pto-assignments/lib/xml.js

Library and CLI tool for parsing USPTO assignment data

var Assignment = require('./assignment');

var XMLDoc = require('./xml_doc');

var Assignment = module.exports = function(xml) {

  // <assignment-record>
  this.reel = parseInt(xml.descendantWithPath('assignment-record.reel-no').val, 10);
  this.frame = parseInt(xml.descendantWithPath('assignment-record.frame-no').val, 10);
  this.lastUpdate = parseDate(xml.descendantWithPath('assignment-record.last-update-date.date').val);
  this.purge = xml.descendantWithPath('assignment-record.purge-indicator').val;
  this.recorded = parseDate(xml.descendantWithPath('assignment-record.recorded-date.date').val);
  this.pages = parseInt(xml.descendantWithPath('assignment-record.page-count').val, 10);

  var correspondent = xml.descendantWithPath('assignment-record.correspondent');
  this.correspondent = {};
  this.correspondent.name = correspondent.descendantWithPath('name').val;

  this.correspondent.address = [];

  for (var i = 0; i < 4; i++) { // see DTD, maximum of 4 address lines, minimum of 1
    var addr = correspondent.descendantWithPath('address-' + i);
    if (addr) {
      this.correspondent.address.push(addr.val);
    }
  }

  this.description = xml.descendantWithPath('assignment-record.conveyance-text').val;

  // <patent-assignors>
  var assignors = this.assignors = [];
  xml.descendantWithPath('patent-assignors').eachChild(function(child) {

    var assignor = {
      name: child.descendantWithPath('name').val
    };



    if (child.descendantWithPath('execution-date.date')) {
      assignor.executed = parseDate(child.descendantWithPath('execution-date.date').val);
    }

    if (child.descendantWithPath('date-acknowledged.date')) { // TODO: rare
      assignor.acknowledged = parseDate(child.descendantWithPath('date-acknowledged.date').val);
    }

    if (child.descendantWithPath('address-1')) {
      assignor.address = [];
      assignor.address.push(child.descendantWithPath('address-1').val);
    }

    if (child.descendantWithPath('address-2')) {
      assignor.address.push(child.descendantWithPath('address-2').val);
    }

    assignors.push(assignor);
  });


  // <patent-assignees>

  var assignees = this.assignees = [];
  xml.descendantWithPath('patent-assignees').eachChild(function(child) {

    var assignee = {
      name: child.descendantWithPath('name').val
    };

    if (child.descendantWithPath('address-1')) {
      assignee.address = [];
      assignee.address.push(child.descendantWithPath('address-1').val);
    }

    if (child.descendantWithPath('address-2')) {
      assignee.address = assignee.address || [];
      assignee.address.push(child.descendantWithPath('address-2').val);
    }

    if (child.descendantWithPath('city')) {
      assignee.city = child.descendantWithPath('city').val;
    }

    if (child.descendantWithPath('state')) {
      assignee.state = child.descendantWithPath('state').val;
    }

    if (child.descendantWithPath('country-name')) {
      assignee.country = child.descendantWithPath('country-name').val;
    } else {
      assignee.country = 'UNITED STATES';
    }

    if (child.descendantWithPath('postcode')) {
      assignee.postcode = child.descendantWithPath('postcode').val;
    }

    assignees.push(assignee);
  });


  // <patent-properties>
  var properties = this.properties = [];
  xml.descendantWithPath('patent-properties').eachChild(function(child) {

    var property = {};

    if (child.descendantWithPath('invention-title')) {
      property.title = child.descendantWithPath('invention-title').val;
    }

    property.documents = [];

    child.eachChild(function(grandchild) {
      if (grandchild.name === 'document-id') {
        var doc = {
          country: grandchild.descendantWithPath('country').val,
          number: grandchild.descendantWithPath('doc-number').val
        };

        if (grandchild.descendantWithPath('kind')) {
          doc.kindCode = grandchild.descendantWithPath('kind').val;
        }

        if (grandchild.descendantWithPath('name')) {
          doc.name = grandchild.descendantWithPath('name').val;
        }

        if (grandchild.descendantWithPath('date')) {
          doc.date = parseDate(grandchild.descendantWithPath('date').val);
        }


        property.documents.push(doc);
      }
    });

    properties.push(property);
  });
};

var XML = module.exports = function(xml, options) {

  // Valid XML 

  if (!xml)
    throw Error('Missing argument: an XML string is required as the first parameters');
  if ('string' !== typeof(xml))
    throw Error('Invalid argument: first argument should be an XML string');

  var parser = new XMLDoc(xml);


  this.dtd = parser.attribute('dtd-version', {
    post: parseFloat
  });
  this.produced = parser.attribute('date-produced', {
    post: parseDate
  });
  this.actionKeyCode = parser.text('action-key-code');
  this.transacted = parser.text('transaction-date.date', {
    post: parseDate
  });

  var assignments = this.assignments = [];

  var emptyFile = parser.doc.descendantWithPath('patent-assignments.data-available-code');
  
  if (emptyFile && emptyFile.val === 'N') {
    this.assignments = null;
    return;
  }

  parser.set('patent-assignments').eachChild(function(child) {
    assignments.push(new Assignment(child));
  });


};






function parseDate(str) {
  var parts = str.match(/(\d\d\d\d)(\d\d)(\d\d)/).splice(-3);
  var month = parseInt(parts[1], 10) - 1;
  return +new Date(parts[0], month, parts[2]);
}