Find elements in 50 HTML files, extract data and create Jekyll post files with YAML metadata using Node.js

extract_data.js

// 1. Get list of files
// 2. For each file:
// 2.1 Read a file
// 2.2 Find required data
// 2.3 Put the information in a JSON object
// 3. Convert JSON object to YAML
// 4. Write new Markdown file in _posts folder

var fs = require('fs')


function getFileList(callback) {
  fs.readdir('.', function (err, files) {
    var hasHTMLextension = /\.html$/;
    if (err)
      throw err;
    for (var index in files) {
      //console.log(files[index]);
      if (hasHTMLextension.test(files[index])) {
        callback(null, files[index]);
      }
    }
  });
};

function readFile(err, fileName) {
  fs.readFile(fileName, 'utf8', function (err,data) {
    if (err) {
      return console.log(err);
    }
    //console.log(data);
    console.log('File' + fileName + ' successfully read');

    parseData(data, fileName);
  });
};

function parseData(data, fileName) {
  // Run some jQuery on a html fragment
  var jsdom = require("jsdom");

  jsdom.env(
    data,
    ["http://code.jquery.com/jquery.js"],
    function (errors, window) {
      var YAML = require('json2yaml');
      var yml;
      var membersJSON = new Object();
      var divTitle;
      var descriptionTitle;
      var divDescription, pDescription, description, memberLinkElement;
      var divInfo, spanInfoTitle, spanInfoTitleParentDiv, spanData;
      var divMemberLogo, imgTag, imgFullPath, imgFileNameOnly;
      var divEmbedContainer, iframeTag, youtubeURL;
      var i;

      membersJSON.layout = 'members';
      membersJSON.published = true;

      divTitle = window.document.getElementsByClassName('title');
      if (divTitle[0]) {
        membersJSON.title = divTitle[0].textContent;
      }

      // description_title
      divDescription = window.document.getElementsByClassName('description');
      if (divDescription[0]) {
        descriptionTitle = divDescription[0].getElementsByTagName('h3');
        if (descriptionTitle[0]) {
          membersJSON.description_title = descriptionTitle[0].textContent;
        }
      }

      // description
      divDescription = window.document.getElementsByClassName('description');
      if (divDescription[0]) {
        pDescription = divDescription[0].getElementsByTagName('p');
        description = pDescription[0].textContent;
        //membersJSON.description = pDescription[0].textContent;
        if (pDescription[1]) {
          memberLinkElement = pDescription[1].getElementsByTagName('a');
          if (memberLinkElement[0]) {
            membersJSON.member_url = memberLinkElement[0].href;
          } else {
            membersJSON.member_url = '';
          }
        }
      }

      // info block
      divInfo = window.document.getElementsByClassName('info');
      if (divInfo[0]) {
        spanInfoTitle = divInfo[0].getElementsByTagName('span');

        membersJSON.original_timestamp = spanInfoTitle[1].textContent;   
        membersJSON.contact = spanInfoTitle[3].textContent;   
        membersJSON.telephone = spanInfoTitle[5].textContent;   
        membersJSON.email = spanInfoTitle[7].textContent;   
        membersJSON.branch = spanInfoTitle[9].textContent;   
      }

      // Member Logo
      divMemberLogo = window.document.getElementsByClassName('member-logo');
      if (divMemberLogo[0]) {
        imgTag = divMemberLogo[0].getElementsByTagName('img');
          if (imgTag[0]) {
            imgFullPath = imgTag[0].src;
            imgFileNameOnly = imgFullPath.replace(/^file:\/\/\/Users\/hex\/Dropbox\/Sites\/study\/images\//i, '');
            membersJSON.member_logo = imgFileNameOnly;
        }
      }

      // YouTube URL
      divEmbedContainer = window.document.getElementsByClassName('embed-container');
      if (divEmbedContainer[0]) {
        iframeTag = divEmbedContainer[0].getElementsByTagName('iframe');
          if (iframeTag[0]) {
            youtubeURL = iframeTag[0].src;
            membersJSON.youtube_url = youtubeURL;
        }
      }
      
      yml = YAML.stringify(membersJSON);
      yml = yml.replace(/^---/, "---\n");
      yml = yml + "\n---" + "\n\n" + description + "\n";

      //console.log(membersJSON);
      console.log(yml);

      writeNewFile(fileName, yml);

    }
  );
};

function writeNewFile(fileName, yml) {
  var fileNameMarkdown = fileName.replace(/\.html$/i, ".markdown");
  var fullPath = './_posts/2014-01-09-' + fileNameMarkdown;
  fs.writeFile(fullPath, yml, function (err) {
    if (err) throw err;
    console.log(fileName + ' is saved!');
  });
};

getFileList(readFile);