@devrev/ts-adaas 1.20.2-beta.4 → 1.20.2-beta.6

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 (39) hide show
  1. package/dist/common/constants.d.ts.map +1 -1
  2. package/dist/common/constants.js +1 -0
  3. package/dist/common/event-type-translation.d.ts.map +1 -1
  4. package/dist/common/event-type-translation.js +1 -0
  5. package/dist/common/event-type-translation.test.js +1 -0
  6. package/dist/http/axios-client.d.ts +6 -2
  7. package/dist/http/axios-client.d.ts.map +1 -1
  8. package/dist/http/axios-client.js +7 -4
  9. package/dist/mock-server/mock-server.d.ts +4 -0
  10. package/dist/mock-server/mock-server.d.ts.map +1 -1
  11. package/dist/mock-server/mock-server.interfaces.d.ts +9 -0
  12. package/dist/mock-server/mock-server.interfaces.d.ts.map +1 -1
  13. package/dist/mock-server/mock-server.js +39 -6
  14. package/dist/multithreading/process-task.d.ts.map +1 -1
  15. package/dist/multithreading/process-task.js +21 -6
  16. package/dist/multithreading/process-task.test.js +28 -1
  17. package/dist/multithreading/spawn/spawn.d.ts.map +1 -1
  18. package/dist/multithreading/spawn/spawn.helpers.d.ts.map +1 -1
  19. package/dist/multithreading/spawn/spawn.helpers.js +9 -0
  20. package/dist/multithreading/spawn/spawn.helpers.test.js +34 -0
  21. package/dist/multithreading/spawn/spawn.js +1 -0
  22. package/dist/multithreading/worker-adapter/worker-adapter.d.ts +13 -0
  23. package/dist/multithreading/worker-adapter/worker-adapter.d.ts.map +1 -1
  24. package/dist/multithreading/worker-adapter/worker-adapter.emit.test.js +51 -5
  25. package/dist/multithreading/worker-adapter/worker-adapter.js +131 -79
  26. package/dist/multithreading/worker-adapter/worker-adapter.upload-failure.test.d.ts +2 -0
  27. package/dist/multithreading/worker-adapter/worker-adapter.upload-failure.test.d.ts.map +1 -0
  28. package/dist/multithreading/worker-adapter/worker-adapter.upload-failure.test.js +135 -0
  29. package/dist/repo/repo.d.ts +1 -2
  30. package/dist/repo/repo.d.ts.map +1 -1
  31. package/dist/repo/repo.js +6 -12
  32. package/dist/repo/repo.test.js +32 -1
  33. package/dist/state/state.d.ts.map +1 -1
  34. package/dist/state/state.js +9 -5
  35. package/dist/state/state.time-value-resolution.test.js +31 -2
  36. package/dist/types/extraction.d.ts +6 -3
  37. package/dist/types/extraction.d.ts.map +1 -1
  38. package/dist/types/extraction.js +3 -0
  39. package/package.json +1 -1
@@ -22,6 +22,7 @@ const workers_1 = require("../../types/workers");
22
22
  const uploader_1 = require("../../uploader/uploader");
23
23
  const event_type_translation_1 = require("../../common/event-type-translation");
24
24
  const helpers_1 = require("../../common/helpers");
25
+ const spawn_helpers_1 = require("../spawn/spawn.helpers");
25
26
  function createWorkerAdapter({ event, adapterState, options, }) {
26
27
  return new WorkerAdapter({
27
28
  event,
@@ -159,93 +160,49 @@ class WorkerAdapter {
159
160
  */
160
161
  async emit(newEventType, data) {
161
162
  return (0, logger_context_1.runWithSdkLogContext)(async () => {
162
- var _a, _b;
163
+ var _a;
163
164
  newEventType = (0, event_type_translation_1.translateOutgoingEventType)(newEventType);
164
165
  if (this.hasWorkerEmitted) {
165
166
  console.warn(`Trying to emit event with event type: ${newEventType}. Ignoring emit request because it has already been emitted.`);
166
167
  return;
167
168
  }
168
- // If the event is ExternalSyncUnitExtractionDone, upload external sync units via a Repo before emitting
169
- // TODO: Remove in v2.0.0
170
- if (newEventType === extraction_1.ExtractorEventType.ExternalSyncUnitExtractionDone &&
171
- (data === null || data === void 0 ? void 0 : data.external_sync_units) &&
172
- data.external_sync_units.length > 0) {
173
- console.log(`Uploading ${data.external_sync_units.length} external sync units via repo before emitting event.`);
174
- this.initializeRepos([
175
- {
176
- itemType: constants_1.AirSyncDefaultItemTypes.EXTERNAL_SYNC_UNITS,
177
- overridenOptions: {
178
- batchSize: 25000,
179
- skipConfirmation: true,
180
- },
181
- },
182
- ]);
183
- await ((_a = this.getRepo(constants_1.AirSyncDefaultItemTypes.EXTERNAL_SYNC_UNITS)) === null || _a === void 0 ? void 0 : _a.push(data.external_sync_units));
184
- // Remove inline external_sync_units from data to avoid SQS size issues
185
- delete data.external_sync_units;
186
- }
187
- // Upload all repos before emitting the event
188
- console.log(`Uploading all repos before emitting event with event type: ${newEventType}.`);
169
+ let eventPayload;
189
170
  try {
190
- await this.uploadAllRepos();
171
+ // If the event is ExternalSyncUnitExtractionDone, upload external sync units via a Repo before emitting.
172
+ // TODO: Remove in v2.0.0
173
+ if (newEventType === extraction_1.ExtractorEventType.ExternalSyncUnitExtractionDone &&
174
+ (data === null || data === void 0 ? void 0 : data.external_sync_units) &&
175
+ data.external_sync_units.length > 0) {
176
+ console.log(`Uploading ${data.external_sync_units.length} external sync units via repo before emitting event.`);
177
+ this.initializeRepos([
178
+ {
179
+ itemType: constants_1.AirSyncDefaultItemTypes.EXTERNAL_SYNC_UNITS,
180
+ overridenOptions: {
181
+ batchSize: 25000,
182
+ skipConfirmation: true,
183
+ },
184
+ },
185
+ ]);
186
+ await ((_a = this.getRepo(constants_1.AirSyncDefaultItemTypes.EXTERNAL_SYNC_UNITS)) === null || _a === void 0 ? void 0 : _a.push(data.external_sync_units));
187
+ // Remove inline external_sync_units from data to avoid SQS size issues
188
+ delete data.external_sync_units;
189
+ }
190
+ await this.beforeEmit(newEventType);
191
+ eventPayload = this.buildEmitPayload(newEventType, data);
191
192
  }
192
193
  catch (error) {
193
- console.error('Error while uploading repos', error);
194
- node_worker_threads_1.parentPort === null || node_worker_threads_1.parentPort === void 0 ? void 0 : node_worker_threads_1.parentPort.postMessage(workers_1.WorkerMessageSubject.WorkerMessageExit);
195
- this.hasWorkerEmitted = true;
194
+ console.error('Error while preparing event for emission.', (0, logger_1.serializeError)(error));
195
+ await this.emitError(error);
196
196
  return;
197
197
  }
198
- // If the extraction is done, we want to save the timestamp of the last successful sync
199
- if (newEventType === extraction_1.ExtractorEventType.AttachmentExtractionDone) {
200
- console.log(`Overwriting lastSuccessfulSyncStarted with lastSyncStarted (${this.state.lastSyncStarted}).`);
201
- this.state.lastSuccessfulSyncStarted = this.state.lastSyncStarted;
202
- this.state.lastSyncStarted = '';
203
- // Clear pending extraction boundaries now that the cycle is complete
204
- this.state.pendingWorkersOldest = '';
205
- this.state.pendingWorkersNewest = '';
206
- // Update workersOldest and workersNewest boundaries from resolved extraction timestamps.
207
- // Expand boundaries: workersOldest gets the earliest timestamp, workersNewest gets the latest.
208
- const extractionStart = this.event.payload.event_context.extract_from;
209
- const extractionEnd = this.event.payload.event_context.extract_to;
210
- if (extractionStart &&
211
- (!this.state.workersOldest ||
212
- extractionStart < this.state.workersOldest)) {
213
- console.log(`Updating workersOldest from '${this.state.workersOldest}' to '${extractionStart}'.`);
214
- this.state.workersOldest = extractionStart;
215
- }
216
- if (extractionEnd &&
217
- (!this.state.workersNewest ||
218
- extractionEnd > this.state.workersNewest)) {
219
- console.log(`Updating workersNewest from '${this.state.workersNewest}' to '${extractionEnd}'.`);
220
- this.state.workersNewest = extractionEnd;
221
- }
222
- }
223
- // We want to save the state every time we emit an event, except for the start and delete events
224
- if (!constants_1.STATELESS_EVENT_TYPES.includes(this.event.payload.event_type)) {
225
- console.log(`Saving state before emitting event with event type: ${newEventType}.`);
226
- try {
227
- await this.adapterState.postState(this.state);
228
- }
229
- catch (error) {
230
- console.error('Error while posting state', error);
231
- node_worker_threads_1.parentPort === null || node_worker_threads_1.parentPort === void 0 ? void 0 : node_worker_threads_1.parentPort.postMessage(workers_1.WorkerMessageSubject.WorkerMessageExit);
232
- this.hasWorkerEmitted = true;
233
- return;
234
- }
235
- }
236
198
  try {
237
- // Always prune error messages to make them shorter before emit
238
- if ((_b = data === null || data === void 0 ? void 0 : data.error) === null || _b === void 0 ? void 0 : _b.message) {
239
- data.error.message = (0, helpers_1.truncateMessage)(data.error.message);
240
- }
241
199
  const isExtractionEvent = Object.values(extraction_1.ExtractorEventType).includes(newEventType);
242
- const isLoaderEvent = Object.values(loading_1.LoaderEventType).includes(newEventType);
243
200
  const progressData = {};
244
201
  if (isExtractionEvent &&
245
- (newEventType == extraction_1.ExtractorEventType.DataExtractionDone ||
246
- newEventType == extraction_1.ExtractorEventType.DataExtractionProgress ||
247
- newEventType == extraction_1.ExtractorEventType.AttachmentExtractionDone ||
248
- newEventType == extraction_1.ExtractorEventType.AttachmentExtractionProgress)) {
202
+ (newEventType === extraction_1.ExtractorEventType.DataExtractionDone ||
203
+ newEventType === extraction_1.ExtractorEventType.DataExtractionProgress ||
204
+ newEventType === extraction_1.ExtractorEventType.AttachmentExtractionDone ||
205
+ newEventType === extraction_1.ExtractorEventType.AttachmentExtractionProgress)) {
249
206
  const repo = this.lastExtractedItemType
250
207
  ? this.repos.find((r) => r.itemType === this.lastExtractedItemType)
251
208
  : undefined;
@@ -260,9 +217,7 @@ class WorkerAdapter {
260
217
  await (0, control_protocol_1.emit)({
261
218
  eventType: newEventType,
262
219
  event: this.event,
263
- data: Object.assign(Object.assign(Object.assign({}, data), (isExtractionEvent ? { artifacts: this.artifacts } : {})), (isLoaderEvent
264
- ? { reports: this.reports, processed_files: this.processedFiles }
265
- : {})),
220
+ data: eventPayload,
266
221
  worker_metadata: Object.assign({}, progressData),
267
222
  });
268
223
  const message = {
@@ -280,14 +235,111 @@ class WorkerAdapter {
280
235
  }
281
236
  });
282
237
  }
238
+ /**
239
+ * Emits a phase-specific error after task or pre-emit execution fails. This
240
+ * path avoids flushing repos again, so artifacts uploaded before the failure
241
+ * are still reported without retrying the failed batch.
242
+ */
243
+ async emitError(error) {
244
+ return (0, logger_context_1.runWithSdkLogContext)(async () => {
245
+ if (this.hasWorkerEmitted) {
246
+ return;
247
+ }
248
+ // Include artifacts uploaded before a failure so they are not orphaned
249
+ // and can be associated with the emitted phase error.
250
+ for (const repo of this.repos) {
251
+ this.artifacts = [...this.artifacts, ...repo.uploadedArtifacts];
252
+ }
253
+ const { eventType } = (0, spawn_helpers_1.getTimeoutErrorEventType)(this.event.payload.event_type);
254
+ try {
255
+ await (0, control_protocol_1.emit)({
256
+ eventType,
257
+ event: this.event,
258
+ data: this.buildEmitPayload(eventType, {
259
+ error: { message: (0, logger_1.serializeError)(error) },
260
+ }),
261
+ });
262
+ const message = {
263
+ subject: workers_1.WorkerMessageSubject.WorkerMessageEmitted,
264
+ payload: { eventType },
265
+ };
266
+ this.artifacts = [];
267
+ node_worker_threads_1.parentPort === null || node_worker_threads_1.parentPort === void 0 ? void 0 : node_worker_threads_1.parentPort.postMessage(message);
268
+ this.hasWorkerEmitted = true;
269
+ }
270
+ catch (emitError) {
271
+ console.error(`Error while emitting error event with event type: ${eventType}.`, (0, logger_1.serializeError)(emitError));
272
+ }
273
+ });
274
+ }
283
275
  async uploadAllRepos() {
284
276
  for (const repo of this.repos) {
285
- const error = await repo.upload();
277
+ await repo.upload();
286
278
  this.artifacts.push(...repo.uploadedArtifacts);
287
- if (error) {
288
- throw error;
279
+ }
280
+ }
281
+ /**
282
+ * Pre-emit hook: uploads all repos and updates extraction boundaries/state.
283
+ * Throws if uploading repos or persisting state fails. The caller emits a
284
+ * compensating error event with whatever artifacts did upload.
285
+ */
286
+ async beforeEmit(eventType) {
287
+ // Upload all repos before emitting the event.
288
+ console.log(`Uploading all repos before emitting event with event type: ${eventType}.`);
289
+ await this.uploadAllRepos();
290
+ // If the extraction is done, save the timestamp of the last successful sync.
291
+ if (eventType === extraction_1.ExtractorEventType.AttachmentExtractionDone) {
292
+ console.log(`Overwriting lastSuccessfulSyncStarted with lastSyncStarted (${this.state.lastSyncStarted}).`);
293
+ this.state.lastSuccessfulSyncStarted = this.state.lastSyncStarted;
294
+ this.state.lastSyncStarted = '';
295
+ // Clear pending extraction boundaries now that the cycle is complete.
296
+ this.state.pendingWorkersOldest = '';
297
+ this.state.pendingWorkersNewest = '';
298
+ // Update workersOldest and workersNewest boundaries from resolved extraction timestamps.
299
+ const extractionStart = this.event.payload.event_context.extract_from;
300
+ const extractionEnd = this.event.payload.event_context.extract_to;
301
+ if (extractionStart &&
302
+ (!this.state.workersOldest ||
303
+ extractionStart < this.state.workersOldest)) {
304
+ console.log(`Updating workersOldest from '${this.state.workersOldest}' to '${extractionStart}'.`);
305
+ this.state.workersOldest = extractionStart;
306
+ }
307
+ if (extractionEnd &&
308
+ (!this.state.workersNewest || extractionEnd > this.state.workersNewest)) {
309
+ console.log(`Updating workersNewest from '${this.state.workersNewest}' to '${extractionEnd}'.`);
310
+ this.state.workersNewest = extractionEnd;
289
311
  }
290
312
  }
313
+ if (constants_1.STATELESS_EVENT_TYPES.includes(this.event.payload.event_type)) {
314
+ return;
315
+ }
316
+ let stateSizeKb = 0;
317
+ try {
318
+ stateSizeKb =
319
+ Buffer.byteLength(JSON.stringify(this.state), 'utf8') / 1024;
320
+ }
321
+ catch (_a) {
322
+ stateSizeKb = 0;
323
+ }
324
+ console.log(`Saving ${stateSizeKb.toFixed(2)} KB state before emitting event with event type: ${eventType}. Current state`, (0, logger_1.getPrintableState)(this.state));
325
+ try {
326
+ await this.adapterState.postState(this.state);
327
+ }
328
+ catch (error) {
329
+ console.error('Error while posting state.', (0, logger_1.serializeError)(error));
330
+ throw error;
331
+ }
332
+ }
333
+ buildEmitPayload(eventType, data) {
334
+ var _a;
335
+ if ((_a = data === null || data === void 0 ? void 0 : data.error) === null || _a === void 0 ? void 0 : _a.message) {
336
+ data.error.message = (0, helpers_1.truncateMessage)(data.error.message);
337
+ }
338
+ const isExtractionEvent = Object.values(extraction_1.ExtractorEventType).includes(eventType);
339
+ const isLoaderEvent = Object.values(loading_1.LoaderEventType).includes(eventType);
340
+ return Object.assign(Object.assign(Object.assign({}, data), (isExtractionEvent ? { artifacts: this.artifacts } : {})), (isLoaderEvent
341
+ ? { reports: this.reports, processed_files: this.processedFiles }
342
+ : {}));
291
343
  }
292
344
  async loadItemTypes({ itemTypesToLoad, }) {
293
345
  return (0, logger_context_1.runWithSdkLogContext)(async () => {
@@ -0,0 +1,2 @@
1
+ export {};
2
+ //# sourceMappingURL=worker-adapter.upload-failure.test.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"worker-adapter.upload-failure.test.d.ts","sourceRoot":"","sources":["../../../src/multithreading/worker-adapter/worker-adapter.upload-failure.test.ts"],"names":[],"mappings":""}
@@ -0,0 +1,135 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ const state_1 = require("../../state/state");
4
+ const jest_setup_1 = require("../../tests/jest.setup");
5
+ const test_helpers_1 = require("../../tests/test-helpers");
6
+ const test_utils_1 = require("../../common/test-utils");
7
+ const types_1 = require("../../types");
8
+ const worker_adapter_1 = require("./worker-adapter");
9
+ /* eslint-disable @typescript-eslint/no-require-imports */
10
+ jest.mock('../../common/control-protocol', () => ({
11
+ emit: jest.fn().mockResolvedValue({}),
12
+ }));
13
+ jest.mock('../../mappers/mappers');
14
+ jest.mock('node:worker_threads', () => ({
15
+ parentPort: { postMessage: jest.fn() },
16
+ }));
17
+ jest.mock('../../attachments-streaming/attachments-streaming-pool', () => ({
18
+ AttachmentsStreamingPool: jest.fn().mockImplementation(() => ({
19
+ streamAll: jest.fn().mockResolvedValue(undefined),
20
+ })),
21
+ }));
22
+ const UPLOAD_URL_PATH = '/internal/airdrop.artifacts.upload-url';
23
+ function makeAdapter() {
24
+ const event = (0, test_utils_1.createMockEvent)(jest_setup_1.mockServer.baseUrl, {
25
+ payload: { event_type: types_1.EventType.StartExtractingData },
26
+ });
27
+ const initialState = {
28
+ attachments: { completed: false },
29
+ lastSyncStarted: '',
30
+ lastSuccessfulSyncStarted: '',
31
+ snapInVersionId: '',
32
+ toDevRev: {
33
+ attachmentsMetadata: {
34
+ artifactIds: [],
35
+ lastProcessed: 0,
36
+ lastProcessedAttachmentsIdsList: [],
37
+ },
38
+ },
39
+ };
40
+ const adapterState = new state_1.State({ event, initialState });
41
+ const adapter = new worker_adapter_1.WorkerAdapter({ event, adapterState });
42
+ const workerThreads = require('node:worker_threads');
43
+ const mockPostMessage = jest.fn();
44
+ if (workerThreads.parentPort) {
45
+ jest
46
+ .spyOn(workerThreads.parentPort, 'postMessage')
47
+ .mockImplementation(mockPostMessage);
48
+ }
49
+ else {
50
+ workerThreads.parentPort = { postMessage: mockPostMessage };
51
+ }
52
+ return { adapter, mockPostMessage };
53
+ }
54
+ describe(`${worker_adapter_1.WorkerAdapter.name} upload failure (near-integration)`, () => {
55
+ let adapter;
56
+ let mockPostMessage;
57
+ beforeEach(() => {
58
+ jest.clearAllMocks();
59
+ jest_setup_1.mockServer.resetRoutes();
60
+ ({ adapter, mockPostMessage } = makeAdapter());
61
+ adapter['adapterState'].postState = jest.fn().mockResolvedValue(undefined);
62
+ });
63
+ afterEach(() => {
64
+ jest.restoreAllMocks();
65
+ });
66
+ it('should throw from repo.push when upload-url returns 400 during batch upload', async () => {
67
+ var _a, _b;
68
+ jest_setup_1.mockServer.setRoute({
69
+ path: UPLOAD_URL_PATH,
70
+ method: 'GET',
71
+ status: 400,
72
+ });
73
+ adapter.initializeRepos([
74
+ {
75
+ itemType: 'tasks',
76
+ overridenOptions: { batchSize: 10 },
77
+ },
78
+ ]);
79
+ await expect((_a = adapter.getRepo('tasks')) === null || _a === void 0 ? void 0 : _a.push((0, test_helpers_1.createItems)(20))).rejects.toThrow('artifact upload URL');
80
+ expect((_b = adapter.getRepo('tasks')) === null || _b === void 0 ? void 0 : _b.getItems().length).toBe(20);
81
+ });
82
+ it('should emit DataExtractionError when uploadAllRepos fails', async () => {
83
+ var _a;
84
+ const { emit: mockEmit } = require('../../common/control-protocol');
85
+ jest_setup_1.mockServer.setRoute({
86
+ path: UPLOAD_URL_PATH,
87
+ method: 'GET',
88
+ status: 400,
89
+ });
90
+ adapter.initializeRepos([{ itemType: 'tasks' }]);
91
+ await ((_a = adapter.getRepo('tasks')) === null || _a === void 0 ? void 0 : _a.push((0, test_helpers_1.createItems)(5)));
92
+ await adapter.emit(types_1.ExtractorEventType.DataExtractionDone);
93
+ expect(mockEmit).toHaveBeenCalledWith(expect.objectContaining({
94
+ eventType: types_1.ExtractorEventType.DataExtractionError,
95
+ data: expect.objectContaining({
96
+ error: expect.objectContaining({
97
+ message: expect.stringContaining('artifact upload URL'),
98
+ }),
99
+ }),
100
+ }));
101
+ expect(mockPostMessage).toHaveBeenCalledWith(expect.objectContaining({
102
+ subject: 'emit',
103
+ payload: { eventType: types_1.ExtractorEventType.DataExtractionError },
104
+ }));
105
+ });
106
+ it('should include partial artifacts when push batch succeeded before uploadAllRepos fails', async () => {
107
+ var _a;
108
+ const { emit: mockEmit } = require('../../common/control-protocol');
109
+ jest_setup_1.mockServer.setRoute({
110
+ path: UPLOAD_URL_PATH,
111
+ method: 'GET',
112
+ status: 200,
113
+ succeedThenFail: {
114
+ successCount: 1,
115
+ errorStatus: 400,
116
+ },
117
+ });
118
+ adapter.initializeRepos([
119
+ {
120
+ itemType: 'tasks',
121
+ overridenOptions: { batchSize: 10 },
122
+ },
123
+ ]);
124
+ await ((_a = adapter.getRepo('tasks')) === null || _a === void 0 ? void 0 : _a.push((0, test_helpers_1.createItems)(15)));
125
+ await adapter.emit(types_1.ExtractorEventType.DataExtractionDone);
126
+ expect(mockEmit).toHaveBeenCalledWith(expect.objectContaining({
127
+ eventType: types_1.ExtractorEventType.DataExtractionError,
128
+ data: expect.objectContaining({
129
+ artifacts: expect.arrayContaining([
130
+ expect.objectContaining({ item_count: 10, item_type: 'tasks' }),
131
+ ]),
132
+ }),
133
+ }));
134
+ });
135
+ });
@@ -1,5 +1,4 @@
1
1
  import { Item } from '../repo/repo.interfaces';
2
- import { ErrorRecord } from '../types/common';
3
2
  import { Artifact } from '../uploader/uploader.interfaces';
4
3
  import { NormalizedAttachment, NormalizedItem, RepoFactoryInterface } from './repo.interfaces';
5
4
  export declare class Repo {
@@ -22,7 +21,7 @@ export declare class Repo {
22
21
  };
23
22
  constructor({ event, itemType, normalize, onUpload, options, }: RepoFactoryInterface);
24
23
  getItems(): (NormalizedItem | NormalizedAttachment | Item)[];
25
- upload(batch?: (NormalizedItem | NormalizedAttachment | Item)[]): Promise<void | ErrorRecord>;
24
+ upload(batch?: (NormalizedItem | NormalizedAttachment | Item)[]): Promise<void>;
26
25
  push(items: Item[]): Promise<boolean>;
27
26
  }
28
27
  //# sourceMappingURL=repo.d.ts.map
@@ -1 +1 @@
1
- {"version":3,"file":"repo.d.ts","sourceRoot":"","sources":["../../src/repo/repo.ts"],"names":[],"mappings":"AAKA,OAAO,EAAE,IAAI,EAAE,MAAM,yBAAyB,CAAC;AAC/C,OAAO,EAAE,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAE9C,OAAO,EAAE,QAAQ,EAAE,MAAM,iCAAiC,CAAC;AAI3D,OAAO,EACL,oBAAoB,EACpB,cAAc,EACd,oBAAoB,EACrB,MAAM,mBAAmB,CAAC;AAG3B,qBAAa,IAAI;IACf,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,OAAO,CAAC,KAAK,CAAmD;IAChE,OAAO,CAAC,SAAS,CAAC,CAAwD;IAC1E,OAAO,CAAC,QAAQ,CAAW;IAC3B,OAAO,CAAC,QAAQ,CAA+B;IAC/C,OAAO,CAAC,OAAO,CAAC,CAAuB;IAChC,iBAAiB,EAAE,QAAQ,EAAE,CAAC;IAC9B,UAAU,EAAE;QACjB,YAAY,EAAE;YAAE,MAAM,CAAC,EAAE,MAAM,CAAC;YAAC,MAAM,CAAC,EAAE,MAAM,CAAA;SAAE,CAAC;QACnD,YAAY,EAAE;YAAE,MAAM,CAAC,EAAE,MAAM,CAAC;YAAC,MAAM,CAAC,EAAE,MAAM,CAAA;SAAE,CAAC;KACpD,CAGC;gBAEU,EACV,KAAK,EACL,QAAQ,EACR,SAAS,EACT,QAAQ,EACR,OAAO,GACR,EAAE,oBAAoB;IAUvB,QAAQ,IAAI,CAAC,cAAc,GAAG,oBAAoB,GAAG,IAAI,CAAC,EAAE;IAItD,MAAM,CACV,KAAK,CAAC,EAAE,CAAC,cAAc,GAAG,oBAAoB,GAAG,IAAI,CAAC,EAAE,GACvD,OAAO,CAAC,IAAI,GAAG,WAAW,CAAC;IAsDxB,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC,OAAO,CAAC;CAyC5C"}
1
+ {"version":3,"file":"repo.d.ts","sourceRoot":"","sources":["../../src/repo/repo.ts"],"names":[],"mappings":"AAKA,OAAO,EAAE,IAAI,EAAE,MAAM,yBAAyB,CAAC;AAE/C,OAAO,EAAE,QAAQ,EAAE,MAAM,iCAAiC,CAAC;AAI3D,OAAO,EACL,oBAAoB,EACpB,cAAc,EACd,oBAAoB,EACrB,MAAM,mBAAmB,CAAC;AAG3B,qBAAa,IAAI;IACf,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,OAAO,CAAC,KAAK,CAAmD;IAChE,OAAO,CAAC,SAAS,CAAC,CAAwD;IAC1E,OAAO,CAAC,QAAQ,CAAW;IAC3B,OAAO,CAAC,QAAQ,CAA+B;IAC/C,OAAO,CAAC,OAAO,CAAC,CAAuB;IAChC,iBAAiB,EAAE,QAAQ,EAAE,CAAC;IAC9B,UAAU,EAAE;QACjB,YAAY,EAAE;YAAE,MAAM,CAAC,EAAE,MAAM,CAAC;YAAC,MAAM,CAAC,EAAE,MAAM,CAAA;SAAE,CAAC;QACnD,YAAY,EAAE;YAAE,MAAM,CAAC,EAAE,MAAM,CAAC;YAAC,MAAM,CAAC,EAAE,MAAM,CAAA;SAAE,CAAC;KACpD,CAGC;gBAEU,EACV,KAAK,EACL,QAAQ,EACR,SAAS,EACT,QAAQ,EACR,OAAO,GACR,EAAE,oBAAoB;IAUvB,QAAQ,IAAI,CAAC,cAAc,GAAG,oBAAoB,GAAG,IAAI,CAAC,EAAE;IAItD,MAAM,CACV,KAAK,CAAC,EAAE,CAAC,cAAc,GAAG,oBAAoB,GAAG,IAAI,CAAC,EAAE,GACvD,OAAO,CAAC,IAAI,CAAC;IA4DV,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC,OAAO,CAAC;CAkC5C"}
package/dist/repo/repo.js CHANGED
@@ -23,6 +23,7 @@ class Repo {
23
23
  return this.items;
24
24
  }
25
25
  async upload(batch) {
26
+ var _a;
26
27
  const itemsToUpload = batch || this.items;
27
28
  if (itemsToUpload.length > 0) {
28
29
  for (const item of itemsToUpload) {
@@ -44,8 +45,8 @@ class Repo {
44
45
  console.log(`Uploading ${itemsToUpload.length} items of type ${this.itemType}. `);
45
46
  const { artifact, error } = await this.uploader.upload(this.itemType, itemsToUpload);
46
47
  if (error || !artifact) {
47
- console.error('Error while uploading batch', error);
48
- return error;
48
+ console.error(`Error while uploading batch of ${itemsToUpload.length} items of type ${this.itemType}.`, error);
49
+ throw new Error((_a = error === null || error === void 0 ? void 0 : error.message) !== null && _a !== void 0 ? _a : `Upload failed for item type "${this.itemType}" without artifact.`);
49
50
  }
50
51
  this.onUpload(artifact);
51
52
  this.uploadedArtifacts.push(artifact);
@@ -80,16 +81,9 @@ class Repo {
80
81
  // Upload in batches while the number of items exceeds the batch size
81
82
  const batchSize = ((_a = this.options) === null || _a === void 0 ? void 0 : _a.batchSize) || constants_1.ARTIFACT_BATCH_SIZE;
82
83
  while (this.items.length >= batchSize) {
83
- // Slice out a batch of batchSize items to upload
84
- const batch = this.items.splice(0, batchSize);
85
- try {
86
- // Upload the batch
87
- await this.upload(batch);
88
- }
89
- catch (error) {
90
- console.error('Error while uploading batch', error);
91
- return false;
92
- }
84
+ const batch = this.items.slice(0, batchSize);
85
+ await this.upload(batch);
86
+ this.items.splice(0, batchSize);
93
87
  }
94
88
  return true;
95
89
  }
@@ -313,7 +313,7 @@ describe(repo_1.Repo.name, () => {
313
313
  error: new Error('fail'),
314
314
  artifact: null,
315
315
  });
316
- await repo.upload([itemWithDate('1', '2022-01-01T00:00:00.000Z')]);
316
+ await expect(repo.upload([itemWithDate('1', '2022-01-01T00:00:00.000Z')])).rejects.toThrow('fail');
317
317
  expect(repo.dateRanges.creationDate.oldest).toBe(ts('2022-01-01T00:00:00.000Z'));
318
318
  expect(repo.dateRanges.creationDate.newest).toBe(ts('2022-01-01T00:00:00.000Z'));
319
319
  });
@@ -343,4 +343,35 @@ describe(repo_1.Repo.name, () => {
343
343
  expect(repo.dateRanges.modifiedDate.newest).toBe(ts('2023-01-01T00:00:00.000Z'));
344
344
  });
345
345
  });
346
+ it('should throw when upload fails', async () => {
347
+ mockUploadFn.mockResolvedValueOnce({
348
+ error: { message: 'upload failed' },
349
+ });
350
+ await expect(repo.upload((0, test_helpers_1.createItems)(1))).rejects.toThrow('upload failed');
351
+ });
352
+ it('should retain items in repo when batch upload fails during push', async () => {
353
+ repo = new repo_1.Repo({
354
+ event: (0, test_utils_1.createMockEvent)(jest_setup_1.mockServer.baseUrl, {
355
+ payload: { event_type: types_1.EventType.ExtractionDataStart },
356
+ }),
357
+ itemType: 'test_item_type',
358
+ normalize,
359
+ onUpload: jest.fn(),
360
+ options: { batchSize: 10 },
361
+ });
362
+ const items = (0, test_helpers_1.createItems)(20);
363
+ mockUploadFn
364
+ .mockResolvedValueOnce({
365
+ artifact: {
366
+ id: 'artifact-1',
367
+ item_type: 'test_item_type',
368
+ item_count: 10,
369
+ },
370
+ })
371
+ .mockResolvedValueOnce({
372
+ error: { message: 'second batch failed' },
373
+ });
374
+ await expect(repo.push(items)).rejects.toThrow('second batch failed');
375
+ expect(repo.getItems().length).toBe(10);
376
+ });
346
377
  });
@@ -1 +1 @@
1
- {"version":3,"file":"state.d.ts","sourceRoot":"","sources":["../../src/state/state.ts"],"names":[],"mappings":"AAYA,OAAO,EACL,YAAY,EAIZ,cAAc,EACf,MAAM,oBAAoB,CAAC;AAC5B,OAAO,EAAE,eAAe,EAAE,MAAM,kBAAkB,CAAC;AAEnD,wBAAsB,kBAAkB,CAAC,cAAc,EAAE,EACvD,KAAK,EACL,YAAY,EACZ,oBAAoB,EACpB,OAAO,GACR,EAAE,cAAc,CAAC,cAAc,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CA8KjE;AAED,qBAAa,KAAK,CAAC,cAAc;IAC/B,OAAO,CAAC,MAAM,CAA+B;IAC7C,OAAO,CAAC,gBAAgB,CAAuB;IAC/C,OAAO,CAAC,eAAe,CAAW;IAClC,OAAO,CAAC,SAAS,CAAS;IAC1B,OAAO,CAAC,WAAW,CAAS;IAC5B,OAAO,CAAC,UAAU,CAAS;IAC3B,OAAO,CAAC,SAAS,CAAS;gBAEd,EAAE,KAAK,EAAE,YAAY,EAAE,EAAE,cAAc,CAAC,cAAc,CAAC;IAenE,IAAI,KAAK,IAAI,YAAY,CAAC,cAAc,CAAC,CAExC;IAED,IAAI,KAAK,CAAC,KAAK,EAAE,YAAY,CAAC,cAAc,CAAC,EAE5C;IAED,IAAI,eAAe,IAAI,eAAe,CAErC;IAED;;;;OAIG;IACG,IAAI,CAAC,YAAY,EAAE,cAAc,GAAG,OAAO,CAAC,IAAI,CAAC;IAiDvD;;;OAGG;IACG,SAAS,CAAC,KAAK,CAAC,EAAE,YAAY,CAAC,cAAc,CAAC;IAqDpD;;;OAGG;IACG,UAAU,IAAI,OAAO,CAAC;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,OAAO,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;CAqBjE"}
1
+ {"version":3,"file":"state.d.ts","sourceRoot":"","sources":["../../src/state/state.ts"],"names":[],"mappings":"AAYA,OAAO,EACL,YAAY,EAIZ,cAAc,EACf,MAAM,oBAAoB,CAAC;AAC5B,OAAO,EAAE,eAAe,EAAE,MAAM,kBAAkB,CAAC;AAEnD,wBAAsB,kBAAkB,CAAC,cAAc,EAAE,EACvD,KAAK,EACL,YAAY,EACZ,oBAAoB,EACpB,OAAO,GACR,EAAE,cAAc,CAAC,cAAc,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CAqLjE;AAED,qBAAa,KAAK,CAAC,cAAc;IAC/B,OAAO,CAAC,MAAM,CAA+B;IAC7C,OAAO,CAAC,gBAAgB,CAAuB;IAC/C,OAAO,CAAC,eAAe,CAAW;IAClC,OAAO,CAAC,SAAS,CAAS;IAC1B,OAAO,CAAC,WAAW,CAAS;IAC5B,OAAO,CAAC,UAAU,CAAS;IAC3B,OAAO,CAAC,SAAS,CAAS;gBAEd,EAAE,KAAK,EAAE,YAAY,EAAE,EAAE,cAAc,CAAC,cAAc,CAAC;IAenE,IAAI,KAAK,IAAI,YAAY,CAAC,cAAc,CAAC,CAExC;IAED,IAAI,KAAK,CAAC,KAAK,EAAE,YAAY,CAAC,cAAc,CAAC,EAE5C;IAED,IAAI,eAAe,IAAI,eAAe,CAErC;IAED;;;;OAIG;IACG,IAAI,CAAC,YAAY,EAAE,cAAc,GAAG,OAAO,CAAC,IAAI,CAAC;IAiDvD;;;OAGG;IACG,SAAS,CAAC,KAAK,CAAC,EAAE,YAAY,CAAC,cAAc,CAAC;IAqDpD;;;OAGG;IACG,UAAU,IAAI,OAAO,CAAC;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,OAAO,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;CAqBjE"}
@@ -67,7 +67,7 @@ async function createAdapterState({ event, initialState, initialDomainMapping, o
67
67
  const eventContext = event.payload.event_context;
68
68
  const overrideExtractFrom = (value) => {
69
69
  if (eventContext.extract_from !== undefined) {
70
- eventContext.platform_extract_from = eventContext.extract_from;
70
+ eventContext.original_extract_from = eventContext.extract_from;
71
71
  }
72
72
  eventContext.extract_from = value;
73
73
  };
@@ -126,10 +126,14 @@ async function createAdapterState({ event, initialState, initialDomainMapping, o
126
126
  }
127
127
  else {
128
128
  // Non-StartExtractingMetadata events: reuse pending values from state
129
- if (eventContext.extract_from === undefined &&
130
- as.state.pendingWorkersOldest) {
131
- overrideExtractFrom(as.state.pendingWorkersOldest);
132
- console.log(`Reusing pendingWorkersOldest as extract_from: ${as.state.pendingWorkersOldest}.`);
129
+ const pendingExtractFrom = as.state.pendingWorkersOldest || as.state.lastSuccessfulSyncStarted;
130
+ if (pendingExtractFrom &&
131
+ (eventContext.extract_from === undefined ||
132
+ eventContext.extract_from !== pendingExtractFrom)) {
133
+ overrideExtractFrom(pendingExtractFrom);
134
+ console.log(`Using ${as.state.pendingWorkersOldest
135
+ ? 'pendingWorkersOldest'
136
+ : 'lastSuccessfulSyncStarted'} as extract_from: ${pendingExtractFrom}.`);
133
137
  }
134
138
  else {
135
139
  console.log('pendingWorkersOldest is not set in state. extract_from will not be populated for this invocation.');
@@ -102,7 +102,7 @@ describe('State — TimeValue resolution', () => {
102
102
  // Assert: the successful sync boundary takes precedence and is retained
103
103
  // for later phases.
104
104
  expect(event.payload.event_context.extract_from).toBe('2025-05-01T00:00:00.000Z');
105
- expect(event.payload.event_context.platform_extract_from).toBe(platformExtractFrom);
105
+ expect(event.payload.event_context.original_extract_from).toBe(platformExtractFrom);
106
106
  expect(state.state.pendingWorkersOldest).toBe('2025-05-01T00:00:00.000Z');
107
107
  });
108
108
  it('should preserve the platform extract_from when resolving a CPv2 start time', async () => {
@@ -134,7 +134,7 @@ describe('State — TimeValue resolution', () => {
134
134
  initialDomainMapping: {},
135
135
  });
136
136
  expect(event.payload.event_context.extract_from).toBe(resolvedExtractFrom);
137
- expect(event.payload.event_context.platform_extract_from).toBe(platformExtractFrom);
137
+ expect(event.payload.event_context.original_extract_from).toBe(platformExtractFrom);
138
138
  });
139
139
  it('should use the platform extract_from when lastSuccessfulSyncStarted is not set', async () => {
140
140
  // Arrange: CPv1.1 sends a concrete extract_from instead of a TimeValue.
@@ -228,6 +228,35 @@ describe('State — TimeValue resolution', () => {
228
228
  expect(event.payload.event_context.extract_from).toBe(lastSuccessfulSyncStarted);
229
229
  expect(state.state.pendingWorkersOldest).toBe(lastSuccessfulSyncStarted);
230
230
  });
231
+ it('should override platform extract_from on StartExtractingData and preserve its original value', async () => {
232
+ const platformExtractFrom = '2026-08-24T13:26:15.388Z';
233
+ const lastSuccessfulSyncStarted = '2026-08-25T13:26:15.388Z';
234
+ const event = (0, test_utils_1.createMockEvent)(jest_setup_1.mockServer.baseUrl, {
235
+ context: {
236
+ snap_in_version_id: 'test_snap_in_version_id',
237
+ },
238
+ payload: {
239
+ event_type: extraction_1.EventType.StartExtractingData,
240
+ event_context: {
241
+ extract_from: platformExtractFrom,
242
+ extraction_start_time: {},
243
+ },
244
+ },
245
+ });
246
+ fetchStateSpy.mockResolvedValue({
247
+ state: JSON.stringify({
248
+ snapInVersionId: 'test_snap_in_version_id',
249
+ lastSuccessfulSyncStarted,
250
+ }),
251
+ });
252
+ await (0, state_1.createAdapterState)({
253
+ event,
254
+ initialState: {},
255
+ initialDomainMapping: {},
256
+ });
257
+ expect(event.payload.event_context.extract_from).toBe(lastSuccessfulSyncStarted);
258
+ expect(event.payload.event_context.original_extract_from).toBe(platformExtractFrom);
259
+ });
231
260
  it('should skip resolution when extraction_end_time has no type', async () => {
232
261
  // Arrange: platform sends extraction_end_time without a type field
233
262
  const event = (0, test_utils_1.createMockEvent)(jest_setup_1.mockServer.baseUrl, {
@@ -56,7 +56,8 @@ export declare enum EventType {
56
56
  StartDeletingExtractorState = "START_DELETING_EXTRACTOR_STATE",
57
57
  StartExtractingAttachments = "START_EXTRACTING_ATTACHMENTS",
58
58
  ContinueExtractingAttachments = "CONTINUE_EXTRACTING_ATTACHMENTS",
59
- StartDeletingExtractorAttachmentsState = "START_DELETING_EXTRACTOR_ATTACHMENTS_STATE"
59
+ StartDeletingExtractorAttachmentsState = "START_DELETING_EXTRACTOR_ATTACHMENTS_STATE",
60
+ ContinueExtractingMetadata = "CONTINUE_EXTRACTING_METADATA"
60
61
  }
61
62
  /**
62
63
  * ExtractorEventType is an enum that defines the different types of events that can be sent from the external extractor to ADaaS.
@@ -143,7 +144,9 @@ export declare enum ExtractorEventType {
143
144
  AttachmentExtractionDone = "ATTACHMENT_EXTRACTION_DONE",
144
145
  AttachmentExtractionError = "ATTACHMENT_EXTRACTION_ERROR",
145
146
  ExtractorAttachmentsStateDeletionDone = "EXTRACTOR_ATTACHMENTS_STATE_DELETION_DONE",
146
- ExtractorAttachmentsStateDeletionError = "EXTRACTOR_ATTACHMENTS_STATE_DELETION_ERROR"
147
+ ExtractorAttachmentsStateDeletionError = "EXTRACTOR_ATTACHMENTS_STATE_DELETION_ERROR",
148
+ MetadataExtractionProgress = "METADATA_EXTRACTION_PROGRESS",
149
+ MetadataExtractionDelayed = "METADATA_EXTRACTION_DELAYED"
147
150
  }
148
151
  /**
149
152
  * @deprecated
@@ -303,7 +306,7 @@ export interface EventContext {
303
306
  * Original extraction start timestamp supplied by the platform before the SDK
304
307
  * overrides extract_from.
305
308
  */
306
- platform_extract_from?: string;
309
+ original_extract_from?: string;
307
310
  import_slug: string;
308
311
  initial_sync_scope?: InitialSyncScope;
309
312
  mode: string;