-
Notifications
You must be signed in to change notification settings - Fork 2.1k
/
parent_document.ts
212 lines (193 loc) Β· 7.12 KB
/
parent_document.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
import * as uuid from "uuid";
import {
type VectorStoreInterface,
type VectorStoreRetrieverInterface,
} from "@langchain/core/vectorstores";
import { Document } from "@langchain/core/documents";
import type { BaseDocumentCompressor } from "./document_compressors/index.js";
import {
TextSplitter,
TextSplitterChunkHeaderOptions,
} from "../text_splitter.js";
import {
MultiVectorRetriever,
type MultiVectorRetrieverInput,
} from "./multi_vector.js";
// eslint-disable-next-line @typescript-eslint/no-explicit-any
export type SubDocs = Document<Record<string, any>>[];
/**
* Interface for the fields required to initialize a
* ParentDocumentRetriever instance.
*/
export type ParentDocumentRetrieverFields = MultiVectorRetrieverInput & {
childSplitter: TextSplitter;
parentSplitter?: TextSplitter;
/**
* A custom retriever to use when retrieving instead of
* the `.similaritySearch` method of the vectorstore.
*/
childDocumentRetriever?: VectorStoreRetrieverInterface<VectorStoreInterface>;
documentCompressor?: BaseDocumentCompressor | undefined;
documentCompressorFilteringFn?: (docs: SubDocs) => SubDocs;
};
/**
* A type of document retriever that splits input documents into smaller chunks
* while separately storing and preserving the original documents.
* The small chunks are embedded, then on retrieval, the original
* "parent" documents are retrieved.
*
* This strikes a balance between better targeted retrieval with small documents
* and the more context-rich larger documents.
* @example
* ```typescript
* const retriever = new ParentDocumentRetriever({
* vectorstore: new MemoryVectorStore(new OpenAIEmbeddings()),
* byteStore: new InMemoryStore<Uint8Array>(),
* parentSplitter: new RecursiveCharacterTextSplitter({
* chunkOverlap: 0,
* chunkSize: 500,
* }),
* childSplitter: new RecursiveCharacterTextSplitter({
* chunkOverlap: 0,
* chunkSize: 50,
* }),
* childK: 20,
* parentK: 5,
* });
*
* const parentDocuments = await getDocuments();
* await retriever.addDocuments(parentDocuments);
* const retrievedDocs = await retriever.getRelevantDocuments("justice breyer");
* ```
*/
export class ParentDocumentRetriever extends MultiVectorRetriever {
static lc_name() {
return "ParentDocumentRetriever";
}
lc_namespace = ["langchain", "retrievers", "parent_document"];
vectorstore: VectorStoreInterface;
protected childSplitter: TextSplitter;
protected parentSplitter?: TextSplitter;
protected idKey = "doc_id";
protected childK?: number;
protected parentK?: number;
childDocumentRetriever:
| VectorStoreRetrieverInterface<VectorStoreInterface>
| undefined;
documentCompressor: BaseDocumentCompressor | undefined;
documentCompressorFilteringFn?: ParentDocumentRetrieverFields["documentCompressorFilteringFn"];
constructor(fields: ParentDocumentRetrieverFields) {
super(fields);
this.vectorstore = fields.vectorstore;
this.childSplitter = fields.childSplitter;
this.parentSplitter = fields.parentSplitter;
this.idKey = fields.idKey ?? this.idKey;
this.childK = fields.childK;
this.parentK = fields.parentK;
this.childDocumentRetriever = fields.childDocumentRetriever;
this.documentCompressor = fields.documentCompressor;
this.documentCompressorFilteringFn = fields.documentCompressorFilteringFn;
}
async _getRelevantDocuments(query: string): Promise<Document[]> {
let subDocs: SubDocs = [];
if (this.childDocumentRetriever) {
subDocs = await this.childDocumentRetriever.getRelevantDocuments(query);
} else {
subDocs = await this.vectorstore.similaritySearch(query, this.childK);
}
if (this.documentCompressor && subDocs.length) {
subDocs = await this.documentCompressor.compressDocuments(subDocs, query);
if (this.documentCompressorFilteringFn) {
subDocs = this.documentCompressorFilteringFn(subDocs);
}
}
// Maintain order
const parentDocIds: string[] = [];
for (const doc of subDocs) {
if (!parentDocIds.includes(doc.metadata[this.idKey])) {
parentDocIds.push(doc.metadata[this.idKey]);
}
}
const parentDocs: Document[] = [];
const storedParentDocs = await this.docstore.mget(parentDocIds);
const retrievedDocs: Document[] = storedParentDocs.filter(
(doc?: Document): doc is Document => doc !== undefined
);
parentDocs.push(...retrievedDocs);
return parentDocs.slice(0, this.parentK);
}
/**
* Adds documents to the docstore and vectorstores.
* If a retriever is provided, it will be used to add documents instead of the vectorstore.
* @param docs The documents to add
* @param config.ids Optional list of ids for documents. If provided should be the same
* length as the list of documents. Can provided if parent documents
* are already in the document store and you don't want to re-add
* to the docstore. If not provided, random UUIDs will be used as ids.
* @param config.addToDocstore Boolean of whether to add documents to docstore.
* This can be false if and only if `ids` are provided. You may want
* to set this to False if the documents are already in the docstore
* and you don't want to re-add them.
* @param config.chunkHeaderOptions Object with options for adding Contextual chunk headers
*/
async addDocuments(
docs: Document[],
config?: {
ids?: string[];
addToDocstore?: boolean;
childDocChunkHeaderOptions?: TextSplitterChunkHeaderOptions;
}
): Promise<void> {
const {
ids,
addToDocstore = true,
childDocChunkHeaderOptions = {},
} = config ?? {};
const parentDocs = this.parentSplitter
? await this.parentSplitter.splitDocuments(docs)
: docs;
let parentDocIds;
if (ids === undefined) {
if (!addToDocstore) {
throw new Error(
`If ids are not passed in, "config.addToDocstore" MUST be true`
);
}
parentDocIds = parentDocs.map((_doc: Document) => uuid.v4());
} else {
parentDocIds = ids;
}
if (parentDocs.length !== parentDocIds.length) {
throw new Error(
`Got uneven list of documents and ids.\nIf "ids" is provided, should be same length as "documents".`
);
}
const embeddedDocs: Document[] = [];
const fullDocs: Record<string, Document> = {};
for (let i = 0; i < parentDocs.length; i += 1) {
const parentDoc = parentDocs[i];
const parentDocId = parentDocIds[i];
const subDocs = await this.childSplitter.splitDocuments(
[parentDoc],
childDocChunkHeaderOptions
);
const taggedSubDocs = subDocs.map(
(subDoc: Document) =>
new Document({
pageContent: subDoc.pageContent,
metadata: { ...subDoc.metadata, [this.idKey]: parentDocId },
})
);
embeddedDocs.push(...taggedSubDocs);
fullDocs[parentDocId] = parentDoc;
}
if (this.childDocumentRetriever) {
await this.childDocumentRetriever.addDocuments(embeddedDocs);
} else {
await this.vectorstore.addDocuments(embeddedDocs);
}
if (addToDocstore) {
await this.docstore.mset(Object.entries(fullDocs));
}
}
}