mstf-kit
Version:
一个现代化的 JavaScript/TypeScript 工具库，提供了丰富的常用工具函数
153 lines (152 loc) • 3.16 kB
JavaScript
const UYGHUR_CHARS = [
"ئ",
"ا",
"ە",
"ب",
"پ",
"ت",
"ج",
"چ",
"خ",
"د",
"ر",
"ز",
"ژ",
"س",
"ش",
"غ",
"ف",
"ق",
"ك",
"گ",
"ڭ",
"ل",
"م",
"ن",
"ھ",
"و",
"ۇ",
"ۆ",
"ۈ",
"ۋ",
"ې",
"ى",
"ي"
];
const UYGHUR_COMMON_WORDS = [
"مەن",
"سەن",
"ئۇ",
"بىز",
"سىلەر",
"ئۇلار",
// 人称代词
"ۋە",
"بىلەن",
"ئۈچۈن",
"لېكىن",
"چۈنكى",
// 连词
"بار",
"يوق",
"كەلدى",
"كەتتى",
"ياخشى",
"چوڭ",
// 常用词
"قانداق",
"نېمە",
"قاچان",
"قەيەر",
"كىم"
// 疑问词
];
const CHINESE_CHAR_PATTERN = /[\u4e00-\u9fa5]/;
function isUyghurText(text, threshold = 0.4) {
if (!text || text.trim().length === 0) {
return false;
}
const cleanText = text.replace(/[\s\p{P}]/gu, "");
if (cleanText.length === 0) {
return false;
}
let uyghurCharCount = 0;
for (const char of cleanText) {
if (UYGHUR_CHARS.includes(char)) {
uyghurCharCount++;
}
}
const uyghurCharRatio = uyghurCharCount / cleanText.length;
const containsCommonWords = UYGHUR_COMMON_WORDS.some((word) => text.includes(word));
if (uyghurCharRatio >= threshold) {
return true;
} else if (uyghurCharRatio >= 0.3 && containsCommonWords) {
return true;
}
return false;
}
function extractUyghurText(text) {
if (!text || text.trim().length === 0) {
return [];
}
const uyghurPattern = new RegExp(`[${UYGHUR_CHARS.join("")}]+`, "g");
const matches = text.match(uyghurPattern);
return matches ? matches.filter((match) => match.length > 1) : [];
}
function compareUyghurAndChineseRatio(text, threshold = 1.5) {
if (!text || text.trim().length === 0) {
return {
uyghurCharCount: 0,
chineseCharCount: 0,
ratio: 0,
isUyghurDominant: false,
dominantScript: "Balanced"
};
}
const cleanText = text.replace(/[\s\p{P}]/gu, "");
if (cleanText.length === 0) {
return {
uyghurCharCount: 0,
chineseCharCount: 0,
ratio: 0,
isUyghurDominant: false,
dominantScript: "Balanced"
};
}
let uyghurCharCount = 0;
let chineseCharCount = 0;
for (const char of cleanText) {
if (UYGHUR_CHARS.includes(char)) {
uyghurCharCount++;
} else if (CHINESE_CHAR_PATTERN.test(char)) {
chineseCharCount++;
}
}
const ratio = chineseCharCount > 0 ? uyghurCharCount / chineseCharCount : uyghurCharCount > 0 ? Infinity : 0;
let dominantScript;
const isUyghurDominant2 = ratio >= threshold;
if (ratio >= threshold) {
dominantScript = "Uyghur";
} else if (ratio <= 1 / threshold) {
dominantScript = "Chinese";
} else {
dominantScript = "Balanced";
}
return {
uyghurCharCount,
chineseCharCount,
ratio,
isUyghurDominant: isUyghurDominant2,
dominantScript
};
}
function isUyghurDominant(text, ratio = 1.5) {
const result = compareUyghurAndChineseRatio(text, ratio);
return result.isUyghurDominant;
}
export {
compareUyghurAndChineseRatio,
extractUyghurText,
isUyghurDominant,
isUyghurText
};