arrFilterByKeywords.mjs

  1. import map from 'lodash-es/map.js'
  2. import size from 'lodash-es/size.js'
  3. import trim from 'lodash-es/trim.js'
  4. import isearr from './isearr.mjs'
  5. import isestr from './isestr.mjs'
  6. import sep from './sep.mjs'
  7. import strleft from './strleft.mjs'
  8. import strdelleft from './strdelleft.mjs'
  11. /**
  12.  * 對arr陣列內各字串進行關鍵字計算，計算是否含有關鍵字與程度，多關鍵字用空白區隔，必要關鍵字可於字首添加「+」，不要關鍵字可於字首添加「-」
  13.  *
  14.  * Unit Test: {@link https://github.com/yuda-lyu/wsemi/blob/master/test/arrFilterByKeywords.test.mjs Github}
  15.  * @memberOf wsemi
  16.  * @param {Array} arr 輸入要被尋找的字串陣列
  17.  * @param {String|Array} keywords 輸入要尋找的關鍵字字串或陣列
  18.  * @returns {Array} 回傳結果物件陣列，陣列內各物件包含hasKeyword與weight欄位，其中hasKeyword代表有無關鍵字為布林值，weight代表權重為浮點數，其值可大於1
  19.  * @example
  20.  *
  21.  * let arr = [
  22.  *     'abc def xyz',
  23.  *     '測試abc中文mnop',
  24.  *     'Instead of creating yet another opinionated application',
  25.  *     'Node.js module which can be integrated into a larger application',
  26.  * ]
  27.  * let kws = null
  28.  * let r = null
  29.  *
  30.  * kws = 'abc'
  31.  * r = arrFilterByKeywords(arr, kws)
  32.  * console.log(r)
  33.  * // => [
  34.  * //     { hasKeyword: true, weight: 1 },
  35.  * //     { hasKeyword: true, weight: 1 },
  36.  * //     { hasKeyword: false, weight: 0 },
  37.  * //     { hasKeyword: false, weight: 0 }
  38.  * // ]
  39.  *
  40.  * kws = 'def'
  41.  * r = arrFilterByKeywords(arr, kws)
  42.  * console.log(r)
  43.  * // => [
  44.  * //     { hasKeyword: true, weight: 1 },
  45.  * //     { hasKeyword: false, weight: 0 },
  46.  * //     { hasKeyword: false, weight: 0 },
  47.  * //     { hasKeyword: false, weight: 0 }
  48.  * // ]
  49.  *
  50.  * kws = 'def 中文'
  51.  * r = arrFilterByKeywords(arr, kws)
  52.  * console.log(r)
  53.  * // => [
  54.  * //     { hasKeyword: true, weight: 1 },
  55.  * //     { hasKeyword: true, weight: 0.25 },
  56.  * //     { hasKeyword: false, weight: 0 },
  57.  * //     { hasKeyword: false, weight: 0 }
  58.  * // ]
  59.  *
  60.  * kws = 'def 中文 mnop'
  61.  * r = arrFilterByKeywords(arr, kws)
  62.  * console.log(r)
  63.  * // => [
  64.  * //     { hasKeyword: true, weight: 1 },
  65.  * //     { hasKeyword: true, weight: 0.5555555555555557 },
  66.  * //     { hasKeyword: false, weight: 0 },
  67.  * //     { hasKeyword: false, weight: 0 }
  68.  * // ]
  69.  *
  70.  * kws = 'def +yet'
  71.  * r = arrFilterByKeywords(arr, kws)
  72.  * console.log(r)
  73.  * // => [
  74.  * //     { hasKeyword: false, weight: 0 },
  75.  * //     { hasKeyword: false, weight: 0 },
  76.  * //     { hasKeyword: true, weight: 1 },
  77.  * //     { hasKeyword: false, weight: 0 }
  78.  * // ]
  79.  *
  80.  * kws = 'def of module -yet'
  81.  * r = arrFilterByKeywords(arr, kws)
  82.  * console.log(r)
  83.  * // => [
  84.  * //     { hasKeyword: true, weight: 1 },
  85.  * //     { hasKeyword: true, weight: 1 },
  86.  * //     { hasKeyword: false, weight: 0 },
  87.  * //     { hasKeyword: true, weight: 1 }
  88.  * // ]
  89.  *
  90.  * kws = '+'
  91.  * r = arrFilterByKeywords(arr, kws)
  92.  * console.log(r)
  93.  * // => [
  94.  * //     { hasKeyword: false, weight: 0 },
  95.  * //     { hasKeyword: false, weight: 0 },
  96.  * //     { hasKeyword: false, weight: 0 },
  97.  * //     { hasKeyword: false, weight: 0 }
  98.  * // ]
  99.  *
  100.  * kws = '-'
  101.  * r = arrFilterByKeywords(arr, kws)
  102.  * console.log(r)
  103.  * // => [
  104.  * //     { hasKeyword: false, weight: 0 },
  105.  * //     { hasKeyword: false, weight: 0 },
  106.  * //     { hasKeyword: false, weight: 0 },
  107.  * //     { hasKeyword: false, weight: 0 }
  108.  * // ]
  109.  *
  110.  * kws = 'def +'
  111.  * r = arrFilterByKeywords(arr, kws)
  112.  * console.log(r)
  113.  * // => [
  114.  * //     { hasKeyword: true, weight: 1 },
  115.  * //     { hasKeyword: false, weight: 0 },
  116.  * //     { hasKeyword: false, weight: 0 },
  117.  * //     { hasKeyword: false, weight: 0 }
  118.  * // ]
  119.  *
  120.  * kws = ['def', 'of', 'module', '-yet']
  121.  * r = arrFilterByKeywords(arr, kws)
  122.  * console.log(r)
  123.  * // => [
  124.  * //     { hasKeyword: true, weight: 1 },
  125.  * //     { hasKeyword: true, weight: 1 },
  126.  * //     { hasKeyword: false, weight: 0 },
  127.  * //     { hasKeyword: true, weight: 1 }
  128.  * // ]
  129.  *
  130.  * kws = ['can be', 'def']
  131.  * r = arrFilterByKeywords(arr, kws)
  132.  * console.log(r)
  133.  * // => [
  134.  * //     { hasKeyword: true, weight: 0.25 },
  135.  * //     { hasKeyword: false, weight: 0 },
  136.  * //     { hasKeyword: false, weight: 0 },
  137.  * //     { hasKeyword: true, weight: 1 }
  138.  * // ]
  139.  *
  140.  * kws = ['+abc']
  141.  * r = arrFilterByKeywords(arr, kws)
  142.  * console.log(r)
  143.  * // => [
  144.  * //     { hasKeyword: true, weight: 1 },
  145.  * //     { hasKeyword: true, weight: 1 },
  146.  * //     { hasKeyword: false, weight: 0 },
  147.  * //     { hasKeyword: false, weight: 0 }
  148.  * // ]
  149.  *
  150.  * kws = ['-abc']
  151.  * r = arrFilterByKeywords(arr, kws)
  152.  * console.log(r)
  153.  * // => [
  154.  * //     { hasKeyword: false, weight: 0 },
  155.  * //     { hasKeyword: false, weight: 0 },
  156.  * //     { hasKeyword: true, weight: 1 },
  157.  * //     { hasKeyword: true, weight: 1 }
  158.  * // ]
  159.  *
  160.  */
  161. function arrFilterByKeywords(arr, keywords) {
  163.     //check
  164.     if (!isearr(arr)) {
  165.         return []
  166.     }
  168.     //check
  169.     if (!isestr(keywords) && !isearr(keywords)) {
  170.         return []
  171.     }
  173.     //kws
  174.     let kws = null
  175.     if (isestr(keywords)) {
  176.         kws = sep(keywords, ' ') //若為字串則用空白切分出各關鍵字
  177.     }
  178.     else {
  179.         kws = keywords //可支援關鍵字內含空白
  180.     }
  182.     //n
  183.     let n = size(kws)
  185.     function core(c) {
  187.         //for
  188.         let b = false
  189.         let weight = 0
  190.         for (let k = 0; k < n; k++) {
  191.             let kw = kws[k]
  193.             //ekw
  194.             let ekw = kw
  195.             let bInclude = strleft(kw, 1) === '+'
  196.             let bExclude = strleft(kw, 1) === '-'
  197.             if (bInclude || bExclude) {
  198.                 ekw = strdelleft(kw, 1)
  199.             }
  201.             //check
  202.             if (trim(ekw) === '') {
  203.                 continue
  204.             }
  206.             //bHas
  207.             let bHas = c.indexOf(ekw) >= 0
  208.             // console.log('c', c, ':ekw', ekw, ':bHas', bHas)
  210.             //必有關鍵字
  211.             if (bInclude) {
  212.                 if (!bHas) {
  213.                     b = false //若無必有關鍵字, 強制視為找不到
  214.                     weight = 0
  215.                     break
  216.                 }
  217.                 else {
  218.                     b = true //找到必有關鍵字, 視為找到
  219.                     weight = 1
  220.                     break
  221.                 }
  222.             }
  224.             //不能有關鍵字
  225.             if (bExclude) {
  226.                 if (bHas) {
  227.                     b = false //找到不能有的關鍵字, 強制視為找不到
  228.                     weight = 0
  229.                     break
  230.                 }
  231.                 else {
  232.                     b = true //找不到不能有關鍵字, 視為找到
  233.                     weight = 1
  234.                     break
  235.                 }
  236.             }
  238.             //含有關建字則依照順序給予權重
  239.             if (bHas) {
  240.                 b = true
  241.                 let w = 1 - k / n
  242.                 w = w ** 2 //非線性遞減, 有第1關鍵字權重1, 只有2+3關鍵字權重也不超過第1關鍵字權重
  243.                 weight += w //權重累加可超過1
  244.             }
  246.         }
  248.         return {
  249.             hasKeyword: b,
  250.             weight,
  251.         }
  252.     }
  254.     //rs
  255.     let rs = map(arr, (c) => {
  256.         return core(c)
  257.     })
  259.     return rs
  260. }
  263. export default arrFilterByKeywords