@chainlesschain/personal-data-hub 0.2.0 → 0.2.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/__tests__/adapters/ai-chat-cookie-capture-spec.test.js +211 -0
- package/__tests__/adapters/ai-chat-health-checker.test.js +262 -0
- package/__tests__/adapters/ai-chat-history.test.js +8 -7
- package/__tests__/adapters/ai-chat-vendors.test.js +149 -8
- package/__tests__/adapters/social-toutiao-kuaishou-scaffold.test.js +269 -0
- package/__tests__/adapters/system-data-android-ingest.test.js +144 -0
- package/__tests__/adapters/system-data-android.test.js +387 -0
- package/__tests__/adapters/wechat-bootstrap.test.js +240 -0
- package/__tests__/adapters/wechat-env-probe.test.js +162 -0
- package/__tests__/adapters/wechat-frida-agent.test.js +322 -0
- package/__tests__/adapters/wechat-frida-integration.test.js +149 -0
- package/__tests__/adapters/wechat-frida-key-provider.test.js +188 -0
- package/__tests__/adapters/wechat-md5-key-provider.test.js +101 -0
- package/__tests__/analysis-skills.test.js +147 -0
- package/__tests__/analysis.test.js +329 -1
- package/__tests__/e2e/ai-chat-cross-source-journey.test.js +213 -0
- package/__tests__/e2e/full-user-journey.test.js +188 -0
- package/__tests__/integration/ai-chat-history-registry.test.js +228 -0
- package/__tests__/integration/aichat-wizard-end-to-end.test.js +282 -0
- package/__tests__/integration/cross-adapter-pipelines.test.js +396 -0
- package/__tests__/integration/social-bilibili-pipeline.test.js +261 -0
- package/__tests__/integration/wechat-bootstrap-end-to-end.test.js +390 -0
- package/__tests__/registry.test.js +4 -2
- package/__tests__/social-adapters.test.js +63 -14
- package/__tests__/social-bilibili-snapshot.test.js +278 -0
- package/__tests__/wechat-adapter.test.js +118 -0
- package/lib/adapters/ai-chat-history/ai-chat-adapter.js +55 -16
- package/lib/adapters/ai-chat-history/cookie-capture-spec.js +331 -0
- package/lib/adapters/ai-chat-history/health-checker.js +210 -0
- package/lib/adapters/ai-chat-history/schema-map.js +42 -5
- package/lib/adapters/ai-chat-history/vendor-spec.js +1 -0
- package/lib/adapters/ai-chat-history/vendors/doubao.js +255 -0
- package/lib/adapters/ai-chat-history/wizard-controller.js +473 -0
- package/lib/adapters/alipay-bill/alipay-bill-adapter.js +4 -0
- package/lib/adapters/social-bilibili/adapter.js +500 -0
- package/lib/adapters/social-bilibili/index.js +21 -169
- package/lib/adapters/social-kuaishou/index.js +237 -0
- package/lib/adapters/social-toutiao/index.js +236 -0
- package/lib/adapters/system-data-android/adapter.js +348 -0
- package/lib/adapters/system-data-android/index.js +76 -0
- package/lib/adapters/wechat/bootstrap.js +146 -0
- package/lib/adapters/wechat/content-parser.js +11 -2
- package/lib/adapters/wechat/db-reader.js +88 -10
- package/lib/adapters/wechat/env-probe.js +218 -0
- package/lib/adapters/wechat/frida-agent/loader.js +74 -0
- package/lib/adapters/wechat/frida-agent/wechat-key-hook.js +248 -0
- package/lib/adapters/wechat/index.js +9 -0
- package/lib/adapters/wechat/key-providers/frida-key-provider.js +252 -0
- package/lib/adapters/wechat/key-providers/index.js +22 -0
- package/lib/adapters/wechat/key-providers/key-provider-base.js +44 -0
- package/lib/adapters/wechat/key-providers/md5-key-provider.js +81 -0
- package/lib/adapters/wechat/normalize.js +12 -3
- package/lib/analysis-skills/spending.js +4 -1
- package/lib/analysis.js +191 -2
- package/lib/index.js +16 -0
- package/lib/prompt-builder.js +11 -1
- package/lib/query-parser.js +7 -1
- package/lib/vault.js +77 -0
- package/package.json +8 -1
|
@@ -37,9 +37,47 @@ describe("BilibiliAdapter", () => {
|
|
|
37
37
|
expect(a.extractMode).toBe("device-pull");
|
|
38
38
|
});
|
|
39
39
|
|
|
40
|
-
it("
|
|
41
|
-
|
|
42
|
-
|
|
40
|
+
it("accepts stateless construction (snapshot mode added in A8)", () => {
|
|
41
|
+
// Before A8: constructor required opts.account.uid. After A8 the adapter
|
|
42
|
+
// is stateless when running snapshot mode (in-APK Android cc reads a JSON
|
|
43
|
+
// produced by the phone). Sqlite mode still needs account.uid but the
|
|
44
|
+
// check moved into _syncViaSqlite where it actually matters.
|
|
45
|
+
expect(() => new BilibiliAdapter({})).not.toThrow();
|
|
46
|
+
expect(() => new BilibiliAdapter({ account: {} })).not.toThrow();
|
|
47
|
+
expect(() => new BilibiliAdapter()).not.toThrow();
|
|
48
|
+
});
|
|
49
|
+
|
|
50
|
+
it("sqlite mode rejects missing account.uid at sync time", async () => {
|
|
51
|
+
const a = new BilibiliAdapter({ dbPath: "/tmp/bili.db" });
|
|
52
|
+
// Path-existence check happens before account.uid validation, so we
|
|
53
|
+
// exercise the guard via dbPath=null + account=null which falls to
|
|
54
|
+
// "sync needs inputPath OR dbPath" first. Use a real-looking dbPath
|
|
55
|
+
// with no account to surface the account.uid throw deterministically.
|
|
56
|
+
const fs = require("node:fs");
|
|
57
|
+
const path = require("node:path");
|
|
58
|
+
const os = require("node:os");
|
|
59
|
+
const dir = fs.mkdtempSync(path.join(os.tmpdir(), "bili-no-acct-"));
|
|
60
|
+
const dbPath = path.join(dir, "bili.db");
|
|
61
|
+
fs.writeFileSync(dbPath, "fake");
|
|
62
|
+
try {
|
|
63
|
+
const b = new BilibiliAdapter({
|
|
64
|
+
dbPath,
|
|
65
|
+
dbDriverFactory: () => () => ({
|
|
66
|
+
prepare: () => ({ all: () => [] }),
|
|
67
|
+
close() {},
|
|
68
|
+
}),
|
|
69
|
+
});
|
|
70
|
+
let threw = null;
|
|
71
|
+
try {
|
|
72
|
+
for await (const _r of b.sync()) { /* drain */ }
|
|
73
|
+
} catch (err) {
|
|
74
|
+
threw = err;
|
|
75
|
+
}
|
|
76
|
+
expect(threw).toBeTruthy();
|
|
77
|
+
expect(String(threw.message)).toMatch(/account\.uid/);
|
|
78
|
+
} finally {
|
|
79
|
+
fs.rmSync(dir, { recursive: true, force: true });
|
|
80
|
+
}
|
|
43
81
|
});
|
|
44
82
|
|
|
45
83
|
it("sync yields history + favourite records via mocked driver", async () => {
|
|
@@ -83,26 +121,34 @@ describe("BilibiliAdapter", () => {
|
|
|
83
121
|
}
|
|
84
122
|
});
|
|
85
123
|
|
|
86
|
-
it("
|
|
124
|
+
it("throws when neither inputPath nor dbPath provided (A8: surface config errors)", async () => {
|
|
125
|
+
// Before A8: sync silently yielded 0 if dbPath missing — masked typos and
|
|
126
|
+
// misconfigured callers. After A8 we throw so callers see the problem.
|
|
87
127
|
const a = new BilibiliAdapter({ account: { uid: "1234" } });
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
128
|
+
let threw = null;
|
|
129
|
+
try {
|
|
130
|
+
for await (const _r of a.sync()) { /* drain */ }
|
|
131
|
+
} catch (err) {
|
|
132
|
+
threw = err;
|
|
133
|
+
}
|
|
134
|
+
expect(threw).toBeTruthy();
|
|
135
|
+
expect(String(threw.message)).toMatch(/inputPath|dbPath/);
|
|
91
136
|
});
|
|
92
137
|
|
|
93
|
-
it("normalize captures bvid/avid/uploader into extra", async () => {
|
|
138
|
+
it("normalize captures bvid/avid/uploader into extra (flat payload, A8 shape)", async () => {
|
|
94
139
|
const a = new BilibiliAdapter({ account: { uid: "1234" } });
|
|
95
140
|
const raw = {
|
|
96
141
|
adapter: "social-bilibili",
|
|
97
|
-
|
|
142
|
+
kind: "history",
|
|
143
|
+
originalId: "bilibili:history:BV1abc",
|
|
98
144
|
capturedAt: 1700000000000,
|
|
99
145
|
payload: {
|
|
100
146
|
kind: "history",
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
147
|
+
title: "Test",
|
|
148
|
+
bvid: "BV1abc",
|
|
149
|
+
avid: "1234",
|
|
150
|
+
uploader: "UpA",
|
|
151
|
+
duration: 300,
|
|
106
152
|
},
|
|
107
153
|
};
|
|
108
154
|
const batch = a.normalize(raw);
|
|
@@ -110,6 +156,9 @@ describe("BilibiliAdapter", () => {
|
|
|
110
156
|
expect(batch.events[0].extra.avid).toBe("1234");
|
|
111
157
|
expect(batch.events[0].extra.uploader).toBe("UpA");
|
|
112
158
|
expect(batch.events[0].extra.duration).toBe(300);
|
|
159
|
+
// A8: history also yields an item entity (video) for KG linkage
|
|
160
|
+
expect(batch.items).toHaveLength(1);
|
|
161
|
+
expect(batch.items[0].extra.bvid).toBe("BV1abc");
|
|
113
162
|
});
|
|
114
163
|
});
|
|
115
164
|
|
|
@@ -0,0 +1,278 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
|
|
3
|
+
import { describe, it, expect, beforeEach } from "vitest";
|
|
4
|
+
|
|
5
|
+
const fs = require("node:fs");
|
|
6
|
+
const path = require("node:path");
|
|
7
|
+
const os = require("node:os");
|
|
8
|
+
|
|
9
|
+
const {
|
|
10
|
+
BilibiliAdapter,
|
|
11
|
+
SNAPSHOT_SCHEMA_VERSION,
|
|
12
|
+
VALID_KINDS,
|
|
13
|
+
} = require("../lib/adapters/social-bilibili");
|
|
14
|
+
const { validateBatch } = require("../lib/batch");
|
|
15
|
+
|
|
16
|
+
// A8 v0.1 (2026-05-22) — snapshot-mode tests, mirroring system-data-android.
|
|
17
|
+
//
|
|
18
|
+
// Why a separate file? `social-adapters.test.js` covers the legacy sqlite
|
|
19
|
+
// path (Phase 7.5 device-pull). Snapshot mode is a brand-new ingestion path
|
|
20
|
+
// driven by in-APK Android cc reading JSON from the phone's own WebView+OkHttp
|
|
21
|
+
// pipeline. Keeping tests separated makes it obvious which mode a regression
|
|
22
|
+
// belongs to.
|
|
23
|
+
|
|
24
|
+
function writeSnapshot(dir, snapshot) {
|
|
25
|
+
const p = path.join(dir, "social-bilibili.json");
|
|
26
|
+
fs.writeFileSync(p, JSON.stringify(snapshot), "utf-8");
|
|
27
|
+
return p;
|
|
28
|
+
}
|
|
29
|
+
|
|
30
|
+
describe("BilibiliAdapter snapshot mode", () => {
|
|
31
|
+
let tmpDir;
|
|
32
|
+
beforeEach(() => {
|
|
33
|
+
tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "bili-snap-"));
|
|
34
|
+
});
|
|
35
|
+
|
|
36
|
+
it("exports SNAPSHOT_SCHEMA_VERSION = 1 + 4 VALID_KINDS", () => {
|
|
37
|
+
expect(SNAPSHOT_SCHEMA_VERSION).toBe(1);
|
|
38
|
+
expect(VALID_KINDS).toEqual(["history", "favourite", "dynamic", "follow"]);
|
|
39
|
+
});
|
|
40
|
+
|
|
41
|
+
it("authenticate(inputPath) ok when readable", async () => {
|
|
42
|
+
const p = writeSnapshot(tmpDir, {
|
|
43
|
+
schemaVersion: 1,
|
|
44
|
+
snapshottedAt: Date.now(),
|
|
45
|
+
events: [],
|
|
46
|
+
});
|
|
47
|
+
const a = new BilibiliAdapter();
|
|
48
|
+
const res = await a.authenticate({ inputPath: p });
|
|
49
|
+
expect(res.ok).toBe(true);
|
|
50
|
+
expect(res.mode).toBe("snapshot-file");
|
|
51
|
+
});
|
|
52
|
+
|
|
53
|
+
it("authenticate(inputPath) fails when path unreadable", async () => {
|
|
54
|
+
const a = new BilibiliAdapter();
|
|
55
|
+
const res = await a.authenticate({ inputPath: path.join(tmpDir, "missing.json") });
|
|
56
|
+
expect(res.ok).toBe(false);
|
|
57
|
+
expect(res.reason).toBe("INPUT_PATH_UNREADABLE");
|
|
58
|
+
});
|
|
59
|
+
|
|
60
|
+
it("authenticate() with neither inputPath nor dbPath returns NO_INPUT", async () => {
|
|
61
|
+
const a = new BilibiliAdapter();
|
|
62
|
+
const res = await a.authenticate({});
|
|
63
|
+
expect(res.ok).toBe(false);
|
|
64
|
+
expect(res.reason).toBe("NO_INPUT");
|
|
65
|
+
});
|
|
66
|
+
|
|
67
|
+
it("rejects schemaVersion mismatch", async () => {
|
|
68
|
+
const p = writeSnapshot(tmpDir, {
|
|
69
|
+
schemaVersion: 99,
|
|
70
|
+
snapshottedAt: Date.now(),
|
|
71
|
+
events: [],
|
|
72
|
+
});
|
|
73
|
+
const a = new BilibiliAdapter();
|
|
74
|
+
let threw = null;
|
|
75
|
+
try {
|
|
76
|
+
for await (const _r of a.sync({ inputPath: p })) { /* drain */ }
|
|
77
|
+
} catch (err) {
|
|
78
|
+
threw = err;
|
|
79
|
+
}
|
|
80
|
+
expect(threw).toBeTruthy();
|
|
81
|
+
expect(String(threw.message)).toMatch(/schemaVersion mismatch/);
|
|
82
|
+
});
|
|
83
|
+
|
|
84
|
+
it("empty events array yields nothing (no crash)", async () => {
|
|
85
|
+
const p = writeSnapshot(tmpDir, {
|
|
86
|
+
schemaVersion: 1,
|
|
87
|
+
snapshottedAt: Date.now(),
|
|
88
|
+
events: [],
|
|
89
|
+
});
|
|
90
|
+
const a = new BilibiliAdapter();
|
|
91
|
+
const raws = [];
|
|
92
|
+
for await (const r of a.sync({ inputPath: p })) raws.push(r);
|
|
93
|
+
expect(raws).toHaveLength(0);
|
|
94
|
+
});
|
|
95
|
+
|
|
96
|
+
it("yields all 4 kinds + normalize produces valid batches", async () => {
|
|
97
|
+
const p = writeSnapshot(tmpDir, {
|
|
98
|
+
schemaVersion: 1,
|
|
99
|
+
snapshottedAt: 1716000000000,
|
|
100
|
+
account: { uid: "12345", displayName: "alice" },
|
|
101
|
+
events: [
|
|
102
|
+
{
|
|
103
|
+
kind: "history",
|
|
104
|
+
id: "BV1abc",
|
|
105
|
+
capturedAt: 1715000000000,
|
|
106
|
+
title: "Rust 异步学习",
|
|
107
|
+
bvid: "BV1abc",
|
|
108
|
+
avid: 42,
|
|
109
|
+
duration: 600,
|
|
110
|
+
uploader: "技术UP主",
|
|
111
|
+
uploaderMid: 100,
|
|
112
|
+
part: "01 介绍",
|
|
113
|
+
},
|
|
114
|
+
{
|
|
115
|
+
kind: "favourite",
|
|
116
|
+
id: "fav-BV2def",
|
|
117
|
+
capturedAt: 1714000000000,
|
|
118
|
+
title: "前端架构",
|
|
119
|
+
bvid: "BV2def",
|
|
120
|
+
folderName: "学习",
|
|
121
|
+
uploader: "码农UP",
|
|
122
|
+
},
|
|
123
|
+
{
|
|
124
|
+
kind: "dynamic",
|
|
125
|
+
id: "dyn-99",
|
|
126
|
+
capturedAt: 1713000000000,
|
|
127
|
+
summary: "今天发了一个新视频",
|
|
128
|
+
dynamicType: "video",
|
|
129
|
+
rid: 99,
|
|
130
|
+
authorMid: 200,
|
|
131
|
+
authorName: "我关注的UP",
|
|
132
|
+
},
|
|
133
|
+
{
|
|
134
|
+
kind: "follow",
|
|
135
|
+
id: "follow-300",
|
|
136
|
+
capturedAt: 1712000000000,
|
|
137
|
+
mid: 300,
|
|
138
|
+
uname: "美食UP",
|
|
139
|
+
face: "https://i0.hdslb.com/...",
|
|
140
|
+
sign: "好吃的视频",
|
|
141
|
+
},
|
|
142
|
+
],
|
|
143
|
+
});
|
|
144
|
+
const a = new BilibiliAdapter();
|
|
145
|
+
const raws = [];
|
|
146
|
+
for await (const r of a.sync({ inputPath: p })) raws.push(r);
|
|
147
|
+
|
|
148
|
+
expect(raws).toHaveLength(4);
|
|
149
|
+
expect(raws.map((r) => r.kind).sort()).toEqual([
|
|
150
|
+
"dynamic",
|
|
151
|
+
"favourite",
|
|
152
|
+
"follow",
|
|
153
|
+
"history",
|
|
154
|
+
]);
|
|
155
|
+
// Stable originalId format
|
|
156
|
+
expect(raws.find((r) => r.kind === "history").originalId).toBe("bilibili:history:BV1abc");
|
|
157
|
+
expect(raws.find((r) => r.kind === "favourite").originalId).toBe("bilibili:favourite:fav-BV2def");
|
|
158
|
+
expect(raws.find((r) => r.kind === "dynamic").originalId).toBe("bilibili:dynamic:dyn-99");
|
|
159
|
+
expect(raws.find((r) => r.kind === "follow").originalId).toBe("bilibili:follow:follow-300");
|
|
160
|
+
|
|
161
|
+
// Normalize each + validate
|
|
162
|
+
for (const raw of raws) {
|
|
163
|
+
const batch = a.normalize(raw);
|
|
164
|
+
const v = validateBatch(batch);
|
|
165
|
+
expect(v.valid).toBe(true);
|
|
166
|
+
|
|
167
|
+
if (raw.kind === "history") {
|
|
168
|
+
expect(batch.events[0].subtype).toBe("browse");
|
|
169
|
+
expect(batch.events[0].extra.bvid).toBe("BV1abc");
|
|
170
|
+
expect(batch.events[0].extra.duration).toBe(600);
|
|
171
|
+
expect(batch.items).toHaveLength(1);
|
|
172
|
+
expect(batch.items[0].name).toBe("Rust 异步学习");
|
|
173
|
+
} else if (raw.kind === "favourite") {
|
|
174
|
+
expect(batch.events[0].subtype).toBe("like");
|
|
175
|
+
expect(batch.events[0].extra.folderName).toBe("学习");
|
|
176
|
+
expect(batch.items).toHaveLength(1);
|
|
177
|
+
} else if (raw.kind === "dynamic") {
|
|
178
|
+
expect(batch.events[0].subtype).toBe("browse");
|
|
179
|
+
expect(batch.events[0].extra.dynamicType).toBe("video");
|
|
180
|
+
expect(batch.events[0].extra.authorName).toBe("我关注的UP");
|
|
181
|
+
} else if (raw.kind === "follow") {
|
|
182
|
+
// Follow yields a person, not an event
|
|
183
|
+
expect(batch.events).toHaveLength(0);
|
|
184
|
+
expect(batch.persons).toHaveLength(1);
|
|
185
|
+
expect(batch.persons[0].names[0]).toBe("美食UP");
|
|
186
|
+
expect(batch.persons[0].identifiers["bilibili-mid"]).toEqual(["300"]);
|
|
187
|
+
}
|
|
188
|
+
}
|
|
189
|
+
});
|
|
190
|
+
|
|
191
|
+
it("per-kind include filter (e.g. include.follow=false drops follows)", async () => {
|
|
192
|
+
const p = writeSnapshot(tmpDir, {
|
|
193
|
+
schemaVersion: 1,
|
|
194
|
+
snapshottedAt: Date.now(),
|
|
195
|
+
events: [
|
|
196
|
+
{ kind: "history", id: "h1", title: "x" },
|
|
197
|
+
{ kind: "follow", id: "f1", mid: 1, uname: "u" },
|
|
198
|
+
],
|
|
199
|
+
});
|
|
200
|
+
const a = new BilibiliAdapter();
|
|
201
|
+
const raws = [];
|
|
202
|
+
for await (const r of a.sync({ inputPath: p, include: { follow: false } })) {
|
|
203
|
+
raws.push(r);
|
|
204
|
+
}
|
|
205
|
+
expect(raws).toHaveLength(1);
|
|
206
|
+
expect(raws[0].kind).toBe("history");
|
|
207
|
+
});
|
|
208
|
+
|
|
209
|
+
it("limit caps emission", async () => {
|
|
210
|
+
const p = writeSnapshot(tmpDir, {
|
|
211
|
+
schemaVersion: 1,
|
|
212
|
+
snapshottedAt: Date.now(),
|
|
213
|
+
events: [
|
|
214
|
+
{ kind: "history", id: "1", title: "a" },
|
|
215
|
+
{ kind: "history", id: "2", title: "b" },
|
|
216
|
+
{ kind: "history", id: "3", title: "c" },
|
|
217
|
+
],
|
|
218
|
+
});
|
|
219
|
+
const a = new BilibiliAdapter();
|
|
220
|
+
const raws = [];
|
|
221
|
+
for await (const r of a.sync({ inputPath: p, limit: 2 })) raws.push(r);
|
|
222
|
+
expect(raws).toHaveLength(2);
|
|
223
|
+
});
|
|
224
|
+
|
|
225
|
+
it("skips unknown kinds (forward-compat with future event types)", async () => {
|
|
226
|
+
const p = writeSnapshot(tmpDir, {
|
|
227
|
+
schemaVersion: 1,
|
|
228
|
+
snapshottedAt: Date.now(),
|
|
229
|
+
events: [
|
|
230
|
+
{ kind: "history", id: "1", title: "ok" },
|
|
231
|
+
{ kind: "fancy-new-kind-from-future", id: "x", data: "?" },
|
|
232
|
+
{ kind: "favourite", id: "f", title: "also ok" },
|
|
233
|
+
],
|
|
234
|
+
});
|
|
235
|
+
const a = new BilibiliAdapter();
|
|
236
|
+
const raws = [];
|
|
237
|
+
for await (const r of a.sync({ inputPath: p })) raws.push(r);
|
|
238
|
+
expect(raws).toHaveLength(2);
|
|
239
|
+
expect(raws.map((r) => r.kind).sort()).toEqual(["favourite", "history"]);
|
|
240
|
+
});
|
|
241
|
+
|
|
242
|
+
it("uses fallback originalId when event.id absent (no crash, still ingestable)", async () => {
|
|
243
|
+
const p = writeSnapshot(tmpDir, {
|
|
244
|
+
schemaVersion: 1,
|
|
245
|
+
snapshottedAt: Date.now(),
|
|
246
|
+
events: [
|
|
247
|
+
// Missing id — adapter should derive from bvid/mid/rid or generate fallback
|
|
248
|
+
{ kind: "history", bvid: "BV1xyz", title: "no-id" },
|
|
249
|
+
{ kind: "follow", mid: 999, uname: "with-mid-no-id" },
|
|
250
|
+
{ kind: "dynamic", summary: "no id no rid" },
|
|
251
|
+
],
|
|
252
|
+
});
|
|
253
|
+
const a = new BilibiliAdapter();
|
|
254
|
+
const raws = [];
|
|
255
|
+
for await (const r of a.sync({ inputPath: p })) raws.push(r);
|
|
256
|
+
expect(raws).toHaveLength(3);
|
|
257
|
+
// history derives from bvid
|
|
258
|
+
expect(raws[0].originalId).toBe("bilibili:history:BV1xyz");
|
|
259
|
+
// follow derives from mid
|
|
260
|
+
expect(raws[1].originalId).toBe("bilibili:follow:999");
|
|
261
|
+
// dynamic with no id/bvid/mid/rid → fallback unknown- prefix
|
|
262
|
+
expect(raws[2].originalId).toMatch(/^bilibili:dynamic:unknown-/);
|
|
263
|
+
});
|
|
264
|
+
|
|
265
|
+
it("snapshot account propagates to payload (Path Y can re-attribute later)", async () => {
|
|
266
|
+
const p = writeSnapshot(tmpDir, {
|
|
267
|
+
schemaVersion: 1,
|
|
268
|
+
snapshottedAt: Date.now(),
|
|
269
|
+
account: { uid: "55555", displayName: "tester" },
|
|
270
|
+
events: [{ kind: "history", id: "1", title: "x" }],
|
|
271
|
+
});
|
|
272
|
+
const a = new BilibiliAdapter();
|
|
273
|
+
const raws = [];
|
|
274
|
+
for await (const r of a.sync({ inputPath: p })) raws.push(r);
|
|
275
|
+
expect(raws).toHaveLength(1);
|
|
276
|
+
expect(raws[0].payload.account.uid).toBe("55555");
|
|
277
|
+
});
|
|
278
|
+
});
|
|
@@ -331,6 +331,29 @@ describe("normalizeWeChatContact", () => {
|
|
|
331
331
|
const b = normalizeWeChatContact({});
|
|
332
332
|
expect(b.persons).toHaveLength(0);
|
|
333
333
|
});
|
|
334
|
+
|
|
335
|
+
// sjqz parity audit follow-up (post-Phase 12.6.10) — classify
|
|
336
|
+
// 公众号 / Official Accounts (gh_*) as merchant subtype so the Ask
|
|
337
|
+
// flow / EntityResolver can filter them out of human contacts.
|
|
338
|
+
it("gh_* username → subtype merchant (公众号 / Official Account)", () => {
|
|
339
|
+
const b = normalizeWeChatContact({
|
|
340
|
+
username: "gh_abc123def",
|
|
341
|
+
nickname: "某品牌官方",
|
|
342
|
+
type: 3,
|
|
343
|
+
});
|
|
344
|
+
expect(b.persons).toHaveLength(1);
|
|
345
|
+
expect(b.persons[0].subtype).toBe("merchant");
|
|
346
|
+
expect(b.persons[0].identifiers.wechatId).toBe("gh_abc123def");
|
|
347
|
+
});
|
|
348
|
+
|
|
349
|
+
it("regular wxid_* → subtype contact (default)", () => {
|
|
350
|
+
const b = normalizeWeChatContact({
|
|
351
|
+
username: "wxid_realfriend",
|
|
352
|
+
nickname: "好友",
|
|
353
|
+
type: 1,
|
|
354
|
+
});
|
|
355
|
+
expect(b.persons[0].subtype).toBe("contact");
|
|
356
|
+
});
|
|
334
357
|
});
|
|
335
358
|
|
|
336
359
|
// ─── WechatAdapter contract + sync flow ──────────────────────────────────
|
|
@@ -441,6 +464,101 @@ describe("WechatAdapter.sync with mocked DB reader", () => {
|
|
|
441
464
|
}
|
|
442
465
|
});
|
|
443
466
|
|
|
467
|
+
// sjqz parity audit follow-up — fetchContacts must exclude
|
|
468
|
+
// @stranger and fake_* by default (vault pollution prevention).
|
|
469
|
+
it("fetchContacts excludes @stranger and fake_* by default", async () => {
|
|
470
|
+
// Pure DI smoke — capture the SQL passed to .prepare() to verify the
|
|
471
|
+
// junk filter is in the query. We mock just enough of better-sqlite3's
|
|
472
|
+
// shape: db.prepare(sql) → { all(limit) → rows }, exec, pragma.
|
|
473
|
+
const seenSql = [];
|
|
474
|
+
const fakeDriver = function Database(_path, _opts) {
|
|
475
|
+
return {
|
|
476
|
+
pragma: () => undefined,
|
|
477
|
+
exec: () => undefined,
|
|
478
|
+
prepare(sql) {
|
|
479
|
+
seenSql.push(sql);
|
|
480
|
+
return {
|
|
481
|
+
all: () => {
|
|
482
|
+
if (sql.startsWith("PRAGMA table_info")) {
|
|
483
|
+
return [
|
|
484
|
+
{ name: "username" },
|
|
485
|
+
{ name: "alias" },
|
|
486
|
+
{ name: "nickname" },
|
|
487
|
+
{ name: "conRemark" },
|
|
488
|
+
{ name: "type" },
|
|
489
|
+
];
|
|
490
|
+
}
|
|
491
|
+
if (sql.startsWith("SELECT count")) return [{ n: 5 }];
|
|
492
|
+
if (sql.includes("FROM rcontact")) return [];
|
|
493
|
+
return [];
|
|
494
|
+
},
|
|
495
|
+
get: () => ({ n: 5 }),
|
|
496
|
+
};
|
|
497
|
+
},
|
|
498
|
+
close: () => undefined,
|
|
499
|
+
};
|
|
500
|
+
};
|
|
501
|
+
const dir = fs.mkdtempSync(path.join(os.tmpdir(), "wechat-junkfilt-"));
|
|
502
|
+
const dbPath = path.join(dir, "EnMicroMsg.db");
|
|
503
|
+
fs.writeFileSync(dbPath, "fake");
|
|
504
|
+
try {
|
|
505
|
+
const reader = new WeChatDBReader({
|
|
506
|
+
dbPath,
|
|
507
|
+
keyProvider: { getKey: async () => "0".repeat(64) },
|
|
508
|
+
driver: fakeDriver,
|
|
509
|
+
});
|
|
510
|
+
await reader.open();
|
|
511
|
+
reader.fetchContacts({ limit: 100 });
|
|
512
|
+
const contactsSql = seenSql.find((s) => s.includes("FROM rcontact"));
|
|
513
|
+
expect(contactsSql).toBeDefined();
|
|
514
|
+
expect(contactsSql).toMatch(/NOT LIKE '%@stranger'/);
|
|
515
|
+
expect(contactsSql).toMatch(/NOT LIKE 'fake_%'/);
|
|
516
|
+
} finally {
|
|
517
|
+
fs.rmSync(dir, { recursive: true, force: true });
|
|
518
|
+
}
|
|
519
|
+
});
|
|
520
|
+
|
|
521
|
+
it("fetchContacts with includeJunk:true drops the filter (forensic mode)", async () => {
|
|
522
|
+
const seenSql = [];
|
|
523
|
+
const fakeDriver = function Database() {
|
|
524
|
+
return {
|
|
525
|
+
pragma: () => undefined,
|
|
526
|
+
exec: () => undefined,
|
|
527
|
+
prepare(sql) {
|
|
528
|
+
seenSql.push(sql);
|
|
529
|
+
return {
|
|
530
|
+
all: () => {
|
|
531
|
+
if (sql.startsWith("PRAGMA table_info")) {
|
|
532
|
+
return ["username", "alias", "nickname", "conRemark", "type"].map((name) => ({ name }));
|
|
533
|
+
}
|
|
534
|
+
if (sql.startsWith("SELECT count")) return [{ n: 5 }];
|
|
535
|
+
return [];
|
|
536
|
+
},
|
|
537
|
+
get: () => ({ n: 5 }),
|
|
538
|
+
};
|
|
539
|
+
},
|
|
540
|
+
close: () => undefined,
|
|
541
|
+
};
|
|
542
|
+
};
|
|
543
|
+
const dir = fs.mkdtempSync(path.join(os.tmpdir(), "wechat-incljunk-"));
|
|
544
|
+
const dbPath = path.join(dir, "EnMicroMsg.db");
|
|
545
|
+
fs.writeFileSync(dbPath, "fake");
|
|
546
|
+
try {
|
|
547
|
+
const reader = new WeChatDBReader({
|
|
548
|
+
dbPath,
|
|
549
|
+
keyProvider: { getKey: async () => "0".repeat(64) },
|
|
550
|
+
driver: fakeDriver,
|
|
551
|
+
});
|
|
552
|
+
await reader.open();
|
|
553
|
+
reader.fetchContacts({ limit: 100, includeJunk: true });
|
|
554
|
+
const contactsSql = seenSql.find((s) => s.includes("FROM rcontact"));
|
|
555
|
+
expect(contactsSql).toBeDefined();
|
|
556
|
+
expect(contactsSql).not.toMatch(/NOT LIKE/);
|
|
557
|
+
} finally {
|
|
558
|
+
fs.rmSync(dir, { recursive: true, force: true });
|
|
559
|
+
}
|
|
560
|
+
});
|
|
561
|
+
|
|
444
562
|
it("idle no-op when DB path missing", async () => {
|
|
445
563
|
const a = new WechatAdapter({
|
|
446
564
|
account: { uin: "self123" },
|
|
@@ -45,6 +45,7 @@ const { SPEC: hunyuanSpec } = require("./vendors/hunyuan");
|
|
|
45
45
|
const { SPEC: qianfanSpec } = require("./vendors/qianfan");
|
|
46
46
|
const { SPEC: cozeSpec } = require("./vendors/coze");
|
|
47
47
|
const { SPEC: dreaminaSpec } = require("./vendors/dreamina");
|
|
48
|
+
const { SPEC: doubaoSpec } = require("./vendors/doubao");
|
|
48
49
|
|
|
49
50
|
const DEFAULT_VENDOR_SPECS = Object.freeze({
|
|
50
51
|
deepseek: deepseekSpec,
|
|
@@ -55,6 +56,7 @@ const DEFAULT_VENDOR_SPECS = Object.freeze({
|
|
|
55
56
|
qianfan: qianfanSpec,
|
|
56
57
|
coze: cozeSpec,
|
|
57
58
|
dreamina: dreaminaSpec,
|
|
59
|
+
doubao: doubaoSpec,
|
|
58
60
|
});
|
|
59
61
|
|
|
60
62
|
class AIChatHistoryAdapter {
|
|
@@ -186,13 +188,19 @@ class AIChatHistoryAdapter {
|
|
|
186
188
|
/**
|
|
187
189
|
* Stream conversation + message envelopes across all configured vendors.
|
|
188
190
|
*
|
|
189
|
-
* Yields
|
|
190
|
-
* {
|
|
191
|
-
* { kind: "message", vendor, message: RawMessage }
|
|
191
|
+
* Yields AdapterRegistry-compliant envelopes:
|
|
192
|
+
* { originalId, capturedAt, payload: { kind, vendor, conversation|message } }
|
|
192
193
|
*
|
|
193
|
-
* The
|
|
194
|
-
*
|
|
195
|
-
*
|
|
194
|
+
* The inner `payload.kind` distinguishes:
|
|
195
|
+
* - "conversation" → emit Topic + vendor Person (no Event yet)
|
|
196
|
+
* - "message" → emit Event + items + vendor Person
|
|
197
|
+
* - "vendor-not-wired" → no-op normalize (Phase 10.1 stub trace)
|
|
198
|
+
* - "vendor-cookie-expired" → no-op normalize (401/403 trace)
|
|
199
|
+
* - "vendor-rate-limited" → no-op normalize (429 trace after retries)
|
|
200
|
+
*
|
|
201
|
+
* The registry calls `normalize(raw)` per yielded envelope. One yield per
|
|
202
|
+
* conversation/message keeps registry batches small so a slow vendor
|
|
203
|
+
* doesn't block faster ones at the registry boundary.
|
|
196
204
|
*
|
|
197
205
|
* @param {object} [opts]
|
|
198
206
|
* @param {string[]} [opts.vendors] restrict to a subset
|
|
@@ -216,28 +224,54 @@ class AIChatHistoryAdapter {
|
|
|
216
224
|
|
|
217
225
|
try {
|
|
218
226
|
for await (const conv of spec.listConversations(ctx, { since: vendorWatermark })) {
|
|
219
|
-
yield {
|
|
227
|
+
yield {
|
|
228
|
+
originalId: `${vendor}:conv:${conv.originalId}`,
|
|
229
|
+
capturedAt: Number(conv.updatedAt) || Number(conv.createdAt) || Date.now(),
|
|
230
|
+
payload: { kind: "conversation", vendor, conversation: conv },
|
|
231
|
+
};
|
|
220
232
|
|
|
221
233
|
for await (const msg of spec.listMessages(ctx, conv.originalId, {})) {
|
|
222
|
-
yield {
|
|
234
|
+
yield {
|
|
235
|
+
originalId: `${vendor}:msg:${msg.originalId}`,
|
|
236
|
+
capturedAt: Number(msg.createdAt) || Date.now(),
|
|
237
|
+
payload: { kind: "message", vendor, message: msg },
|
|
238
|
+
};
|
|
223
239
|
}
|
|
224
240
|
}
|
|
225
241
|
} catch (err) {
|
|
242
|
+
const traceCapturedAt = Date.now();
|
|
226
243
|
if (err instanceof NotImplementedYetError) {
|
|
227
244
|
this._logger.warn(
|
|
228
245
|
`[ai-chat] vendor=${vendor} not wired (Phase 10.2+ work): ${err.message}`,
|
|
229
246
|
);
|
|
230
|
-
yield {
|
|
247
|
+
yield {
|
|
248
|
+
originalId: `${vendor}:trace:not-wired:${traceCapturedAt}`,
|
|
249
|
+
capturedAt: traceCapturedAt,
|
|
250
|
+
payload: { kind: "vendor-not-wired", vendor, error: err.code },
|
|
251
|
+
};
|
|
231
252
|
continue;
|
|
232
253
|
}
|
|
233
254
|
if (err instanceof CookieExpiredError) {
|
|
234
255
|
this._logger.warn(`[ai-chat] vendor=${vendor} cookie expired: ${err.message}`);
|
|
235
|
-
yield {
|
|
256
|
+
yield {
|
|
257
|
+
originalId: `${vendor}:trace:cookie-expired:${traceCapturedAt}`,
|
|
258
|
+
capturedAt: traceCapturedAt,
|
|
259
|
+
payload: { kind: "vendor-cookie-expired", vendor, error: err.code },
|
|
260
|
+
};
|
|
236
261
|
continue;
|
|
237
262
|
}
|
|
238
263
|
if (err instanceof RateLimitedError) {
|
|
239
264
|
this._logger.warn(`[ai-chat] vendor=${vendor} rate limited: ${err.message}`);
|
|
240
|
-
yield {
|
|
265
|
+
yield {
|
|
266
|
+
originalId: `${vendor}:trace:rate-limited:${traceCapturedAt}`,
|
|
267
|
+
capturedAt: traceCapturedAt,
|
|
268
|
+
payload: {
|
|
269
|
+
kind: "vendor-rate-limited",
|
|
270
|
+
vendor,
|
|
271
|
+
error: err.code,
|
|
272
|
+
retryAfterMs: err.retryAfterMs,
|
|
273
|
+
},
|
|
274
|
+
};
|
|
241
275
|
continue;
|
|
242
276
|
}
|
|
243
277
|
throw err;
|
|
@@ -256,14 +290,19 @@ class AIChatHistoryAdapter {
|
|
|
256
290
|
if (!raw || typeof raw !== "object") {
|
|
257
291
|
return { events: [], persons: [], places: [], items: [], topics: [] };
|
|
258
292
|
}
|
|
293
|
+
// Registry-compliant envelopes wrap kind inside payload. Adapter-internal
|
|
294
|
+
// tests (Phase 10.1) sometimes pass the inner shape directly — accept
|
|
295
|
+
// both for forward compat.
|
|
296
|
+
const inner = raw.payload && typeof raw.payload === "object" ? raw.payload : raw;
|
|
297
|
+
const kind = inner.kind;
|
|
259
298
|
|
|
260
|
-
if (
|
|
299
|
+
if (kind === "vendor-not-wired" || kind === "vendor-cookie-expired" || kind === "vendor-rate-limited") {
|
|
261
300
|
// Nothing to write; the warning was already logged by sync().
|
|
262
301
|
return { events: [], persons: [], places: [], items: [], topics: [] };
|
|
263
302
|
}
|
|
264
303
|
|
|
265
|
-
if (
|
|
266
|
-
const conv =
|
|
304
|
+
if (kind === "conversation") {
|
|
305
|
+
const conv = inner.conversation;
|
|
267
306
|
const spec = this._vendorSpecs[conv.vendor];
|
|
268
307
|
const displayName = spec ? spec.displayName : conv.vendor;
|
|
269
308
|
return {
|
|
@@ -275,8 +314,8 @@ class AIChatHistoryAdapter {
|
|
|
275
314
|
};
|
|
276
315
|
}
|
|
277
316
|
|
|
278
|
-
if (
|
|
279
|
-
const msg =
|
|
317
|
+
if (kind === "message") {
|
|
318
|
+
const msg = inner.message;
|
|
280
319
|
const spec = this._vendorSpecs[msg.vendor];
|
|
281
320
|
const displayName = spec ? spec.displayName : msg.vendor;
|
|
282
321
|
return {
|