368 lines
8.3 KiB
JavaScript
Executable File
368 lines
8.3 KiB
JavaScript
Executable File
/*
|
||
* Optimized by: EvanNotFound
|
||
*/
|
||
|
||
const articleLibrary = [];
|
||
const corpus = [];
|
||
let flag = null;
|
||
|
||
hexo.extend.filter.register("template_locals", function (localVariables) {
|
||
const cfg = hexo.theme.config.articles.recommendation;
|
||
if (!cfg.enable) {
|
||
return localVariables;
|
||
}
|
||
if (!flag) {
|
||
flag = 1;
|
||
fetchData(localVariables.site.posts, cfg);
|
||
fetchData(localVariables.site.pages, cfg);
|
||
articleRecommendation(cfg);
|
||
}
|
||
return localVariables;
|
||
});
|
||
|
||
function fetchData(s, cfg) {
|
||
s.each(function (p) {
|
||
if (["post", "docs"].includes(p.layout)) {
|
||
articleLibrary.push({
|
||
path: p.path,
|
||
title: p.title || p.seo_title || p.short_title,
|
||
headimg: p.thumbnail || p.banner || p.cover || cfg.placeholder,
|
||
});
|
||
corpus.push(tokenize(p.raw));
|
||
}
|
||
});
|
||
}
|
||
|
||
function cleanData(data) {
|
||
const symbolLists = [
|
||
",",
|
||
".",
|
||
"?",
|
||
"!",
|
||
":",
|
||
";",
|
||
"、",
|
||
"……",
|
||
"~",
|
||
"&",
|
||
"@",
|
||
"#",
|
||
"，",
|
||
"。",
|
||
"？",
|
||
"！",
|
||
"：",
|
||
"；",
|
||
"·",
|
||
"…",
|
||
"～",
|
||
"＆",
|
||
"＠",
|
||
"＃",
|
||
"“",
|
||
"”",
|
||
"‘",
|
||
"’",
|
||
"〝",
|
||
"〞",
|
||
'"',
|
||
"'",
|
||
"＂",
|
||
"＇",
|
||
"´",
|
||
"＇",
|
||
"(",
|
||
")",
|
||
"【",
|
||
"】",
|
||
"《",
|
||
"》",
|
||
"＜",
|
||
"＞",
|
||
"﹝",
|
||
"﹞",
|
||
"<",
|
||
">",
|
||
"(",
|
||
")",
|
||
"[",
|
||
"]",
|
||
"«",
|
||
"»",
|
||
"‹",
|
||
"›",
|
||
"〔",
|
||
"〕",
|
||
"〈",
|
||
"〉",
|
||
"{",
|
||
"}",
|
||
"［",
|
||
"］",
|
||
"「",
|
||
"」",
|
||
"｛",
|
||
"｝",
|
||
"〖",
|
||
"〗",
|
||
"『",
|
||
"』",
|
||
"︵",
|
||
"︷",
|
||
"︹",
|
||
"︿",
|
||
"︽",
|
||
"﹁",
|
||
"﹃",
|
||
"︻",
|
||
"︗",
|
||
"/",
|
||
"|",
|
||
"\\",
|
||
"︶",
|
||
"︸",
|
||
"︺",
|
||
"﹀",
|
||
"︾",
|
||
"﹂",
|
||
"﹄",
|
||
"﹄",
|
||
"︼",
|
||
"︘",
|
||
"／",
|
||
"｜",
|
||
"＼",
|
||
"_",
|
||
"¯",
|
||
"＿",
|
||
"￣",
|
||
"﹏",
|
||
"﹋",
|
||
"﹍",
|
||
"﹉",
|
||
"﹎",
|
||
"﹊",
|
||
"`",
|
||
"ˋ",
|
||
"¦",
|
||
"︴",
|
||
"¡",
|
||
"¿",
|
||
"^",
|
||
"ˇ",
|
||
"",
|
||
"¨",
|
||
"ˊ",
|
||
" ",
|
||
" ",
|
||
"%",
|
||
"*",
|
||
"-",
|
||
"+",
|
||
"=",
|
||
"￥",
|
||
"$",
|
||
"（",
|
||
"）",
|
||
];
|
||
data = data.replace(/\s/g, " ");
|
||
data = data.replace(/\!\[(.*?)\]\(.*?\)/g, (_a, b) => {
|
||
return b;
|
||
});
|
||
data = data.replace(
|
||
/(http|ftp|https):\/\/[\w\-_]+(\.[\w\-_]+)+([\w\-\.,@?^=%&:/~\+#]*[\w\-\@?^=%&/~\+#])?/g,
|
||
" ",
|
||
);
|
||
for (const symbol of symbolLists) {
|
||
data = data.replace(new RegExp("\\" + symbol, "g"), " ");
|
||
}
|
||
data = data.replace(/\d+/g, " ");
|
||
data = data.replace(/\s/g, " ");
|
||
return data;
|
||
}
|
||
|
||
function tokenize(data) {
|
||
const jieba = require("nodejieba");
|
||
return jieba
|
||
.cut(cleanData(data), true)
|
||
.filter((word) => word !== " " && !/^[0-9]*$/.test(word));
|
||
}
|
||
|
||
function cosineSimilarity(vector1, vector2) {
|
||
let numerator = 0;
|
||
let sqrt1 = 0;
|
||
let sqrt2 = 0;
|
||
if (vector1.length == vector2.length) {
|
||
for (let i = 0; i < vector1.length; i++) {
|
||
numerator += vector1[i] * vector2[i];
|
||
sqrt1 += vector1[i] * vector1[i];
|
||
sqrt2 += vector2[i] * vector2[i];
|
||
}
|
||
return numerator / (Math.sqrt(sqrt1) * Math.sqrt(sqrt2));
|
||
}
|
||
}
|
||
|
||
function createWordLibrary(allWords) {
|
||
const wordLibrary = {};
|
||
allWords.forEach((word) => {
|
||
wordLibrary[word] = 0;
|
||
});
|
||
return wordLibrary;
|
||
}
|
||
|
||
function calculateWordFrequency(wordLibrary, wordsInArticle) {
|
||
const wordOccurrenceLibrary = wordsInArticle.reduce(
|
||
(wordCountObj, wordName) => {
|
||
if (wordName in wordCountObj) {
|
||
wordCountObj[wordName]++;
|
||
}
|
||
return wordCountObj;
|
||
},
|
||
JSON.parse(JSON.stringify(wordLibrary)),
|
||
);
|
||
|
||
const wordFrequency = JSON.parse(JSON.stringify(wordLibrary));
|
||
for (const word of Object.keys(wordLibrary)) {
|
||
wordFrequency[word] = wordOccurrenceLibrary[word] / wordsInArticle.length;
|
||
}
|
||
return wordFrequency;
|
||
}
|
||
|
||
function articleRecommendation(cfg) {
|
||
const dataSet = {};
|
||
const similaritySet = {};
|
||
const recommendationSet = {};
|
||
let allWordsInAllArticles = [];
|
||
|
||
for (const wordList of corpus) {
|
||
allWordsInAllArticles = [
|
||
...new Set(allWordsInAllArticles.concat(wordList)),
|
||
];
|
||
}
|
||
|
||
const wordLibrary = createWordLibrary(allWordsInAllArticles);
|
||
const documentCountLibrary = JSON.parse(JSON.stringify(wordLibrary));
|
||
|
||
for (let i = 0; i < corpus.length; i++) {
|
||
const articlePath = articleLibrary[i].path;
|
||
const wordsInArticle = corpus[i];
|
||
|
||
const wordFrequency = calculateWordFrequency(wordLibrary, wordsInArticle);
|
||
dataSet[articlePath] = { wordFrequency };
|
||
|
||
for (const word of Object.keys(wordLibrary)) {
|
||
if (wordFrequency[word]) {
|
||
documentCountLibrary[word]++;
|
||
}
|
||
}
|
||
}
|
||
|
||
for (let i = 0; i < corpus.length; i++) {
|
||
const articlePath = articleLibrary[i].path;
|
||
dataSet[articlePath]["inverseDocumentFrequency"] = JSON.parse(
|
||
JSON.stringify(wordLibrary),
|
||
);
|
||
dataSet[articlePath]["wordFrequency-inverseDocumentFrequency"] = JSON.parse(
|
||
JSON.stringify(wordLibrary),
|
||
);
|
||
dataSet[articlePath]["wordFrequencyVector"] = [];
|
||
for (const word of Object.keys(wordLibrary)) {
|
||
const inverseDocumentFrequency = Math.log(
|
||
corpus.length / (documentCountLibrary[word] + 1),
|
||
);
|
||
const wordFrequencyInverseDocumentFrequency =
|
||
dataSet[articlePath]["wordFrequency"][word] * inverseDocumentFrequency;
|
||
dataSet[articlePath]["wordFrequencyVector"].push(
|
||
wordFrequencyInverseDocumentFrequency,
|
||
);
|
||
}
|
||
}
|
||
|
||
for (let i = 0; i < corpus.length; i++) {
|
||
const articlePath1 = articleLibrary[i].path;
|
||
similaritySet[articlePath1] = {};
|
||
for (let j = 0; j < corpus.length; j++) {
|
||
const articlePath2 = articleLibrary[j].path;
|
||
similaritySet[articlePath1][articlePath2] = cosineSimilarity(
|
||
dataSet[articlePath1]["wordFrequencyVector"],
|
||
dataSet[articlePath2]["wordFrequencyVector"],
|
||
);
|
||
}
|
||
for (let j = 0; j < corpus.length; j++) {
|
||
recommendationSet[articlePath1] = Object.keys(
|
||
similaritySet[articlePath1],
|
||
).sort(function (a, b) {
|
||
return similaritySet[articlePath1][b] - similaritySet[articlePath1][a]; // Descending order
|
||
});
|
||
}
|
||
const index = recommendationSet[articlePath1].indexOf(articlePath1);
|
||
if (index > -1) {
|
||
recommendationSet[articlePath1].splice(index, 1);
|
||
}
|
||
recommendationSet[articlePath1] = recommendationSet[articlePath1].slice(
|
||
0,
|
||
cfg.limit,
|
||
);
|
||
for (let j = 0; j < recommendationSet[articlePath1].length; j++) {
|
||
const e = recommendationSet[articlePath1][j];
|
||
recommendationSet[articlePath1][j] = articleLibrary.filter(
|
||
(w) => w.path == e,
|
||
)[0];
|
||
}
|
||
}
|
||
hexo.locals.set("recommendationSet", function () {
|
||
return recommendationSet;
|
||
});
|
||
}
|
||
|
||
hexo.extend.helper.register("articleRecommendationGenerator", function (post) {
|
||
if (!post) return "";
|
||
const cfg = hexo.theme.config.articles.recommendation;
|
||
if (!cfg.enable) {
|
||
return "";
|
||
}
|
||
for (const dir of cfg.skip_dirs) {
|
||
if (new RegExp("^" + dir, "g").test(post.path)) {
|
||
return "";
|
||
}
|
||
}
|
||
const recommendationSet = hexo.locals.get("recommendationSet");
|
||
const recommendedArticles = recommendationSet[post.path];
|
||
return userInterface(recommendedArticles, cfg);
|
||
});
|
||
|
||
function userInterface(recommendedArticles, cfg) {
|
||
let html = "";
|
||
let htmlMobile = "";
|
||
for (const item of recommendedArticles) {
|
||
html += itemInterface(item);
|
||
}
|
||
for (const itemMobile of recommendedArticles.slice(0, cfg.mobile_limit)) {
|
||
htmlMobile += itemInterface(itemMobile);
|
||
}
|
||
return `
|
||
<div class="recommended-article px-2 sm:px-6 md:px-8">
|
||
<div class="recommended-desktop">
|
||
<div class="recommended-article-header text-xl md:text-3xl font-bold mt-10">
|
||
<i aria-hidden="true"></i><span>${cfg.title}</span>
|
||
</div>
|
||
<div class="recommended-article-group">${html}</div>
|
||
</div>
|
||
<div class="recommended-mobile">
|
||
<div class="recommended-article-header text-xl md:text-3xl font-bold mt-10">
|
||
<i aria-hidden="true"></i><span>${cfg.title}</span>
|
||
</div>
|
||
<div class="recommended-article-group">${htmlMobile}</div>
|
||
</div>
|
||
</div>`;
|
||
}
|
||
|
||
function itemInterface(item) {
|
||
return `<a class="recommended-article-item" href="${
|
||
hexo.config.root + item.path
|
||
}" title="${item.title}" rel="bookmark">
|
||
<img src="${item.headimg}" alt="${item.title}" class="!max-w-none">
|
||
<span class="title">${item.title}</span>
|
||
</a>`;
|
||
}
|