Hexo使用Gulp压缩静态资源

1. Gulp简介

gulp是一个自动化构建工具，能够强化我们的前端工作流。

gulp is an open-source JavaScript toolkit created by Eric Schoffstall used as a streaming build system (similar to a more package-focussed Make) in front-end web development.

It is a task runner built on Node.js and npm, used for automation of time-consuming and repetitive tasks involved in web development like minification, concatenation, cache busting, unit testing, linting, optimization, etc.

gulp uses a code-over-configuration approach to define its tasks and relies on its small, single-purpose plugins to carry them out. The gulp ecosystem includes more than 3500 such plugins.

更多内容，参考wikipedia-gulp.js、Gulp官网和Gulp中文网。
本文中，会使用gulp来压缩hexo生成的静态资源文件，加快站点的访问速度。

2. 安装配置gulp

1、安装gulp
npm install --global gulp-cli

2、安装gulp模块

npm install gulp --save
npm install gulp-htmlclean gulp-htmlmin gulp-clean-css gulp-uglify gulp-imagemin --save
npm install gulp-babel babel-preset-env babel-preset-mobx --save
npm install -D @babel/core @babel/preset-react @babel/preset-env --save

最终生成的package.json为：

{
  "name": "hexo-site",
  "version": "0.0.0",
  "private": true,
  "scripts": {
    "build": "hexo generate",
    "clean": "hexo clean",
    "deploy": "hexo deploy",
    "server": "hexo server"
  },
  "hexo": {
    "version": "4.2.0"
  },
  "dependencies": {
    "babel-preset-env": "^1.7.0",
    "babel-preset-mobx": "^2.0.0",
    "gulp": "^4.0.2",
    "gulp-babel": "^8.0.0",
    "gulp-clean-css": "^4.2.0",
    "gulp-htmlclean": "^2.7.22",
    "gulp-htmlmin": "^5.0.1",
    "gulp-imagemin": "^7.1.0",
    "gulp-uglify": "^3.0.2",
    "hexo": "^4.0.0",
    "hexo-generator-archive": "^1.0.0",
    "hexo-generator-baidu-sitemap": "^0.1.6",
    "hexo-generator-category": "^1.0.0",
    "hexo-generator-feed": "^2.2.0",
    "hexo-generator-index": "^1.0.0",
    "hexo-generator-searchdb": "^1.2.0",
    "hexo-generator-sitemap": "^2.0.0",
    "hexo-generator-tag": "^1.0.0",
    "hexo-neat": "^1.0.4",
    "hexo-renderer-ejs": "^1.0.0",
    "hexo-renderer-marked": "^2.0.0",
    "hexo-renderer-stylus": "^1.1.0",
    "hexo-server": "^1.0.0"
  },
  "devDependencies": {
    "@babel/core": "^7.8.4",
    "@babel/preset-env": "^7.8.4",
    "@babel/preset-react": "^7.8.3"
  }
}

4、在hexo目录创建gulpfile.js，内容为：

let gulp = require('gulp')
let cleanCSS = require('gulp-clean-css')
let htmlmin = require('gulp-htmlmin')
let htmlclean = require('gulp-htmlclean')
let babel = require('gulp-babel') /* 转换为es2015 */
let uglify = require('gulp-uglify')
let imagemin = require('gulp-imagemin')

// 设置根目录
const root = './public'

// 匹配模式， **/*代表匹配所有目录下的所有文件
const pattern = '**/*'

// 压缩html
gulp.task('minify-html', function() {
  return gulp
    // 匹配所有 .html结尾的文件
    .src(`${root}/${pattern}.html`)
    .pipe(htmlclean())
    .pipe(
      htmlmin({
        removeComments: true,
        minifyJS: true,
        minifyCSS: true,
        minifyURLs: true
      })
    )
    .pipe(gulp.dest('./public'))
})

// 压缩css
gulp.task('minify-css', function() {
  return gulp
    // 匹配所有 .css结尾的文件
    .src(`${root}/${pattern}.css`)
    .pipe(
      cleanCSS({
        compatibility: 'ie8'
      })
    )
    .pipe(gulp.dest('./public'))
})

// 压缩js
gulp.task('minify-js', function() {
  return gulp
    // 匹配所有 .js结尾的文件
    .src(`${root}/${pattern}.js`)
    .pipe(
      babel({
        presets: ['env']
      })
    )
    .pipe(uglify())
    .pipe(gulp.dest('./public'))
})

// 压缩图片
gulp.task('minify-images', function() {
  return gulp
    // 匹配public/images目录下的所有文件
    .src(`${root}/images/${pattern}`)
    .pipe(
      imagemin(
        [
          imagemin.gifsicle({ optimizationLevel: 3 }),
          imagemin.jpegtran({ progressive: true }),
          imagemin.optipng({ optimizationLevel: 7 }),
          imagemin.svgo()
        ],
        { verbose: true }
      )
    )
    .pipe(gulp.dest('./public/images'))
})

gulp.task('default', gulp.series('minify-html', 'minify-css', 'minify-js'))

4、执行压缩
gulp

3. 命令精简

使用了gulp时候，构建发布需要四个命令：

hexo clean
hexo g
gulp
hexo d

这四个命令，可以都写在package.json。

"scripts": {
  "build": "hexo clean && hexo g && gulp",
  "deploy": "hexo clean && hexo g && gulp && hexo d"
}

构建只需要执行npm run build，构建发布只需要执行npm run deploy。

4. travis配置

对应的，修改.travis.yml配置为：

sudo: false
language: node_js
node_js:
- 10.16.3
cache: npm
branches:
  only:
  - master # build master branch only

env:
  global:
  - GIT_USER: voidking
  - HEXO_BACKUP_REPO: github.com/voidking/hexo-backup.git
  - HEXO_THEME_REPO: github.com/voidking/hexo-theme-next.git
  - GITHUB_PAGES_REPO: github.com/voidking/voidking.github.io.git
  - VOIDKING_REPO: github.com/voidking/voidking.git

before_install:
- export TZ='Asia/Shanghai'
- npm install hexo -g
- npm install gulp-cli -g

install:
- npm install

script:
- git clone https://${HEXO_THEME_REPO} themes/next
- git clone https://${GIT_USER}:${GITHUB_TOKEN}@${HEXO_BACKUP_REPO} hexo-backup
- mv hexo-backup/source .
- rm -rf source/private
- npm run build

after_success:
- git config --global user.name "voidking"
- git config --global user.email "voidking@qq.com"
- git clone https://${GIT_USER}:${GITHUB_TOKEN}@${GITHUB_PAGES_REPO} voidking
- unalias cp
- cp -rf public/. voidking
- cd voidking
- git add .
- git commit -m "Travis CI Auto Builder"
- git push --force --quiet "https://${GIT_USER}:${GITHUB_TOKEN}@${GITHUB_PAGES_REPO}" master:master

5. 精简search.xml

html、css和js都压缩了，很开心。但是，还有一个大文件没有压缩，就是本地搜索的DB文件search.xml。我的博客有接近600篇文章，这个search.xml文件的大小为7.5M，很大。
打开search.xml文件，发现里面不止包含文章内容，还包含html标签。参考hexo-generator-searchdb，发现可以设置不生成标签。

5.1. 修改localsearch配置

1、修改hexo/_config.yml的localsearch配置为：

# local search
search:
  path: search.xml
  field: post
  format: striptags
  limit: 10000

2、重新生成search.xml文件，新的文件只有3.5M。

3、修改local-search.js
为了正常使用搜索功能，需要修改hexo主题的local-search.js。如果不修改的话，搜索时只搜索标题，不会搜索内容。编辑next/source/js/local-search.js，如下修改：

// line 120
// let content = data.content ? data.content.trim().replace(/<[^>]+>/g, '') : '';
let content = data.content;

以上，实现了search.xml的精简，nice。

5.2. 自定义精简

但是，3.5M依然很大，能不能再精简一下？可以。
1、localsearch的format改回html。

2、修改xml_generator.js
编辑hexo/node_modules/hexo-generator-searchdb/lib/xml_generator.js，定义自己想要删除的字符：

'use strict';

const path = require('path');
const fs = require('fs');
const nunjucks = require('nunjucks');
var env = new nunjucks.Environment();

var searchTmplSrc = path.join(__dirname, '../templates/search.xml');
var searchTmpl = nunjucks.compile(fs.readFileSync(searchTmplSrc, 'utf8'), env);

var stripe_code_line_num = function(str) { // 去除代码
    return str.replace(/<figure class="highlight.*?<\/figure>/ig, '');
}
var stripe = function (str) { // 去除html标签
    return str.replace(/(<([^>]+)>)/ig, '');
}
var minify = function (str) { // 压缩成一行
    return str.trim().replace(/\n/g, ' ').replace(/\s+/g, ' ');
}

module.exports = function(locals) {
  var config = this.config;
  var database = require('./database')(locals, config);
  database.forEach( function(element, index) {
    element.content = minify(stripe(stripe_code_line_num(element.content)));
  });
  var xml = searchTmpl.render({
    articles: database,
    config  : config.search
  });
  return {
    path: config.search.path,
    data: xml
  };
};

3、重新生成search.xml，这次只有2.5M，nice。

以上两种精简search.xml的方法都很好，这里我选择使用修改localsearch format的方法。因为如果在xml_generator.js不删除文章中的代码，两种方法的压缩结果基本相同，而方法一通用性更好。