Bookhoard/foliate-js
1
0
Fork 0
mirror of https://github.com/john-okeefe/foliate-js.git synced 2026-09-09 11:29:14 -04:00
Code Issues Packages Projects Releases Wiki Activity

Add non-collator/segmenter-based search

Browse Source
This commit is contained in:
John Factotum
2023-03-26 03:53:56 +08:00
parent 45bfe6ee71
commit 42a410571c
1 changed files with 24 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
search.js
+24 -8
View File
@@ -20,23 +20,31 @@ const makeExcerpt = (strs, { startIndex, startOffset, endIndex, endOffset }) =>
return { pre, match, post } return { pre, match, post }
} }
// TODO: maybe use this for exact matches as it would be faster const simpleSearch = function* (strs, query, options = {}) {
/* const { locales = 'en', sensitivity } = options
export const simpleSearch = function* (strs, query, locales = 'en') { const matchCase = sensitivity === 'variant'
const haystack = strs.join('') const haystack = strs.join('')
const lowerHaystack = haystack.toLocaleLowerCase(locales) const lowerHaystack = matchCase ? haystack : haystack.toLocaleLowerCase(locales)
const needle = query.toLocaleLowerCase(locales) const needle = matchCase ? query : query.toLocaleLowerCase(locales)
const needleLength = needle.length const needleLength = needle.length
let index = -1 let index = -1
let strIndex = -1
let sum = 0
do { do {
index = lowerHaystack.indexOf(needle, index + 1) index = lowerHaystack.indexOf(needle, index + 1)
if (index > -1) { if (index > -1) {
while (sum <= index) sum += strs[++strIndex].length
const startIndex = strIndex
const startOffset = index - (sum - strs[strIndex].length)
const end = index + needleLength const end = index + needleLength
// TODO while (sum <= end) sum += strs[++strIndex].length
const endIndex = strIndex
const endOffset = end - (sum - strs[strIndex].length)
const range = { startIndex, startOffset, endIndex, endOffset }
yield { range, excerpt: makeExcerpt(strs, range) }
} }
} while (index > -1) } while (index > -1)
} }
*/
const segmenterSearch = function* (strs, query, options = {}) { const segmenterSearch = function* (strs, query, options = {}) {
const { locales = 'en', granularity = 'word', sensitivity = 'base' } = options const { locales = 'en', granularity = 'word', sensitivity = 'base' } = options
@@ -92,11 +100,19 @@ const segmenterSearch = function* (strs, query, options = {}) {
} }
} }
export const search = (strs, query, options) => {
const { granularity = 'grapheme', sensitivity = 'base' } = options
if (!Intl?.Segmenter || granularity === 'grapheme'
&& (sensitivity === 'variant' || sensitivity === 'accent'))
return simpleSearch(strs, query, options)
return segmenterSearch(strs, query, options)
}
export const searchMatcher = (textWalker, opts) => { export const searchMatcher = (textWalker, opts) => {
const { defalutLocale, matchCase, matchDiacritics, matchWholeWords } = opts const { defalutLocale, matchCase, matchDiacritics, matchWholeWords } = opts
return function* (doc, query) { return function* (doc, query) {
const iter = textWalker(doc, function* (strs, makeRange) { const iter = textWalker(doc, function* (strs, makeRange) {
for (const result of segmenterSearch(strs, query, { for (const result of search(strs, query, {
locales: doc.body.lang || doc.documentElement.lang || defalutLocale || 'en', locales: doc.body.lang || doc.documentElement.lang || defalutLocale || 'en',
granularity: matchWholeWords ? 'word' : 'grapheme', granularity: matchWholeWords ? 'word' : 'grapheme',
sensitivity: matchDiacritics && matchCase ? 'variant' sensitivity: matchDiacritics && matchCase ? 'variant'