Michael Stapelberg
102 lines
3.1 KiB
JavaScript
102 lines
3.1 KiB
JavaScript
import { create, search as oramaSearch, insertMultiple } from './orama/orama.js';
|
|
import { pluginQPS } from './orama/plugin-qps.js';
|
|
import { pluginPT15 } from './orama/plugin-pt15.js';
|
|
//import { pluginEmbeddings } from 'https://cdn.jsdelivr.net/npm/@orama/plugin-embeddings@3.0.8.js'
|
|
//import * as tf from 'https://cdn.jsdelivr.net/npm/@tensorflow/tfjs-core';
|
|
//import 'https://cdn.jsdelivr.net/npm/@tensorflow/tfjs-backend-webgl';
|
|
|
|
//import { createTokenizer } from '@orama/tokenizers/japanese'
|
|
//import { stopwords as japaneseStopwords } from "@orama/stopwords/japanese";
|
|
|
|
let searchEngine = null;
|
|
|
|
async function init() {
|
|
async function initIndex(index) {
|
|
/*
|
|
const embeddings = await pluginEmbeddings({
|
|
embeddings: {
|
|
// Property used to store generated embeddings. Must be defined in the schema.
|
|
defaultProperty: 'embeddings',
|
|
onInsert: {
|
|
// Generate embeddings at insert-time.
|
|
// Turn off if you're inserting documents with embeddings already generated.
|
|
generate: true,
|
|
// Properties to use for generating embeddings at insert time.
|
|
// These properties will be concatenated and used to generate embeddings.
|
|
properties: ['description'],
|
|
verbose: true,
|
|
}
|
|
}
|
|
});
|
|
*/
|
|
searchEngine = await create({
|
|
schema: {
|
|
title: 'string',
|
|
content: 'string',
|
|
uri: 'string',
|
|
breadcrumb: 'string',
|
|
description: 'string',
|
|
tags: 'string[]',
|
|
// embeddings: 'vector[1]'
|
|
},
|
|
plugins: [
|
|
// embeddings,
|
|
// pluginQPS()
|
|
pluginPT15(),
|
|
],
|
|
/*
|
|
defaultLanguage: 'french',
|
|
components: {
|
|
tokenizer: {
|
|
stemmingFn: stemmer,
|
|
},
|
|
},
|
|
*/
|
|
});
|
|
await insertMultiple(searchEngine, index);
|
|
|
|
window.relearn.isSearchEngineReady = true;
|
|
window.relearn.executeInitialSearch();
|
|
}
|
|
|
|
if (window.relearn.index_js_url) {
|
|
var js = document.createElement('script');
|
|
js.src = window.relearn.index_js_url;
|
|
js.setAttribute('async', '');
|
|
js.onload = function () {
|
|
initIndex(relearn_searchindex);
|
|
};
|
|
js.onerror = function (e) {
|
|
console.error('Error getting Hugo index file');
|
|
};
|
|
document.head.appendChild(js);
|
|
}
|
|
}
|
|
|
|
async function search(term) {
|
|
const searchResponse = await oramaSearch(searchEngine, {
|
|
// mode: 'hybrid', // vector search seems not to work
|
|
term: term,
|
|
properties: '*',
|
|
threshold: 0, // only show results where all keywords were found
|
|
limit: 99,
|
|
boost: {
|
|
// doesn't seem to make a difference in score
|
|
tags: 1.8,
|
|
title: 1.5,
|
|
description: 1.3,
|
|
breadcrumb: 1.2,
|
|
},
|
|
// distinctOn: 'title', // just to filter out changelog/releasenotes if having the same title
|
|
// exact: true, // not for PT15
|
|
// tolerance: 1, // not for PT15
|
|
});
|
|
console.log('new term', term);
|
|
searchResponse.hits.forEach((hit) => console.log(hit.score, hit.document.uri));
|
|
return searchResponse.hits.map((hit) => ({ matches: [term, ...term.split(' ')], page: hit.document }));
|
|
}
|
|
|
|
window.relearn = window.relearn ?? {};
|
|
window.relearn.search = window.relearn.search ?? {};
|
|
window.relearn.search.adapter = { init, search };
|