add stream.Transform API implementation

parser.js

@@ -1,34 +1,43 @@
 var expat = require('node-expat')
 var _ = require('lodash')
+var util = require('util')
+var stream = require('stream')
+
 var ParserState = require('./parserState')
 
-function XmlParser (xmlStream, opts) {
+function XmlParser (opts) {
   this.opts = opts || {}
   this.parserState = new ParserState()
   this.parser = new expat.Parser('UTF-8')
-  var scope = this
-  this.parser.pause = function () {
-    xmlStream.pause()
-    scope.parser.stop()
+  var transformOpts = { readableObjectMode: true }
+  stream.Transform.call(this, transformOpts)
 }
-  this.parser.restart = function () {
-    scope.parser.resume()
-    xmlStream.resume()
-  }
-  process.nextTick(function () {
-    parse.call(scope, xmlStream)
-  })
-  return this.parser
+util.inherits(XmlParser, stream.Transform)
+
+XmlParser.prototype._transform = function (chunk, encoding, callback) {
+  if (!this.opts.resourcePath) this.emit('error', new Error('resourcePath missing'))
+  if (encoding !== 'buffer') this.emit('error', new Error('unsupported encoding'))
+
+  this.parse(chunk)
+  callback()
 }
 
-function parse (xmlStream) {
-  if (!this.opts.resourcePath) this.parser.emit('error', new Error('resourcePath missing'))
+XmlParser.prototype.parse = function (chunk) {
   var scope = this
-  var parser = scope.parser
+  var parser = this.parser
   var state = this.parserState
   var lastIndex
   var resourcePath = this.opts.resourcePath
 
+  if (state.isRootNode) registerEvents()
+
+  if (typeof chunk === 'string') {
+    parser.parse('', true)
+  } else {
+    parser.parse(chunk.toString())
+  }
+
+  function registerEvents () {
     parser.on('startElement', function (name, attrs) {
       if (state.isRootNode) validateResourcePath(name)
       state.currentPath = state.currentPath + '/' + name
@@ -48,10 +57,14 @@ function parse (xmlStream) {
       if (state.isPathfound) processText(text)
     })
 
-  parser.on('end', function () {
-    parser.emit('finish')
+    parser.on('error', function (err) {
+      scope.emit('error', new Error(err + 'at line no:' + parser.getCurrentLineNumber() + ' on column no:' + parser.getCurrentColumnNumber()))
     })
 
+    parser.on('end', function () {
+    })
+  }
+
   function processStartElement (name, attrs) {
     if (!name) return
     var obj = {}
@@ -81,8 +94,8 @@ function parse (xmlStream) {
     var rpath = resourcePath.substring(0, index)
 
     if (rpath === state.currentPath) {
-      if (scope.opts.emitEventsOnNodeName) parser.emit(name, state.object)
-      parser.emit('data', state.object)
+      if (scope.opts.emitEventsOnNodeName) scope.emit(name, state.object)
+      scope.push(state.object)
       state.object = {}
     }
   }
@@ -146,10 +159,15 @@ function parse (xmlStream) {
     temp = temp.substring(0, index)
 
     if (temp !== name) {
-      xmlStream.end()
+      this.end()
     }
   }
 }
 
+XmlParser.prototype._flush = function (callback) {
+  this.parse('')
+  callback()
+}
+
 module.exports = XmlParser
 

test/test.js

@@ -7,7 +7,7 @@ describe('Tests', function () {
   describe('simple behaviour testing', function () {
     it('should properly parse a simple file.', function (done) {
       var xmlStream = fs.createReadStream('./test/TestFiles/item.xml')
-      var parser = new ParserFactory(xmlStream, {resourcePath: '/items/item'})
+      var parser = new ParserFactory({resourcePath: '/items/item'})
       var expectedData = [
                            { '$': { id: '1', test: 'hello' },
                              subitem:
@@ -39,7 +39,7 @@ describe('Tests', function () {
 
     it('should properly parse a medium size file.', function (done) {
       var xmlStream = fs.createReadStream('./test/TestFiles/medium.xml')
-      var parser = new ParserFactory(xmlStream, {resourcePath: '/items/item'})
+      var parser = new ParserFactory({resourcePath: '/items/item'})
 
       var dataEventCount = 0
 
@@ -61,7 +61,7 @@ describe('Tests', function () {
 
     it('should properly parse a file containing many nodes.', function (done) {
       var xmlStream = fs.createReadStream('./test/TestFiles/manyItems.xml')
-      var parser = new ParserFactory(xmlStream, {resourcePath: '/items/item'})
+      var parser = new ParserFactory({resourcePath: '/items/item'})
 
       var dataEventCount = 0
 
@@ -83,7 +83,7 @@ describe('Tests', function () {
 
     it('should properly parse a xml simple file in which nodes contain text values randomly.', function (done) {
       var xmlStream = fs.createReadStream('./test/TestFiles/randomText.xml')
-      var parser = new ParserFactory(xmlStream, {resourcePath: '/items/item'})
+      var parser = new ParserFactory({resourcePath: '/items/item'})
       var expectedData = [ { '$': { 'id': '1', 'test': 'hello' }, '_': ' item  one  two',
                              'subitem': [ { '$': { 'sub': 'TESTING SUB' }, '_': 'one' },
                                           { '$': { 'sub': '2' }, '_': 'two' } ] },
@@ -111,5 +111,48 @@ describe('Tests', function () {
       })
       xmlStream.pipe(parser)
     })
+
+    it('should properly parse a file containing many nodes.', function (done) {
+      var xmlStream = fs.createReadStream('./test/TestFiles/manyItems.xml')
+      var parser = new ParserFactory({resourcePath: '/items/item'})
+      // console.log(parser)
+      var dataEventCount = 0
+
+      parser.on('data', function (data) {
+        dataEventCount++
+      })
+
+      parser.on('error', function (err) {
+        done(err)
+      })
+
+      parser.on('end', function () {
+        // console.log('dataEventCount=', dataEventCount)
+        dataEventCount.should.equal(296)
+        done()
+      })
+      xmlStream.pipe(parser)
+    })
+
+    it('should properly parse a huge file.', function (done) {
+      var xmlStream = fs.createReadStream('./test/TestFiles/hugeFile.xml')
+      var parser = new ParserFactory({resourcePath: '/items/item'})
+      // console.log(parser)
+      var dataEventCount = 0
+      parser.on('data', function (data) {
+        dataEventCount++
+      })
+
+      parser.on('error', function (err) {
+        done(err)
+      })
+
+      parser.on('end', function () {
+        // console.log('dataEventCount=', dataEventCount)
+        dataEventCount.should.equal(2072)
+        done()
+      })
+      xmlStream.pipe(parser)
+    })
   })
 })