-
Notifications
You must be signed in to change notification settings - Fork 295
/
MetadataExtractors.ts
467 lines (390 loc) · 11.7 KB
/
MetadataExtractors.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
import type { BaseNode } from "../Node.js";
import { MetadataMode, TextNode } from "../Node.js";
import type { LLM } from "../llm/index.js";
import { OpenAI } from "../llm/index.js";
import {
defaultKeywordExtractorPromptTemplate,
defaultQuestionAnswerPromptTemplate,
defaultSummaryExtractorPromptTemplate,
defaultTitleCombinePromptTemplate,
defaultTitleExtractorPromptTemplate,
} from "./prompts.js";
import { BaseExtractor } from "./types.js";
const STRIP_REGEX = /(\r\n|\n|\r)/gm;
type KeywordExtractArgs = {
llm?: LLM;
keywords?: number;
};
type ExtractKeyword = {
excerptKeywords: string;
};
/**
* Extract keywords from a list of nodes.
*/
export class KeywordExtractor extends BaseExtractor {
/**
* LLM instance.
* @type {LLM}
*/
llm: LLM;
/**
* Number of keywords to extract.
* @type {number}
* @default 5
*/
keywords: number = 5;
/**
* Constructor for the KeywordExtractor class.
* @param {LLM} llm LLM instance.
* @param {number} keywords Number of keywords to extract.
* @throws {Error} If keywords is less than 1.
*/
constructor(options?: KeywordExtractArgs) {
if (options?.keywords && options.keywords < 1)
throw new Error("Keywords must be greater than 0");
super();
this.llm = options?.llm ?? new OpenAI();
this.keywords = options?.keywords ?? 5;
}
/**
*
* @param node Node to extract keywords from.
* @returns Keywords extracted from the node.
*/
async extractKeywordsFromNodes(node: BaseNode): Promise<ExtractKeyword | {}> {
if (this.isTextNodeOnly && !(node instanceof TextNode)) {
return {};
}
const completion = await this.llm.complete({
prompt: defaultKeywordExtractorPromptTemplate({
contextStr: node.getContent(MetadataMode.ALL),
keywords: this.keywords,
}),
});
return {
excerptKeywords: completion.text,
};
}
/**
*
* @param nodes Nodes to extract keywords from.
* @returns Keywords extracted from the nodes.
*/
async extract(nodes: BaseNode[]): Promise<Array<ExtractKeyword> | Array<{}>> {
const results = await Promise.all(
nodes.map((node) => this.extractKeywordsFromNodes(node)),
);
return results;
}
}
type TitleExtractorsArgs = {
llm?: LLM;
nodes?: number;
nodeTemplate?: string;
combineTemplate?: string;
};
type ExtractTitle = {
documentTitle: string;
};
/**
* Extract title from a list of nodes.
*/
export class TitleExtractor extends BaseExtractor {
/**
* LLM instance.
* @type {LLM}
*/
llm: LLM;
/**
* Can work for mixture of text and non-text nodes
* @type {boolean}
* @default false
*/
isTextNodeOnly: boolean = false;
/**
* Number of nodes to extrct titles from.
* @type {number}
* @default 5
*/
nodes: number = 5;
/**
* The prompt template to use for the title extractor.
* @type {string}
*/
nodeTemplate: string;
/**
* The prompt template to merge title with..
* @type {string}
*/
combineTemplate: string;
/**
* Constructor for the TitleExtractor class.
* @param {LLM} llm LLM instance.
* @param {number} nodes Number of nodes to extract titles from.
* @param {string} nodeTemplate The prompt template to use for the title extractor.
* @param {string} combineTemplate The prompt template to merge title with..
*/
constructor(options?: TitleExtractorsArgs) {
super();
this.llm = options?.llm ?? new OpenAI();
this.nodes = options?.nodes ?? 5;
this.nodeTemplate =
options?.nodeTemplate ?? defaultTitleExtractorPromptTemplate();
this.combineTemplate =
options?.combineTemplate ?? defaultTitleCombinePromptTemplate();
}
/**
* Extract titles from a list of nodes.
* @param {BaseNode[]} nodes Nodes to extract titles from.
* @returns {Promise<BaseNode<ExtractTitle>[]>} Titles extracted from the nodes.
*/
async extract(nodes: BaseNode[]): Promise<Array<ExtractTitle>> {
const nodesToExtractTitle = this.filterNodes(nodes);
if (!nodesToExtractTitle.length) {
return [];
}
const nodesByDocument = this.separateNodesByDocument(nodesToExtractTitle);
const titlesByDocument = await this.extractTitles(nodesByDocument);
return nodesToExtractTitle.map((node) => {
return {
documentTitle: titlesByDocument[node.sourceNode?.nodeId ?? ""],
};
});
}
private filterNodes(nodes: BaseNode[]): BaseNode[] {
return nodes.filter((node) => {
if (this.isTextNodeOnly && !(node instanceof TextNode)) {
return false;
}
return true;
});
}
private separateNodesByDocument(
nodes: BaseNode[],
): Record<string, BaseNode[]> {
const nodesByDocument: Record<string, BaseNode[]> = {};
for (const node of nodes) {
const parentNode = node.sourceNode?.nodeId;
if (!parentNode) {
continue;
}
if (!nodesByDocument[parentNode]) {
nodesByDocument[parentNode] = [];
}
nodesByDocument[parentNode].push(node);
}
return nodesByDocument;
}
private async extractTitles(
nodesByDocument: Record<string, BaseNode[]>,
): Promise<Record<string, string>> {
const titlesByDocument: Record<string, string> = {};
for (const [key, nodes] of Object.entries(nodesByDocument)) {
const titleCandidates = await this.getTitlesCandidates(nodes);
const combinedTitles = titleCandidates.join(", ");
const completion = await this.llm.complete({
prompt: defaultTitleCombinePromptTemplate({
contextStr: combinedTitles,
}),
});
titlesByDocument[key] = completion.text;
}
return titlesByDocument;
}
private async getTitlesCandidates(nodes: BaseNode[]): Promise<string[]> {
const titleJobs = nodes.map(async (node) => {
const completion = await this.llm.complete({
prompt: defaultTitleExtractorPromptTemplate({
contextStr: node.getContent(MetadataMode.ALL),
}),
});
return completion.text;
});
return await Promise.all(titleJobs);
}
}
type QuestionAnswerExtractArgs = {
llm?: LLM;
questions?: number;
promptTemplate?: string;
embeddingOnly?: boolean;
};
type ExtractQuestion = {
questionsThisExcerptCanAnswer: string;
};
/**
* Extract questions from a list of nodes.
*/
export class QuestionsAnsweredExtractor extends BaseExtractor {
/**
* LLM instance.
* @type {LLM}
*/
llm: LLM;
/**
* Number of questions to generate.
* @type {number}
* @default 5
*/
questions: number = 5;
/**
* The prompt template to use for the question extractor.
* @type {string}
*/
promptTemplate: string;
/**
* Wheter to use metadata for embeddings only
* @type {boolean}
* @default false
*/
embeddingOnly: boolean = false;
/**
* Constructor for the QuestionsAnsweredExtractor class.
* @param {LLM} llm LLM instance.
* @param {number} questions Number of questions to generate.
* @param {string} promptTemplate The prompt template to use for the question extractor.
* @param {boolean} embeddingOnly Wheter to use metadata for embeddings only.
*/
constructor(options?: QuestionAnswerExtractArgs) {
if (options?.questions && options.questions < 1)
throw new Error("Questions must be greater than 0");
super();
this.llm = options?.llm ?? new OpenAI();
this.questions = options?.questions ?? 5;
this.promptTemplate =
options?.promptTemplate ??
defaultQuestionAnswerPromptTemplate({
numQuestions: this.questions,
contextStr: "",
});
this.embeddingOnly = options?.embeddingOnly ?? false;
}
/**
* Extract answered questions from a node.
* @param {BaseNode} node Node to extract questions from.
* @returns {Promise<Array<ExtractQuestion> | Array<{}>>} Questions extracted from the node.
*/
async extractQuestionsFromNode(
node: BaseNode,
): Promise<ExtractQuestion | {}> {
if (this.isTextNodeOnly && !(node instanceof TextNode)) {
return {};
}
const contextStr = node.getContent(this.metadataMode);
const prompt = defaultQuestionAnswerPromptTemplate({
contextStr,
numQuestions: this.questions,
});
const questions = await this.llm.complete({
prompt,
});
return {
questionsThisExcerptCanAnswer: questions.text.replace(STRIP_REGEX, ""),
};
}
/**
* Extract answered questions from a list of nodes.
* @param {BaseNode[]} nodes Nodes to extract questions from.
* @returns {Promise<Array<ExtractQuestion> | Array<{}>>} Questions extracted from the nodes.
*/
async extract(
nodes: BaseNode[],
): Promise<Array<ExtractQuestion> | Array<{}>> {
const results = await Promise.all(
nodes.map((node) => this.extractQuestionsFromNode(node)),
);
return results;
}
}
type SummaryExtractArgs = {
llm?: LLM;
summaries?: string[];
promptTemplate?: string;
};
type ExtractSummary = {
sectionSummary: string;
prevSectionSummary: string;
nextSectionSummary: string;
};
/**
* Extract summary from a list of nodes.
*/
export class SummaryExtractor extends BaseExtractor {
/**
* LLM instance.
* @type {LLM}
*/
llm: LLM;
/**
* List of summaries to extract: 'self', 'prev', 'next'
* @type {string[]}
*/
summaries: string[];
/**
* The prompt template to use for the summary extractor.
* @type {string}
*/
promptTemplate: string;
private selfSummary: boolean;
private prevSummary: boolean;
private nextSummary: boolean;
constructor(options?: SummaryExtractArgs) {
const summaries = options?.summaries ?? ["self"];
if (
summaries &&
!summaries.some((s) => ["self", "prev", "next"].includes(s))
)
throw new Error("Summaries must be one of 'self', 'prev', 'next'");
super();
this.llm = options?.llm ?? new OpenAI();
this.summaries = summaries;
this.promptTemplate =
options?.promptTemplate ?? defaultSummaryExtractorPromptTemplate();
this.selfSummary = summaries?.includes("self") ?? false;
this.prevSummary = summaries?.includes("prev") ?? false;
this.nextSummary = summaries?.includes("next") ?? false;
}
/**
* Extract summary from a node.
* @param {BaseNode} node Node to extract summary from.
* @returns {Promise<string>} Summary extracted from the node.
*/
async generateNodeSummary(node: BaseNode): Promise<string> {
if (this.isTextNodeOnly && !(node instanceof TextNode)) {
return "";
}
const contextStr = node.getContent(this.metadataMode);
const prompt = defaultSummaryExtractorPromptTemplate({
contextStr,
});
const summary = await this.llm.complete({
prompt,
});
return summary.text.replace(STRIP_REGEX, "");
}
/**
* Extract summaries from a list of nodes.
* @param {BaseNode[]} nodes Nodes to extract summaries from.
* @returns {Promise<Array<ExtractSummary> | Arry<{}>>} Summaries extracted from the nodes.
*/
async extract(nodes: BaseNode[]): Promise<Array<ExtractSummary> | Array<{}>> {
if (!nodes.every((n) => n instanceof TextNode))
throw new Error("Only `TextNode` is allowed for `Summary` extractor");
const nodeSummaries = await Promise.all(
nodes.map((node) => this.generateNodeSummary(node)),
);
const metadataList: any[] = nodes.map(() => ({}));
for (let i = 0; i < nodes.length; i++) {
if (i > 0 && this.prevSummary && nodeSummaries[i - 1]) {
metadataList[i]["prevSectionSummary"] = nodeSummaries[i - 1];
}
if (i < nodes.length - 1 && this.nextSummary && nodeSummaries[i + 1]) {
metadataList[i]["nextSectionSummary"] = nodeSummaries[i + 1];
}
if (this.selfSummary && nodeSummaries[i]) {
metadataList[i]["sectionSummary"] = nodeSummaries[i];
}
}
return metadataList;
}
}