inquery/index.js

A portable indexing and searching library.

const BM25    = require( '@spacebartech/bm25' );
const Globals = require( './globals' );

const HasProperty = function ( a, b ) {
	return Object.prototype.hasOwnProperty.call( a, b );
};

class Inquery {

	constructor() {

		this.index = new BM25();

	}

	search( query ) {

		if ( !query ) {
			return [];
		}

		const searchTerms    = query.split( /\ +/ ); // eslint-disable-line
		const stripStopWords = ( searchTerms.length > 2 );

		const tokenizedTerms = this.index.tokenize( query, stripStopWords );
		const tokens         = [];

		const documentTokens = Object.keys( this.index.terms );
		const terms          = this.index.terms;

		tokenizedTerms.forEach( ( term ) => {

			// find perfect match
			if ( HasProperty( terms, term ) ) {

				tokens.push( {
					term,
					relevance : 3,
				} );

			}

			// find match of substring
			documentTokens.forEach( ( token ) => {

				const longest  = token.length > term.length ? token : term;
				const shortest = longest === token ? term : token;

				const match = Globals.findMatch( term, token );

				if ( match !== -1 ) {

					tokens.push( {
						term      : token,
						relevance : shortest.length / ( shortest.length + match ),
					} );

				}

			} );

		} );

		const documents = this.index.documents; // eslint-disable-line
		const addedIn   = {};
		const results   = [];

		tokens.forEach( ( token ) => {

			const term = this.index.terms[token.term];

			term.foundIn.forEach( ( id ) => {

				if ( HasProperty( addedIn, id ) ) {

					// decrease the relevance of the token
					if ( token.relevance < 1 ) {
						token.relevance *= token.relevance; // eslint-disable-line
					}

					// add this decreased value to the total
					// sum of all relevance in results
					const docIndex = addedIn[id];

					results[docIndex].relevance += token.relevance;

				}

				else {

					// otherwise, make a note of where it is in
					// the results array
					addedIn[id] = results.length;

					const result = {
						relevance : token.relevance,
						document  : documents[id],
						id
					};

					results.push( result );

				}

			} );

		} );

		return results;

	}

	addDocument( document, key ) {

		this.index.addDocument( document, key );

	}

	getIndex() {

		return this.index.terms;

	}

}

module.exports = Inquery;