtraKinds.map((_v, i) => `@xk${i}`);
+      q.excludeExtraKinds.forEach((v, i) => { params[`xk${i}`] = v; });
+      where.push(
+        "(json_extract(extra, '$.kind') IS NULL OR json_extract(extra, '$.kind') NOT IN (" +
+        placeholders.join(", ") + "))",
+      );
+    }
 
     const limit = Number.isInteger(q.limit) && q.limit > 0 ? Math.min(q.limit, 10000) : 100;
     const offset = Number.isInteger(q.offset) && q.offset >= 0 ? q.offset : 0;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chainlesschain/personal-data-hub",
-  "version": "0.4.27",
+  "version": "0.4.28",
   "description": "Personal Data Hub — UnifiedSchema + validators + KG ingest helpers for the data-back-to-the-individual middleware",
   "type": "commonjs",
   "main": "lib/index.js",