A NodeJS command to convert html files which are generated with the theme "next" by "hexo" to markdown files. Should use npm to install [ cheerio to-markdown walk ] firstly.

README.md

A NodeJS command to convert html files which are generated with the theme next by hexo to markdown files.

You should use npm to install cheerio to-markdown walk firstly, and then use node command to run the script.

hexo_next_html_2_md.js

#!/usr/bin/env node

"use strict";

// 检查 node 版本
const execSync = require('child_process').execSync;
const nodeVer = execSync('node -v').toString().replace(/\r|\n/ig, "").replace('v', '');
if (nodeVer < '8.0.0') {
  Utils.logRed('当前 node 版本：' + nodeVer + '，请更新至 8.0.0 或以上（建议使用 nvm 管理 node）');
  process.exit();
}

var ascii = "\
\n   ('-.             ('-.              (`-.                         .-') _           \
\n  ( OO ).-.       _(  OO)           _(OO  )_                      (  OO) )          \
\n  / . --. /      (,------.      ,--(_/   ,. \\        ,-.-')       /     '._        \
\n  | \\-.  \\        |  .---'      \\   \\   /(__/        |  |OO)      |'--...__)    \
\n.-'-'  |  |       |  |           \\   \\ /   /         |  |  \\      '--.  .--'     \
\n \\| |_.'  |      (|  '--.         \\   '   /,         |  |(_/         |  |         \
\n  |  .-.  |       |  .--'          \\     /__)       ,|  |_.'         |  |          \
\n  |  | |  |       |  `---.          \\   /          (_|  |            |  |          \
\n  `--' `--'       `------'           `-'             `--'            `--'           \
";

console.log(ascii);



var fs = require('fs');
var walk = require ('walk');
var toMarkdown = require('to-markdown');
var cheerio = require('cheerio');
var URL = require('url').URL;

var mdDirName = '_posts_bk';


walkFiles();


var files = [];
function walkFiles() {
  var walker = walk.walk('..', { filters: ['hexo_html_2_md', 'node_modules', mdDirName, 'about'] });
  walker.on('file', function (root, stat, next) {
    if (stat.name === 'index.html') {
      files.push(root + '/' + stat.name);
    }
    next();
  });
  walker.on('end', function () {
    try {
      for (var i in files) {
        startConvert(files[i]);
      }
    } catch (error) {
      console.error('error ' + i + ': ' + files[i] + ': ', error.message);
    }
  });
}

function startConvert(filePath) {
  var elements = getElememts(filePath);
  var header = getHeader(elements);
  var mdStr = getMarkdown(elements.articleBody, elements.articleCodeTags);
  saveFile(elements, header, mdStr);
}




// ------------ private method
function getElememts(htmlFilePath) {
  var htmlStr = fs.readFileSync(htmlFilePath, 'utf8');
  var $ = cheerio.load(htmlStr);

  var elements = {};

  // article body
  var allBody = convertCodeTag($);
  elements.articleBody = allBody.body;
  elements.articleCodeTags = allBody.codeTags;

  // file name
  var url = $('meta[property="og:url"]').attr('content');
  if (url.indexOf('/about/index.html') !== -1) {
    elements.fileName = 'about';
  } else {
    url = new URL(url);
    var pathnames = url.pathname.split('/');
    elements.fileName = decodeURI(pathnames[4]);
  }

  // title
  var title = $('meta[property="og:title"]').attr('content').replace('%CF%80', 'π');
  elements.title = title;

  // date
  var date = $('time[itemprop="dateCreated"]').attr('datetime');
  date = date.replace(/T/, ' ').replace(/\+(.+)/, '');
  elements.date = date;

  // tags
  var tags = '[' + $('meta[name="keywords"]').attr('content').replace(/,$/gi, '') + ']';
  elements.tags = tags;

  // category
  var category = $('a[itemprop="url"][rel="index"] span[itemprop="name"]').text();
  elements.category = category;

  // layout
  var layout = 'post';
  elements.layout = layout;

  return elements;
}

function getHeader(elements) {
  var header = '---\n' +
    'title: ' + elements.title + '  \n' +
    'date: ' + elements.date + '  \n' +
    'tags: ' + elements.tags + '  \n' +
    'category: ' + elements.category + '  \n' +
    'layout: ' + elements.layout + '  \n' +
    '\n---\n\n';
  return header;
}

function convertCodeTag($) {
  var codeTags = [];
  $('figure').each(function (i0, el0) {
    // console.log($(el0).find('td[class="code"] pre').html());

    var codeTagHtml = $(el0).find('td[class="code"] pre').html();
    if (codeTagHtml) {
      var arr = codeTagHtml.split('<br>');

      var code = '';
      for (var i in arr) {
        var _ = cheerio.load(arr[i]);
        code += (_.text() + (i < arr.length - 1 ? '\n' : ''));
      }
      code = '\n```\n' + code + '```\n';

      var key = '__code__tag__' + i0;
      codeTags[key] = code;
      $(el0).replaceWith(key);
    }
  })
  convertTable($);
  return { body: $('span[itemprop="articleBody"]').html(), codeTags: codeTags };
}

/*
// 以 <span class="line"> 来作为一行的分隔，但是发现有些是最外层有 <span class="line">，里面还有 <span class="line">，所以这样转会有问题
// 即这样转出来是有问题的： <span class="line"> test1 <span class="line"> test2 </span> </span>
// 这样转出来就没问题： <span class="line"> test1 </span> <span class="line"> test2 </span>
// 经观察，还是以 <br> 来作为一行的分隔才没问题，所以使用以上方法
function convertCodeTag_old($) {
  var codeTags = [];
  $('figure').each(function(i0, el0) {
    $(el0).find('td[class="code"] pre').each(function (i1, el1) {

      var code = '';
      $(el1).find('span[class="line"]').each(function (i2, el2) {
        code = code + $(el2).text() + '\n';
      });

      code = '\n```\n' + code + '```\n';
      var key = '__code__tag__' + i0;
      codeTags[key] = code;
      $(el0).replaceWith(key);
    });
  })
  convertTable($);
  return { body: $('span[itemprop="articleBody"]').html(), codeTags: codeTags };
}
*/

function convertTable($) {
  $('table').each(function (i0, el0) {
    
    var headers = [];
    var bodies = [];

    if ($(el0).has('thead')) {
      $(el0).find('thead').each(function (i1, el1) {
        $(el0).find('tr th').each(function(i11, el11) {
          headers.push($(el11).text());
        });
        var rows = $(el0).find('tbody tr').each(function (i2, el2) {
          var aRow = [];
          $(el2).find('td').each(function (i3, el3) {
            aRow.push($(el3).text());
          });
          bodies.push(aRow);
        });
      });
    }

    if (headers.length > 0) {

      var headerStr = '|';
      var sep = '|';
      for (var aHeader in headers) {
        headerStr += (headers[aHeader] + '|');
        sep += '---|';
      }

      var rowStr = '';
      for (var aRow in bodies) {
        var aRowStr = '|';
        for (var one in bodies[aRow]) {
          aRowStr += (bodies[aRow][one] + '|');
        }
        rowStr += (aRowStr + '<br>');
      }

      var result = '<br>' + headerStr + '<br>' + sep + '<br>' + rowStr;
      // console.log(result);

      $(el0).replaceWith(result);
    }
  });
}

function getMarkdown(articleBody, articleCodeTags) {
  var mdStr = toMarkdown(articleBody);
  for (var key in articleCodeTags) {
    if (articleCodeTags.hasOwnProperty(key)) {
      var element = articleCodeTags[key];
      mdStr = mdStr.replace(key, element);
    }
  }
  mdStr = mdStr.replace('<a id="more"></a>', '<!--more-->').replace(/\[\]\(\#(.+)\)/g, '').replace(/```  /g, '```').replace(/<figure class="half">|<\/figure>/g, '');
  return mdStr; walk
}

function saveFile(elements, header, mdStr) {
  var dir = '../' + mdDirName + '/';
  if (!fs.existsSync(dir)) {
    fs.mkdirSync(dir);
  }
  fs.writeFileSync(dir + elements.fileName + '.md', header + mdStr, 'utf8');
  console.log('Saved ' + elements.fileName);
}