Text Chunks

The Text Chunks module provides a function to extract specific text segments from documents, with optional highlighting. This is useful for rendering relevant extracts, entity highlights, or contextual sentences around a match.

info

Offset and length values used in textChunks come from the document's Record object returned after a search (e.g., from extractslocations, matchlocations, or entity fields).

Functions

fetchTextChunks()

Fetches text chunks for a given document at the specified locations, with optional context sentences.

Parameters

Parameter	Type	Required	Description
id	string	✓	The document's unique identifier
textChunks	TextLocation[]	✓	Array of { offset, length } locations to extract
highlights	string[]	✓	Array of highlight category names to apply (e.g. 'extractslocations', 'person')
query	Query	✓	The current query context
leftSentencesCount	number	✓	Number of context sentences to include before each chunk
rightSentencesCount	number	✓	Number of context sentences to include after each chunk

Returns Promise<{ chunks: TextChunk[] }> — object containing the array of extracted text chunks.

Example

fetch-text-chunks.ts

import { fetchTextChunks } from '@sinequa/atomic';

const response = await fetchTextChunks(
  'record-id-123',
  [
    { offset: 14937, length: 147 }, // extract location
    { offset: 1069, length: 4 },   // entity chunk
    { offset: 1188, length: 13 },  // person chunk
  ],
  ['extractslocations', 'matchlocations', 'person', 'geo'],
  { name: '_query', text: 'tesla' },
  1,  // 1 sentence before each chunk
  3   // 3 sentences after each chunk
);

response.chunks.forEach(chunk => console.log(chunk.text));