search: refactor to swap search library #409

see lunr-adapter.js to see how other search engines can be integrated
This commit is contained in:
Sören Weber 2024-10-25 17:48:04 +02:00
parent 28fce6b04c
commit 2574cb8340
No known key found for this signature in database
GPG key ID: BEC6D55545451B6D
6 changed files with 210 additions and 143 deletions

View file

@ -6,7 +6,7 @@
{{ partial "heading-pre.html" . }}{{ partial "heading.html" . }}{{ partial "heading-post.html" . }}
<search>
<form action="javascript:triggerSearch()">
<form action="javascript:executeTriggeredSearch()">
<div class="searchform">
<label class="a11y-only" for="R-search-by-detail">{{ T "Search" }}</label>
<input data-search-input id="R-search-by-detail" class="search-by" name="search-by" type="search" placeholder="{{ T "Search-placeholder" }}">

View file

@ -69,5 +69,5 @@
{{- $file := (printf "js/lunr/lunr.%s.min.js" .) }}
<script src="{{ $file | relURL}}{{ $assetBusting }}" defer></script>
{{- end }}
<script src="{{ "js/search.js" | relURL }}{{ $assetBusting }}" defer></script>
<script src="{{ "js/search.js" | relURL }}{{ $assetBusting }}" type="module" defer></script>
{{- end }}

View file

@ -1 +1 @@
7.1.1
7.1.1+28fce6b04c414523280c53ee02f9f3a94d9d23da

136
static/js/lunr-adapter.js Normal file
View file

@ -0,0 +1,136 @@
/*
# Adapter Interface
The search adapter needs to provide the following functions that are called from search.js
## init()
Initialize the search engine and the search index
### Parameters
one
### Returns
none
### Remarks
Once successfully completed, needs to call
````
window.relearn.isSearchEngineReady = true;
window.relearn.executeInitialSearch();
````
## search()
Performs the search and returs found results
### Parameters
**term**: The search term that was typed in by the user
### Returns
Must return an array of found pages, sorted with the most relevant pages first.
Each array item needs the following layout:
````
{
index: string, // optional, id of the page in the search index
matches: string[], // optional, TODO: have to find out what it does
page: {
breadcrumb: string,
title: string,
uri: string,
content: string,
tags: string[]
}
}
````
*/
let lunrIndex, pagesIndex;
function init() {
function initLunrIndex( index ){
pagesIndex = index;
// Set up Lunr by declaring the fields we use
// Also provide their boost level for the ranking
lunrIndex = lunr(function() {
this.use(lunr.multiLanguage.apply(null, contentLangs));
this.ref('index');
this.field('title', {
boost: 15
});
this.field('tags', {
boost: 10
});
this.field('content', {
boost: 5
});
this.pipeline.remove(lunr.stemmer);
this.searchPipeline.remove(lunr.stemmer);
// Feed Lunr with each file and let index them
pagesIndex.forEach(function(page, idx) {
page.index = idx;
this.add(page);
}, this);
});
window.relearn.isSearchEngineReady = true;
window.relearn.executeInitialSearch();
}
if( window.index_js_url ){
var js = document.createElement("script");
js.src = index_js_url;
js.setAttribute("async", "");
js.onload = function(){
initLunrIndex(relearn_searchindex);
};
js.onerror = function(e){
console.error('Error getting Hugo index file');
};
document.head.appendChild(js);
}
}
function search(term) {
function searchPatterns(word) {
// for short words high amounts of typos doesn't make sense
// for long words we allow less typos because this largly increases search time
var typos = [
{ len: -1, typos: 1 },
{ len: 60, typos: 2 },
{ len: 40, typos: 3 },
{ len: 20, typos: 4 },
{ len: 16, typos: 3 },
{ len: 12, typos: 2 },
{ len: 8, typos: 1 },
{ len: 4, typos: 0 },
];
return [
word + '^100',
word + '*^10',
'*' + word + '^10',
word + '~' + typos.reduce( function( a, c, i ){ return word.length < c.len ? c : a; } ).typos + '^1'
];
}
// Find the item in our index corresponding to the Lunr one to have more info
// Remove Lunr special search characters: https://lunrjs.com/guides/searching.html
term = term.replace( /[*:^~+-]/g, ' ' );
var searchTerm = lunr.tokenizer( term ).reduce( function(a,token){return a.concat(searchPatterns(token.str))}, []).join(' ');
return !searchTerm || !lunrIndex ? [] : lunrIndex.search(searchTerm).map(function(result) {
return { index: result.ref, matches: Object.keys(result.matchData.metadata), page: pagesIndex[ result.ref ] };
});
}
export { init, search };

View file

@ -1,56 +1,28 @@
import { init, search } from './lunr-adapter.js';
(function(){
window.relearn = window.relearn || {};
window.relearn.runInitialSearch = function(){
if( window.relearn.isSearchInit && window.relearn.isLunrInit ){
window.relearn.executeInitialSearch =
function executeInitialSearch(){
if( window.relearn.isSearchInterfaceReady && window.relearn.isSearchEngineReady ){
var input = document.querySelector('#R-search-by-detail');
if( !input ){
return;
}
var value = input.value;
searchDetail( value );
executeSearch( value );
}
}
var lunrIndex, pagesIndex;
function initLunrIndex( index ){
pagesIndex = index;
// Set up Lunr by declaring the fields we use
// Also provide their boost level for the ranking
lunrIndex = lunr(function() {
this.use(lunr.multiLanguage.apply(null, contentLangs));
this.ref('index');
this.field('title', {
boost: 15
});
this.field('tags', {
boost: 10
});
this.field('content', {
boost: 5
});
this.pipeline.remove(lunr.stemmer);
this.searchPipeline.remove(lunr.stemmer);
// Feed Lunr with each file and let LUnr actually index them
pagesIndex.forEach(function(page, idx) {
page.index = idx;
this.add(page);
}, this);
});
window.relearn.isLunrInit = true;
window.relearn.runInitialSearch();
}
function triggerSearch(){
function executeTriggeredSearch(){
var input = document.querySelector('#R-search-by-detail');
if( !input ){
return;
}
var value = input.value;
searchDetail( value );
executeSearch( value );
// add a new entry to the history after the user
// changed the term; this does not reload the page
@ -70,7 +42,7 @@ function triggerSearch(){
}
}
window.addEventListener( 'popstate', function ( event ){
function executeHistorySearch( event ){
// restart search if browsed through history
if( event.state ){
var state = window.history.state || {};
@ -91,78 +63,14 @@ window.addEventListener( 'popstate', function ( event ){
// recreate the last search results and eventually
// restore the previous scrolling position
searchDetail( search );
executeSearch( search );
}
}
}
}
});
function executeSearch( value ) {
var input = document.querySelector('#R-search-by-detail');
if( input ){
input.addEventListener( 'keydown', function(event) {
// if we are pressing ESC in the searchdetail our focus will
// be stolen by the other event handlers, so we have to refocus
// here after a short while
if (event.key == "Escape") {
setTimeout( function(){ input.focus(); }, 0 );
}
});
}
function initLunrJs() {
// new way to load our search index
if( window.index_js_url ){
var js = document.createElement("script");
js.src = index_js_url;
js.setAttribute("async", "");
js.onload = function(){
initLunrIndex(relearn_searchindex);
};
js.onerror = function(e){
console.error('Error getting Hugo index file');
};
document.head.appendChild(js);
}
}
/**
* Trigger a search in Lunr and transform the result
*
* @param {String} term
* @return {Array} results
*/
function search(term) {
// Find the item in our index corresponding to the Lunr one to have more info
// Remove Lunr special search characters: https://lunrjs.com/guides/searching.html
term = term.replace( /[*:^~+-]/g, ' ' );
var searchTerm = lunr.tokenizer( term ).reduce( function(a,token){return a.concat(searchPatterns(token.str))}, []).join(' ');
return !searchTerm || !lunrIndex ? [] : lunrIndex.search(searchTerm).map(function(result) {
return { index: result.ref, matches: Object.keys(result.matchData.metadata) }
});
}
function searchPatterns(word) {
// for short words high amounts of typos doesn't make sense
// for long words we allow less typos because this largly increases search time
var typos = [
{ len: -1, typos: 1 },
{ len: 60, typos: 2 },
{ len: 40, typos: 3 },
{ len: 20, typos: 4 },
{ len: 16, typos: 3 },
{ len: 12, typos: 2 },
{ len: 8, typos: 1 },
{ len: 4, typos: 0 },
];
return [
word + '^100',
word + '*^10',
'*' + word + '^10',
word + '~' + typos.reduce( function( a, c, i ){ return word.length < c.len ? c : a; } ).typos + '^1'
];
}
function resolvePlaceholders( s, args ) {
var args = args || [];
// use replace to iterate over the string
@ -174,7 +82,6 @@ function resolvePlaceholders( s, args ) {
});
};
function searchDetail( value ) {
var results = document.querySelector('#R-searchresults');
var hint = document.querySelector('.searchhint');
hint.innerText = '';
@ -183,12 +90,15 @@ function searchDetail( value ) {
if( a.length ){
hint.innerText = resolvePlaceholders( window.T_N_results_found, [ value, a.length ] );
a.forEach( function(item){
var page = pagesIndex[item.index];
var page = item.page;
var context = [];
if( item.matches ){
var numContextWords = 10;
var contextPattern = '(?:\\S+ +){0,' + numContextWords + '}\\S*\\b(?:' +
item.matches.map( function(match){return match.replace(/\W/g, '\\$&')} ).join('|') +
')\\b\\S*(?: +\\S+){0,' + numContextWords + '}';
var context = page.content.match(new RegExp(contextPattern, 'i'));
context = page.content.match(new RegExp(contextPattern, 'i'));
}
var divsuggestion = document.createElement('a');
divsuggestion.className = 'autocomplete-suggestion';
divsuggestion.setAttribute('data-term', value);
@ -233,9 +143,7 @@ function searchDetail( value ) {
}
}
initLunrJs();
function startSearch(){
function initSearchAfterLoad(){
var input = document.querySelector('#R-search-by-detail');
if( input ){
var state = window.history.state || {};
@ -244,7 +152,7 @@ function startSearch(){
window.history.replaceState( state, '', window.location );
}
var searchList = new autoComplete({
new autoComplete({
/* selector for the search box element */
selectorToInsert: 'search:has(.searchbox)',
selector: '#R-search-by',
@ -254,12 +162,15 @@ function startSearch(){
},
/* renderItem displays individual search results */
renderItem: function( item, term ) {
var page = pagesIndex[item.index];
var page = item.page;
var context = [];
if( item.matches ){
var numContextWords = 2;
var contextPattern = '(?:\\S+ +){0,' + numContextWords + '}\\S*\\b(?:' +
item.matches.map( function(match){return match.replace(/\W/g, '\\$&')} ).join('|') +
')\\b\\S*(?: +\\S+){0,' + numContextWords + '}';
var context = page.content.match(new RegExp(contextPattern, 'i'));
context = page.content.match(new RegExp(contextPattern, 'i'));
}
var divsuggestion = document.createElement('div');
divsuggestion.className = 'autocomplete-suggestion';
divsuggestion.setAttribute('data-term', term);
@ -286,4 +197,24 @@ function startSearch(){
});
};
ready( startSearch );
function initSearch(){
init();
window.addEventListener( 'popstate', executeHistorySearch );
var input = document.querySelector('#R-search-by-detail');
if( input ){
input.addEventListener( 'keydown', function(event) {
// if we are pressing ESC in the search-by-detail our focus will
// be stolen by the other event handlers, so we have to refocus
// here after a short while
if (event.key == "Escape") {
setTimeout( function(){ input.focus(); }, 0 );
}
});
}
ready( initSearchAfterLoad );
}
initSearch();
})();

View file

@ -1531,7 +1531,7 @@ function searchInputHandler( value ){
}
function initSearch() {
// sync input/escape between searchbox and searchdetail
// sync input/escape between searchbox and executeSearch
var inputs = document.querySelectorAll( 'input.search-by' );
inputs.forEach( function( e ){
e.addEventListener( 'keydown', function( event ){
@ -1592,8 +1592,8 @@ function initSearch() {
});
}
window.relearn.isSearchInit = true;
window.relearn.runInitialSearch && window.relearn.runInitialSearch();
window.relearn.isSearchInterfaceReady = true;
window.relearn.executeInitialSearch && window.relearn.executeInitialSearch();
}
function updateTheme( detail ){