@chainlesschain/personal-data-hub 0.4.3 → 0.4.4

package/__tests__/adapters/travel-tencent-map.test.js

@@ -0,0 +1,207 @@
+"use strict";
+
+import { describe, it, expect } from "vitest";
+const fs = require("node:fs");
+const path = require("node:path");
+const os = require("node:os");
+const crypto = require("node:crypto");
+
+const {
+  TencentMapAdapter,
+  NAME,
+  VERSION,
+  SNAPSHOT_SCHEMA_VERSION,
+} = require("../../lib/adapters/travel-tencent-map");
+
+function writeTmp(content, ext = "json") {
+  const p = path.join(
+    os.tmpdir(),
+    `cc-tencentmap-test-${crypto.randomUUID()}.${ext}`,
+  );
+  fs.writeFileSync(p, content, "utf-8");
+  return p;
+}
+
+async function collect(gen) {
+  const out = [];
+  for await (const x of gen) out.push(x);
+  return out;
+}
+
+function makeFakeDriverFactory(tables, log = {}) {
+  return () =>
+    class FakeDb {
+      constructor(dbPath, opts) {
+        log.opened = { dbPath, opts };
+      }
+      prepare(sql) {
+        for (const [needle, rows] of Object.entries(tables)) {
+          if (sql.includes(needle)) return { all: () => rows };
+        }
+        throw new Error(`no such table in: ${sql}`);
+      }
+      close() {
+        log.closed = true;
+      }
+    };
+}
+
+const SNAPSHOT = {
+  schemaVersion: SNAPSHOT_SCHEMA_VERSION,
+  snapshottedAt: 1716383021000,
+  vendor: "tencent-map",
+  account: { uid: "U1" },
+  events: [
+    {
+      kind: "favourite",
+      id: "fav-1",
+      capturedAt: 1716383021000,
+      name: "公司",
+      lat: 31.2,
+      lng: 121.44,
+      category: "company",
+    },
+    {
+      kind: "route",
+      id: "route-2",
+      capturedAt: 1716383021000,
+      from: { name: "公司" },
+      to: { name: "体育馆" },
+      mode: "bike",
+    },
+  ],
+};
+
+describe("constants", () => {
+  it("exposes name/version/schema", () => {
+    expect(NAME).toBe("travel-tencent-map");
+    expect(VERSION).toBe("0.2.0");
+    expect(SNAPSHOT_SCHEMA_VERSION).toBe(1);
+  });
+});
+
+describe("authenticate", () => {
+  it("mirrors baidu: snapshot / sqlite-needs-deviceId / NO_INPUT", async () => {
+    const p = writeTmp("{}");
+    try {
+      const a = new TencentMapAdapter();
+      expect((await a.authenticate({ inputPath: p })).mode).toBe(
+        "snapshot-file",
+      );
+      expect(
+        (
+          await new TencentMapAdapter({ dbPath: "x.db" }).authenticate({})
+        ).reason,
+      ).toBe("NO_ACCOUNT_DEVICE_ID");
+      expect((await a.authenticate({})).reason).toBe("NO_INPUT");
+    } finally {
+      fs.unlinkSync(p);
+    }
+  });
+});
+
+describe("sync — snapshot mode", () => {
+  it("yields events with tencent-map: prefixed originalId", async () => {
+    const p = writeTmp(JSON.stringify(SNAPSHOT));
+    try {
+      const a = new TencentMapAdapter();
+      const items = await collect(a.sync({ inputPath: p }));
+      expect(items.map((i) => i.originalId)).toEqual([
+        "tencent-map:favourite:fav-1",
+        "tencent-map:route:route-2",
+      ]);
+      expect(items[0].payload.account).toEqual({ uid: "U1" });
+    } finally {
+      fs.unlinkSync(p);
+    }
+  });
+
+  it("throws on schemaVersion mismatch", async () => {
+    const p = writeTmp(JSON.stringify({ schemaVersion: 2, events: [] }));
+    try {
+      await expect(
+        collect(new TencentMapAdapter().sync({ inputPath: p })),
+      ).rejects.toThrow(/schemaVersion mismatch/);
+    } finally {
+      fs.unlinkSync(p);
+    }
+  });
+});
+
+describe("sync — sqlite mode (fake driver)", () => {
+  it("yields rows + falls back to tencent_* legacy tables (keyword alias)", async () => {
+    const p = writeTmp("fake", "db");
+    const log = {};
+    try {
+      const a = new TencentMapAdapter({
+        dbPath: p,
+        account: { deviceId: "DEV1" },
+        dbDriverFactory: makeFakeDriverFactory(
+          {
+            tencent_route_history: [
+              {
+                _id: 5,
+                mode: "walk",
+                start_name: "家",
+                end_name: "菜场",
+                time: 1716383021,
+              },
+            ],
+            tencent_search_history: [
+              { _id: 6, keyword: "奶茶", city: "深圳", time: 1716383021000 },
+            ],
+          },
+          log,
+        ),
+      });
+      const items = await collect(a.sync({}));
+      expect(items).toHaveLength(2);
+      expect(items[0].payload.record).toMatchObject({
+        vendorId: "tencentmap",
+        recordId: "route-5",
+        vehicleType: "walk",
+        carrier: "腾讯地图",
+        departureMs: 1716383021 * 1000,
+      });
+      // searchRowToRecord accepts the tencent `keyword` alias
+      expect(items[1].payload.record.to).toMatchObject({
+        name: "奶茶",
+        city: "深圳",
+      });
+      expect(log.closed).toBe(true);
+    } finally {
+      fs.unlinkSync(p);
+    }
+  });
+
+  it("requires account.deviceId at sync time", async () => {
+    const a = new TencentMapAdapter({ dbPath: "x.db" });
+    await expect(collect(a.sync({}))).rejects.toThrow(
+      /account\.deviceId required/,
+    );
+  });
+});
+
+describe("normalize", () => {
+  it("snapshot route → bike trip titled by place names", async () => {
+    const p = writeTmp(JSON.stringify(SNAPSHOT));
+    try {
+      const a = new TencentMapAdapter();
+      const [fav, route] = await collect(a.sync({ inputPath: p }));
+      expect(a.normalize(fav).events[0].content.title).toBe("visit: → 公司");
+      const batch = a.normalize(route);
+      expect(batch.events[0].content.title).toBe("bike: 公司 → 体育馆");
+      expect(
+        batch.persons.find((x) => x.subtype === "merchant").names,
+      ).toEqual(["腾讯地图"]);
+    } finally {
+      fs.unlinkSync(p);
+    }
+  });
+
+  it("throws on missing payload", () => {
+    expect(() => new TencentMapAdapter().normalize(null)).toThrow(
+      /payload missing/,
+    );
+  });
+});

package/lib/adapters/social-kuaishou/index.js

@@ -54,8 +54,13 @@ const KIND_PROFILE = "profile";
 const KIND_WATCH = "watch";
 const KIND_COLLECT = "collect";
 const KIND_SEARCH = "search";
-// v0.2.1 — KIND_PROFILE added (mirrors Douyin/Toutiao); v0.3 will add watch/
-// collect/search via NS_sig3. SNAPSHOT_SCHEMA_VERSION stays at 1 — additive.
+// v0.2.1 — KIND_PROFILE added (mirrors Douyin/Toutiao). The watch/collect/
+// search producers LANDED since (verified 2026-06-11): Android
+// KuaishouLocalCollector emits all 4 kinds via the NS_sig3 WebSignBridge
+// path, KuaishouRootDbExtractor emits watch/collect/search, and the PC ADB
+// KuaishouApiClient fetches them through its injected signProvider (signed
+// GraphQL). This adapter normalizes whatever the snapshot carries.
+// SNAPSHOT_SCHEMA_VERSION stays at 1 — additive.
 const VALID_SNAPSHOT_KINDS = Object.freeze([
   KIND_PROFILE,
   KIND_WATCH,

package/lib/adapters/social-kuaishou-adb/api-client.js

@@ -119,23 +119,19 @@ class KuaishouApiClient {
       );
       return null;
     }
-    let decoded;
-    try {
-      decoded = decodeURIComponent(cpMatch[1]);
-    } catch {
-      decoded = cpMatch[1];
-    }
-    const trimmed = decoded.trimStart();
-    if (!trimmed.startsWith("{")) {
+    const jsonText = apiPhDecodeCandidates(cpMatch[1]).find((c) =>
+      c.trimStart().startsWith("{"),
+    );
+    if (!jsonText) {
       this._setLastError(
         -9,
-        "kuaishou.web.cp.api_ph 解码后非 JSON (likely base64 — v0.3 加 fallback)",
+        "kuaishou.web.cp.api_ph 解码后非 JSON (urlencoded + base64 fallback 均失败)",
       );
       return null;
     }
     let obj;
     try {
-      obj = JSON.parse(decoded);
+      obj = JSON.parse(jsonText);
     } catch (e) {
       this._setLastError(-3, "parse: " + (e.message || String(e)));
       return null;
@@ -359,20 +355,43 @@ function extractPhotoList(feeds, limit, build) {
   return out;
 }
 
-function extractEmbeddedUid(cpRaw) {
+/**
+ * api_ph payload decode chain (v0.3): newer Kuaishou builds write the
+ * `kuaishou.web.cp.api_ph` cookie as base64(JSON) instead of urlencoded
+ * JSON. Yields the URI-decoded string first; when that doesn't look like
+ * JSON but matches the base64 charset (std or url-safe), also yields the
+ * base64-decoded form — gated on the result starting with `{` so lenient
+ * Buffer decoding of arbitrary text can't surface garbage.
+ */
+function apiPhDecodeCandidates(cpRaw) {
   let decoded;
   try {
     decoded = decodeURIComponent(cpRaw);
   } catch {
     decoded = cpRaw;
   }
-  for (const pat of [
-    /"?user_id"?\s*:\s*"?(\d+)"?/,
-    /"?uid"?\s*:\s*"?(\d+)"?/,
-    /"?userId"?\s*:\s*"?(\d+)"?/,
-  ]) {
-    const m = pat.exec(decoded);
-    if (m && m[1] && m[1] !== "0") return m[1];
+  const out = [decoded];
+  const trimmed = decoded.trim();
+  if (!trimmed.startsWith("{") && /^[A-Za-z0-9+/\-_]+={0,2}$/.test(trimmed)) {
+    const b64 = Buffer.from(
+      trimmed.replace(/-/g, "+").replace(/_/g, "/"),
+      "base64",
+    ).toString("utf-8");
+    if (b64.trimStart().startsWith("{")) out.push(b64);
+  }
+  return out;
+}
+
+function extractEmbeddedUid(cpRaw) {
+  for (const decoded of apiPhDecodeCandidates(cpRaw)) {
+    for (const pat of [
+      /"?user_id"?\s*:\s*"?(\d+)"?/,
+      /"?uid"?\s*:\s*"?(\d+)"?/,
+      /"?userId"?\s*:\s*"?(\d+)"?/,
+    ]) {
+      const m = pat.exec(decoded);
+      if (m && m[1] && m[1] !== "0") return m[1];
+    }
   }
   return null;
 }
@@ -393,5 +412,6 @@ module.exports = {
     normalizeMs,
     extractPhotoList,
     extractEmbeddedUid,
+    apiPhDecodeCandidates,
   },
 };

package/lib/adapters/social-kuaishou-adb/cookies-extension.js

@@ -34,6 +34,9 @@ const crypto = require("node:crypto");
 const {
   readChromiumCookies,
 } = require("../social-bilibili-adb/chromium-cookies-reader");
+const {
+  _internals: { apiPhDecodeCandidates },
+} = require("./api-client");
 
 const KUAISHOU_COOKIES_REMOTE_PATH =
   "/data/data/com.smile.gifmaker/app_webview/Default/Cookies";
@@ -137,22 +140,20 @@ function pickUidFromCookieMap(byName) {
   }
   const cpRaw = byName.get("kuaishou.web.cp.api_ph")?.value;
   if (cpRaw) {
-    let decoded;
-    try {
-      decoded = decodeURIComponent(cpRaw);
-    } catch {
-      decoded = cpRaw;
-    }
     // Try nested user_id / uid / userId regex (don't require strict JSON
-    // — api_ph format isn't documented and varies)
-    for (const pat of [
-      /"?user_id"?\s*:\s*"?(\d+)"?/,
-      /"?uid"?\s*:\s*"?(\d+)"?/,
-      /"?userId"?\s*:\s*"?(\d+)"?/,
-    ]) {
-      const m = pat.exec(decoded);
-      if (m && m[1] && m[1] !== "0") {
-        return m[1];
+    // — api_ph format isn't documented and varies). v0.3: candidates
+    // include the base64-decoded form for newer Kuaishou builds that
+    // write api_ph as base64(JSON).
+    for (const decoded of apiPhDecodeCandidates(cpRaw)) {
+      for (const pat of [
+        /"?user_id"?\s*:\s*"?(\d+)"?/,
+        /"?uid"?\s*:\s*"?(\d+)"?/,
+        /"?userId"?\s*:\s*"?(\d+)"?/,
+      ]) {
+        const m = pat.exec(decoded);
+        if (m && m[1] && m[1] !== "0") {
+          return m[1];
+        }
       }
     }
   }

package/lib/adapters/social-toutiao/index.js

@@ -57,10 +57,14 @@ const KIND_PROFILE = "profile";
 const KIND_READ = "read";
 const KIND_COLLECTION = "collection";
 const KIND_SEARCH = "search";
-// v0.2.1 — KIND_PROFILE added (mirrors Douyin); v0.3 will add read/collection
-// /search once _signature path is wired. SNAPSHOT_SCHEMA_VERSION stays at 1:
-// old (events-only) snapshots remain compatible; new profile events are an
-// additive extension.
+// v0.2.1 — KIND_PROFILE added (mirrors Douyin). The read/collection/search
+// producers LANDED since (verified 2026-06-11): Android ToutiaoLocalCollector
+// emits all 4 kinds via the _signature WebSignBridge path, ToutiaoRootDbExtractor
+// emits read/collection/search, and the PC ADB ToutiaoApiClient fetches
+// feed/collection/search through its injected signProvider. This adapter
+// normalizes whatever the snapshot carries. SNAPSHOT_SCHEMA_VERSION stays
+// at 1: old (events-only) snapshots remain compatible; profile events are
+// an additive extension.
 const VALID_SNAPSHOT_KINDS = Object.freeze([
   KIND_PROFILE,
   KIND_READ,

package/lib/adapters/travel-base/index.js

@@ -150,8 +150,15 @@ function normalizeTravelRecord(rec, ctx = {}) {
 
 function buildTitle(rec) {
   const vt = rec.vehicleType || "trip";
-  const from = rec.from ? (rec.from.station || rec.from.city || "?") : "";
-  const to = rec.to ? (rec.to.station || rec.to.city || "?") : "";
+  // station > city > name — name matters for Amap route/search records,
+  // which carry ONLY p.name (no station/city); without it every Amap trip
+  // event was titled "car: ? → ?".
+  const from = rec.from
+    ? (rec.from.station || rec.from.city || rec.from.name || "?")
+    : "";
+  const to = rec.to
+    ? (rec.to.station || rec.to.city || rec.to.name || "?")
+    : "";
   if (from && to) return `${vt}: ${from} → ${to}`;
   if (to) return `${vt}: → ${to}`;
   return `${vt}: ${rec.carrier || rec.recordId}`;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chainlesschain/personal-data-hub",
-  "version": "0.4.3",
+  "version": "0.4.4",
   "description": "Personal Data Hub — UnifiedSchema + validators + KG ingest helpers for the data-back-to-the-individual middleware",
   "type": "commonjs",
   "main": "lib/index.js",