Entity Tool

search.js

/**
 * Saebekassebil <Jakob Miland> 2012
 * GeekShadow <Antoine Turmel> 2012
 *
 * Put this file in top of the project directory
 * $ node search.js
 *
 **/

var fs = require('fs'),
    url = require('url'),
    path = require('path'),
    child_process = require('child_process'),
    exec = child_process.exec, child, dtds = {}, entitylist = {};

function processFile(path, callback) {
  fs.readFile(path, 'utf8', function(error, data) {
    if(error) {
      return console.error('Error reading file: ' + error);
    }

    // Search for included DTDs
    var matches = data.match(/SYSTEM\s+\"(.+?\.dtd)\"/g);
    if(matches) {
      matches.forEach(function(match) {
        var dtdfile = match.match(/SYSTEM\s+\"(.+?\.dtd)\"/)[1];
        if(!(dtdfile in dtds)) {
          var dtd = processDTDFile(dtdfile);
          if(dtd) {
            dtds[dtdfile] = dtd;
            console.log("Parsed DTD file: " + dtdfile);
          }
        }
      });
    }

    // Search for entities
    matches = data.match(/\&([a-zA-Z0-9\._]+)/g);
    if(matches) {
      matches.forEach(function(entity) {
        entity = entity.substr(1); // cut off &
        var adtd = hasEntity(entity);
        if(entity in entitylist) {
          entitylist[entity]++;
        }

        if(adtd) {
          //console.log(adtd + ' has ' + entity);
        } else {
          //console.warn(entity + ' not found');
        }
      });
    }

    if(typeof callback === 'function') {
      callback();
    }
  });
}

function hasEntity(entity) {
  for(var i in dtds) {
    if(dtds.hasOwnProperty(i)) {
      if(entity in dtds[i]) {
        return i;
      }
    }
  }

  return false;
}

function processDTDFile(file) {
  // Translate chrome URL to absolute URL
  var parsedURL = url.parse(file), filepath;
  if(parsedURL.protocol !== 'chrome:') {
    return null;
  }

  if(parsedURL.host === 'songbird') {
    filepath = './locales/en-US/songbird/' + path.basename(parsedURL.pathname);
  } else if(parsedURL.host === 'branding') {
    filepath = './locales/branding/' + path.basename(parsedURL.pathname);
  } else {
    return null;
  }

  // Open and parse DTD file
  var dtd = fs.readFileSync(filepath, 'utf8');
  var lines = dtd.split('\n');
  var regex = /<\!ENTITY\s+([a-zA-Z0-9\._]+)\s+\"(.+?)\"\s*?>/;
  var entities = {};
  for(var i = 0, length = lines.length; i < length; i++) {
    var match = lines[i].match(regex);
    if(match) {
      if(match[1] in entities) {
        console.warn('Overwritting entity ' + match[1]);
      }

      entities[match[1]] = match[2];
      entitylist[match[1]] = 0;
    }
  }

  return entities;
}

child = exec('find ./ | grep \'[.]xul\\|[.]xml\\|[.]xhtml\'', function(error, stdout, stderr) {
  if(error) {
    return console.error('Error finding files: ' + error);
  }

  var files = stdout.split('\n');
  files = files.slice(0, files.length-1);

  var active = -1;
  function fileCallback() {
    active++;
    if(active <= files.length-1) {
      processFile(files[active], fileCallback);
    } else {
      console.log('Unused entities:');
      var count = 0;
      for(var i in entitylist) {
        if(entitylist[i] === 0) {
          count++;
          console.log(i);
        }
      }

      console.log('A total of ' + count + ' unused entities');
    }
  }

  fileCallback();
});