preprocessor.js

var path  = require('path');
var helper = require('./helper');


var preprocessor = {

  /**
   * Execute preprocessor on main, and embedded document
   * @param  {Object}   template
   * @param  {Function} callback
   */
  execute : function (template, callback) {
    if (template === null || template.files === undefined) {
      return callback(null, template);
    }
    for (var i = -1; i < template.embeddings.length; i++) {
      var _mainOrEmbeddedTemplate = template.filename;
      var _parentFilter = '';
      if (i > -1) {
        _mainOrEmbeddedTemplate = _parentFilter = template.embeddings[i];
      }
      var _fileType = path.extname(_mainOrEmbeddedTemplate);
      switch (_fileType) {
        case '.xlsx':
          preprocessor.convertSharedStringToInlineString(template, _parentFilter);
          break;
        default:
          break;
      }
    }
    return callback(null, template);
  },

  /**
   * [XLSX] Convert shared string to inline string in Excel format in order to be compatible with Carbone algorithm
   * @param  {Object} template     (modified)
   * @param  {String} parentFilter apply the transformation on a specific embedded file
   * @return {Object}              the modified template
   */
  convertSharedStringToInlineString : function (template, parentFilter) {
    var _sharedStrings = [];
    var _filesToConvert = [];
    var _sharedStringIndex = -1;
    // parse all files and find shared strings first
    for (var i = 0; i < template.files.length; i++) {
      var _file = template.files[i];
      if (_file.parent === parentFilter) {
        if (/sharedStrings\.xml$/.test(_file.name) === true) {
          _sharedStringIndex = i;
          _sharedStrings = preprocessor.readSharedString(_file.data);
        }
        else if (/\.xml$/.test(_file.name) === true) {
          if (_file.name.indexOf('sheet') !== -1) {
            _filesToConvert.push(_file);
          }
        }
      }
    }
    preprocessor.removeOneFile(template, _sharedStringIndex, parentFilter);
    // once shared string is found, convert files
    for (var f = 0; f < _filesToConvert.length; f++) {
      var _modifiedFile = _filesToConvert[f];
      _modifiedFile.data = preprocessor.removeRowCounterInWorksheet(
                    preprocessor.convertToInlineString(_modifiedFile.data, _sharedStrings)
                  );
    }
    return template;
  },
  
  /**
   * [XLSX] Remove one file in the template and its relations
   * @param  {Object} template
   * @param  {Integer} indexOfFileToRemove index of the file to remove
   * @param  {String} parentFilter         filter to modify only an embedded document
   * @return {}                            it modifies the template directly
   */
  removeOneFile : function (template, indexOfFileToRemove, parentFilter) {
    if (indexOfFileToRemove < 0 || indexOfFileToRemove >= template.files.length) {
      return;
    }
    var _fileToRemove = template.files[indexOfFileToRemove];
    var _dirname  = path.dirname(_fileToRemove.name);
    var _basename = path.basename(_fileToRemove.name);
    template.files.splice(indexOfFileToRemove, 1);
    for (var i = 0; i < template.files.length; i++) {
      var _file = template.files[i];
      if (_file.parent === parentFilter) {
        // remove relations
        if (_dirname + '/_rels/workbook.xml.rels' === _file.name) {
          var _regExp = new RegExp('<Relationship [^>]*Target="' + helper.regexEscape(_basename) + '"[^>]*\/>');
          _file.data = _file.data.replace(_regExp, '');
        }
      }
    }
  },

  /**
   * [XLSX] Parse and generate an array of shared string
   * @param  {String} sharedStringXml shared string content
   * @return {Array}                  array
   */
  readSharedString : function (sharedStringXml) {
    var _sharedStrings = [];
    if (sharedStringXml === null || sharedStringXml === undefined) {
      return _sharedStrings;
    }
    var _tagRegex = new RegExp('<si>(.+?)<\/si>','g');
    var _tag = _tagRegex.exec(sharedStringXml);
    while (_tag !== null) {
      _sharedStrings.push(_tag[1]);
      _tag = _tagRegex.exec(sharedStringXml);
    }      
    return _sharedStrings;
  },

  /**
   * [XLSX] Inject shared string in sheets
   * @param  {String} xml           sheets where to insert shared strings
   * @param  {Array} sharedStrings  shared string
   * @return {String}               updated xml
   */
  convertToInlineString : function (xml, sharedStrings) {
    if (typeof(xml) !== 'string') {
      return xml;
    }
    // find all tags which have attribute t="s" (type = shared string) 
    var _inlinedXml = xml.replace(/(<(\w)[^>]*t="s"[^>]*>)(.*?)(<\/\2>)/g, function (m, openTag, tagName, content, closeTag) {
      // change type of tag to "inline string"
      var _newXml = openTag.replace('t="s"', 't="inlineStr"');
      // get the index of shared string
      var _tab = /<v>(\d+?)<\/v>/.exec(content);
      if (_tab instanceof Array && _tab.length > 0) {
        // replace the index by the string
        var _sharedStringIndex = parseInt(_tab[1], 10);
        _newXml += '<is>' + sharedStrings[_sharedStringIndex] + '</is>' + closeTag;
        return _newXml;
      }
      // if something goes wrong, do nothing
      return m;
    });
    return _inlinedXml;
  },

  /**
   * [XLSX] Remove row and column counter (r=1, c=A1) in sheet (should be added in post-processing)
   * Carbone Engine cannot update these counter itself
   * @param  {String} xml sheet
   * @return {String}     sheet updated
   */
  removeRowCounterInWorksheet : function (xml) {
    if (typeof(xml) !== 'string') {
      return xml;
    }
    return xml.replace(/<(?:c|row)[^>]*(r="\S+")[^>]*>/g, function (m, rowValue) {
      return m.replace(rowValue, '');
    }).replace(/<(?:c|row)[^>]*(spans="\S+")[^>]*>/g, function (m, rowValue) {
      return m.replace(rowValue, '');
    }); 
  }
};

module.exports = preprocessor;