|
| 1 | +import { ToWords } from "to-words"; |
| 2 | +import { getSample } from "./services/elasticsearch.js"; |
| 3 | + |
| 4 | +const toWords = new ToWords({ |
| 5 | + localeCode: "en-US", |
| 6 | + converterOptions: { |
| 7 | + ignoreDecimal: false, |
| 8 | + doNotAddOnly: true, |
| 9 | + }, |
| 10 | +}); |
| 11 | + |
| 12 | +function stringToWordArray(string) { |
| 13 | + let symbolRegex = |
| 14 | + /_|\+|=|\)|\(|\[|{|}|]|;|:|"|'|<|>|\.|,|\/|\?|\||\\|!|@|#|\$|%|\^|&|\*/g; |
| 15 | + let workingString = string.replaceAll("-", " "); |
| 16 | + workingString = workingString.replaceAll(symbolRegex, " "); |
| 17 | + let stringArray = workingString.split(" "); |
| 18 | + return stringArray.filter((entry) => entry.trim() != ""); |
| 19 | +} |
| 20 | + |
| 21 | +function kwProcessor(terms, kwArr) { |
| 22 | + for (let term in terms) { |
| 23 | + terms[term] = terms[term].toLowerCase(); |
| 24 | + } |
| 25 | + let foundKws = []; |
| 26 | + |
| 27 | + for (let word in terms) { |
| 28 | + for (let group in kwArr) { |
| 29 | + let currentGroup = kwArr[group]; |
| 30 | + for (let index in currentGroup) { |
| 31 | + if (currentGroup[index] == terms[word]) { |
| 32 | + foundKws.push(...currentGroup); |
| 33 | + break; |
| 34 | + } |
| 35 | + } |
| 36 | + } |
| 37 | + } |
| 38 | + if (foundKws) return [...new Set(foundKws)]; |
| 39 | +} |
| 40 | + |
| 41 | +async function getNumerals(stringArr) { |
| 42 | + let numerals = []; |
| 43 | + let nameWordLen = 0; |
| 44 | + for (let word in stringArr) { |
| 45 | + let curWord = stringArr[word]; |
| 46 | + if (validateRomanNumeral(curWord)) { |
| 47 | + nameWordLen = word; |
| 48 | + let numeral = parseNumeral(curWord); |
| 49 | + if (numeral) numerals.push(numeral); |
| 50 | + } |
| 51 | + } |
| 52 | + //Guard clause, exits when we didn't find a valid numeral |
| 53 | + if (!nameWordLen) return; |
| 54 | + let searchQuery = stringArr.slice(0, nameWordLen).join(" ").trim(); |
| 55 | + //Check if this is a series |
| 56 | + let results = await getSample(searchQuery); |
| 57 | + let series = false; |
| 58 | + //always return ii if it's available |
| 59 | + for (let x in numerals) { |
| 60 | + if (numerals[x] == 2) return [...new Set(numerals)]; |
| 61 | + } |
| 62 | + if (results.length > 1) { |
| 63 | + for (let x in results) { |
| 64 | + let seriesNumeral = []; |
| 65 | + let words = stringToWordArray(results[x].sample); |
| 66 | + for (let word in words) { |
| 67 | + let numeral = parseNumeral(words[word]); |
| 68 | + if (numeral) seriesNumeral.push(numeral); |
| 69 | + } |
| 70 | + if (seriesNumeral > 0) { |
| 71 | + for (let x in numerals) { |
| 72 | + for (let y in seriesNumeral) { |
| 73 | + if (numerals[x] != seriesNumeral[y]) { |
| 74 | + series = true; |
| 75 | + } |
| 76 | + } |
| 77 | + } |
| 78 | + } |
| 79 | + } |
| 80 | + if (!series) return; |
| 81 | + numerals.push(getNumberNames(numerals)); |
| 82 | + return [...new Set(numerals)]; |
| 83 | + } |
| 84 | +} |
| 85 | + |
| 86 | +function parseNumeral(string) { |
| 87 | + //Keep these upper case to reduce the number of false positives. Make sure the input isn't tolower |
| 88 | + const romanNumerals = { |
| 89 | + /*M: 1000, |
| 90 | + CM: 900, |
| 91 | + D: 500, |
| 92 | + CD: 400, |
| 93 | + C: 100, |
| 94 | + XC: 90, |
| 95 | + L: 50, |
| 96 | + XL: 40,*/ |
| 97 | + X: 10, |
| 98 | + IX: 9, |
| 99 | + V: 5, |
| 100 | + IV: 4, |
| 101 | + I: 1, |
| 102 | + }; |
| 103 | + if (validateRomanNumeral(string)) { |
| 104 | + let numeralSum = 0; |
| 105 | + string = string.toUpperCase(); |
| 106 | + for (let numeral in romanNumerals) { |
| 107 | + while (string.startsWith(numeral)) { |
| 108 | + numeralSum += romanNumerals[numeral]; |
| 109 | + string = string.substring(numeral.length); |
| 110 | + } |
| 111 | + } |
| 112 | + if (string.length > 0) return 0; |
| 113 | + return numeralSum; |
| 114 | + } |
| 115 | +} |
| 116 | + |
| 117 | +function getNumberNames(stringArr) { |
| 118 | + let numbers = []; |
| 119 | + for (let number in stringArr) { |
| 120 | + let curNum = stringArr[number]; |
| 121 | + if (/^\d+$/.test(curNum)) { |
| 122 | + let numberName = toWords.convert(parseInt(curNum)); |
| 123 | + if (numberName) numbers.push(numberName.trim()); |
| 124 | + } |
| 125 | + } |
| 126 | + return [...new Set(numbers)]; |
| 127 | +} |
| 128 | + |
| 129 | +function validateRomanNumeral(string) { |
| 130 | + if (!string) return false; |
| 131 | + if (string == "vim") return false; |
| 132 | + let romanRegex = /i|v|x|l|c|d|m/gi; |
| 133 | + return !string.replaceAll(romanRegex, ""); |
| 134 | +} |
| 135 | + |
| 136 | +export async function optimizeKws(object) { |
| 137 | + for (let column in object.keywords) { |
| 138 | + if (!object.data[column]) continue; |
| 139 | + let wordArr = stringToWordArray(object.data[column]); |
| 140 | + let workKws = kwProcessor(wordArr, object.keywords[column]); |
| 141 | + //special case for filenames |
| 142 | + if (column == "filename") { |
| 143 | + let numerals = await getNumerals(wordArr); |
| 144 | + if (numerals) { |
| 145 | + workKws.push(...numerals); |
| 146 | + } |
| 147 | + workKws.push(...getNumberNames(wordArr)); |
| 148 | + } |
| 149 | + object.data[column + "kws"] = workKws.join(" ").trim(); |
| 150 | + } |
| 151 | + return object.data; |
| 152 | +} |
0 commit comments