vectra 0.9.0 → 0.10.0

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 (74) hide show
  1. package/README.md +10 -10
  2. package/package.json +7 -6
  3. package/src/ItemSelector.ts +7 -1
  4. package/src/LocalDocumentIndex.ts +7 -1
  5. package/src/LocalDocumentResult.ts +70 -22
  6. package/src/LocalIndex.ts +70 -3
  7. package/src/internals/Colorize.ts +3 -3
  8. package/src/internals/wink-bm25-text-search.d.ts +4 -0
  9. package/src/types.ts +1 -0
  10. package/src/vectra-cli.ts +13 -2
  11. package/lib/FileFetcher.d.ts +0 -5
  12. package/lib/FileFetcher.d.ts.map +0 -1
  13. package/lib/FileFetcher.js +0 -69
  14. package/lib/FileFetcher.js.map +0 -1
  15. package/lib/GPT3Tokenizer.d.ts +0 -9
  16. package/lib/GPT3Tokenizer.d.ts.map +0 -1
  17. package/lib/GPT3Tokenizer.js +0 -17
  18. package/lib/GPT3Tokenizer.js.map +0 -1
  19. package/lib/ItemSelector.d.ts +0 -41
  20. package/lib/ItemSelector.d.ts.map +0 -1
  21. package/lib/ItemSelector.js +0 -162
  22. package/lib/ItemSelector.js.map +0 -1
  23. package/lib/LocalDocument.d.ts +0 -54
  24. package/lib/LocalDocument.d.ts.map +0 -1
  25. package/lib/LocalDocument.js +0 -146
  26. package/lib/LocalDocument.js.map +0 -1
  27. package/lib/LocalDocumentIndex.d.ts +0 -128
  28. package/lib/LocalDocumentIndex.d.ts.map +0 -1
  29. package/lib/LocalDocumentIndex.js +0 -446
  30. package/lib/LocalDocumentIndex.js.map +0 -1
  31. package/lib/LocalDocumentResult.d.ts +0 -45
  32. package/lib/LocalDocumentResult.d.ts.map +0 -1
  33. package/lib/LocalDocumentResult.js +0 -282
  34. package/lib/LocalDocumentResult.js.map +0 -1
  35. package/lib/LocalIndex.d.ts +0 -136
  36. package/lib/LocalIndex.d.ts.map +0 -1
  37. package/lib/LocalIndex.js +0 -413
  38. package/lib/LocalIndex.js.map +0 -1
  39. package/lib/OpenAIEmbeddings.d.ts +0 -126
  40. package/lib/OpenAIEmbeddings.d.ts.map +0 -1
  41. package/lib/OpenAIEmbeddings.js +0 -174
  42. package/lib/OpenAIEmbeddings.js.map +0 -1
  43. package/lib/TextSplitter.d.ts +0 -20
  44. package/lib/TextSplitter.d.ts.map +0 -1
  45. package/lib/TextSplitter.js +0 -537
  46. package/lib/TextSplitter.js.map +0 -1
  47. package/lib/WebFetcher.d.ts +0 -15
  48. package/lib/WebFetcher.d.ts.map +0 -1
  49. package/lib/WebFetcher.js +0 -224
  50. package/lib/WebFetcher.js.map +0 -1
  51. package/lib/index.d.ts +0 -12
  52. package/lib/index.d.ts.map +0 -1
  53. package/lib/index.js +0 -28
  54. package/lib/index.js.map +0 -1
  55. package/lib/internals/Colorize.d.ts +0 -14
  56. package/lib/internals/Colorize.d.ts.map +0 -1
  57. package/lib/internals/Colorize.js +0 -64
  58. package/lib/internals/Colorize.js.map +0 -1
  59. package/lib/internals/index.d.ts +0 -3
  60. package/lib/internals/index.d.ts.map +0 -1
  61. package/lib/internals/index.js +0 -19
  62. package/lib/internals/index.js.map +0 -1
  63. package/lib/internals/types.d.ts +0 -43
  64. package/lib/internals/types.d.ts.map +0 -1
  65. package/lib/internals/types.js +0 -3
  66. package/lib/internals/types.js.map +0 -1
  67. package/lib/types.d.ts +0 -145
  68. package/lib/types.d.ts.map +0 -1
  69. package/lib/types.js +0 -3
  70. package/lib/types.js.map +0 -1
  71. package/lib/vectra-cli.d.ts +0 -2
  72. package/lib/vectra-cli.d.ts.map +0 -1
  73. package/lib/vectra-cli.js +0 -303
  74. package/lib/vectra-cli.js.map +0 -1
@@ -1,537 +0,0 @@
1
- "use strict";
2
- Object.defineProperty(exports, "__esModule", { value: true });
3
- exports.TextSplitter = void 0;
4
- const GPT3Tokenizer_1 = require("./GPT3Tokenizer");
5
- const ALPHANUMERIC_CHARS = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789';
6
- class TextSplitter {
7
- constructor(config) {
8
- this._config = Object.assign({
9
- keepSeparators: false,
10
- chunkSize: 400,
11
- chunkOverlap: 40,
12
- }, config);
13
- // Create a default tokenizer if none is provided
14
- if (!this._config.tokenizer) {
15
- this._config.tokenizer = new GPT3Tokenizer_1.GPT3Tokenizer();
16
- }
17
- // Use default separators if none are provided
18
- if (!this._config.separators || this._config.separators.length === 0) {
19
- this._config.separators = this.getSeparators(this._config.docType);
20
- }
21
- // Validate the config settings
22
- if (this._config.chunkSize < 1) {
23
- throw new Error("chunkSize must be >= 1");
24
- }
25
- else if (this._config.chunkOverlap < 0) {
26
- throw new Error("chunkOverlap must be >= 0");
27
- }
28
- else if (this._config.chunkOverlap > this._config.chunkSize) {
29
- throw new Error("chunkOverlap must be <= chunkSize");
30
- }
31
- }
32
- split(text) {
33
- // Get basic chunks
34
- const chunks = this.recursiveSplit(text, this._config.separators, 0);
35
- const that = this;
36
- function getOverlapTokens(tokens) {
37
- if (tokens != undefined) {
38
- const len = tokens.length > that._config.chunkOverlap ? that._config.chunkOverlap : tokens.length;
39
- return tokens.slice(0, len);
40
- }
41
- else {
42
- return [];
43
- }
44
- }
45
- // Add overlap tokens and text to the start and end of each chunk
46
- if (this._config.chunkOverlap > 0) {
47
- for (let i = 1; i < chunks.length; i++) {
48
- const previousChunk = chunks[i - 1];
49
- const chunk = chunks[i];
50
- const nextChunk = i < chunks.length - 1 ? chunks[i + 1] : undefined;
51
- chunk.startOverlap = getOverlapTokens(previousChunk.tokens.reverse()).reverse();
52
- chunk.endOverlap = getOverlapTokens(nextChunk === null || nextChunk === void 0 ? void 0 : nextChunk.tokens);
53
- }
54
- }
55
- return chunks;
56
- }
57
- recursiveSplit(text, separators, startPos) {
58
- const chunks = [];
59
- if (text.length > 0) {
60
- // Split text into parts
61
- let parts;
62
- let separator = '';
63
- const nextSeparators = separators.length > 1 ? separators.slice(1) : [];
64
- if (separators.length > 0) {
65
- // Split by separator
66
- separator = separators[0];
67
- parts = separator == ' ' ? this.splitBySpaces(text) : text.split(separator);
68
- }
69
- else {
70
- // Cut text in half
71
- const half = Math.floor(text.length / 2);
72
- parts = [text.substring(0, half), text.substring(half)];
73
- }
74
- // Iterate over parts
75
- for (let i = 0; i < parts.length; i++) {
76
- const lastChunk = (i === parts.length - 1);
77
- // Get chunk text and endPos
78
- let chunk = parts[i];
79
- const endPos = (startPos + (chunk.length - 1)) + (lastChunk ? 0 : separator.length);
80
- if (this._config.keepSeparators && !lastChunk) {
81
- chunk += separator;
82
- }
83
- // Ensure chunk contains text
84
- if (!this.containsAlphanumeric(chunk)) {
85
- continue;
86
- }
87
- // Optimization to avoid encoding really large chunks
88
- if (chunk.length / 6 > this._config.chunkSize) {
89
- // Break the text into smaller chunks
90
- const subChunks = this.recursiveSplit(chunk, nextSeparators, startPos);
91
- chunks.push(...subChunks);
92
- }
93
- else {
94
- // Encode chunk text
95
- const tokens = this._config.tokenizer.encode(chunk);
96
- if (tokens.length > this._config.chunkSize) {
97
- // Break the text into smaller chunks
98
- const subChunks = this.recursiveSplit(chunk, nextSeparators, startPos);
99
- chunks.push(...subChunks);
100
- }
101
- else {
102
- // Append chunk to output
103
- chunks.push({
104
- text: chunk,
105
- tokens: tokens,
106
- startPos: startPos,
107
- endPos: endPos,
108
- startOverlap: [],
109
- endOverlap: [],
110
- });
111
- }
112
- }
113
- // Update startPos
114
- startPos = endPos + 1;
115
- }
116
- }
117
- return this.combineChunks(chunks);
118
- }
119
- combineChunks(chunks) {
120
- const combinedChunks = [];
121
- let currentChunk;
122
- let currentLength = 0;
123
- const separator = this._config.keepSeparators ? '' : ' ';
124
- for (let i = 0; i < chunks.length; i++) {
125
- const chunk = chunks[i];
126
- if (currentChunk) {
127
- const length = currentChunk.tokens.length + chunk.tokens.length;
128
- if (length > this._config.chunkSize) {
129
- combinedChunks.push(currentChunk);
130
- currentChunk = chunk;
131
- currentLength = chunk.tokens.length;
132
- }
133
- else {
134
- currentChunk.text += separator + chunk.text;
135
- currentChunk.endPos = chunk.endPos;
136
- currentChunk.tokens.push(...chunk.tokens);
137
- currentLength += chunk.tokens.length;
138
- }
139
- }
140
- else {
141
- currentChunk = chunk;
142
- currentLength = chunk.tokens.length;
143
- }
144
- }
145
- if (currentChunk) {
146
- combinedChunks.push(currentChunk);
147
- }
148
- return combinedChunks;
149
- }
150
- containsAlphanumeric(text) {
151
- for (let i = 0; i < text.length; i++) {
152
- if (ALPHANUMERIC_CHARS.includes(text[i])) {
153
- return true;
154
- }
155
- }
156
- return false;
157
- }
158
- splitBySpaces(text) {
159
- // Split text by tokens and return parts
160
- const parts = [];
161
- let tokens = this._config.tokenizer.encode(text);
162
- do {
163
- if (tokens.length <= this._config.chunkSize) {
164
- parts.push(this._config.tokenizer.decode(tokens));
165
- break;
166
- }
167
- else {
168
- const span = tokens.splice(0, this._config.chunkSize);
169
- parts.push(this._config.tokenizer.decode(span));
170
- }
171
- } while (true);
172
- return parts;
173
- }
174
- getSeparators(docType) {
175
- switch (docType !== null && docType !== void 0 ? docType : '') {
176
- case "cpp":
177
- return [
178
- // Split along class definitions
179
- "\nclass ",
180
- // Split along function definitions
181
- "\nvoid ",
182
- "\nint ",
183
- "\nfloat ",
184
- "\ndouble ",
185
- // Split along control flow statements
186
- "\nif ",
187
- "\nfor ",
188
- "\nwhile ",
189
- "\nswitch ",
190
- "\ncase ",
191
- // Split by the normal type of lines
192
- "\n\n",
193
- "\n",
194
- " "
195
- ];
196
- case "go":
197
- return [
198
- // Split along function definitions
199
- "\nfunc ",
200
- "\nvar ",
201
- "\nconst ",
202
- "\ntype ",
203
- // Split along control flow statements
204
- "\nif ",
205
- "\nfor ",
206
- "\nswitch ",
207
- "\ncase ",
208
- // Split by the normal type of lines
209
- "\n\n",
210
- "\n",
211
- " "
212
- ];
213
- case "java":
214
- case "c#":
215
- case "csharp":
216
- case "cs":
217
- case "ts":
218
- case "tsx":
219
- case "typescript":
220
- return [
221
- // split along regions
222
- "// LLM-REGION",
223
- "/* LLM-REGION",
224
- "/** LLM-REGION",
225
- // Split along class definitions
226
- "\nclass ",
227
- // Split along method definitions
228
- "\npublic ",
229
- "\nprotected ",
230
- "\nprivate ",
231
- "\nstatic ",
232
- // Split along control flow statements
233
- "\nif ",
234
- "\nfor ",
235
- "\nwhile ",
236
- "\nswitch ",
237
- "\ncase ",
238
- // Split by the normal type of lines
239
- "\n\n",
240
- "\n",
241
- " "
242
- ];
243
- case "js":
244
- case "jsx":
245
- case "javascript":
246
- return [
247
- // split along regions
248
- "// LLM-REGION",
249
- "/* LLM-REGION",
250
- "/** LLM-REGION",
251
- // Split along class definitions
252
- "\nclass ",
253
- // Split along function definitions
254
- "\nfunction ",
255
- "\nconst ",
256
- "\nlet ",
257
- "\nvar ",
258
- "\nclass ",
259
- // Split along control flow statements
260
- "\nif ",
261
- "\nfor ",
262
- "\nwhile ",
263
- "\nswitch ",
264
- "\ncase ",
265
- "\ndefault ",
266
- // Split by the normal type of lines
267
- "\n\n",
268
- "\n",
269
- " "
270
- ];
271
- case "php":
272
- return [
273
- // Split along function definitions
274
- "\nfunction ",
275
- // Split along class definitions
276
- "\nclass ",
277
- // Split along control flow statements
278
- "\nif ",
279
- "\nforeach ",
280
- "\nwhile ",
281
- "\ndo ",
282
- "\nswitch ",
283
- "\ncase ",
284
- // Split by the normal type of lines
285
- "\n\n",
286
- "\n",
287
- " "
288
- ];
289
- case "proto":
290
- return [
291
- // Split along message definitions
292
- "\nmessage ",
293
- // Split along service definitions
294
- "\nservice ",
295
- // Split along enum definitions
296
- "\nenum ",
297
- // Split along option definitions
298
- "\noption ",
299
- // Split along import statements
300
- "\nimport ",
301
- // Split along syntax declarations
302
- "\nsyntax ",
303
- // Split by the normal type of lines
304
- "\n\n",
305
- "\n",
306
- " "
307
- ];
308
- case "python":
309
- case "py":
310
- return [
311
- // First, try to split along class definitions
312
- "\nclass ",
313
- "\ndef ",
314
- "\n\tdef ",
315
- // Now split by the normal type of lines
316
- "\n\n",
317
- "\n",
318
- " "
319
- ];
320
- case "rst":
321
- return [
322
- // Split along section titles
323
- "\n===\n",
324
- "\n---\n",
325
- "\n***\n",
326
- // Split along directive markers
327
- "\n.. ",
328
- // Split by the normal type of lines
329
- "\n\n",
330
- "\n",
331
- " "
332
- ];
333
- case "ruby":
334
- return [
335
- // Split along method definitions
336
- "\ndef ",
337
- "\nclass ",
338
- // Split along control flow statements
339
- "\nif ",
340
- "\nunless ",
341
- "\nwhile ",
342
- "\nfor ",
343
- "\ndo ",
344
- "\nbegin ",
345
- "\nrescue ",
346
- // Split by the normal type of lines
347
- "\n\n",
348
- "\n",
349
- " "
350
- ];
351
- case "rust":
352
- return [
353
- // Split along function definitions
354
- "\nfn ",
355
- "\nconst ",
356
- "\nlet ",
357
- // Split along control flow statements
358
- "\nif ",
359
- "\nwhile ",
360
- "\nfor ",
361
- "\nloop ",
362
- "\nmatch ",
363
- "\nconst ",
364
- // Split by the normal type of lines
365
- "\n\n",
366
- "\n",
367
- " "
368
- ];
369
- case "scala":
370
- return [
371
- // Split along class definitions
372
- "\nclass ",
373
- "\nobject ",
374
- // Split along method definitions
375
- "\ndef ",
376
- "\nval ",
377
- "\nvar ",
378
- // Split along control flow statements
379
- "\nif ",
380
- "\nfor ",
381
- "\nwhile ",
382
- "\nmatch ",
383
- "\ncase ",
384
- // Split by the normal type of lines
385
- "\n\n",
386
- "\n",
387
- " "
388
- ];
389
- case "swift":
390
- return [
391
- // Split along function definitions
392
- "\nfunc ",
393
- // Split along class definitions
394
- "\nclass ",
395
- "\nstruct ",
396
- "\nenum ",
397
- // Split along control flow statements
398
- "\nif ",
399
- "\nfor ",
400
- "\nwhile ",
401
- "\ndo ",
402
- "\nswitch ",
403
- "\ncase ",
404
- // Split by the normal type of lines
405
- "\n\n",
406
- "\n",
407
- " "
408
- ];
409
- case "md":
410
- case "markdown":
411
- return [
412
- // First, try to split along Markdown headings (starting with level 2)
413
- "\n## ",
414
- "\n### ",
415
- "\n#### ",
416
- "\n##### ",
417
- "\n###### ",
418
- // Note the alternative syntax for headings (below) is not handled here
419
- // Heading level 2
420
- // ---------------
421
- // End of code block
422
- "```\n\n",
423
- // Horizontal lines
424
- "\n\n***\n\n",
425
- "\n\n---\n\n",
426
- "\n\n___\n\n",
427
- // Note that this splitter doesn't handle horizontal lines defined
428
- // by *three or more* of ***, ---, or ___, but this is not handled
429
- // Github tables
430
- "<table>",
431
- // "<tr>",
432
- // "<td>",
433
- // "<td ",
434
- "\n\n",
435
- "\n",
436
- " "
437
- ];
438
- case "latex":
439
- return [
440
- // First, try to split along Latex sections
441
- "\n\\chapter{",
442
- "\n\\section{",
443
- "\n\\subsection{",
444
- "\n\\subsubsection{",
445
- // Now split by environments
446
- "\n\\begin{enumerate}",
447
- "\n\\begin{itemize}",
448
- "\n\\begin{description}",
449
- "\n\\begin{list}",
450
- "\n\\begin{quote}",
451
- "\n\\begin{quotation}",
452
- "\n\\begin{verse}",
453
- "\n\\begin{verbatim}",
454
- // Now split by math environments
455
- "\n\\begin{align}",
456
- "$$",
457
- "$",
458
- // Now split by the normal type of lines
459
- "\n\n",
460
- "\n",
461
- " "
462
- ];
463
- case "html":
464
- return [
465
- // First, try to split along HTML tags
466
- "<body>",
467
- "<div>",
468
- "<p>",
469
- "<br>",
470
- "<li>",
471
- "<h1>",
472
- "<h2>",
473
- "<h3>",
474
- "<h4>",
475
- "<h5>",
476
- "<h6>",
477
- "<span>",
478
- "<table>",
479
- "<tr>",
480
- "<td>",
481
- "<th>",
482
- "<ul>",
483
- "<ol>",
484
- "<header>",
485
- "<footer>",
486
- "<nav>",
487
- // Head
488
- "<head>",
489
- "<style>",
490
- "<script>",
491
- "<meta>",
492
- "<title>",
493
- // Normal type of lines
494
- " "
495
- ];
496
- case "sol":
497
- return [
498
- // Split along compiler informations definitions
499
- "\npragma ",
500
- "\nusing ",
501
- // Split along contract definitions
502
- "\ncontract ",
503
- "\ninterface ",
504
- "\nlibrary ",
505
- // Split along method definitions
506
- "\nconstructor ",
507
- "\ntype ",
508
- "\nfunction ",
509
- "\nevent ",
510
- "\nmodifier ",
511
- "\nerror ",
512
- "\nstruct ",
513
- "\nenum ",
514
- // Split along control flow statements
515
- "\nif ",
516
- "\nfor ",
517
- "\nwhile ",
518
- "\ndo while ",
519
- "\nassembly ",
520
- // Split by the normal type of lines
521
- "\n\n",
522
- "\n",
523
- " "
524
- ];
525
- default:
526
- return [
527
- // Split by the normal type of lines
528
- "\n\n",
529
- "\n",
530
- " ",
531
- "",
532
- ];
533
- }
534
- }
535
- }
536
- exports.TextSplitter = TextSplitter;
537
- //# sourceMappingURL=TextSplitter.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"TextSplitter.js","sourceRoot":"","sources":["../src/TextSplitter.ts"],"names":[],"mappings":";;;AAAA,mDAAgD;AAGhD,MAAM,kBAAkB,GAAG,gEAAgE,CAAC;AAW5F,MAAa,YAAY;IAGrB,YAAmB,MAAoC;QACnD,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC;YACzB,cAAc,EAAE,KAAK;YACrB,SAAS,EAAE,GAAG;YACd,YAAY,EAAE,EAAE;SACG,EAAE,MAAM,CAAC,CAAC;QAEjC,iDAAiD;QACjD,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;YACzB,IAAI,CAAC,OAAO,CAAC,SAAS,GAAG,IAAI,6BAAa,EAAE,CAAC;SAChD;QAED,8CAA8C;QAC9C,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,UAAU,IAAI,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE;YAClE,IAAI,CAAC,OAAO,CAAC,UAAU,GAAG,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;SACtE;QAED,+BAA+B;QAC/B,IAAI,IAAI,CAAC,OAAO,CAAC,SAAS,GAAG,CAAC,EAAE;YAC5B,MAAM,IAAI,KAAK,CAAC,wBAAwB,CAAC,CAAC;SAC7C;aAAM,IAAI,IAAI,CAAC,OAAO,CAAC,YAAY,GAAG,CAAC,EAAE;YACtC,MAAM,IAAI,KAAK,CAAC,2BAA2B,CAAC,CAAC;SAChD;aAAM,IAAI,IAAI,CAAC,OAAO,CAAC,YAAY,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;YAC3D,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;SACxD;IACL,CAAC;IAEM,KAAK,CAAC,IAAY;QACrB,mBAAmB;QACnB,MAAM,MAAM,GAAG,IAAI,CAAC,cAAc,CAAC,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC;QAErE,MAAM,IAAI,GAAG,IAAI,CAAC;QAClB,SAAS,gBAAgB,CAAC,MAAiB;YACvC,IAAI,MAAM,IAAI,SAAS,EAAE;gBACrB,MAAM,GAAG,GAAG,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC;gBAClG,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;aAC/B;iBAAM;gBACH,OAAO,EAAE,CAAC;aACb;QACL,CAAC;QAED,iEAAiE;QACjE,IAAI,IAAI,CAAC,OAAO,CAAC,YAAY,GAAG,CAAC,EAAE;YAC/B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACpC,MAAM,aAAa,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;gBACpC,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;gBACxB,MAAM,SAAS,GAAG,CAAC,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;gBACpE,KAAK,CAAC,YAAY,GAAG,gBAAgB,CAAC,aAAa,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC;gBAChF,KAAK,CAAC,UAAU,GAAG,gBAAgB,CAAC,SAAS,aAAT,SAAS,uBAAT,SAAS,CAAE,MAAM,CAAC,CAAC;aAC1D;SACJ;QAED,OAAO,MAAM,CAAC;IAClB,CAAC;IAEO,cAAc,CAAC,IAAY,EAAE,UAAoB,EAAE,QAAgB;QACvE,MAAM,MAAM,GAAgB,EAAE,CAAC;QAC/B,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE;YACjB,wBAAwB;YACxB,IAAI,KAAe,CAAC;YACpB,IAAI,SAAS,GAAG,EAAE,CAAC;YACnB,MAAM,cAAc,GAAG,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YACxE,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE;gBACvB,qBAAqB;gBACrB,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;gBAC1B,KAAK,GAAG,SAAS,IAAI,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;aAC/E;iBAAM;gBACH,mBAAmB;gBACnB,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;gBACzC,KAAK,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC;aAC3D;YAED,qBAAqB;YACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACnC,MAAM,SAAS,GAAG,CAAC,CAAC,KAAK,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;gBAE3C,4BAA4B;gBAC5B,IAAI,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBACrB,MAAM,MAAM,GAAG,CAAC,QAAQ,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;gBACpF,IAAI,IAAI,CAAC,OAAO,CAAC,cAAc,IAAI,CAAC,SAAS,EAAE;oBAC3C,KAAK,IAAI,SAAS,CAAC;iBACtB;gBAED,6BAA6B;gBAC7B,IAAI,CAAC,IAAI,CAAC,oBAAoB,CAAC,KAAK,CAAC,EAAE;oBACnC,SAAS;iBACZ;gBAED,qDAAqD;gBACrD,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;oBAC3C,qCAAqC;oBACrC,MAAM,SAAS,GAAG,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,cAAc,EAAE,QAAQ,CAAC,CAAC;oBACvE,MAAM,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC;iBAC7B;qBAAM;oBACH,oBAAoB;oBACpB,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;oBACpD,IAAI,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;wBACxC,qCAAqC;wBACrC,MAAM,SAAS,GAAG,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,cAAc,EAAE,QAAQ,CAAC,CAAC;wBACvE,MAAM,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAC;qBAC7B;yBAAM;wBACH,yBAAyB;wBACzB,MAAM,CAAC,IAAI,CAAC;4BACR,IAAI,EAAE,KAAK;4BACX,MAAM,EAAE,MAAM;4BACd,QAAQ,EAAE,QAAQ;4BAClB,MAAM,EAAE,MAAM;4BACd,YAAY,EAAE,EAAE;4BAChB,UAAU,EAAE,EAAE;yBACjB,CAAC,CAAC;qBACN;iBAEJ;gBAGD,kBAAkB;gBAClB,QAAQ,GAAG,MAAM,GAAG,CAAC,CAAC;aACzB;SACJ;QAED,OAAO,IAAI,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;IACtC,CAAC;IAEO,aAAa,CAAC,MAAmB;QACrC,MAAM,cAAc,GAAgB,EAAE,CAAC;QACvC,IAAI,YAAiC,CAAC;QACtC,IAAI,aAAa,GAAG,CAAC,CAAC;QACtB,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC;QACzD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACpC,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;YACxB,IAAI,YAAY,EAAE;gBACd,MAAM,MAAM,GAAG,YAAY,CAAC,MAAM,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC;gBAChE,IAAI,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;oBACjC,cAAc,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;oBAClC,YAAY,GAAG,KAAK,CAAC;oBACrB,aAAa,GAAG,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC;iBACvC;qBAAM;oBACH,YAAY,CAAC,IAAI,IAAI,SAAS,GAAG,KAAK,CAAC,IAAI,CAAC;oBAC5C,YAAY,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC;oBACnC,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;oBAC1C,aAAa,IAAI,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC;iBACxC;aACJ;iBAAM;gBACH,YAAY,GAAG,KAAK,CAAC;gBACrB,aAAa,GAAG,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC;aACvC;SACJ;QACD,IAAI,YAAY,EAAE;YACd,cAAc,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;SACrC;QACD,OAAO,cAAc,CAAC;IAC1B,CAAC;IAEO,oBAAoB,CAAC,IAAY;QACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAClC,IAAI,kBAAkB,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE;gBACtC,OAAO,IAAI,CAAC;aACf;SACJ;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAEO,aAAa,CAAC,IAAY;QAC9B,wCAAwC;QACxC,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QACjD,GAAG;YACC,IAAI,MAAM,CAAC,MAAM,IAAI,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;gBACzC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC;gBAClD,MAAM;aACT;iBAAM;gBACH,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,EAAE,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;gBACtD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;aACnD;SACJ,QAAQ,IAAI,EAAE;QAEf,OAAO,KAAK,CAAC;IACjB,CAAC;IAEO,aAAa,CAAC,OAAgB;QAClC,QAAQ,OAAO,aAAP,OAAO,cAAP,OAAO,GAAI,EAAE,EAAE;YACnB,KAAK,KAAK;gBACN,OAAO;oBACH,gCAAgC;oBAChC,UAAU;oBACV,mCAAmC;oBACnC,SAAS;oBACT,QAAQ;oBACR,UAAU;oBACV,WAAW;oBACX,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,WAAW;oBACX,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,IAAI;gBACL,OAAO;oBACH,mCAAmC;oBACnC,SAAS;oBACT,QAAQ;oBACR,UAAU;oBACV,SAAS;oBACT,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,WAAW;oBACX,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,MAAM,CAAC;YACZ,KAAK,IAAI,CAAC;YACV,KAAK,QAAQ,CAAC;YACd,KAAK,IAAI,CAAC;YACV,KAAK,IAAI,CAAC;YACV,KAAK,KAAK,CAAC;YACX,KAAK,YAAY;gBACb,OAAO;oBACH,sBAAsB;oBACtB,eAAe;oBACf,eAAe;oBACf,gBAAgB;oBAChB,gCAAgC;oBAChC,UAAU;oBACV,iCAAiC;oBACjC,WAAW;oBACX,cAAc;oBACd,YAAY;oBACZ,WAAW;oBACX,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,WAAW;oBACX,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,IAAI,CAAC;YACV,KAAK,KAAK,CAAC;YACX,KAAK,YAAY;gBACb,OAAO;oBACH,sBAAsB;oBACtB,eAAe;oBACf,eAAe;oBACf,gBAAgB;oBAChB,gCAAgC;oBAChC,UAAU;oBACV,mCAAmC;oBACnC,aAAa;oBACb,UAAU;oBACV,QAAQ;oBACR,QAAQ;oBACR,UAAU;oBACV,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,WAAW;oBACX,SAAS;oBACT,YAAY;oBACZ,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,KAAK;gBACN,OAAO;oBACH,mCAAmC;oBACnC,aAAa;oBACb,gCAAgC;oBAChC,UAAU;oBACV,sCAAsC;oBACtC,OAAO;oBACP,YAAY;oBACZ,UAAU;oBACV,OAAO;oBACP,WAAW;oBACX,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,OAAO;gBACR,OAAO;oBACH,kCAAkC;oBAClC,YAAY;oBACZ,kCAAkC;oBAClC,YAAY;oBACZ,+BAA+B;oBAC/B,SAAS;oBACT,iCAAiC;oBACjC,WAAW;oBACX,gCAAgC;oBAChC,WAAW;oBACX,kCAAkC;oBAClC,WAAW;oBACX,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,QAAQ,CAAC;YACd,KAAK,IAAI;gBACL,OAAO;oBACH,8CAA8C;oBAC9C,UAAU;oBACV,QAAQ;oBACR,UAAU;oBACV,wCAAwC;oBACxC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,KAAK;gBACN,OAAO;oBACH,6BAA6B;oBAC7B,SAAS;oBACT,SAAS;oBACT,SAAS;oBACT,gCAAgC;oBAChC,OAAO;oBACP,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,MAAM;gBACP,OAAO;oBACH,iCAAiC;oBACjC,QAAQ;oBACR,UAAU;oBACV,sCAAsC;oBACtC,OAAO;oBACP,WAAW;oBACX,UAAU;oBACV,QAAQ;oBACR,OAAO;oBACP,UAAU;oBACV,WAAW;oBACX,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,MAAM;gBACP,OAAO;oBACH,mCAAmC;oBACnC,OAAO;oBACP,UAAU;oBACV,QAAQ;oBACR,sCAAsC;oBACtC,OAAO;oBACP,UAAU;oBACV,QAAQ;oBACR,SAAS;oBACT,UAAU;oBACV,UAAU;oBACV,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,OAAO;gBACR,OAAO;oBACH,gCAAgC;oBAChC,UAAU;oBACV,WAAW;oBACX,iCAAiC;oBACjC,QAAQ;oBACR,QAAQ;oBACR,QAAQ;oBACR,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,UAAU;oBACV,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,OAAO;gBACR,OAAO;oBACH,mCAAmC;oBACnC,SAAS;oBACT,gCAAgC;oBAChC,UAAU;oBACV,WAAW;oBACX,SAAS;oBACT,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,OAAO;oBACP,WAAW;oBACX,SAAS;oBACT,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,IAAI,CAAC;YACV,KAAK,UAAU;gBACX,OAAO;oBACH,sEAAsE;oBACtE,OAAO;oBACP,QAAQ;oBACR,SAAS;oBACT,UAAU;oBACV,WAAW;oBACX,uEAAuE;oBACvE,kBAAkB;oBAClB,kBAAkB;oBAClB,oBAAoB;oBACpB,SAAS;oBACT,mBAAmB;oBACnB,aAAa;oBACb,aAAa;oBACb,aAAa;oBACb,kEAAkE;oBAClE,kEAAkE;oBAClE,gBAAgB;oBAChB,SAAS;oBACT,UAAU;oBACV,UAAU;oBACV,UAAU;oBACV,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,OAAO;gBACR,OAAO;oBACH,2CAA2C;oBAC3C,cAAc;oBACd,cAAc;oBACd,iBAAiB;oBACjB,oBAAoB;oBAEpB,4BAA4B;oBAC5B,sBAAsB;oBACtB,oBAAoB;oBACpB,wBAAwB;oBACxB,iBAAiB;oBACjB,kBAAkB;oBAClB,sBAAsB;oBACtB,kBAAkB;oBAClB,qBAAqB;oBAErB,iCAAiC;oBACjC,kBAAkB;oBAClB,IAAI;oBACJ,GAAG;oBAEH,wCAAwC;oBACxC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN,KAAK,MAAM;gBACP,OAAO;oBACH,sCAAsC;oBACtC,QAAQ;oBACR,OAAO;oBACP,KAAK;oBACL,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,QAAQ;oBACR,SAAS;oBACT,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,MAAM;oBACN,UAAU;oBACV,UAAU;oBACV,OAAO;oBACP,OAAO;oBACP,QAAQ;oBACR,SAAS;oBACT,UAAU;oBACV,QAAQ;oBACR,SAAS;oBACT,uBAAuB;oBACvB,GAAG;iBACN,CAAC;YACN,KAAK,KAAK;gBACN,OAAO;oBACH,gDAAgD;oBAChD,WAAW;oBACX,UAAU;oBACV,mCAAmC;oBACnC,aAAa;oBACb,cAAc;oBACd,YAAY;oBACZ,iCAAiC;oBACjC,gBAAgB;oBAChB,SAAS;oBACT,aAAa;oBACb,UAAU;oBACV,aAAa;oBACb,UAAU;oBACV,WAAW;oBACX,SAAS;oBACT,sCAAsC;oBACtC,OAAO;oBACP,QAAQ;oBACR,UAAU;oBACV,aAAa;oBACb,aAAa;oBACb,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;iBACN,CAAC;YACN;gBACI,OAAO;oBACH,oCAAoC;oBACpC,MAAM;oBACN,IAAI;oBACJ,GAAG;oBACH,EAAE;iBACL,CAAC;SACT;IACL,CAAC;CACJ;AAliBD,oCAkiBC"}
@@ -1,15 +0,0 @@
1
- import { AxiosRequestConfig } from "axios";
2
- import { TextFetcher } from './types';
3
- export interface WebFetcherConfig {
4
- headers?: Record<string, string>;
5
- requestConfig?: AxiosRequestConfig;
6
- htmlToMarkdown: boolean;
7
- summarizeHtml: boolean;
8
- }
9
- export declare class WebFetcher implements TextFetcher {
10
- private readonly _config;
11
- constructor(config?: Partial<WebFetcherConfig>);
12
- fetch(uri: string, onDocument: (uri: string, text: string, docType?: string) => Promise<boolean>): Promise<boolean>;
13
- private htmlToMarkdown;
14
- }
15
- //# sourceMappingURL=WebFetcher.d.ts.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"WebFetcher.d.ts","sourceRoot":"","sources":["../src/WebFetcher.ts"],"names":[],"mappings":"AAAA,OAAc,EAAE,kBAAkB,EAAE,MAAM,OAAO,CAAC;AAClD,OAAO,EAAE,WAAW,EAAE,MAAM,SAAS,CAAC;AA2BtC,MAAM,WAAW,gBAAgB;IAC7B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAC,MAAM,CAAC,CAAC;IAChC,aAAa,CAAC,EAAE,kBAAkB,CAAC;IACnC,cAAc,EAAE,OAAO,CAAC;IACxB,aAAa,EAAE,OAAO,CAAC;CAC1B;AAED,qBAAa,UAAW,YAAW,WAAW;IAC1C,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAmB;gBAExB,MAAM,CAAC,EAAE,OAAO,CAAC,gBAAgB,CAAC;IAOxC,KAAK,CAAC,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,GAAG,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,MAAM,KAAK,OAAO,CAAC,OAAO,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC;IAyChI,OAAO,CAAC,cAAc;CAmCzB"}