vectra 0.7.6 → 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.
- package/README.md +10 -10
- package/package.json +7 -6
- package/src/ItemSelector.ts +7 -1
- package/src/LocalDocumentIndex.ts +10 -4
- package/src/LocalDocumentResult.ts +70 -22
- package/src/LocalIndex.ts +77 -10
- package/src/TextSplitter.ts +10 -15
- package/src/internals/Colorize.ts +3 -3
- package/src/internals/wink-bm25-text-search.d.ts +4 -0
- package/src/types.ts +2 -1
- package/src/vectra-cli.ts +13 -2
- package/lib/FileFetcher.d.ts +0 -5
- package/lib/FileFetcher.d.ts.map +0 -1
- package/lib/FileFetcher.js +0 -69
- package/lib/FileFetcher.js.map +0 -1
- package/lib/GPT3Tokenizer.d.ts +0 -9
- package/lib/GPT3Tokenizer.d.ts.map +0 -1
- package/lib/GPT3Tokenizer.js +0 -17
- package/lib/GPT3Tokenizer.js.map +0 -1
- package/lib/ItemSelector.d.ts +0 -41
- package/lib/ItemSelector.d.ts.map +0 -1
- package/lib/ItemSelector.js +0 -162
- package/lib/ItemSelector.js.map +0 -1
- package/lib/LocalDocument.d.ts +0 -54
- package/lib/LocalDocument.d.ts.map +0 -1
- package/lib/LocalDocument.js +0 -146
- package/lib/LocalDocument.js.map +0 -1
- package/lib/LocalDocumentIndex.d.ts +0 -128
- package/lib/LocalDocumentIndex.d.ts.map +0 -1
- package/lib/LocalDocumentIndex.js +0 -446
- package/lib/LocalDocumentIndex.js.map +0 -1
- package/lib/LocalDocumentResult.d.ts +0 -45
- package/lib/LocalDocumentResult.d.ts.map +0 -1
- package/lib/LocalDocumentResult.js +0 -282
- package/lib/LocalDocumentResult.js.map +0 -1
- package/lib/LocalIndex.d.ts +0 -136
- package/lib/LocalIndex.d.ts.map +0 -1
- package/lib/LocalIndex.js +0 -413
- package/lib/LocalIndex.js.map +0 -1
- package/lib/OpenAIEmbeddings.d.ts +0 -126
- package/lib/OpenAIEmbeddings.d.ts.map +0 -1
- package/lib/OpenAIEmbeddings.js +0 -174
- package/lib/OpenAIEmbeddings.js.map +0 -1
- package/lib/TextSplitter.d.ts +0 -20
- package/lib/TextSplitter.d.ts.map +0 -1
- package/lib/TextSplitter.js +0 -543
- package/lib/TextSplitter.js.map +0 -1
- package/lib/WebFetcher.d.ts +0 -15
- package/lib/WebFetcher.d.ts.map +0 -1
- package/lib/WebFetcher.js +0 -224
- package/lib/WebFetcher.js.map +0 -1
- package/lib/index.d.ts +0 -12
- package/lib/index.d.ts.map +0 -1
- package/lib/index.js +0 -28
- package/lib/index.js.map +0 -1
- package/lib/internals/Colorize.d.ts +0 -14
- package/lib/internals/Colorize.d.ts.map +0 -1
- package/lib/internals/Colorize.js +0 -64
- package/lib/internals/Colorize.js.map +0 -1
- package/lib/internals/index.d.ts +0 -3
- package/lib/internals/index.d.ts.map +0 -1
- package/lib/internals/index.js +0 -19
- package/lib/internals/index.js.map +0 -1
- package/lib/internals/types.d.ts +0 -43
- package/lib/internals/types.d.ts.map +0 -1
- package/lib/internals/types.js +0 -3
- package/lib/internals/types.js.map +0 -1
- package/lib/types.d.ts +0 -145
- package/lib/types.d.ts.map +0 -1
- package/lib/types.js +0 -3
- package/lib/types.js.map +0 -1
- package/lib/vectra-cli.d.ts +0 -2
- package/lib/vectra-cli.d.ts.map +0 -1
- package/lib/vectra-cli.js +0 -303
- package/lib/vectra-cli.js.map +0 -1
package/lib/TextSplitter.js
DELETED
|
@@ -1,543 +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
|
-
const parts = [];
|
|
160
|
-
const words = text.split(' ');
|
|
161
|
-
if (words.length > 0) {
|
|
162
|
-
let part = words[0];
|
|
163
|
-
for (let i = 1; i < words.length; i++) {
|
|
164
|
-
const nextWord = words[i];
|
|
165
|
-
if (this._config.tokenizer.encode(part + ' ' + nextWord).length <= this._config.chunkSize) {
|
|
166
|
-
part += ' ' + nextWord;
|
|
167
|
-
}
|
|
168
|
-
else {
|
|
169
|
-
parts.push(part);
|
|
170
|
-
part = nextWord;
|
|
171
|
-
}
|
|
172
|
-
}
|
|
173
|
-
parts.push(part);
|
|
174
|
-
}
|
|
175
|
-
else {
|
|
176
|
-
parts.push(text);
|
|
177
|
-
}
|
|
178
|
-
return parts;
|
|
179
|
-
}
|
|
180
|
-
getSeparators(docType) {
|
|
181
|
-
switch (docType !== null && docType !== void 0 ? docType : '') {
|
|
182
|
-
case "cpp":
|
|
183
|
-
return [
|
|
184
|
-
// Split along class definitions
|
|
185
|
-
"\nclass ",
|
|
186
|
-
// Split along function definitions
|
|
187
|
-
"\nvoid ",
|
|
188
|
-
"\nint ",
|
|
189
|
-
"\nfloat ",
|
|
190
|
-
"\ndouble ",
|
|
191
|
-
// Split along control flow statements
|
|
192
|
-
"\nif ",
|
|
193
|
-
"\nfor ",
|
|
194
|
-
"\nwhile ",
|
|
195
|
-
"\nswitch ",
|
|
196
|
-
"\ncase ",
|
|
197
|
-
// Split by the normal type of lines
|
|
198
|
-
"\n\n",
|
|
199
|
-
"\n",
|
|
200
|
-
" "
|
|
201
|
-
];
|
|
202
|
-
case "go":
|
|
203
|
-
return [
|
|
204
|
-
// Split along function definitions
|
|
205
|
-
"\nfunc ",
|
|
206
|
-
"\nvar ",
|
|
207
|
-
"\nconst ",
|
|
208
|
-
"\ntype ",
|
|
209
|
-
// Split along control flow statements
|
|
210
|
-
"\nif ",
|
|
211
|
-
"\nfor ",
|
|
212
|
-
"\nswitch ",
|
|
213
|
-
"\ncase ",
|
|
214
|
-
// Split by the normal type of lines
|
|
215
|
-
"\n\n",
|
|
216
|
-
"\n",
|
|
217
|
-
" "
|
|
218
|
-
];
|
|
219
|
-
case "java":
|
|
220
|
-
case "c#":
|
|
221
|
-
case "csharp":
|
|
222
|
-
case "cs":
|
|
223
|
-
case "ts":
|
|
224
|
-
case "tsx":
|
|
225
|
-
case "typescript":
|
|
226
|
-
return [
|
|
227
|
-
// split along regions
|
|
228
|
-
"// LLM-REGION",
|
|
229
|
-
"/* LLM-REGION",
|
|
230
|
-
"/** LLM-REGION",
|
|
231
|
-
// Split along class definitions
|
|
232
|
-
"\nclass ",
|
|
233
|
-
// Split along method definitions
|
|
234
|
-
"\npublic ",
|
|
235
|
-
"\nprotected ",
|
|
236
|
-
"\nprivate ",
|
|
237
|
-
"\nstatic ",
|
|
238
|
-
// Split along control flow statements
|
|
239
|
-
"\nif ",
|
|
240
|
-
"\nfor ",
|
|
241
|
-
"\nwhile ",
|
|
242
|
-
"\nswitch ",
|
|
243
|
-
"\ncase ",
|
|
244
|
-
// Split by the normal type of lines
|
|
245
|
-
"\n\n",
|
|
246
|
-
"\n",
|
|
247
|
-
" "
|
|
248
|
-
];
|
|
249
|
-
case "js":
|
|
250
|
-
case "jsx":
|
|
251
|
-
case "javascript":
|
|
252
|
-
return [
|
|
253
|
-
// split along regions
|
|
254
|
-
"// LLM-REGION",
|
|
255
|
-
"/* LLM-REGION",
|
|
256
|
-
"/** LLM-REGION",
|
|
257
|
-
// Split along class definitions
|
|
258
|
-
"\nclass ",
|
|
259
|
-
// Split along function definitions
|
|
260
|
-
"\nfunction ",
|
|
261
|
-
"\nconst ",
|
|
262
|
-
"\nlet ",
|
|
263
|
-
"\nvar ",
|
|
264
|
-
"\nclass ",
|
|
265
|
-
// Split along control flow statements
|
|
266
|
-
"\nif ",
|
|
267
|
-
"\nfor ",
|
|
268
|
-
"\nwhile ",
|
|
269
|
-
"\nswitch ",
|
|
270
|
-
"\ncase ",
|
|
271
|
-
"\ndefault ",
|
|
272
|
-
// Split by the normal type of lines
|
|
273
|
-
"\n\n",
|
|
274
|
-
"\n",
|
|
275
|
-
" "
|
|
276
|
-
];
|
|
277
|
-
case "php":
|
|
278
|
-
return [
|
|
279
|
-
// Split along function definitions
|
|
280
|
-
"\nfunction ",
|
|
281
|
-
// Split along class definitions
|
|
282
|
-
"\nclass ",
|
|
283
|
-
// Split along control flow statements
|
|
284
|
-
"\nif ",
|
|
285
|
-
"\nforeach ",
|
|
286
|
-
"\nwhile ",
|
|
287
|
-
"\ndo ",
|
|
288
|
-
"\nswitch ",
|
|
289
|
-
"\ncase ",
|
|
290
|
-
// Split by the normal type of lines
|
|
291
|
-
"\n\n",
|
|
292
|
-
"\n",
|
|
293
|
-
" "
|
|
294
|
-
];
|
|
295
|
-
case "proto":
|
|
296
|
-
return [
|
|
297
|
-
// Split along message definitions
|
|
298
|
-
"\nmessage ",
|
|
299
|
-
// Split along service definitions
|
|
300
|
-
"\nservice ",
|
|
301
|
-
// Split along enum definitions
|
|
302
|
-
"\nenum ",
|
|
303
|
-
// Split along option definitions
|
|
304
|
-
"\noption ",
|
|
305
|
-
// Split along import statements
|
|
306
|
-
"\nimport ",
|
|
307
|
-
// Split along syntax declarations
|
|
308
|
-
"\nsyntax ",
|
|
309
|
-
// Split by the normal type of lines
|
|
310
|
-
"\n\n",
|
|
311
|
-
"\n",
|
|
312
|
-
" "
|
|
313
|
-
];
|
|
314
|
-
case "python":
|
|
315
|
-
case "py":
|
|
316
|
-
return [
|
|
317
|
-
// First, try to split along class definitions
|
|
318
|
-
"\nclass ",
|
|
319
|
-
"\ndef ",
|
|
320
|
-
"\n\tdef ",
|
|
321
|
-
// Now split by the normal type of lines
|
|
322
|
-
"\n\n",
|
|
323
|
-
"\n",
|
|
324
|
-
" "
|
|
325
|
-
];
|
|
326
|
-
case "rst":
|
|
327
|
-
return [
|
|
328
|
-
// Split along section titles
|
|
329
|
-
"\n===\n",
|
|
330
|
-
"\n---\n",
|
|
331
|
-
"\n***\n",
|
|
332
|
-
// Split along directive markers
|
|
333
|
-
"\n.. ",
|
|
334
|
-
// Split by the normal type of lines
|
|
335
|
-
"\n\n",
|
|
336
|
-
"\n",
|
|
337
|
-
" "
|
|
338
|
-
];
|
|
339
|
-
case "ruby":
|
|
340
|
-
return [
|
|
341
|
-
// Split along method definitions
|
|
342
|
-
"\ndef ",
|
|
343
|
-
"\nclass ",
|
|
344
|
-
// Split along control flow statements
|
|
345
|
-
"\nif ",
|
|
346
|
-
"\nunless ",
|
|
347
|
-
"\nwhile ",
|
|
348
|
-
"\nfor ",
|
|
349
|
-
"\ndo ",
|
|
350
|
-
"\nbegin ",
|
|
351
|
-
"\nrescue ",
|
|
352
|
-
// Split by the normal type of lines
|
|
353
|
-
"\n\n",
|
|
354
|
-
"\n",
|
|
355
|
-
" "
|
|
356
|
-
];
|
|
357
|
-
case "rust":
|
|
358
|
-
return [
|
|
359
|
-
// Split along function definitions
|
|
360
|
-
"\nfn ",
|
|
361
|
-
"\nconst ",
|
|
362
|
-
"\nlet ",
|
|
363
|
-
// Split along control flow statements
|
|
364
|
-
"\nif ",
|
|
365
|
-
"\nwhile ",
|
|
366
|
-
"\nfor ",
|
|
367
|
-
"\nloop ",
|
|
368
|
-
"\nmatch ",
|
|
369
|
-
"\nconst ",
|
|
370
|
-
// Split by the normal type of lines
|
|
371
|
-
"\n\n",
|
|
372
|
-
"\n",
|
|
373
|
-
" "
|
|
374
|
-
];
|
|
375
|
-
case "scala":
|
|
376
|
-
return [
|
|
377
|
-
// Split along class definitions
|
|
378
|
-
"\nclass ",
|
|
379
|
-
"\nobject ",
|
|
380
|
-
// Split along method definitions
|
|
381
|
-
"\ndef ",
|
|
382
|
-
"\nval ",
|
|
383
|
-
"\nvar ",
|
|
384
|
-
// Split along control flow statements
|
|
385
|
-
"\nif ",
|
|
386
|
-
"\nfor ",
|
|
387
|
-
"\nwhile ",
|
|
388
|
-
"\nmatch ",
|
|
389
|
-
"\ncase ",
|
|
390
|
-
// Split by the normal type of lines
|
|
391
|
-
"\n\n",
|
|
392
|
-
"\n",
|
|
393
|
-
" "
|
|
394
|
-
];
|
|
395
|
-
case "swift":
|
|
396
|
-
return [
|
|
397
|
-
// Split along function definitions
|
|
398
|
-
"\nfunc ",
|
|
399
|
-
// Split along class definitions
|
|
400
|
-
"\nclass ",
|
|
401
|
-
"\nstruct ",
|
|
402
|
-
"\nenum ",
|
|
403
|
-
// Split along control flow statements
|
|
404
|
-
"\nif ",
|
|
405
|
-
"\nfor ",
|
|
406
|
-
"\nwhile ",
|
|
407
|
-
"\ndo ",
|
|
408
|
-
"\nswitch ",
|
|
409
|
-
"\ncase ",
|
|
410
|
-
// Split by the normal type of lines
|
|
411
|
-
"\n\n",
|
|
412
|
-
"\n",
|
|
413
|
-
" "
|
|
414
|
-
];
|
|
415
|
-
case "md":
|
|
416
|
-
case "markdown":
|
|
417
|
-
return [
|
|
418
|
-
// First, try to split along Markdown headings (starting with level 2)
|
|
419
|
-
"\n## ",
|
|
420
|
-
"\n### ",
|
|
421
|
-
"\n#### ",
|
|
422
|
-
"\n##### ",
|
|
423
|
-
"\n###### ",
|
|
424
|
-
// Note the alternative syntax for headings (below) is not handled here
|
|
425
|
-
// Heading level 2
|
|
426
|
-
// ---------------
|
|
427
|
-
// End of code block
|
|
428
|
-
"```\n\n",
|
|
429
|
-
// Horizontal lines
|
|
430
|
-
"\n\n***\n\n",
|
|
431
|
-
"\n\n---\n\n",
|
|
432
|
-
"\n\n___\n\n",
|
|
433
|
-
// Note that this splitter doesn't handle horizontal lines defined
|
|
434
|
-
// by *three or more* of ***, ---, or ___, but this is not handled
|
|
435
|
-
// Github tables
|
|
436
|
-
"<table>",
|
|
437
|
-
// "<tr>",
|
|
438
|
-
// "<td>",
|
|
439
|
-
// "<td ",
|
|
440
|
-
"\n\n",
|
|
441
|
-
"\n",
|
|
442
|
-
" "
|
|
443
|
-
];
|
|
444
|
-
case "latex":
|
|
445
|
-
return [
|
|
446
|
-
// First, try to split along Latex sections
|
|
447
|
-
"\n\\chapter{",
|
|
448
|
-
"\n\\section{",
|
|
449
|
-
"\n\\subsection{",
|
|
450
|
-
"\n\\subsubsection{",
|
|
451
|
-
// Now split by environments
|
|
452
|
-
"\n\\begin{enumerate}",
|
|
453
|
-
"\n\\begin{itemize}",
|
|
454
|
-
"\n\\begin{description}",
|
|
455
|
-
"\n\\begin{list}",
|
|
456
|
-
"\n\\begin{quote}",
|
|
457
|
-
"\n\\begin{quotation}",
|
|
458
|
-
"\n\\begin{verse}",
|
|
459
|
-
"\n\\begin{verbatim}",
|
|
460
|
-
// Now split by math environments
|
|
461
|
-
"\n\\begin{align}",
|
|
462
|
-
"$$",
|
|
463
|
-
"$",
|
|
464
|
-
// Now split by the normal type of lines
|
|
465
|
-
"\n\n",
|
|
466
|
-
"\n",
|
|
467
|
-
" "
|
|
468
|
-
];
|
|
469
|
-
case "html":
|
|
470
|
-
return [
|
|
471
|
-
// First, try to split along HTML tags
|
|
472
|
-
"<body>",
|
|
473
|
-
"<div>",
|
|
474
|
-
"<p>",
|
|
475
|
-
"<br>",
|
|
476
|
-
"<li>",
|
|
477
|
-
"<h1>",
|
|
478
|
-
"<h2>",
|
|
479
|
-
"<h3>",
|
|
480
|
-
"<h4>",
|
|
481
|
-
"<h5>",
|
|
482
|
-
"<h6>",
|
|
483
|
-
"<span>",
|
|
484
|
-
"<table>",
|
|
485
|
-
"<tr>",
|
|
486
|
-
"<td>",
|
|
487
|
-
"<th>",
|
|
488
|
-
"<ul>",
|
|
489
|
-
"<ol>",
|
|
490
|
-
"<header>",
|
|
491
|
-
"<footer>",
|
|
492
|
-
"<nav>",
|
|
493
|
-
// Head
|
|
494
|
-
"<head>",
|
|
495
|
-
"<style>",
|
|
496
|
-
"<script>",
|
|
497
|
-
"<meta>",
|
|
498
|
-
"<title>",
|
|
499
|
-
// Normal type of lines
|
|
500
|
-
" "
|
|
501
|
-
];
|
|
502
|
-
case "sol":
|
|
503
|
-
return [
|
|
504
|
-
// Split along compiler informations definitions
|
|
505
|
-
"\npragma ",
|
|
506
|
-
"\nusing ",
|
|
507
|
-
// Split along contract definitions
|
|
508
|
-
"\ncontract ",
|
|
509
|
-
"\ninterface ",
|
|
510
|
-
"\nlibrary ",
|
|
511
|
-
// Split along method definitions
|
|
512
|
-
"\nconstructor ",
|
|
513
|
-
"\ntype ",
|
|
514
|
-
"\nfunction ",
|
|
515
|
-
"\nevent ",
|
|
516
|
-
"\nmodifier ",
|
|
517
|
-
"\nerror ",
|
|
518
|
-
"\nstruct ",
|
|
519
|
-
"\nenum ",
|
|
520
|
-
// Split along control flow statements
|
|
521
|
-
"\nif ",
|
|
522
|
-
"\nfor ",
|
|
523
|
-
"\nwhile ",
|
|
524
|
-
"\ndo while ",
|
|
525
|
-
"\nassembly ",
|
|
526
|
-
// Split by the normal type of lines
|
|
527
|
-
"\n\n",
|
|
528
|
-
"\n",
|
|
529
|
-
" "
|
|
530
|
-
];
|
|
531
|
-
default:
|
|
532
|
-
return [
|
|
533
|
-
// Split by the normal type of lines
|
|
534
|
-
"\n\n",
|
|
535
|
-
"\n",
|
|
536
|
-
" ",
|
|
537
|
-
"",
|
|
538
|
-
];
|
|
539
|
-
}
|
|
540
|
-
}
|
|
541
|
-
}
|
|
542
|
-
exports.TextSplitter = TextSplitter;
|
|
543
|
-
//# sourceMappingURL=TextSplitter.js.map
|
package/lib/TextSplitter.js.map
DELETED
|
@@ -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,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC9B,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE;YAClB,IAAI,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACnC,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC1B,IAAI,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,GAAG,GAAG,GAAG,QAAQ,CAAC,CAAC,MAAM,IAAI,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE;oBACvF,IAAI,IAAI,GAAG,GAAG,QAAQ,CAAC;iBAC1B;qBAAM;oBACH,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;oBACjB,IAAI,GAAG,QAAQ,CAAC;iBACnB;aACJ;YACD,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SACpB;aAAM;YACH,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SACpB;QAED,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;AAviBD,oCAuiBC"}
|
package/lib/WebFetcher.d.ts
DELETED
|
@@ -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
|
package/lib/WebFetcher.d.ts.map
DELETED
|
@@ -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"}
|