@chainlesschain/personal-data-hub 0.4.28 → 0.4.30

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.
Files changed (204) hide show
  1. package/README.md +13 -5
  2. package/lib/adapters/social-douyin-adb/usage-profile-reader.js +253 -0
  3. package/lib/adapters/social-douyin-adb/watch-history-reader.js +104 -31
  4. package/lib/adapters/social-toutiao-adb/article-reader.js +202 -0
  5. package/lib/analysis-skills/overview.js +24 -4
  6. package/lib/analysis-skills/spending.js +63 -2
  7. package/lib/analysis-skills/timeline.js +11 -6
  8. package/lib/prompt-builder.js +15 -1
  9. package/lib/query-parser.js +38 -8
  10. package/package.json +4 -1
  11. package/__tests__/adapter-guide.test.js +0 -47
  12. package/__tests__/adapter-spec.test.js +0 -78
  13. package/__tests__/adapters/ai-chat-cookie-capture-spec.test.js +0 -211
  14. package/__tests__/adapters/ai-chat-health-checker.test.js +0 -262
  15. package/__tests__/adapters/ai-chat-history.test.js +0 -396
  16. package/__tests__/adapters/ai-chat-http-client.test.js +0 -242
  17. package/__tests__/adapters/ai-chat-vendors.test.js +0 -874
  18. package/__tests__/adapters/alipay-bill-adapter.test.js +0 -538
  19. package/__tests__/adapters/apple-health.test.js +0 -95
  20. package/__tests__/adapters/bank-family.test.js +0 -125
  21. package/__tests__/adapters/biz-tianyancha.test.js +0 -159
  22. package/__tests__/adapters/browser-history-chrome.test.js +0 -377
  23. package/__tests__/adapters/browser-history-edge.test.js +0 -159
  24. package/__tests__/adapters/car-mercedesme.test.js +0 -74
  25. package/__tests__/adapters/doc-baidu-netdisk.test.js +0 -102
  26. package/__tests__/adapters/doc-camscanner.test.js +0 -147
  27. package/__tests__/adapters/doc-platforms.test.js +0 -177
  28. package/__tests__/adapters/edu-huawei-learning-live.test.js +0 -198
  29. package/__tests__/adapters/edu-zuoyebang-live.test.js +0 -226
  30. package/__tests__/adapters/email-adapter-snapshot.test.js +0 -237
  31. package/__tests__/adapters/email-adapter.test.js +0 -742
  32. package/__tests__/adapters/email-classifier.test.js +0 -347
  33. package/__tests__/adapters/email-imap-session.test.js +0 -334
  34. package/__tests__/adapters/email-parser.test.js +0 -244
  35. package/__tests__/adapters/email-pdf-extractor.test.js +0 -529
  36. package/__tests__/adapters/email-providers.test.js +0 -84
  37. package/__tests__/adapters/email-retry-progress.test.js +0 -294
  38. package/__tests__/adapters/email-templates.test.js +0 -822
  39. package/__tests__/adapters/family-23-collectors-scaffold.test.js +0 -182
  40. package/__tests__/adapters/finance-alipay-live.test.js +0 -258
  41. package/__tests__/adapters/finance-dcep.test.js +0 -74
  42. package/__tests__/adapters/fitness-joyrun.test.js +0 -82
  43. package/__tests__/adapters/game-genshin-live.test.js +0 -238
  44. package/__tests__/adapters/game-genshin-scaffold.test.js +0 -108
  45. package/__tests__/adapters/game-honor-of-kings-live.test.js +0 -230
  46. package/__tests__/adapters/git-activity.test.js +0 -222
  47. package/__tests__/adapters/gov-12123.test.js +0 -103
  48. package/__tests__/adapters/gov-ixiamen.test.js +0 -150
  49. package/__tests__/adapters/gov-tax.test.js +0 -135
  50. package/__tests__/adapters/health-meiyou.test.js +0 -125
  51. package/__tests__/adapters/local-files.test.js +0 -264
  52. package/__tests__/adapters/local-im-pc.test.js +0 -154
  53. package/__tests__/adapters/messaging-whatsapp.test.js +0 -289
  54. package/__tests__/adapters/music-kugou.test.js +0 -187
  55. package/__tests__/adapters/music-qq.test.js +0 -112
  56. package/__tests__/adapters/netease-music-live.test.js +0 -244
  57. package/__tests__/adapters/netease-music.test.js +0 -74
  58. package/__tests__/adapters/pc-local-discovery.test.js +0 -141
  59. package/__tests__/adapters/qq-pc-direct-read.test.js +0 -227
  60. package/__tests__/adapters/reading-family.test.js +0 -108
  61. package/__tests__/adapters/recruit-boss.test.js +0 -180
  62. package/__tests__/adapters/shell-history.test.js +0 -180
  63. package/__tests__/adapters/shopping-base.test.js +0 -179
  64. package/__tests__/adapters/shopping-dianping.test.js +0 -239
  65. package/__tests__/adapters/social-bilibili-adb-api-client.test.js +0 -721
  66. package/__tests__/adapters/social-bilibili-adb-chromium-cookies-reader.test.js +0 -346
  67. package/__tests__/adapters/social-bilibili-adb-collector.test.js +0 -284
  68. package/__tests__/adapters/social-bilibili-adb-cookies-extension.test.js +0 -343
  69. package/__tests__/adapters/social-bilibili-adb-snapshot-builder.test.js +0 -296
  70. package/__tests__/adapters/social-csdn.test.js +0 -175
  71. package/__tests__/adapters/social-dongchedi.test.js +0 -165
  72. package/__tests__/adapters/social-douyin-adb-aweme-detail.test.js +0 -165
  73. package/__tests__/adapters/social-douyin-adb-collector.test.js +0 -254
  74. package/__tests__/adapters/social-douyin-adb-db-extension.test.js +0 -114
  75. package/__tests__/adapters/social-douyin-adb-im-db-parser.test.js +0 -304
  76. package/__tests__/adapters/social-douyin-adb-snapshot-builder.test.js +0 -216
  77. package/__tests__/adapters/social-douyin-adb-watch-history.test.js +0 -192
  78. package/__tests__/adapters/social-kuaishou-adb-api-client.test.js +0 -496
  79. package/__tests__/adapters/social-kuaishou-adb-collector.test.js +0 -276
  80. package/__tests__/adapters/social-kuaishou-adb-cookies-extension.test.js +0 -152
  81. package/__tests__/adapters/social-kuaishou-adb-snapshot-builder.test.js +0 -178
  82. package/__tests__/adapters/social-toutiao-adb-account-reader.test.js +0 -135
  83. package/__tests__/adapters/social-toutiao-adb-api-client.test.js +0 -626
  84. package/__tests__/adapters/social-toutiao-adb-collector.test.js +0 -378
  85. package/__tests__/adapters/social-toutiao-adb-cookies-extension.test.js +0 -193
  86. package/__tests__/adapters/social-toutiao-adb-snapshot-builder.test.js +0 -196
  87. package/__tests__/adapters/social-toutiao-kuaishou-scaffold.test.js +0 -311
  88. package/__tests__/adapters/social-weibo-adb-api-client.test.js +0 -362
  89. package/__tests__/adapters/social-weibo-adb-collector.test.js +0 -201
  90. package/__tests__/adapters/social-weibo-adb-cookies-extension.test.js +0 -167
  91. package/__tests__/adapters/social-weibo-adb-snapshot-builder.test.js +0 -189
  92. package/__tests__/adapters/social-xiaohongshu-adb-api-client.test.js +0 -431
  93. package/__tests__/adapters/social-xiaohongshu-adb-collector.test.js +0 -207
  94. package/__tests__/adapters/social-xiaohongshu-adb-cookies-extension.test.js +0 -0
  95. package/__tests__/adapters/social-xiaohongshu-adb-sign-provider-injection.test.js +0 -351
  96. package/__tests__/adapters/social-xiaohongshu-adb-sign.test.js +0 -130
  97. package/__tests__/adapters/social-xiaohongshu-adb-snapshot-builder.test.js +0 -200
  98. package/__tests__/adapters/social-zhihu.test.js +0 -246
  99. package/__tests__/adapters/system-data-adapter.test.js +0 -443
  100. package/__tests__/adapters/system-data-android-ingest.test.js +0 -144
  101. package/__tests__/adapters/system-data-android.test.js +0 -519
  102. package/__tests__/adapters/system-data-disclosure.test.js +0 -153
  103. package/__tests__/adapters/travel-12306.test.js +0 -512
  104. package/__tests__/adapters/travel-amap.test.js +0 -219
  105. package/__tests__/adapters/travel-baidu-map.test.js +0 -305
  106. package/__tests__/adapters/travel-base.test.js +0 -205
  107. package/__tests__/adapters/travel-ctrip.test.js +0 -377
  108. package/__tests__/adapters/travel-didi-consumer.test.js +0 -66
  109. package/__tests__/adapters/travel-didi.test.js +0 -204
  110. package/__tests__/adapters/travel-tencent-map.test.js +0 -207
  111. package/__tests__/adapters/travel-tongcheng.test.js +0 -289
  112. package/__tests__/adapters/video-platforms.test.js +0 -152
  113. package/__tests__/adapters/video-xigua.test.js +0 -106
  114. package/__tests__/adapters/vscode.test.js +0 -299
  115. package/__tests__/adapters/wechat-bootstrap.test.js +0 -240
  116. package/__tests__/adapters/wechat-env-probe.test.js +0 -162
  117. package/__tests__/adapters/wechat-frida-agent.test.js +0 -322
  118. package/__tests__/adapters/wechat-frida-integration.test.js +0 -149
  119. package/__tests__/adapters/wechat-frida-key-provider.test.js +0 -188
  120. package/__tests__/adapters/wechat-md5-key-provider.test.js +0 -101
  121. package/__tests__/adapters/wechat-pc-direct-read.test.js +0 -365
  122. package/__tests__/adapters/wechat-pc-group-topic.test.js +0 -63
  123. package/__tests__/adapters/wechat-pc-v4-sidecar.test.js +0 -72
  124. package/__tests__/adapters/weread.test.js +0 -123
  125. package/__tests__/adapters/wework-pc.test.js +0 -124
  126. package/__tests__/adapters/win-recent.test.js +0 -192
  127. package/__tests__/analysis-skills.test.js +0 -679
  128. package/__tests__/analysis.test.js +0 -1845
  129. package/__tests__/audio-ximalaya-snapshot.test.js +0 -279
  130. package/__tests__/batch.test.js +0 -133
  131. package/__tests__/bridges-cc-kg.test.js +0 -231
  132. package/__tests__/bridges-cc-llm.test.js +0 -191
  133. package/__tests__/bridges-cc-rag.test.js +0 -162
  134. package/__tests__/categories.test.js +0 -92
  135. package/__tests__/e2e/ai-chat-cross-source-journey.test.js +0 -213
  136. package/__tests__/e2e/full-user-journey.test.js +0 -188
  137. package/__tests__/e2e/local-data-adapters-cli.e2e.test.js +0 -146
  138. package/__tests__/entity-resolver-ingest-hook.test.js +0 -177
  139. package/__tests__/entity-resolver-stages.test.js +0 -411
  140. package/__tests__/entity-resolver-vault.test.js +0 -249
  141. package/__tests__/entity-resolver.test.js +0 -526
  142. package/__tests__/fitness-keep-snapshot.test.js +0 -224
  143. package/__tests__/fixtures/entity-resolver-200-mock.json +0 -96
  144. package/__tests__/ids.test.js +0 -45
  145. package/__tests__/integration/ai-chat-history-registry.test.js +0 -228
  146. package/__tests__/integration/aichat-wizard-end-to-end.test.js +0 -282
  147. package/__tests__/integration/cross-adapter-pipelines.test.js +0 -396
  148. package/__tests__/integration/local-data-adapters-pipeline.test.js +0 -373
  149. package/__tests__/integration/social-bilibili-pipeline.test.js +0 -261
  150. package/__tests__/integration/wechat-bootstrap-end-to-end.test.js +0 -390
  151. package/__tests__/key-providers.test.js +0 -126
  152. package/__tests__/kg-derive.test.js +0 -219
  153. package/__tests__/llm-client.test.js +0 -122
  154. package/__tests__/longtail-adapters.test.js +0 -281
  155. package/__tests__/messaging-qq-snapshot.test.js +0 -294
  156. package/__tests__/mobile-extractor-encrypted.test.js +0 -460
  157. package/__tests__/mobile-extractor.test.js +0 -288
  158. package/__tests__/mock-adapter.test.js +0 -93
  159. package/__tests__/prompt-builder.test.js +0 -249
  160. package/__tests__/query-parser.test.js +0 -302
  161. package/__tests__/rag-derive.test.js +0 -169
  162. package/__tests__/registry-readiness.test.js +0 -292
  163. package/__tests__/registry.test.js +0 -420
  164. package/__tests__/salvage-ingest.test.js +0 -97
  165. package/__tests__/schemas.test.js +0 -331
  166. package/__tests__/shopping-adapters.test.js +0 -392
  167. package/__tests__/shopping-eleme-snapshot.test.js +0 -454
  168. package/__tests__/shopping-pinduoduo-snapshot.test.js +0 -484
  169. package/__tests__/shopping-snapshot.test.js +0 -438
  170. package/__tests__/shopping-vipshop-snapshot.test.js +0 -425
  171. package/__tests__/shopping-xianyu-snapshot.test.js +0 -451
  172. package/__tests__/sidecar-contacts-cross-validate.test.js +0 -186
  173. package/__tests__/sidecar-supervisor.test.js +0 -128
  174. package/__tests__/sign-providers.test.js +0 -62
  175. package/__tests__/social-adapters.test.js +0 -280
  176. package/__tests__/social-bilibili-snapshot.test.js +0 -278
  177. package/__tests__/social-douban-snapshot.test.js +0 -351
  178. package/__tests__/social-douyin-im-direct-read.test.js +0 -377
  179. package/__tests__/social-douyin-salvage-collector.test.js +0 -98
  180. package/__tests__/social-douyin-salvage-mapper.test.js +0 -90
  181. package/__tests__/social-douyin-snapshot.test.js +0 -256
  182. package/__tests__/social-kuaishou-snapshot.test.js +0 -362
  183. package/__tests__/social-toutiao-snapshot.test.js +0 -366
  184. package/__tests__/social-weibo-snapshot.test.js +0 -234
  185. package/__tests__/social-weibo-sqlite-device.test.js +0 -174
  186. package/__tests__/social-xiaohongshu-snapshot.test.js +0 -232
  187. package/__tests__/sqlite-leaf-salvage.test.js +0 -97
  188. package/__tests__/travel-adapters.test.js +0 -483
  189. package/__tests__/travel-maps-snapshot.test.js +0 -426
  190. package/__tests__/vault-driver-error.test.js +0 -74
  191. package/__tests__/vault-search-helpers.test.js +0 -104
  192. package/__tests__/vault-search.test.js +0 -423
  193. package/__tests__/vault.test.js +0 -767
  194. package/__tests__/wechat-adapter.test.js +0 -594
  195. package/__tests__/whatsapp-adapter.test.js +0 -138
  196. package/scripts/_make-fixture-all.js +0 -126
  197. package/scripts/_make-fixture-contacts.js +0 -84
  198. package/scripts/evaluate-entity-resolver.js +0 -213
  199. package/scripts/run-native-tests-sandbox.sh +0 -55
  200. package/scripts/smoke-phase-5-5.js +0 -196
  201. package/scripts/smoke-phase-5-7.js +0 -181
  202. package/scripts/smoke-system-data-contacts.js +0 -309
  203. package/scripts/smoke-system-data.js +0 -312
  204. package/vitest.config.js +0 -88
@@ -1,742 +0,0 @@
1
- "use strict";
2
-
3
- import { describe, it, expect } from "vitest";
4
-
5
- const {
6
- EmailAdapter,
7
- parseWatermark,
8
- formatWatermark,
9
- } = require("../../lib/adapters/email-imap/email-adapter");
10
- const { assertAdapter } = require("../../lib/adapter-spec");
11
- const { validateBatch } = require("../../lib/batch");
12
-
13
- function makeMockSession(spec = {}) {
14
- const recorder = {
15
- constructorArgs: null,
16
- connectCalls: 0,
17
- closedCalls: 0,
18
- openedMailboxes: [],
19
- fetchRanges: [],
20
- };
21
- const factory = (opts) => {
22
- recorder.constructorArgs = opts;
23
- let openMb = null;
24
- return {
25
- async connect() {
26
- recorder.connectCalls += 1;
27
- if (spec.connectThrows) throw spec.connectThrows;
28
- },
29
- async openMailbox(name) {
30
- recorder.openedMailboxes.push(name);
31
- const mb = spec.mailboxes && spec.mailboxes[name];
32
- if (!mb) {
33
- const err = new Error(`Mailbox doesn't exist: ${name}`);
34
- err.code = "MAILBOX_NOT_FOUND";
35
- throw err;
36
- }
37
- openMb = { name, ...mb };
38
- return {
39
- uidValidity: mb.uidValidity,
40
- uidNext: mb.uidNext || 9999,
41
- exists: (mb.envelopes || []).length,
42
- };
43
- },
44
- async *fetchEnvelopesSince(sinceUid = 0) {
45
- recorder.fetchRanges.push({ mailbox: openMb && openMb.name, sinceUid, mode: "envelope" });
46
- if (!openMb) return;
47
- for (const env of openMb.envelopes || []) {
48
- if (env.uid > sinceUid) yield env;
49
- }
50
- },
51
- // Phase 5.2 — adapter.sync now calls fetchFullSince. The mock
52
- // emits the same envelopes with an empty source Buffer; the
53
- // adapter parses it via its injected parser (or skips parsing
54
- // when source is empty).
55
- async *fetchFullSince(sinceUid = 0) {
56
- recorder.fetchRanges.push({ mailbox: openMb && openMb.name, sinceUid, mode: "full" });
57
- if (!openMb) return;
58
- for (const env of openMb.envelopes || []) {
59
- if (env.uid > sinceUid) {
60
- yield { ...env, source: env.source || Buffer.alloc(0) };
61
- }
62
- }
63
- },
64
- async close() {
65
- recorder.closedCalls += 1;
66
- },
67
- };
68
- };
69
- return { factory, recorder };
70
- }
71
-
72
- const env = (uid, overrides = {}) => ({
73
- uid,
74
- internalDate: new Date(`2026-04-${String(uid % 30).padStart(2, "0")}T10:00:00Z`),
75
- flags: ["\\Seen"],
76
- messageId: `<msg-${uid}@example.com>`,
77
- subject: `Subject ${uid}`,
78
- from: [{ name: "Alice", address: `alice${uid}@example.com` }],
79
- to: [{ name: "Me", address: "me@example.com" }],
80
- cc: [],
81
- date: new Date(`2026-04-${String(uid % 30).padStart(2, "0")}T10:00:00Z`),
82
- size: 1024,
83
- ...overrides,
84
- });
85
-
86
- describe("EmailAdapter contract", () => {
87
- it("conforms to PersonalDataAdapter spec", () => {
88
- const a = new EmailAdapter({
89
- account: { provider: "qq", email: "u@qq.com", authCode: "abc123" },
90
- sessionFactory: makeMockSession({}).factory,
91
- });
92
- const r = assertAdapter(a);
93
- expect(r.ok).toBe(true);
94
- if (!r.ok) console.log(r.errors);
95
- });
96
-
97
- it("exposes the canonical name + version + capabilities", () => {
98
- const a = new EmailAdapter({
99
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
100
- sessionFactory: makeMockSession({}).factory,
101
- });
102
- expect(a.name).toBe("email-imap");
103
- expect(a.version).toBe("0.7.0"); // Phase 5.8 — snapshot mode for Android in-APK IMAP fetch
104
- expect(a.capabilities).toContain("sync:imap");
105
- expect(a.capabilities).toContain("auth:authcode");
106
- expect(a.capabilities).toContain("parse:mime-body");
107
- expect(a.capabilities).toContain("classify:layer1-rules");
108
- expect(a.dataDisclosure.sensitivity).toBe("high");
109
- });
110
-
111
- it("rejects missing or malformed account", () => {
112
- expect(() => new EmailAdapter()).toThrow();
113
- expect(() => new EmailAdapter({})).toThrow(/account/);
114
- expect(() => new EmailAdapter({ account: {} })).toThrow(/email/);
115
- expect(() => new EmailAdapter({ account: { email: "noatsign" } })).toThrow(/email/);
116
- expect(() => new EmailAdapter({ account: { email: "u@x.com" } })).toThrow(/authCode/);
117
- });
118
- });
119
-
120
- describe("EmailAdapter.authenticate", () => {
121
- it("returns ok:true when connect succeeds", async () => {
122
- const { factory, recorder } = makeMockSession({});
123
- const a = new EmailAdapter({
124
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
125
- sessionFactory: factory,
126
- });
127
- const r = await a.authenticate();
128
- expect(r.ok).toBe(true);
129
- expect(r.account).toBe("u@qq.com");
130
- expect(r.provider).toBe("qq");
131
- expect(recorder.connectCalls).toBe(1);
132
- expect(recorder.closedCalls).toBe(1);
133
- });
134
-
135
- it("returns ok:false reason=AUTH_FAILED on credential error", async () => {
136
- const { ImapAuthFailedError } = require("../../lib/adapters/email-imap/imap-session");
137
- const { factory } = makeMockSession({
138
- connectThrows: new ImapAuthFailedError("bad pass"),
139
- });
140
- const a = new EmailAdapter({
141
- account: { provider: "qq", email: "u@qq.com", authCode: "wrong" },
142
- sessionFactory: factory,
143
- });
144
- const r = await a.authenticate();
145
- expect(r.ok).toBe(false);
146
- expect(r.reason).toBe("AUTH_FAILED");
147
- });
148
-
149
- it("returns ok:false reason=CONNECTION_FAILED on network error", async () => {
150
- const { ImapConnectionFailedError } = require("../../lib/adapters/email-imap/imap-session");
151
- const { factory } = makeMockSession({
152
- connectThrows: new ImapConnectionFailedError("ECONNREFUSED"),
153
- });
154
- const a = new EmailAdapter({
155
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
156
- sessionFactory: factory,
157
- });
158
- const r = await a.authenticate();
159
- expect(r.ok).toBe(false);
160
- expect(r.reason).toBe("CONNECTION_FAILED");
161
- });
162
- });
163
-
164
- describe("EmailAdapter.healthCheck", () => {
165
- it("matches the registry's expected shape on success", async () => {
166
- const { factory } = makeMockSession({});
167
- const a = new EmailAdapter({
168
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
169
- sessionFactory: factory,
170
- });
171
- const r = await a.healthCheck();
172
- expect(r.ok).toBe(true);
173
- expect(r.lastChecked).toBeGreaterThan(0);
174
- });
175
-
176
- it("matches the registry's expected shape on failure", async () => {
177
- const { factory } = makeMockSession({
178
- connectThrows: new Error("Authentication invalid"),
179
- });
180
- const a = new EmailAdapter({
181
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
182
- sessionFactory: factory,
183
- });
184
- const r = await a.healthCheck();
185
- expect(r.ok).toBe(false);
186
- expect(r.reason).toBeDefined();
187
- });
188
- });
189
-
190
- describe("EmailAdapter.sync", () => {
191
- it("yields one RawEvent per envelope across all default folders", async () => {
192
- const { factory, recorder } = makeMockSession({
193
- mailboxes: {
194
- INBOX: { uidValidity: 1, envelopes: [env(1), env(2), env(3)] },
195
- "Sent Messages": { uidValidity: 1, envelopes: [env(10), env(11)] },
196
- },
197
- });
198
- const a = new EmailAdapter({
199
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
200
- sessionFactory: factory,
201
- });
202
- const raws = [];
203
- for await (const r of a.sync()) raws.push(r);
204
- expect(raws).toHaveLength(5);
205
- expect(raws.every((r) => r.adapter === "email-imap")).toBe(true);
206
- expect(raws.every((r) => r.payload.uid > 0)).toBe(true);
207
- expect(recorder.openedMailboxes).toEqual(["INBOX", "Sent Messages"]);
208
- });
209
-
210
- it("originalId uses Message-ID when present", async () => {
211
- const { factory } = makeMockSession({
212
- mailboxes: { INBOX: { uidValidity: 1, envelopes: [env(1)] } },
213
- });
214
- const a = new EmailAdapter({
215
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
216
- sessionFactory: factory,
217
- });
218
- const raws = [];
219
- for await (const r of a.sync()) raws.push(r);
220
- expect(raws[0].originalId).toBe("<msg-1@example.com>");
221
- });
222
-
223
- it("falls back to synthetic originalId when Message-ID missing", async () => {
224
- const { factory } = makeMockSession({
225
- mailboxes: {
226
- INBOX: { uidValidity: 1, envelopes: [env(7, { messageId: "" })] },
227
- },
228
- });
229
- const a = new EmailAdapter({
230
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
231
- sessionFactory: factory,
232
- });
233
- const raws = [];
234
- for await (const r of a.sync()) raws.push(r);
235
- expect(raws[0].originalId).toContain("mid-fallback");
236
- expect(raws[0].originalId).toContain("INBOX");
237
- expect(raws[0].originalId).toContain(":7");
238
- });
239
-
240
- it("respects sinceWatermark (UID > lastUid)", async () => {
241
- const { factory } = makeMockSession({
242
- mailboxes: {
243
- INBOX: { uidValidity: 1, envelopes: [env(1), env(2), env(3), env(4)] },
244
- },
245
- });
246
- const a = new EmailAdapter({
247
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
248
- sessionFactory: factory,
249
- });
250
- const raws = [];
251
- for await (const r of a.sync({ sinceWatermark: "1:2" })) raws.push(r);
252
- expect(raws.map((r) => r.payload.uid)).toEqual([3, 4]);
253
- });
254
-
255
- it("resets sinceUid to 0 when UIDVALIDITY changed (full re-scan)", async () => {
256
- const { factory } = makeMockSession({
257
- mailboxes: {
258
- INBOX: { uidValidity: 99, envelopes: [env(1), env(2)] },
259
- },
260
- });
261
- const a = new EmailAdapter({
262
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
263
- sessionFactory: factory,
264
- });
265
- const raws = [];
266
- for await (const r of a.sync({ sinceWatermark: "42:100" })) raws.push(r);
267
- expect(raws).toHaveLength(2);
268
- expect(raws.map((r) => r.payload.uid).sort()).toEqual([1, 2]);
269
- });
270
-
271
- it("respects maxPerFolder cap", async () => {
272
- const big = Array.from({ length: 50 }, (_, i) => env(i + 1));
273
- const { factory } = makeMockSession({
274
- mailboxes: { INBOX: { uidValidity: 1, envelopes: big } },
275
- });
276
- const a = new EmailAdapter({
277
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
278
- sessionFactory: factory,
279
- });
280
- const raws = [];
281
- for await (const r of a.sync({ maxPerFolder: 7 })) raws.push(r);
282
- expect(raws).toHaveLength(7);
283
- });
284
-
285
- it("closes session even when sync throws mid-stream", async () => {
286
- let closed = false;
287
- const exploding = {
288
- async connect() {},
289
- async openMailbox() {
290
- return { uidValidity: 1, uidNext: 100, exists: 0 };
291
- },
292
- async *fetchFullSince() {
293
- yield { ...env(1), source: Buffer.alloc(0) };
294
- throw new Error("network drop");
295
- },
296
- async close() {
297
- closed = true;
298
- },
299
- };
300
- const factory = () => exploding;
301
- const a = new EmailAdapter({
302
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
303
- sessionFactory: factory,
304
- });
305
- const raws = [];
306
- await expect(async () => {
307
- for await (const r of a.sync()) raws.push(r);
308
- }).rejects.toThrow(/network drop/);
309
- expect(closed).toBe(true);
310
- expect(raws).toHaveLength(1);
311
- });
312
- });
313
-
314
- describe("EmailAdapter.normalize", () => {
315
- it("produces a valid UnifiedSchema batch (1 Event + 1 Person per email)", () => {
316
- const a = new EmailAdapter({
317
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
318
- sessionFactory: makeMockSession({}).factory,
319
- });
320
- const raw = {
321
- adapter: "email-imap",
322
- originalId: "<msg-1@example.com>",
323
- capturedAt: 1700000000000,
324
- payload: { ...env(1), folder: "INBOX" },
325
- };
326
- const batch = a.normalize(raw);
327
- expect(batch.events).toHaveLength(1);
328
- expect(batch.persons).toHaveLength(1);
329
- expect(batch.places).toEqual([]);
330
- expect(batch.items).toEqual([]);
331
-
332
- const ev = batch.events[0];
333
- expect(ev.type).toBe("event");
334
- expect(ev.subtype).toBe("message");
335
- expect(ev.content.title).toBe("Subject 1");
336
- expect(ev.content.text).toContain("alice1@example.com");
337
- expect(ev.extra.emailFolder).toBe("INBOX");
338
- expect(ev.extra.uid).toBe(1);
339
- expect(ev.source.adapter).toBe("email-imap");
340
-
341
- const p = batch.persons[0];
342
- expect(p.id).toBe("person-email-alice1@example.com");
343
- expect(p.identifiers.email).toEqual(["alice1@example.com"]);
344
- expect(p.names).toEqual(["Alice"]);
345
-
346
- const valid = validateBatch(batch);
347
- expect(valid.valid).toBe(true);
348
- if (!valid.valid) console.log(valid.errors);
349
- });
350
-
351
- it("handles missing subject + missing from gracefully", () => {
352
- const a = new EmailAdapter({
353
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
354
- sessionFactory: makeMockSession({}).factory,
355
- });
356
- const raw = {
357
- adapter: "email-imap",
358
- originalId: "<m@x>",
359
- capturedAt: 0,
360
- payload: { ...env(2, { subject: "", from: [] }), folder: "INBOX" },
361
- };
362
- const batch = a.normalize(raw);
363
- expect(batch.events[0].content.title).toBe("(no subject)");
364
- expect(batch.persons).toHaveLength(0);
365
- expect(batch.events[0].actor).toBe("person-self");
366
- const v = validateBatch(batch);
367
- expect(v.valid).toBe(true);
368
- });
369
-
370
- it("dedups same sender across multiple emails (stable person id)", () => {
371
- const a = new EmailAdapter({
372
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
373
- sessionFactory: makeMockSession({}).factory,
374
- });
375
- const senderEnv = (uid) => env(uid, {
376
- from: [{ name: "Bob", address: "bob@example.com" }],
377
- messageId: `<m-${uid}@x>`,
378
- });
379
- const b1 = a.normalize({ adapter: "email-imap", originalId: "<m-1@x>", capturedAt: 0, payload: { ...senderEnv(1), folder: "INBOX" } });
380
- const b2 = a.normalize({ adapter: "email-imap", originalId: "<m-2@x>", capturedAt: 0, payload: { ...senderEnv(2), folder: "INBOX" } });
381
- expect(b1.persons[0].id).toBe(b2.persons[0].id);
382
- expect(b1.persons[0].id).toBe("person-email-bob@example.com");
383
- });
384
-
385
- it("rejects missing raw or payload", () => {
386
- const a = new EmailAdapter({
387
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
388
- sessionFactory: makeMockSession({}).factory,
389
- });
390
- expect(() => a.normalize()).toThrow();
391
- expect(() => a.normalize({})).toThrow(/payload/);
392
- });
393
- });
394
-
395
- // ─── Phase 5.2: body parsing integration ───────────────────────────────
396
-
397
- describe("EmailAdapter — body parsing (Phase 5.2)", () => {
398
- it("sync injects parsedBody into payload when parser succeeds", async () => {
399
- const { factory } = makeMockSession({
400
- mailboxes: {
401
- INBOX: {
402
- uidValidity: 1,
403
- envelopes: [{
404
- ...env(1),
405
- source: Buffer.from("RAW BYTES", "utf8"),
406
- }],
407
- },
408
- },
409
- });
410
- const a = new EmailAdapter({
411
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
412
- sessionFactory: factory,
413
- // Inject a fake parser so test doesn't depend on mailparser
414
- parser: async (raw) => ({
415
- headers: { subject: "Parsed" },
416
- textBody: "this is the parsed text body",
417
- htmlBody: "",
418
- attachments: [
419
- { filename: "a.pdf", contentType: "application/pdf", contentDisposition: "attachment",
420
- size: 42, sha256: "abc123", isInline: false, isEncrypted: false },
421
- ],
422
- contentSha256: "deadbeef",
423
- sourceBytes: raw.length,
424
- subject: "Parsed",
425
- date: new Date("2026-05-19"),
426
- }),
427
- });
428
- const raws = [];
429
- for await (const r of a.sync()) raws.push(r);
430
- expect(raws).toHaveLength(1);
431
- expect(raws[0].payload.parsedBody).toBeDefined();
432
- expect(raws[0].payload.parsedBody.textBody).toBe("this is the parsed text body");
433
- expect(raws[0].payload.parsedBody.attachments).toHaveLength(1);
434
- // Source bytes themselves get stripped from the payload to avoid bloat
435
- expect(raws[0].payload.source).toBeUndefined();
436
- });
437
-
438
- it("normalize uses parsedBody.textBody as event.content.text", () => {
439
- const a = new EmailAdapter({
440
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
441
- sessionFactory: makeMockSession({}).factory,
442
- parser: async () => ({}),
443
- });
444
- const raw = {
445
- adapter: "email-imap",
446
- originalId: "<m@x>",
447
- capturedAt: 0,
448
- payload: {
449
- ...env(1),
450
- folder: "INBOX",
451
- parsedBody: {
452
- textBody: "Dear user, your account statement is attached.",
453
- htmlBody: "",
454
- attachments: [],
455
- contentSha256: "abc",
456
- headers: { "list-unsubscribe": "<mailto:unsub@bank.com>" },
457
- },
458
- },
459
- };
460
- const batch = a.normalize(raw);
461
- expect(batch.events[0].content.text).toBe("Dear user, your account statement is attached.");
462
- expect(batch.events[0].extra.rawSha256).toBe("abc");
463
- expect(batch.events[0].extra.indicatorHeaders["list-unsubscribe"]).toBe("<mailto:unsub@bank.com>");
464
- });
465
-
466
- it("normalize falls back to envelope prose when parsedBody is absent", () => {
467
- const a = new EmailAdapter({
468
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
469
- sessionFactory: makeMockSession({}).factory,
470
- parser: async () => ({}),
471
- });
472
- const raw = {
473
- adapter: "email-imap",
474
- originalId: "<m@x>",
475
- capturedAt: 0,
476
- payload: { ...env(1), folder: "INBOX" }, // no parsedBody
477
- };
478
- const batch = a.normalize(raw);
479
- expect(batch.events[0].content.text).toContain("alice1@example.com");
480
- expect(batch.events[0].content.text).toContain("subject:");
481
- });
482
-
483
- it("normalize HTML-only bodies are stripped to plain text", () => {
484
- const a = new EmailAdapter({
485
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
486
- sessionFactory: makeMockSession({}).factory,
487
- parser: async () => ({}),
488
- });
489
- const raw = {
490
- adapter: "email-imap",
491
- originalId: "<m@x>",
492
- capturedAt: 0,
493
- payload: {
494
- ...env(1),
495
- folder: "INBOX",
496
- parsedBody: {
497
- textBody: "",
498
- htmlBody: "<p>Hi <b>there</b>!</p><script>alert(1)</script>",
499
- attachments: [],
500
- },
501
- },
502
- };
503
- const batch = a.normalize(raw);
504
- expect(batch.events[0].content.text).toContain("Hi");
505
- expect(batch.events[0].content.text).toContain("there");
506
- expect(batch.events[0].content.text).not.toContain("<p>");
507
- expect(batch.events[0].content.text).not.toContain("alert(1)"); // script content stripped
508
- });
509
-
510
- it("normalize caps body at maxBodyChars + appends truncation marker", () => {
511
- const longText = "X".repeat(20_000);
512
- const a = new EmailAdapter({
513
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
514
- sessionFactory: makeMockSession({}).factory,
515
- parser: async () => ({}),
516
- maxBodyChars: 100,
517
- });
518
- const raw = {
519
- adapter: "email-imap",
520
- originalId: "<m@x>",
521
- capturedAt: 0,
522
- payload: {
523
- ...env(1),
524
- folder: "INBOX",
525
- parsedBody: { textBody: longText, htmlBody: "", attachments: [] },
526
- },
527
- };
528
- const batch = a.normalize(raw);
529
- expect(batch.events[0].content.text.length).toBeLessThan(longText.length);
530
- expect(batch.events[0].content.text).toMatch(/truncated/);
531
- });
532
-
533
- it("normalize captures attachment metadata in extra.attachments", () => {
534
- const a = new EmailAdapter({
535
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
536
- sessionFactory: makeMockSession({}).factory,
537
- parser: async () => ({}),
538
- });
539
- const raw = {
540
- adapter: "email-imap",
541
- originalId: "<m@x>",
542
- capturedAt: 0,
543
- payload: {
544
- ...env(1),
545
- folder: "INBOX",
546
- parsedBody: {
547
- textBody: "see attached",
548
- htmlBody: "",
549
- attachments: [
550
- { filename: "stmt.pdf", contentType: "application/pdf", contentDisposition: "attachment",
551
- size: 12345, sha256: "abc", isInline: false, isEncrypted: true },
552
- ],
553
- },
554
- },
555
- };
556
- const batch = a.normalize(raw);
557
- expect(batch.events[0].extra.attachments).toHaveLength(1);
558
- const a0 = batch.events[0].extra.attachments[0];
559
- expect(a0.filename).toBe("stmt.pdf");
560
- expect(a0.isEncrypted).toBe(true);
561
- expect(a0.sha256).toBe("abc");
562
- });
563
-
564
- it("sync degrades gracefully when parser throws (parseError captured)", async () => {
565
- const { factory } = makeMockSession({
566
- mailboxes: {
567
- INBOX: { uidValidity: 1, envelopes: [{ ...env(1), source: Buffer.from("garbage", "utf8") }] },
568
- },
569
- });
570
- const a = new EmailAdapter({
571
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
572
- sessionFactory: factory,
573
- parser: async () => { throw new Error("malformed MIME"); },
574
- });
575
- const raws = [];
576
- for await (const r of a.sync()) raws.push(r);
577
- expect(raws).toHaveLength(1);
578
- expect(raws[0].payload.parsedBody).toBeDefined();
579
- expect(raws[0].payload.parsedBody.parseError).toContain("malformed MIME");
580
- // normalize should still produce a valid event (envelope fallback)
581
- const batch = a.normalize(raws[0]);
582
- expect(batch.events).toHaveLength(1);
583
- expect(batch.events[0].extra.parseError).toContain("malformed MIME");
584
- const v = validateBatch(batch);
585
- expect(v.valid).toBe(true);
586
- });
587
- });
588
-
589
- // ─── Phase 5.3: classification integration ─────────────────────────────
590
-
591
- describe("EmailAdapter — classification (Phase 5.3)", () => {
592
- function bankEnv(uid = 1) {
593
- return env(uid, {
594
- from: [{ address: "ebank@cmbchina.com" }],
595
- subject: "招商银行 11 月对账单",
596
- source: Buffer.from("RAW", "utf8"),
597
- });
598
- }
599
-
600
- it("sync attaches classification to payload (Layer 1 short-circuit)", async () => {
601
- const { factory } = makeMockSession({
602
- mailboxes: { INBOX: { uidValidity: 1, envelopes: [bankEnv()] } },
603
- });
604
- const a = new EmailAdapter({
605
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
606
- sessionFactory: factory,
607
- parser: async () => ({ textBody: "stmt", attachments: [] }),
608
- });
609
- const raws = [];
610
- for await (const r of a.sync()) raws.push(r);
611
- expect(raws[0].payload.classification).toBeDefined();
612
- expect(raws[0].payload.classification.category).toBe("bill_bank");
613
- expect(raws[0].payload.classification.layer).toBe("L1");
614
- });
615
-
616
- it("normalize copies classification into extra.classified + .classification", () => {
617
- const a = new EmailAdapter({
618
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
619
- sessionFactory: makeMockSession({}).factory,
620
- parser: async () => ({}),
621
- });
622
- const raw = {
623
- adapter: "email-imap",
624
- originalId: "<m@x>",
625
- capturedAt: 0,
626
- payload: {
627
- ...env(1, { from: [{ address: "x@cmbchina.com" }], subject: "招商银行账单" }),
628
- folder: "INBOX",
629
- classification: {
630
- category: "bill_bank",
631
- confidence: 0.95,
632
- layer: "L1",
633
- ruleName: "bill_bank.cn-bank-major",
634
- },
635
- },
636
- };
637
- const batch = a.normalize(raw);
638
- expect(batch.events[0].extra.classified).toBe("bill_bank");
639
- expect(batch.events[0].extra.classification.category).toBe("bill_bank");
640
- expect(batch.events[0].extra.classification.layer).toBe("L1");
641
- expect(batch.events[0].extra.classification.ruleName).toContain("bill_bank");
642
- });
643
-
644
- it("ambiguous email triggers Layer 2 when LLM is provided", async () => {
645
- const { MockLLMClient } = require("../../lib/llm-client");
646
- const llm = new MockLLMClient({
647
- reply: '{"category":"register","confidence":0.85,"reason":"verification code"}',
648
- });
649
- const { factory } = makeMockSession({
650
- mailboxes: {
651
- INBOX: {
652
- uidValidity: 1,
653
- envelopes: [env(2, {
654
- from: [{ address: "noreply@unknown-service.example" }],
655
- subject: "Welcome",
656
- source: Buffer.from("RAW", "utf8"),
657
- })],
658
- },
659
- },
660
- });
661
- const a = new EmailAdapter({
662
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
663
- sessionFactory: factory,
664
- parser: async () => ({ textBody: "Welcome aboard, here is your verification link." }),
665
- llm,
666
- });
667
- const raws = [];
668
- for await (const r of a.sync()) raws.push(r);
669
- // Layer 1 likely returned 'register' at ~0.75 (welcome rule) — falls to Layer 2
670
- expect(raws[0].payload.classification.category).toBe("register");
671
- // Either L2 fired (if L1 conf < 0.85) or L1 stuck.
672
- expect(["L1", "L2"]).toContain(raws[0].payload.classification.layer);
673
- });
674
-
675
- it("classifier error inside sync degrades to OTHER (doesn't abort sync)", async () => {
676
- const { factory } = makeMockSession({
677
- mailboxes: { INBOX: { uidValidity: 1, envelopes: [bankEnv()] } },
678
- });
679
- const a = new EmailAdapter({
680
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
681
- sessionFactory: factory,
682
- parser: async () => ({}),
683
- classifier: async () => { throw new Error("classifier exploded"); },
684
- });
685
- const raws = [];
686
- for await (const r of a.sync()) raws.push(r);
687
- expect(raws).toHaveLength(1);
688
- expect(raws[0].payload.classification.category).toBe("other");
689
- expect(raws[0].payload.classification.error).toContain("classifier exploded");
690
- });
691
-
692
- it("disableClassification skips both layers entirely", async () => {
693
- const { factory } = makeMockSession({
694
- mailboxes: { INBOX: { uidValidity: 1, envelopes: [bankEnv()] } },
695
- });
696
- const a = new EmailAdapter({
697
- account: { provider: "qq", email: "u@qq.com", authCode: "x", folders: ["INBOX"] },
698
- sessionFactory: factory,
699
- parser: async () => ({}),
700
- disableClassification: true,
701
- });
702
- const raws = [];
703
- for await (const r of a.sync()) raws.push(r);
704
- expect(raws[0].payload.classification).toBeUndefined();
705
- });
706
-
707
- it("capabilities advertise classifier surface", () => {
708
- const a = new EmailAdapter({
709
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
710
- sessionFactory: makeMockSession({}).factory,
711
- });
712
- expect(a.capabilities).toContain("classify:layer1-rules");
713
- expect(a.capabilities).not.toContain("classify:layer2-llm"); // no LLM provided
714
-
715
- const { MockLLMClient } = require("../../lib/llm-client");
716
- const b = new EmailAdapter({
717
- account: { provider: "qq", email: "u@qq.com", authCode: "x" },
718
- sessionFactory: makeMockSession({}).factory,
719
- llm: new MockLLMClient({ reply: "{}" }),
720
- });
721
- expect(b.capabilities).toContain("classify:layer2-llm");
722
- });
723
- });
724
-
725
- describe("parseWatermark / formatWatermark", () => {
726
- it("parses well-formed strings", () => {
727
- expect(parseWatermark("42:100")).toEqual({ uidValidity: "42", lastUid: 100 });
728
- expect(parseWatermark("abc:0")).toEqual({ uidValidity: "abc", lastUid: 0 });
729
- });
730
-
731
- it("falls back to null/0 for malformed input", () => {
732
- expect(parseWatermark("")).toEqual({ uidValidity: null, lastUid: 0 });
733
- expect(parseWatermark("no-colon")).toEqual({ uidValidity: null, lastUid: 0 });
734
- expect(parseWatermark(null)).toEqual({ uidValidity: null, lastUid: 0 });
735
- });
736
-
737
- it("formats correctly", () => {
738
- expect(formatWatermark(42, 100)).toBe("42:100");
739
- expect(formatWatermark("abc", 0)).toBe("abc:0");
740
- expect(formatWatermark(null, 5)).toBe(":5");
741
- });
742
- });