@ontos-ai/knowhere-sdk 2.0.0 → 2.1.2
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/README.md +98 -21
- package/dist/index.d.mts +403 -176
- package/dist/index.d.ts +403 -176
- package/dist/index.js +1970 -51
- package/dist/index.mjs +1966 -50
- package/package.json +2 -2
package/dist/index.d.mts
CHANGED
|
@@ -118,176 +118,6 @@ interface JobResult {
|
|
|
118
118
|
readonly isFailed: boolean;
|
|
119
119
|
}
|
|
120
120
|
|
|
121
|
-
/**
|
|
122
|
-
* Parsing model options
|
|
123
|
-
*/
|
|
124
|
-
type ParsingModel = 'base' | 'advanced';
|
|
125
|
-
/**
|
|
126
|
-
* Document type options
|
|
127
|
-
*/
|
|
128
|
-
type DocType = 'auto' | 'pdf' | 'docx' | 'txt' | 'md';
|
|
129
|
-
/**
|
|
130
|
-
* Parsing configuration parameters
|
|
131
|
-
*/
|
|
132
|
-
interface ParsingParams {
|
|
133
|
-
/** Parsing model to use (default: 'base') */
|
|
134
|
-
model?: ParsingModel;
|
|
135
|
-
/** Enable OCR for scanned documents (default: false) */
|
|
136
|
-
ocrEnabled?: boolean;
|
|
137
|
-
/** Knowledge base directory */
|
|
138
|
-
kbDir?: string;
|
|
139
|
-
/** Document type hint (default: 'auto') */
|
|
140
|
-
docType?: DocType;
|
|
141
|
-
/** Enable smart title parsing (default: false) */
|
|
142
|
-
smartTitleParse?: boolean;
|
|
143
|
-
/** Generate image summaries (default: false) */
|
|
144
|
-
summaryImage?: boolean;
|
|
145
|
-
/** Generate table summaries (default: false) */
|
|
146
|
-
summaryTable?: boolean;
|
|
147
|
-
/** Generate text summaries (default: false) */
|
|
148
|
-
summaryTxt?: boolean;
|
|
149
|
-
/** Additional fragment description */
|
|
150
|
-
addFragDesc?: string;
|
|
151
|
-
}
|
|
152
|
-
/**
|
|
153
|
-
* Webhook configuration
|
|
154
|
-
*/
|
|
155
|
-
interface WebhookConfig {
|
|
156
|
-
/** Webhook URL to notify on job completion */
|
|
157
|
-
url: string;
|
|
158
|
-
}
|
|
159
|
-
/**
|
|
160
|
-
* Client-provided display metadata copied onto the published document.
|
|
161
|
-
*/
|
|
162
|
-
type DocumentMetadata = Record<string, unknown>;
|
|
163
|
-
/**
|
|
164
|
-
* Job creation parameters
|
|
165
|
-
*/
|
|
166
|
-
interface CreateJobParams {
|
|
167
|
-
/** Source type: 'file' for upload, 'url' for remote document */
|
|
168
|
-
sourceType: 'file' | 'url';
|
|
169
|
-
/** Source URL (required if sourceType is 'url') */
|
|
170
|
-
sourceUrl?: string;
|
|
171
|
-
/** File name (required if sourceType is 'file') */
|
|
172
|
-
fileName?: string;
|
|
173
|
-
/** Optional custom data identifier */
|
|
174
|
-
dataId?: string;
|
|
175
|
-
/** Retrieval namespace for the canonical document */
|
|
176
|
-
namespace?: string;
|
|
177
|
-
/** Existing document identifier when updating a published document */
|
|
178
|
-
documentId?: string;
|
|
179
|
-
/** Display metadata to copy onto the published document */
|
|
180
|
-
documentMetadata?: DocumentMetadata;
|
|
181
|
-
/** Parsing configuration */
|
|
182
|
-
parsingParams?: ParsingParams;
|
|
183
|
-
/** Webhook configuration */
|
|
184
|
-
webhook?: WebhookConfig;
|
|
185
|
-
}
|
|
186
|
-
/**
|
|
187
|
-
* File upload parameters
|
|
188
|
-
*/
|
|
189
|
-
interface UploadParams {
|
|
190
|
-
/** File to upload (path, Buffer, Stream, or Uint8Array) */
|
|
191
|
-
file: string | Buffer | ReadStream | Uint8Array;
|
|
192
|
-
/** Upload progress callback */
|
|
193
|
-
onProgress?: (progress: UploadProgress) => void;
|
|
194
|
-
/** Abort signal for cancellation */
|
|
195
|
-
signal?: AbortSignal;
|
|
196
|
-
}
|
|
197
|
-
/**
|
|
198
|
-
* Wait options for polling
|
|
199
|
-
*/
|
|
200
|
-
interface WaitOptions {
|
|
201
|
-
/** Polling interval in milliseconds (default: 10000) */
|
|
202
|
-
pollInterval?: number;
|
|
203
|
-
/** Maximum wait time in milliseconds (default: 1800000 = 30 minutes) */
|
|
204
|
-
pollTimeout?: number;
|
|
205
|
-
/** Progress callback */
|
|
206
|
-
onProgress?: (status: PollProgress) => void;
|
|
207
|
-
/** Abort signal for cancellation */
|
|
208
|
-
signal?: AbortSignal;
|
|
209
|
-
}
|
|
210
|
-
/**
|
|
211
|
-
* Load options for result parsing
|
|
212
|
-
*/
|
|
213
|
-
interface LoadOptions {
|
|
214
|
-
/** Whether to verify ZIP checksum (default: true) */
|
|
215
|
-
verifyChecksum?: boolean;
|
|
216
|
-
}
|
|
217
|
-
/**
|
|
218
|
-
* High-level parse parameters
|
|
219
|
-
*/
|
|
220
|
-
interface ParseParams {
|
|
221
|
-
/** Source URL (mutually exclusive with file) */
|
|
222
|
-
url?: string;
|
|
223
|
-
/** File to parse (path, Buffer, Stream, or Uint8Array) */
|
|
224
|
-
file?: string | Buffer | ReadStream | Uint8Array;
|
|
225
|
-
/** File name (auto-inferred for file paths and fs.ReadStream with a path) */
|
|
226
|
-
fileName?: string;
|
|
227
|
-
/** Parsing model (default: 'base') */
|
|
228
|
-
model?: ParsingModel;
|
|
229
|
-
/** Enable OCR (default: false) */
|
|
230
|
-
ocr?: boolean;
|
|
231
|
-
/** Document type hint */
|
|
232
|
-
docType?: DocType;
|
|
233
|
-
/** Enable smart title parsing */
|
|
234
|
-
smartTitleParse?: boolean;
|
|
235
|
-
/** Generate image summaries */
|
|
236
|
-
summaryImage?: boolean;
|
|
237
|
-
/** Generate table summaries */
|
|
238
|
-
summaryTable?: boolean;
|
|
239
|
-
/** Generate text summaries */
|
|
240
|
-
summaryTxt?: boolean;
|
|
241
|
-
/** Custom data identifier */
|
|
242
|
-
dataId?: string;
|
|
243
|
-
/** Retrieval namespace for the canonical document */
|
|
244
|
-
namespace?: string;
|
|
245
|
-
/** Existing document identifier when updating a published document */
|
|
246
|
-
documentId?: string;
|
|
247
|
-
/** Display metadata to copy onto the published document */
|
|
248
|
-
documentMetadata?: DocumentMetadata;
|
|
249
|
-
/** Additional fragment description */
|
|
250
|
-
addFragDesc?: string;
|
|
251
|
-
/** Knowledge base directory */
|
|
252
|
-
kbDir?: string;
|
|
253
|
-
/** Polling interval in milliseconds */
|
|
254
|
-
pollInterval?: number;
|
|
255
|
-
/** Maximum wait time in milliseconds */
|
|
256
|
-
pollTimeout?: number;
|
|
257
|
-
/** Whether to verify ZIP checksum (default: true) */
|
|
258
|
-
verifyChecksum?: boolean;
|
|
259
|
-
/** Webhook configuration */
|
|
260
|
-
webhook?: WebhookConfig;
|
|
261
|
-
/** Upload progress callback */
|
|
262
|
-
onUploadProgress?: (progress: UploadProgress) => void;
|
|
263
|
-
/** Poll progress callback */
|
|
264
|
-
onPollProgress?: (status: PollProgress) => void;
|
|
265
|
-
/** Abort signal */
|
|
266
|
-
signal?: AbortSignal;
|
|
267
|
-
}
|
|
268
|
-
/**
|
|
269
|
-
* Upload progress information
|
|
270
|
-
*/
|
|
271
|
-
interface UploadProgress {
|
|
272
|
-
/** Bytes uploaded */
|
|
273
|
-
loaded: number;
|
|
274
|
-
/** Total bytes (may be undefined for streams) */
|
|
275
|
-
total?: number;
|
|
276
|
-
/** Upload percentage (0-100) */
|
|
277
|
-
percent: number;
|
|
278
|
-
}
|
|
279
|
-
/**
|
|
280
|
-
* Polling progress information
|
|
281
|
-
*/
|
|
282
|
-
interface PollProgress {
|
|
283
|
-
/** Current job status */
|
|
284
|
-
status: JobStatus;
|
|
285
|
-
/** Elapsed time in seconds */
|
|
286
|
-
elapsedSeconds: number;
|
|
287
|
-
/** Current job result */
|
|
288
|
-
jobResult: JobResult;
|
|
289
|
-
}
|
|
290
|
-
|
|
291
121
|
/**
|
|
292
122
|
* Statistics about the parsed document
|
|
293
123
|
*/
|
|
@@ -469,6 +299,8 @@ interface ImageChunk extends BaseChunk {
|
|
|
469
299
|
type: 'image';
|
|
470
300
|
/** Relative file path in ZIP */
|
|
471
301
|
filePath: string;
|
|
302
|
+
/** Application-owned or server-provided display URL for the image asset. */
|
|
303
|
+
assetUrl?: string;
|
|
472
304
|
/** Image data buffer */
|
|
473
305
|
data: Buffer;
|
|
474
306
|
/** Image format (derived from file extension) */
|
|
@@ -483,6 +315,8 @@ interface TableChunk extends BaseChunk {
|
|
|
483
315
|
type: 'table';
|
|
484
316
|
/** Relative file path in ZIP */
|
|
485
317
|
filePath: string;
|
|
318
|
+
/** Application-owned or server-provided display URL for the table asset. */
|
|
319
|
+
assetUrl?: string;
|
|
486
320
|
/** HTML representation */
|
|
487
321
|
html: string;
|
|
488
322
|
/** Save table HTML to disk */
|
|
@@ -544,6 +378,281 @@ interface ParseResult {
|
|
|
544
378
|
save(directory: string): Promise<string>;
|
|
545
379
|
}
|
|
546
380
|
|
|
381
|
+
type KnowhereAssetStorageBody = Uint8Array;
|
|
382
|
+
interface KnowhereAssetStorageObject {
|
|
383
|
+
readonly key: string;
|
|
384
|
+
readonly body: KnowhereAssetStorageBody;
|
|
385
|
+
readonly contentType: string;
|
|
386
|
+
readonly metadata?: Readonly<Record<string, string>>;
|
|
387
|
+
}
|
|
388
|
+
interface KnowhereAssetStorageHead {
|
|
389
|
+
readonly key: string;
|
|
390
|
+
readonly contentType?: string;
|
|
391
|
+
readonly contentLength?: number;
|
|
392
|
+
readonly metadata?: Readonly<Record<string, string>>;
|
|
393
|
+
readonly url?: string;
|
|
394
|
+
}
|
|
395
|
+
interface KnowhereAssetStorageWriteResult {
|
|
396
|
+
readonly key: string;
|
|
397
|
+
readonly url?: string;
|
|
398
|
+
}
|
|
399
|
+
interface KnowhereAssetStorageAdapter {
|
|
400
|
+
writeObject(input: KnowhereAssetStorageObject): Promise<KnowhereAssetStorageWriteResult>;
|
|
401
|
+
headObject?(key: string): Promise<KnowhereAssetStorageHead | null>;
|
|
402
|
+
getObjectUrl?(key: string): Promise<string | null>;
|
|
403
|
+
}
|
|
404
|
+
interface KnowhereAssetStorageOptions {
|
|
405
|
+
readonly adapter: KnowhereAssetStorageAdapter;
|
|
406
|
+
readonly keyPrefix: string;
|
|
407
|
+
readonly skipExisting?: boolean;
|
|
408
|
+
}
|
|
409
|
+
interface KnowhereAssetStorageResult {
|
|
410
|
+
readonly result: ParseResult;
|
|
411
|
+
readonly assetUrlsByFilePath: Readonly<Record<string, string>>;
|
|
412
|
+
}
|
|
413
|
+
interface ParsedDocumentStorageDocument {
|
|
414
|
+
readonly documentId: string;
|
|
415
|
+
readonly revisionKey: string;
|
|
416
|
+
}
|
|
417
|
+
type ParsedDocumentRevisionParams = ParsedDocumentStorageDocument;
|
|
418
|
+
interface ParsedDocumentObjectParams extends ParsedDocumentRevisionParams {
|
|
419
|
+
readonly path: string;
|
|
420
|
+
}
|
|
421
|
+
interface ParsedDocumentObjectHead extends ParsedDocumentObjectParams {
|
|
422
|
+
readonly contentType?: string;
|
|
423
|
+
readonly contentLength?: number;
|
|
424
|
+
readonly metadata?: Readonly<Record<string, string>>;
|
|
425
|
+
readonly url?: string;
|
|
426
|
+
}
|
|
427
|
+
interface ParsedDocumentObject extends ParsedDocumentObjectHead {
|
|
428
|
+
readonly body: KnowhereAssetStorageBody;
|
|
429
|
+
}
|
|
430
|
+
interface ParsedDocumentWriteObjectParams extends ParsedDocumentObjectParams {
|
|
431
|
+
readonly body: KnowhereAssetStorageBody;
|
|
432
|
+
readonly contentType: string;
|
|
433
|
+
readonly metadata?: Readonly<Record<string, string>>;
|
|
434
|
+
}
|
|
435
|
+
interface ParsedDocumentWriteObjectResult extends ParsedDocumentObjectParams {
|
|
436
|
+
readonly url?: string;
|
|
437
|
+
}
|
|
438
|
+
type ParsedDocumentCommitSource = 'resultZip' | 'remoteReconstruction';
|
|
439
|
+
interface ParsedDocumentCommit {
|
|
440
|
+
readonly version: 1;
|
|
441
|
+
readonly documentId: string;
|
|
442
|
+
readonly revisionKey: string;
|
|
443
|
+
readonly source: ParsedDocumentCommitSource;
|
|
444
|
+
readonly committedAt: string;
|
|
445
|
+
}
|
|
446
|
+
interface ParsedDocumentSyncProgress {
|
|
447
|
+
readonly documentId: string;
|
|
448
|
+
readonly revisionKey: string;
|
|
449
|
+
readonly nextChunkPage?: number;
|
|
450
|
+
readonly status: 'running' | 'completed' | 'failed';
|
|
451
|
+
readonly updatedAt: string;
|
|
452
|
+
readonly error?: string;
|
|
453
|
+
}
|
|
454
|
+
interface ParsedDocumentStorage {
|
|
455
|
+
readObject(params: ParsedDocumentObjectParams): Promise<ParsedDocumentObject | null>;
|
|
456
|
+
writeObject(params: ParsedDocumentWriteObjectParams): Promise<ParsedDocumentWriteObjectResult>;
|
|
457
|
+
headObject?(params: ParsedDocumentObjectParams): Promise<ParsedDocumentObjectHead | null>;
|
|
458
|
+
getObjectUrl?(params: ParsedDocumentObjectParams): Promise<string | null>;
|
|
459
|
+
deletePrefix?(params: ParsedDocumentRevisionParams): Promise<void>;
|
|
460
|
+
readSyncProgress(params: ParsedDocumentRevisionParams): Promise<ParsedDocumentSyncProgress | null>;
|
|
461
|
+
writeSyncProgress(params: ParsedDocumentSyncProgress): Promise<void>;
|
|
462
|
+
}
|
|
463
|
+
interface ParsedDocumentSyncScheduler {
|
|
464
|
+
schedule(task: () => Promise<void>): void | Promise<void>;
|
|
465
|
+
}
|
|
466
|
+
interface ParsedDocumentStorageLimits {
|
|
467
|
+
readonly remotePageSize?: number;
|
|
468
|
+
readonly maxPagesPerSync?: number;
|
|
469
|
+
readonly syncDeadlineMs?: number;
|
|
470
|
+
readonly grepMaxPages?: number;
|
|
471
|
+
readonly grepDeadlineMs?: number;
|
|
472
|
+
readonly outlineMaxPages?: number;
|
|
473
|
+
readonly outlineDeadlineMs?: number;
|
|
474
|
+
}
|
|
475
|
+
interface ParsedDocumentStorageConfig {
|
|
476
|
+
readonly storage: ParsedDocumentStorage;
|
|
477
|
+
readonly scheduler?: ParsedDocumentSyncScheduler;
|
|
478
|
+
readonly limits?: ParsedDocumentStorageLimits;
|
|
479
|
+
}
|
|
480
|
+
|
|
481
|
+
/**
|
|
482
|
+
* Parsing model options
|
|
483
|
+
*/
|
|
484
|
+
type ParsingModel = 'base' | 'advanced';
|
|
485
|
+
/**
|
|
486
|
+
* Document type options
|
|
487
|
+
*/
|
|
488
|
+
type DocType = 'auto' | 'pdf' | 'docx' | 'txt' | 'md';
|
|
489
|
+
/**
|
|
490
|
+
* Parsing configuration parameters
|
|
491
|
+
*/
|
|
492
|
+
interface ParsingParams {
|
|
493
|
+
/** Parsing model to use (default: 'base') */
|
|
494
|
+
model?: ParsingModel;
|
|
495
|
+
/** Enable OCR for scanned documents (default: false) */
|
|
496
|
+
ocrEnabled?: boolean;
|
|
497
|
+
/** Knowledge base directory */
|
|
498
|
+
kbDir?: string;
|
|
499
|
+
/** Document type hint (default: 'auto') */
|
|
500
|
+
docType?: DocType;
|
|
501
|
+
/** Enable smart title parsing (default: false) */
|
|
502
|
+
smartTitleParse?: boolean;
|
|
503
|
+
/** Generate image summaries (default: false) */
|
|
504
|
+
summaryImage?: boolean;
|
|
505
|
+
/** Generate table summaries (default: false) */
|
|
506
|
+
summaryTable?: boolean;
|
|
507
|
+
/** Generate text summaries (default: false) */
|
|
508
|
+
summaryTxt?: boolean;
|
|
509
|
+
/** Additional fragment description */
|
|
510
|
+
addFragDesc?: string;
|
|
511
|
+
}
|
|
512
|
+
/**
|
|
513
|
+
* Webhook configuration
|
|
514
|
+
*/
|
|
515
|
+
interface WebhookConfig {
|
|
516
|
+
/** Webhook URL to notify on job completion */
|
|
517
|
+
url: string;
|
|
518
|
+
}
|
|
519
|
+
/**
|
|
520
|
+
* Client-provided display metadata copied onto the published document.
|
|
521
|
+
*/
|
|
522
|
+
type DocumentMetadata = Record<string, unknown>;
|
|
523
|
+
/**
|
|
524
|
+
* Job creation parameters
|
|
525
|
+
*/
|
|
526
|
+
interface CreateJobParams {
|
|
527
|
+
/** Source type: 'file' for upload, 'url' for remote document */
|
|
528
|
+
sourceType: 'file' | 'url';
|
|
529
|
+
/** Source URL (required if sourceType is 'url') */
|
|
530
|
+
sourceUrl?: string;
|
|
531
|
+
/** File name (required if sourceType is 'file') */
|
|
532
|
+
fileName?: string;
|
|
533
|
+
/** Optional custom data identifier */
|
|
534
|
+
dataId?: string;
|
|
535
|
+
/** Retrieval namespace for the canonical document */
|
|
536
|
+
namespace?: string;
|
|
537
|
+
/** Existing document identifier when updating a published document */
|
|
538
|
+
documentId?: string;
|
|
539
|
+
/** Display metadata to copy onto the published document */
|
|
540
|
+
documentMetadata?: DocumentMetadata;
|
|
541
|
+
/** Parsing configuration */
|
|
542
|
+
parsingParams?: ParsingParams;
|
|
543
|
+
/** Webhook configuration */
|
|
544
|
+
webhook?: WebhookConfig;
|
|
545
|
+
}
|
|
546
|
+
/**
|
|
547
|
+
* File upload parameters
|
|
548
|
+
*/
|
|
549
|
+
interface UploadParams {
|
|
550
|
+
/** File to upload (path, Buffer, Stream, or Uint8Array) */
|
|
551
|
+
file: string | Buffer | ReadStream | Uint8Array;
|
|
552
|
+
/** Upload progress callback */
|
|
553
|
+
onProgress?: (progress: UploadProgress) => void;
|
|
554
|
+
/** Abort signal for cancellation */
|
|
555
|
+
signal?: AbortSignal;
|
|
556
|
+
}
|
|
557
|
+
/**
|
|
558
|
+
* Wait options for polling
|
|
559
|
+
*/
|
|
560
|
+
interface WaitOptions {
|
|
561
|
+
/** Polling interval in milliseconds (default: 10000) */
|
|
562
|
+
pollInterval?: number;
|
|
563
|
+
/** Maximum wait time in milliseconds (default: 1800000 = 30 minutes) */
|
|
564
|
+
pollTimeout?: number;
|
|
565
|
+
/** Progress callback */
|
|
566
|
+
onProgress?: (status: PollProgress) => void;
|
|
567
|
+
/** Abort signal for cancellation */
|
|
568
|
+
signal?: AbortSignal;
|
|
569
|
+
}
|
|
570
|
+
/**
|
|
571
|
+
* Load options for result parsing
|
|
572
|
+
*/
|
|
573
|
+
interface LoadOptions {
|
|
574
|
+
/** Whether to verify ZIP checksum (default: true) */
|
|
575
|
+
verifyChecksum?: boolean;
|
|
576
|
+
}
|
|
577
|
+
/**
|
|
578
|
+
* High-level parse parameters
|
|
579
|
+
*/
|
|
580
|
+
interface ParseParams {
|
|
581
|
+
/** Source URL (mutually exclusive with file) */
|
|
582
|
+
url?: string;
|
|
583
|
+
/** File to parse (path, Buffer, Stream, or Uint8Array) */
|
|
584
|
+
file?: string | Buffer | ReadStream | Uint8Array;
|
|
585
|
+
/** File name (auto-inferred for file paths and fs.ReadStream with a path) */
|
|
586
|
+
fileName?: string;
|
|
587
|
+
/** Parsing model (default: 'base') */
|
|
588
|
+
model?: ParsingModel;
|
|
589
|
+
/** Enable OCR (default: false) */
|
|
590
|
+
ocr?: boolean;
|
|
591
|
+
/** Document type hint */
|
|
592
|
+
docType?: DocType;
|
|
593
|
+
/** Enable smart title parsing */
|
|
594
|
+
smartTitleParse?: boolean;
|
|
595
|
+
/** Generate image summaries */
|
|
596
|
+
summaryImage?: boolean;
|
|
597
|
+
/** Generate table summaries */
|
|
598
|
+
summaryTable?: boolean;
|
|
599
|
+
/** Generate text summaries */
|
|
600
|
+
summaryTxt?: boolean;
|
|
601
|
+
/** Custom data identifier */
|
|
602
|
+
dataId?: string;
|
|
603
|
+
/** Retrieval namespace for the canonical document */
|
|
604
|
+
namespace?: string;
|
|
605
|
+
/** Existing document identifier when updating a published document */
|
|
606
|
+
documentId?: string;
|
|
607
|
+
/** Display metadata to copy onto the published document */
|
|
608
|
+
documentMetadata?: DocumentMetadata;
|
|
609
|
+
/** Additional fragment description */
|
|
610
|
+
addFragDesc?: string;
|
|
611
|
+
/** Knowledge base directory */
|
|
612
|
+
kbDir?: string;
|
|
613
|
+
/** Polling interval in milliseconds */
|
|
614
|
+
pollInterval?: number;
|
|
615
|
+
/** Maximum wait time in milliseconds */
|
|
616
|
+
pollTimeout?: number;
|
|
617
|
+
/** Whether to verify ZIP checksum (default: true) */
|
|
618
|
+
verifyChecksum?: boolean;
|
|
619
|
+
/**
|
|
620
|
+
* Optional asset storage adapter used to copy parsed media and page citation
|
|
621
|
+
* assets into application-owned storage before the result is returned.
|
|
622
|
+
*/
|
|
623
|
+
storageAdapter?: KnowhereAssetStorageOptions;
|
|
624
|
+
/** Webhook configuration */
|
|
625
|
+
webhook?: WebhookConfig;
|
|
626
|
+
/** Upload progress callback */
|
|
627
|
+
onUploadProgress?: (progress: UploadProgress) => void;
|
|
628
|
+
/** Poll progress callback */
|
|
629
|
+
onPollProgress?: (status: PollProgress) => void;
|
|
630
|
+
/** Abort signal */
|
|
631
|
+
signal?: AbortSignal;
|
|
632
|
+
}
|
|
633
|
+
/**
|
|
634
|
+
* Upload progress information
|
|
635
|
+
*/
|
|
636
|
+
interface UploadProgress {
|
|
637
|
+
/** Bytes uploaded */
|
|
638
|
+
loaded: number;
|
|
639
|
+
/** Total bytes (may be undefined for streams) */
|
|
640
|
+
total?: number;
|
|
641
|
+
/** Upload percentage (0-100) */
|
|
642
|
+
percent: number;
|
|
643
|
+
}
|
|
644
|
+
/**
|
|
645
|
+
* Polling progress information
|
|
646
|
+
*/
|
|
647
|
+
interface PollProgress {
|
|
648
|
+
/** Current job status */
|
|
649
|
+
status: JobStatus;
|
|
650
|
+
/** Elapsed time in seconds */
|
|
651
|
+
elapsedSeconds: number;
|
|
652
|
+
/** Current job result */
|
|
653
|
+
jobResult: JobResult;
|
|
654
|
+
}
|
|
655
|
+
|
|
547
656
|
interface HttpClientOptions {
|
|
548
657
|
baseURL: string;
|
|
549
658
|
apiKey?: string;
|
|
@@ -894,14 +1003,14 @@ interface DocumentChunkListParams {
|
|
|
894
1003
|
pageSize?: number;
|
|
895
1004
|
/** Optional chunk type filter */
|
|
896
1005
|
chunkType?: DocumentChunkType;
|
|
897
|
-
/** Set true to include 7-day asset URLs for media
|
|
1006
|
+
/** Set true to include 7-day asset URLs for media and page citation assets */
|
|
898
1007
|
includeAssetUrls?: boolean;
|
|
899
1008
|
}
|
|
900
1009
|
/**
|
|
901
1010
|
* Query parameters for GET /v2/documents/{document_id}/chunks/{document_chunk_id}.
|
|
902
1011
|
*/
|
|
903
1012
|
interface DocumentChunkGetParams {
|
|
904
|
-
/** Set true to include 7-day asset URLs for media
|
|
1013
|
+
/** Set true to include 7-day asset URLs for media and page citation assets */
|
|
905
1014
|
includeAssetUrls?: boolean;
|
|
906
1015
|
}
|
|
907
1016
|
/**
|
|
@@ -930,11 +1039,25 @@ interface DocumentChunk {
|
|
|
930
1039
|
sortOrder: number;
|
|
931
1040
|
/** Chunk metadata returned by the API */
|
|
932
1041
|
metadata: Record<string, unknown>;
|
|
933
|
-
/** 7-day asset URL for media
|
|
1042
|
+
/** 7-day asset URL for media or page citation assets when available */
|
|
934
1043
|
assetUrl?: string | null;
|
|
935
1044
|
/** Chunk creation timestamp */
|
|
936
1045
|
createdAt?: Date;
|
|
937
1046
|
}
|
|
1047
|
+
/**
|
|
1048
|
+
* Response from GET /v2/documents/{document_id}/files/page-citation-source.
|
|
1049
|
+
*/
|
|
1050
|
+
interface DocumentPageCitationSource {
|
|
1051
|
+
documentId: string;
|
|
1052
|
+
namespace?: string;
|
|
1053
|
+
jobId?: string;
|
|
1054
|
+
jobResultId?: string;
|
|
1055
|
+
variant?: string;
|
|
1056
|
+
fileName: string;
|
|
1057
|
+
contentType: string;
|
|
1058
|
+
url: string;
|
|
1059
|
+
expiresAt?: Date;
|
|
1060
|
+
}
|
|
938
1061
|
/**
|
|
939
1062
|
* Response from GET /v2/documents/{document_id}/chunks.
|
|
940
1063
|
*/
|
|
@@ -988,6 +1111,10 @@ declare class Documents extends BaseResource {
|
|
|
988
1111
|
* Get one current-revision chunk for one canonical document.
|
|
989
1112
|
*/
|
|
990
1113
|
getChunk(documentId: string, documentChunkId: string, params?: DocumentChunkGetParams): Promise<DocumentChunkResponse>;
|
|
1114
|
+
/**
|
|
1115
|
+
* Get a signed normalized source-file URL for compatibility and debugging.
|
|
1116
|
+
*/
|
|
1117
|
+
getPageCitationSource(documentId: string): Promise<DocumentPageCitationSource>;
|
|
991
1118
|
/**
|
|
992
1119
|
* Archive one canonical document by ID.
|
|
993
1120
|
*/
|
|
@@ -998,6 +1125,18 @@ declare class Documents extends BaseResource {
|
|
|
998
1125
|
private createChunkGetRequestConfig;
|
|
999
1126
|
}
|
|
1000
1127
|
|
|
1128
|
+
type PageCitationAssetContentType = 'image/png' | 'image/jpeg';
|
|
1129
|
+
type PageCitationAssetSource = 'knowhere-rendered-page-citation-source';
|
|
1130
|
+
interface PageCitationAsset {
|
|
1131
|
+
pageNum: number;
|
|
1132
|
+
artifactRef: string;
|
|
1133
|
+
assetUrl?: string;
|
|
1134
|
+
contentType: PageCitationAssetContentType;
|
|
1135
|
+
width?: number;
|
|
1136
|
+
height?: number;
|
|
1137
|
+
source: PageCitationAssetSource;
|
|
1138
|
+
}
|
|
1139
|
+
|
|
1001
1140
|
type KnowledgeChunkType = DocumentChunkType;
|
|
1002
1141
|
interface KnowledgeParseParams extends ParseParams {
|
|
1003
1142
|
/** Optional stable local identifier for this parsed result copy. */
|
|
@@ -1028,6 +1167,7 @@ interface LocalKnowledgeDocument {
|
|
|
1028
1167
|
interface LocalKnowledgeParseResponse {
|
|
1029
1168
|
document: LocalKnowledgeDocument;
|
|
1030
1169
|
result: ParseResult;
|
|
1170
|
+
assetUrlsByFilePath?: Readonly<Record<string, string>>;
|
|
1031
1171
|
}
|
|
1032
1172
|
interface KnowledgeAsyncParseParams extends ParseParams {
|
|
1033
1173
|
/** Optional stable local identifier to use when this job result is cached later. */
|
|
@@ -1052,11 +1192,20 @@ interface KnowledgeStartupRecoveryResponse {
|
|
|
1052
1192
|
checkedJobs: number;
|
|
1053
1193
|
results: KnowledgeAsyncJobStatusResponse[];
|
|
1054
1194
|
}
|
|
1055
|
-
interface
|
|
1195
|
+
interface KnowledgeLoadJobResultParams {
|
|
1056
1196
|
jobId: string;
|
|
1057
|
-
localDocumentId?: string;
|
|
1058
1197
|
verifyChecksum?: boolean;
|
|
1198
|
+
storageAdapter?: KnowhereAssetStorageOptions;
|
|
1059
1199
|
}
|
|
1200
|
+
interface KnowledgeImportJobResultParams extends KnowledgeLoadJobResultParams {
|
|
1201
|
+
localDocumentId?: string;
|
|
1202
|
+
}
|
|
1203
|
+
interface KnowledgeJobResultResponse {
|
|
1204
|
+
result: ParseResult;
|
|
1205
|
+
assetUrlsByFilePath?: Readonly<Record<string, string>>;
|
|
1206
|
+
}
|
|
1207
|
+
/** @deprecated Use KnowledgeImportJobResultParams with importJobResult instead. */
|
|
1208
|
+
type KnowledgeCacheJobResultParams = KnowledgeImportJobResultParams;
|
|
1060
1209
|
interface KnowledgeCacheDocumentParams {
|
|
1061
1210
|
/** Canonical Knowhere document identifier from published retrieval results. */
|
|
1062
1211
|
documentId: string;
|
|
@@ -1070,6 +1219,16 @@ interface KnowledgeDocumentReference {
|
|
|
1070
1219
|
documentId?: string;
|
|
1071
1220
|
/** Server parse job identifier whose completed result should be cached before reading. */
|
|
1072
1221
|
jobId?: string;
|
|
1222
|
+
/** Current parsed revision key when the caller already has it. */
|
|
1223
|
+
revisionKey?: string;
|
|
1224
|
+
}
|
|
1225
|
+
type KnowledgeParsedStorageOptions = ParsedDocumentStorageConfig;
|
|
1226
|
+
type KnowledgeSyncParsedDocumentParams = KnowledgeDocumentReference;
|
|
1227
|
+
interface KnowledgeSyncParsedDocumentResponse {
|
|
1228
|
+
documentId: string;
|
|
1229
|
+
revisionKey: string;
|
|
1230
|
+
completed: boolean;
|
|
1231
|
+
commit?: ParsedDocumentCommit;
|
|
1073
1232
|
}
|
|
1074
1233
|
interface KnowledgeSection {
|
|
1075
1234
|
sectionPath: string;
|
|
@@ -1088,8 +1247,12 @@ interface KnowledgeOutline {
|
|
|
1088
1247
|
typeCounts: Record<KnowledgeChunkType, number>;
|
|
1089
1248
|
sections: KnowledgeSection[];
|
|
1090
1249
|
sectionTree: KnowledgeSection[];
|
|
1250
|
+
truncated?: boolean;
|
|
1251
|
+
continuationCursor?: string;
|
|
1091
1252
|
}
|
|
1092
1253
|
interface KnowledgeReadParams extends KnowledgeDocumentReference {
|
|
1254
|
+
page?: number;
|
|
1255
|
+
pageSize?: number;
|
|
1093
1256
|
sectionPath?: string;
|
|
1094
1257
|
startChunk?: number;
|
|
1095
1258
|
endChunk?: number;
|
|
@@ -1109,6 +1272,7 @@ interface KnowledgeReadChunk {
|
|
|
1109
1272
|
sectionPath: string;
|
|
1110
1273
|
sourceChunkPath: string;
|
|
1111
1274
|
filePath?: string;
|
|
1275
|
+
assetUrl?: string;
|
|
1112
1276
|
pageNumbers?: number[];
|
|
1113
1277
|
metadata: Record<string, unknown>;
|
|
1114
1278
|
}
|
|
@@ -1116,9 +1280,14 @@ interface KnowledgeReadResponse {
|
|
|
1116
1280
|
document: LocalKnowledgeDocument;
|
|
1117
1281
|
chunks: KnowledgeReadChunk[];
|
|
1118
1282
|
nextChunk?: number;
|
|
1283
|
+
page?: number;
|
|
1284
|
+
pageSize?: number;
|
|
1285
|
+
totalChunks?: number;
|
|
1286
|
+
totalPages?: number;
|
|
1119
1287
|
}
|
|
1120
1288
|
interface KnowledgeGrepParams extends KnowledgeDocumentReference {
|
|
1121
1289
|
pattern: string;
|
|
1290
|
+
continuationCursor?: string;
|
|
1122
1291
|
isRegex?: boolean;
|
|
1123
1292
|
isCaseSensitive?: boolean;
|
|
1124
1293
|
maxResults?: number;
|
|
@@ -1142,6 +1311,7 @@ interface KnowledgeGrepResponse {
|
|
|
1142
1311
|
matches: KnowledgeGrepMatch[];
|
|
1143
1312
|
scannedChunks: number;
|
|
1144
1313
|
truncated: boolean;
|
|
1314
|
+
continuationCursor?: string;
|
|
1145
1315
|
}
|
|
1146
1316
|
interface KnowledgeSearchParams {
|
|
1147
1317
|
query: string;
|
|
@@ -1180,15 +1350,27 @@ interface KnowledgeSearchResult {
|
|
|
1180
1350
|
declare class Knowledge {
|
|
1181
1351
|
private readonly client;
|
|
1182
1352
|
private readonly store;
|
|
1353
|
+
private readonly cacheDirectory;
|
|
1354
|
+
private readonly parsedStorageConfig;
|
|
1183
1355
|
constructor(client: Knowhere, options?: {
|
|
1184
1356
|
cacheDirectory?: string;
|
|
1357
|
+
parsedStorage?: ParsedDocumentStorageConfig;
|
|
1185
1358
|
});
|
|
1186
1359
|
withCacheDirectory(cacheDirectory: string): Knowledge;
|
|
1360
|
+
withParsedStorage(parsedStorage: KnowledgeParsedStorageOptions): Knowledge;
|
|
1361
|
+
parseToLocalCache(params: KnowledgeParseParams): Promise<LocalKnowledgeParseResponse>;
|
|
1362
|
+
/** @deprecated Use parseToLocalCache for local cache imports or the top-level client.parse for server-safe parsing. */
|
|
1187
1363
|
parse(params: KnowledgeParseParams): Promise<LocalKnowledgeParseResponse>;
|
|
1188
1364
|
startParse(params: KnowledgeAsyncParseParams): Promise<KnowledgeAsyncParseResponse>;
|
|
1189
1365
|
getJobStatus(jobId: string): Promise<KnowledgeAsyncJobStatusResponse>;
|
|
1190
1366
|
recoverPendingAsyncParseJobs(): Promise<KnowledgeStartupRecoveryResponse>;
|
|
1367
|
+
loadJobResult(params: KnowledgeLoadJobResultParams): Promise<KnowledgeJobResultResponse>;
|
|
1368
|
+
importJobResult(params: KnowledgeImportJobResultParams): Promise<LocalKnowledgeParseResponse>;
|
|
1369
|
+
/** @deprecated Use importJobResult for local cache imports or loadJobResult for server-safe loads. */
|
|
1191
1370
|
cacheJobResult(params: KnowledgeCacheJobResultParams): Promise<LocalKnowledgeParseResponse>;
|
|
1371
|
+
syncParsedDocument(params: KnowledgeSyncParsedDocumentParams): Promise<KnowledgeSyncParsedDocumentResponse>;
|
|
1372
|
+
private syncParseResultToConfiguredStorage;
|
|
1373
|
+
private parseWithoutClientStorageAdapter;
|
|
1192
1374
|
cacheDocument(params: KnowledgeCacheDocumentParams): Promise<LocalKnowledgeParseResponse>;
|
|
1193
1375
|
private resolveAsyncCache;
|
|
1194
1376
|
listDocuments(): Promise<LocalKnowledgeDocument[]>;
|
|
@@ -1198,6 +1380,19 @@ declare class Knowledge {
|
|
|
1198
1380
|
search(params: KnowledgeSearchParams): Promise<KnowledgeSearchResponse>;
|
|
1199
1381
|
private resolveSearchDocuments;
|
|
1200
1382
|
private loadReadableResult;
|
|
1383
|
+
private readRemoteDocumentChunks;
|
|
1384
|
+
private scanRemoteDocumentChunks;
|
|
1385
|
+
private grepRemoteDocumentChunks;
|
|
1386
|
+
private getRemoteDocumentOutline;
|
|
1387
|
+
private tryReadChunksFromParsedStorage;
|
|
1388
|
+
private tryGrepChunksFromParsedStorage;
|
|
1389
|
+
private tryReadOutlineFromParsedStorage;
|
|
1390
|
+
private readCommittedParsedStorageResult;
|
|
1391
|
+
private scheduleParsedStorageSync;
|
|
1392
|
+
private syncRemoteDocumentToParsedStorage;
|
|
1393
|
+
private syncRemoteDocumentChunksToParsedStorage;
|
|
1394
|
+
private readRemoteReconstructionChunks;
|
|
1395
|
+
private requireParsedStorageConfig;
|
|
1201
1396
|
private findLocalDocumentByRemoteId;
|
|
1202
1397
|
private getPublishedDocumentJobId;
|
|
1203
1398
|
}
|
|
@@ -1295,6 +1490,38 @@ declare class Knowhere {
|
|
|
1295
1490
|
|
|
1296
1491
|
declare const VERSION = "0.1.0";
|
|
1297
1492
|
|
|
1493
|
+
declare function storeParseResultAssets(result: ParseResult, options?: KnowhereAssetStorageOptions): Promise<KnowhereAssetStorageResult>;
|
|
1494
|
+
|
|
1495
|
+
declare class DiskParsedDocumentStorage implements ParsedDocumentStorage {
|
|
1496
|
+
private readonly rootDirectory;
|
|
1497
|
+
constructor(rootDirectory: string);
|
|
1498
|
+
readObject(params: ParsedDocumentObjectParams): Promise<ParsedDocumentObject | null>;
|
|
1499
|
+
writeObject(params: ParsedDocumentWriteObjectParams): Promise<ParsedDocumentWriteObjectResult>;
|
|
1500
|
+
headObject(params: ParsedDocumentObjectParams): Promise<ParsedDocumentObjectHead | null>;
|
|
1501
|
+
getObjectUrl(params: ParsedDocumentObjectParams): Promise<string | null>;
|
|
1502
|
+
deletePrefix(params: ParsedDocumentRevisionParams): Promise<void>;
|
|
1503
|
+
readSyncProgress(params: ParsedDocumentRevisionParams): Promise<ParsedDocumentSyncProgress | null>;
|
|
1504
|
+
writeSyncProgress(params: ParsedDocumentSyncProgress): Promise<void>;
|
|
1505
|
+
private getObjectPath;
|
|
1506
|
+
private getRevisionDirectory;
|
|
1507
|
+
private readOptionalJson;
|
|
1508
|
+
private readObjectMetadata;
|
|
1509
|
+
private writeObjectMetadata;
|
|
1510
|
+
}
|
|
1511
|
+
|
|
1512
|
+
declare function syncParseResultToParsedDocumentStorage(params: {
|
|
1513
|
+
readonly result: ParseResult;
|
|
1514
|
+
readonly storage: ParsedDocumentStorage;
|
|
1515
|
+
readonly documentId?: string;
|
|
1516
|
+
readonly revisionKey?: string;
|
|
1517
|
+
readonly source?: ParsedDocumentCommitSource;
|
|
1518
|
+
}): Promise<{
|
|
1519
|
+
readonly result: ParseResult;
|
|
1520
|
+
readonly documentId: string;
|
|
1521
|
+
readonly revisionKey: string;
|
|
1522
|
+
readonly commit: ParsedDocumentCommit;
|
|
1523
|
+
}>;
|
|
1524
|
+
|
|
1298
1525
|
/**
|
|
1299
1526
|
* Base error class for all Knowhere SDK errors
|
|
1300
1527
|
*/
|
|
@@ -1424,4 +1651,4 @@ declare class JobFailedError extends KnowhereError {
|
|
|
1424
1651
|
constructor(message: string, code: string, jobResult: JobResult);
|
|
1425
1652
|
}
|
|
1426
1653
|
|
|
1427
|
-
export { APIError, type AuthTokenProvider, AuthenticationError, BadRequestError, type BaseChunk, ChecksumError, type Chunk, ConflictError, type CreateJobParams, type DocType, type Document, type DocumentChunk, type DocumentChunkGetParams, type DocumentChunkListParams, type DocumentChunkListResponse, type DocumentChunkPagination, type DocumentChunkResponse, type DocumentChunkType, type DocumentListPagination, type DocumentListParams, type DocumentListResponse, type DocumentMetadata, Documents, type FileIndex, GatewayTimeoutError, type ImageChunk, InternalServerError, InvalidStateError, type Job, type JobError, JobFailedError, type JobResult, type JobStatus, Jobs, Knowhere, KnowhereError, type KnowhereOptions, Knowledge, type KnowledgeAsyncJobStatusResponse, type KnowledgeAsyncParseParams, type KnowledgeAsyncParseResponse, type KnowledgeCacheDocumentParams, type KnowledgeCacheJobResultParams, type KnowledgeChunkType, type KnowledgeDocumentReference, type KnowledgeGrepMatch, type KnowledgeGrepParams, type KnowledgeGrepResponse, type KnowledgeOutline, type KnowledgeParseParams, type KnowledgeReadChunk, type KnowledgeReadParams, type KnowledgeReadResponse, type KnowledgeSearchParams, type KnowledgeSearchReference, type KnowledgeSearchResponse, type KnowledgeSearchResult, type KnowledgeSection, type KnowledgeStartupRecoveryResponse, type LoadOptions, type LocalKnowledgeDocument, type LocalKnowledgeParseResponse, LocalKnowledgeStore, type Manifest, NetworkError, NotFoundError, type PageChunk, type ParseParams, type ParseResult, type ParsingModel, type ParsingParams, PaymentRequiredError, PermissionDeniedError, type PollProgress, PollingTimeoutError, RateLimitError, Retrieval, type RetrievalChannel, type RetrievalChunkType, type RetrievalFilterMode, type RetrievalQueryParams, type RetrievalQueryResponse, type RetrievalReferencedChunk, type RetrievalResult, type RetrievalSectionExclusion, type RetrievalSource, ServiceUnavailableError, type Statistics, type TableChunk, type TextChunk, TimeoutError, type UploadParams, type UploadProgress, VERSION, ValidationError, type WaitOptions, type WebhookConfig, Knowhere as default };
|
|
1654
|
+
export { APIError, type AuthTokenProvider, AuthenticationError, BadRequestError, type BaseChunk, ChecksumError, type Chunk, ConflictError, type CreateJobParams, DiskParsedDocumentStorage, type DocType, type Document, type DocumentChunk, type DocumentChunkGetParams, type DocumentChunkListParams, type DocumentChunkListResponse, type DocumentChunkPagination, type DocumentChunkResponse, type DocumentChunkType, type DocumentListPagination, type DocumentListParams, type DocumentListResponse, type DocumentMetadata, type DocumentPageCitationSource, Documents, type FileIndex, GatewayTimeoutError, type ImageChunk, InternalServerError, InvalidStateError, type Job, type JobError, JobFailedError, type JobResult, type JobStatus, Jobs, Knowhere, type KnowhereAssetStorageAdapter, type KnowhereAssetStorageBody, type KnowhereAssetStorageHead, type KnowhereAssetStorageObject, type KnowhereAssetStorageOptions, type KnowhereAssetStorageResult, type KnowhereAssetStorageWriteResult, KnowhereError, type KnowhereOptions, Knowledge, type KnowledgeAsyncJobStatusResponse, type KnowledgeAsyncParseParams, type KnowledgeAsyncParseResponse, type KnowledgeCacheDocumentParams, type KnowledgeCacheJobResultParams, type KnowledgeChunkType, type KnowledgeDocumentReference, type KnowledgeGrepMatch, type KnowledgeGrepParams, type KnowledgeGrepResponse, type KnowledgeImportJobResultParams, type KnowledgeJobResultResponse, type KnowledgeLoadJobResultParams, type KnowledgeOutline, type KnowledgeParseParams, type KnowledgeParsedStorageOptions, type KnowledgeReadChunk, type KnowledgeReadParams, type KnowledgeReadResponse, type KnowledgeSearchParams, type KnowledgeSearchReference, type KnowledgeSearchResponse, type KnowledgeSearchResult, type KnowledgeSection, type KnowledgeStartupRecoveryResponse, type KnowledgeSyncParsedDocumentParams, type KnowledgeSyncParsedDocumentResponse, type LoadOptions, type LocalKnowledgeDocument, type LocalKnowledgeParseResponse, LocalKnowledgeStore, type Manifest, NetworkError, NotFoundError, type PageChunk, type PageCitationAsset, type PageCitationAssetContentType, type PageCitationAssetSource, type ParseParams, type ParseResult, type ParsedDocumentCommit, type ParsedDocumentCommitSource, type ParsedDocumentObject, type ParsedDocumentObjectHead, type ParsedDocumentObjectParams, type ParsedDocumentRevisionParams, type ParsedDocumentStorage, type ParsedDocumentStorageConfig, type ParsedDocumentStorageDocument, type ParsedDocumentStorageLimits, type ParsedDocumentSyncProgress, type ParsedDocumentSyncScheduler, type ParsedDocumentWriteObjectParams, type ParsedDocumentWriteObjectResult, type ParsingModel, type ParsingParams, PaymentRequiredError, PermissionDeniedError, type PollProgress, PollingTimeoutError, RateLimitError, Retrieval, type RetrievalChannel, type RetrievalChunkType, type RetrievalFilterMode, type RetrievalQueryParams, type RetrievalQueryResponse, type RetrievalReferencedChunk, type RetrievalResult, type RetrievalSectionExclusion, type RetrievalSource, ServiceUnavailableError, type Statistics, type TableChunk, type TextChunk, TimeoutError, type UploadParams, type UploadProgress, VERSION, ValidationError, type WaitOptions, type WebhookConfig, Knowhere as default, storeParseResultAssets, syncParseResultToParsedDocumentStorage };
|