xml-toolkit 1.0.15 → 1.0.17

package/README.md

@@ -8,6 +8,23 @@ npm i xml-toolkit
 ```
 
 # Using
+
+* [Parsing a small file completely](XMLParser)
+
+```js
+const fs = require ('fs')
+const {XMLParser} = require ('xml-toolkit')
+
+const xml    = fs.readFileSync ('doc.xml')
+const parser = new XMLParser  ({...options})
+
+const document = parser.process (xml)
+
+for (const element of document.detach ().children) {
+  console.log (element.attributes)
+}
+```
+
 * [Reading a Record List](https://github.com/do-/node-xml-toolkit/wiki/Use-Case:-Reading-a-Record-List)
 
 ```js

package/index.js

@@ -8,5 +8,8 @@ for (const name of [
 	'XMLNode',
 	'XMLSchemata',
 	'SOAP11',
+	'EntityResolver',
+	'XMLIterator',
+	'XMLParser',
 
 ]) module.exports [name] = require ('./lib/' + name)

package/lib/AttributesMap.js

@@ -2,15 +2,13 @@ const CH_COLON = ':'.charCodeAt (0)
 
 const AttributesMap = class extends Map {
 
-	constructor (xmlReader, nsMap = null) {
+	constructor (entityResolver = null, nsMap = null) {
 
 		super ()
 
-		this._xmlReader = xmlReader
+		this.entityResolver = entityResolver
 		this._nsMap = nsMap
 
-		this.fixText = xmlReader.fixText
-
 	}
 
 	getLocalName (name) {
@@ -65,7 +63,17 @@ const AttributesMap = class extends Map {
 
 		}
 
-		return super.set (k, v === '' ? null : this.fixText (v))
+		const {entityResolver} = this
+
+		return super.set (k,
+
+			v === ''                ? null : 
+
+			entityResolver === null ? v    :
+			
+			entityResolver.fix (v)
+			
+		)
 
 	}
 

package/lib/SAXEvent.js

@@ -7,7 +7,7 @@ const CH_QUESTION    = '?'.charCodeAt (0)
 const CH_SLASH       = '/'.charCodeAt (0)
 const CH_MINUS       = '-'.charCodeAt (0)
 
-const STR_XML        = 'XML'
+const STR_XML        = 'xml'
 const STR_CDATA      = '[CDATA['
 const STR_DOCTYPE    = 'DOCTYPE'
 
@@ -46,14 +46,8 @@ const SAXEvent = class {
 						return END_ELEMENT
 
 					case CH_QUESTION: 
-
-						switch (src.slice (2, 5).toUpperCase ()) {
-
-							case STR_XML: return START_DOCUMENT
-
-							default: return PROCESSING_INSTRUCTION
-
-						}
+					
+						return src.slice (2, 5) === STR_XML ? START_DOCUMENT : PROCESSING_INSTRUCTION
 
 					case CH_EXCLAMATION:
 
@@ -123,19 +117,19 @@ const SAXEvent = class {
 	
 	get name () {
 	
-		if (this._afterName === 0) this.findAfterName ()
-		
 		const {src} = this, start = src.charCodeAt (1) === CH_SLASH ? 2 : 1
 
+		if (this._afterName === 0) this.findAfterName (start)
+
 		return this.src.slice (start, this._afterName).trim ()
 
 	}
 	
-	findAfterName () {
+	findAfterName (start) {
 	
 		const {src} = this, {length} = src; 
 
-		for (let i = 1; i < length; i ++) {
+		for (let i = start; i < length; i ++) {
 		
 			const c = src.charCodeAt (i)
 			
@@ -228,4 +222,20 @@ SAXEvent.TYPES = {
 	CDATA,
 }
 
+SAXEvent.KET = new Map ([
+
+	[START_DOCUMENT,         '?>'],
+	[PROCESSING_INSTRUCTION, '?>'],
+
+	[START_ELEMENT,           '>'],
+	[END_ELEMENT,             '>'],
+	[DTD,                     '>'],
+
+	[CDATA,                 ']]>'],
+	[COMMENT,               '-->'],
+
+	[CHARACTERS,              '<'],
+
+])
+
 module.exports = SAXEvent

package/lib/XMLIterator.js

@@ -0,0 +1,57 @@
+const XMLNode  = require ('./XMLNode.js')
+const SAXEvent = require ('./SAXEvent.js')
+const EntityResolver = require ('./EntityResolver.js')
+
+const XMLIterator = class {
+
+	constructor (src, options = {}) {
+	
+		this.src          = src
+		this.options      = options
+		this.pos          = 0
+		this.selfEnclosed = null
+		this.entityResolver = options.entityResolver || new EntityResolver ()
+
+	}
+	
+	[Symbol.iterator] () {
+	    return this;
+	}
+	
+	autoClose () {
+	
+		let value = this.selfEnclosed
+
+		value.type = SAXEvent.TYPES.END_ELEMENT
+		
+		this.selfEnclosed = null
+	
+		return {value}
+
+	}
+
+	next () {
+	
+		if (this.selfEnclosed !== null) return this.autoClose ()
+	
+		const {src, pos, entityResolver} = this
+
+		if (src.length - pos < 1) return {done: true}
+		
+		const value = new XMLNode (src.slice (pos), entityResolver)
+
+		value.trim ()
+		
+		const {length} = value.src; if (length === 0) return {done: true}
+
+		this.pos += length
+
+		if (value.isStartElement && value.isSelfEnclosed) this.selfEnclosed = value
+
+		return {value}
+
+	}
+	
+}
+
+module.exports = XMLIterator

package/lib/XMLMarshaller.js

@@ -196,7 +196,9 @@ const XMLMarshaller = class {
 	appendContent (node, data) {
 
 		const {localName, namespaceURI, attributes, children, targetNamespace} = node, {name, type, ref} = attributes
-	
+		
+		if (localName === 'attribute') return
+
 		if (ref) return this.appendContent (this.xs.getByReference (ref), data)
 
 		switch (localName) {

package/lib/XMLNode.js

@@ -3,7 +3,7 @@ const AttributesMap = require ('./AttributesMap')
 const NamespacesMap = require ('./NamespacesMap')
 const MoxyLikeJsonEncoder = require ('./MoxyLikeJsonEncoder')
 
-const XML_READER = Symbol ('_xmlReader')
+const ENTITY_RESOLVER = Symbol ('_entityResolver')
 const ATTRIBUTES = Symbol ('_attributes')
 const TYPE       = Symbol ('_type')
 const PARENT     = Symbol ('_parent')
@@ -16,13 +16,13 @@ const m2o =
 
 const XMLNode = class extends SAXEvent {
 
-	constructor (src, xmlReader, _type) {
+	constructor (src, entityResolver = null, _type) {
 
 		super (src)
 
 		this [TYPE] = _type || super.type
 
-		this [XML_READER] = xmlReader
+		this [ENTITY_RESOLVER] = entityResolver
 		this [PARENT]     = null
 		this [LEVEL]      = 0
 		
@@ -32,7 +32,7 @@ const XMLNode = class extends SAXEvent {
 
 	cloneStart () {
 
-		let e = new XMLNode (this.src, this [XML_READER], SAXEvent.TYPES.START_ELEMENT)
+		let e = new XMLNode (this.src, this [ENTITY_RESOLVER], SAXEvent.TYPES.START_ELEMENT)
 
 		e [PARENT]     = this [PARENT]
 		e [LEVEL]      = this [LEVEL]
@@ -110,7 +110,7 @@ const XMLNode = class extends SAXEvent {
 	
 	readAttributes () {
 
-		const m = new AttributesMap (this [XML_READER], this [NS_MAP])
+		const m = new AttributesMap (this [ENTITY_RESOLVER], this [NS_MAP])
 		
 		this.writeAttributesToMap (m)
 		
@@ -129,22 +129,53 @@ const XMLNode = class extends SAXEvent {
 	get text () {
 
 		let s = super.text
+
+		if (this [TYPE] === SAXEvent.TYPES.CHARACTERS) {
+			
+			const entityResolver = this [ENTITY_RESOLVER]
+		
+			if (entityResolver) return entityResolver.fix (s)
 		
-		if (this [TYPE] === SAXEvent.TYPES.CHARACTERS) s = this [XML_READER].fixText (s)
+		}
 
 		return s
 
 	}
 
+	detach_children (list, o = {}) {
+	
+		if (!Array.isArray (list) || list.length === 0) return []
+
+		let last = null, a = []; for (const i of list) {
+		
+			const d = i.detach (); if (d === null) continue
+			
+			if (d instanceof String && last instanceof String) {
+
+				last += d
+
+			}
+			else {
+
+				a.push (last = d)
+
+			}					
+		
+		}
+		
+		return a
+
+	}
+	
 	detach (o = {}) {
 	
 		switch (this.type) {
-
+		
 			case SAXEvent.TYPES.CHARACTERS:
 			case SAXEvent.TYPES.CDATA:
 				return this.text
 				
-			default:
+			case SAXEvent.TYPES.END_ELEMENT:
 
 				const {localName, namespaceURI, attributes, children, namespacesMap} = this
 				
@@ -152,16 +183,46 @@ const XMLNode = class extends SAXEvent {
 					localName,
 					namespaceURI,
 					attributes    : m2o (attributes),
-					children      : (children || []).map (n => n.detach (o)),
+					children      : this.detach_children (children, o),
 				}
 
 				if (o.nsMap) r.namespacesMap = m2o (namespacesMap || [])
 
 				return r
+				
+			default:
 
+				return null
+				
 		}
 
 	}
+	
+	trim () {
+	
+		const {src, type} = this, ket = SAXEvent.KET.get (type)
+		
+		let pos = src.indexOf (ket)
+		
+		if (ket !== '<') {
+
+			if (pos === -1) {
+
+				const max = 20
+
+				let txt = src; if (txt.length > max) txt = txt.slice (0, max) + '...'
+
+				throw new Error (`Unfinished ${type}: ${txt}`)
+
+			}
+
+			pos += ket.length
+		
+		}
+		
+		this.src = src.slice (0, pos)
+	
+	}
 
 }
 

package/lib/XMLParser.js

@@ -0,0 +1,86 @@
+const assert      = require ('assert')
+const SAXEvent    = require ('./SAXEvent.js')
+const XMLNode     = require ('./XMLNode.js')
+const XMLIterator = require ('./XMLIterator.js')
+
+const XMLReader = class {
+
+	constructor (options = {}) {
+		
+		if (!('stripSpace' in options)) options.stripSpace = true
+		assert (options.stripSpace === true || options.stripSpace === false, 'options.stripSpace must be boolean, not ' + typeof options.stripSpace)
+
+		if (!('useEntities' in options)) options.useEntities = true
+		assert (options.useEntities === true || options.useEntities === false, 'options.useEntities must be boolean, not ' + typeof options.useEntities)
+
+		if (!('useNamespaces' in options)) options.useNamespaces = true
+		assert (options.useNamespaces === true || options.useNamespaces === false, 'options.useNamespaces must be boolean, not ' + typeof options.useNamespaces)
+
+		this.stripSpace    = options.stripSpace
+		this.useEntities   = options.useEntities
+		this.useNamespaces = options.useNamespaces
+
+		if (this.useEntities) this.entityResolver = new (require ('./EntityResolver.js')) ()
+		
+
+	}
+	
+	process (src) {
+	
+		this.text = ''
+		this.document = null
+		this.element = null
+
+		const {entityResolver} = this, nodes = new XMLIterator (src, {entityResolver})
+		
+		for (const node of nodes) {
+
+			let {type} = node
+
+			switch (type) {
+
+				case SAXEvent.TYPES.CHARACTERS:
+				case SAXEvent.TYPES.CDATA:
+
+					this.text += node.text					
+					continue
+
+				default:
+
+					if (this.text.length === 0) break
+					if (this.stripSpace) this.text = this.text.trim ()
+					if (this.text.length === 0) break
+					(new XMLNode (this.text, this.entityResolver, SAXEvent.TYPES.CHARACTERS)).parent = this.element
+					this.text = ''
+
+			}
+
+			switch (type) {
+							
+				case SAXEvent.TYPES.START_ELEMENT:
+					
+					node.children = []
+					if (this.useNamespaces) node.readNamespaces ()
+					node.parent = this.element
+					this.element = node
+					if (this.document === null) this.document = node
+					break
+
+				case SAXEvent.TYPES.END_ELEMENT:
+
+					if (this.element === null) throw new Error (`Unbalanced end element tag "${node.text}" occured at position ${nodes.pos}`)
+					this.element.type = type
+					this.element = this.element.parent
+					break
+
+			}
+
+		}
+				
+		return this.document
+
+	}
+
+}
+
+module.exports = XMLReader

package/lib/XMLReader.js

@@ -123,7 +123,7 @@ const XMLReader = class extends Transform {
 	
 		this.flush_text ()
 
-		this.publish (new XMLNode ('', this), SAXEvent.TYPES.END_DOCUMENT)
+		this.publish (new XMLNode ('', this.entityResolver), SAXEvent.TYPES.END_DOCUMENT)
 
 		callback ()
 
@@ -168,7 +168,7 @@ const XMLReader = class extends Transform {
 		
 		if (text.length !== 0) {
 		
-			let e = new XMLNode (text, this, SAXEvent.TYPES.CHARACTERS)
+			let e = new XMLNode (text, this.entityResolver, SAXEvent.TYPES.CHARACTERS)
 			
 			e.parent = this.element
 			
@@ -179,22 +179,12 @@ const XMLReader = class extends Transform {
 		this.text = ''
 
 	}
-	
-	get fixText () {
-	
-		if (!this.useEntities) return s => s
-		
-		const {entityResolver} = this
-
-		return s => entityResolver.fix (s)
-
-	}
 
 	_transform (chunk, encoding, callback) {
 
 		const {length} = chunk; if (length !== 0) {
 
-			let e = new XMLNode (chunk, this), {type} = e, {element} = this
+			let e = new XMLNode (chunk, this.entityResolver), {type} = e, {element} = this
 
 			switch (type) {
 

package/lib/XMLSchema.js

@@ -23,7 +23,9 @@ const XMLSchema = class extends Map {
 		
 		for (const e of children) {
 		
-			const {name} = e.attributes
+			const {attributes} = e; if (!('name' in attributes)) continue
+			
+			const {name} = attributes
 
 			this.set (name, e)
 			

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xml-toolkit",
-  "version": "1.0.15",
+  "version": "1.0.17",
   "description": "Collection of classes for dealing with XML",
   "main": "index.js",
   "scripts": {

package/test/test.js

@@ -1,6 +1,6 @@
 const fs = require ('fs')
 const assert = require ('assert')
-const {XMLReader, SAXEvent, XMLLexer, AttributesMap, XMLNode, XMLSchemata, SOAP11} = require ('../')
+const {XMLReader, SAXEvent, XMLLexer, AttributesMap, XMLNode, XMLSchemata, SOAP11, EntityResolver, XMLIterator, XMLParser} = require ('../')
 
 async function test_001_lexer_sync (fn) {
 
@@ -15,8 +15,15 @@ console.log (xml)
 //		maxLength: 40,
 //		encoding: 'ascii',
 	})
+	
+
+	lexer.on ('data', data => {
+	
+		const e = new SAXEvent (data)
 
-	lexer.on ('data', data => console.log (new SAXEvent (data).attributes))
+		console.log ([e, e.type])	
+
+	})
 	
 //	for (let c of xml) lexer.write (c); lexer.end ()
 //	for (let c of xml) lexer.write (Buffer.from ([c])); lexer.end ()
@@ -323,115 +330,109 @@ async function test_007_wsdl (fn) {
 
 	console.log (soap.http ({GetForm9Sync: {address: {Region: {Code: 78}}}}))
 
+	console.log (soap)
+
 }
 
 async function test_008_schemata (fn) {
 
-	const xs = await XMLSchemata.fromFile ('test/fns-dohflna-ru-root.xsd')
+	const xs = await XMLSchemata.fromFile ('test/30017.xsd')
 
 	const data = 
 	
-{
-			  "DOHFLNAResponse": {
-				"ИдЗапрос": "000000000000000000000000000000000001",
-
-
-				"ДохФЛ": {
-				  "ОтчетГод": "2015",
-				  "ДохФЛ_НА": [
-					{
-					  "ПолучДох": {
-						"ИННФЛ": "100000000074",
-						"ДатаРожд": "1980-01-01",
-						"ФИО": {
-						  "FamilyName": "ИВАНОВ",
-						  "FirstName": "ИВАН",
-						  "Patronymic": "ИВАНОВИЧ"
-						},
-						"УдЛичнФЛ": {
-						  "DocumentCode": "21",
-						  "SeriesNumber": "0000 000000"
-						}
-					  },
-					  "СвНА": {
-						"СвНАФЛ": {
-						  "ИННФЛ": "100000000074",
-						  "ФИО": {
-							"FamilyName": "ИВАНОВ",
-							"FirstName": "ИВАН",
-							"Patronymic": "ИВАНОВИЧ"
-						  }
-						}
-					  },
-					  "СведДох_2НДФЛ": {
-						"Ставка": "13",
-						"ДохВыч": {
-						  "СвСумДох": [
-							{
-							  "Месяц": "01",
-							  "КодДоход": "2000",
-							  "СумДоход": "1000.12"
-							},
-							{
-							  "Месяц": "02",
-							  "КодДоход": "2000",
-							  "СумДоход": "2000.12"
-							}
-						  ]
-						},
-						"СГДНалПер": {
-						  "СумДохОбщ": "3000.24",
-						  "НалБаза": "3000.24"
-						}
-					  }
+		{"FNSINNSingularRequest":
+			{
+				"СведЮЛ": {
+					"ОГРН":"1037843048880",
+					"ИННЮЛ":"7825497650",
+					"НаимОрг":"Нагрузочное тестирование"
+				},
+				"СведФЛ": {
+					"ДатаРожд":"1973-07-14",
+					"МестоРожд":null,
+					"ФИО":{
+						"Фамилия":"ВАЛЕРЬЕВНА",
+						"Имя":"1973-07-14",
+						"Отчество":"1973-07-14"
 					},
-					{
-					  "ПолучДох": {
-						"ИННФЛ": "100000000074",
-						"ДатаРожд": "1980-01-01",
-						"ФИО": {
-						  "FamilyName": "ИВАНОВ",
-						  "FirstName": "ИВАН",
-						  "Patronymic": "ИВАНОВИЧ"
-						},
-						"УдЛичнФЛ": {
-						  "DocumentCode": "21",
-						  "SeriesNumber": "0000 000000"
-						}
-					  },
-					  "СвНА": {
-						"СвНАЮЛ": {
-						  "НаимОрг": "ООО ТЕСТ",
-						  "ИННЮЛ": "1000000002",
-						  "КПП": "010101001"
-						}
-					  },
-					  "СведДох_НДПриб": {
-						"ДохНалПер": {
-						  "Ставка": "2",
-						  "СумДохОбщ": "5000.11",
-						  "НалБаза": "3000.11"
-						}
-					  }
-					}
-				  ]
-				}
-				
+				"УдЛичнФЛ":{
+					"КодВидДок":"21",
+					"ВыдДок":"11",
+					"КодВыдДок":"111 111",
+					"ДатаДок":"2022-10-01",
+					"СерНомДок":"11 11 111111"}
+				},
+				"ИдЗапрос":"7daf950d-d71c-ea20-eaa9-5096324ca3b3"
+			}
+		}
 
+//	console.log (xs.stringify (data))
 
-			  }				
-		}
+//	console.log (xs)
+
+}
+
+function test_010_node () {
+
+	for (const s of 
+		[
+			'<a />zzz',
+			'zzz<b>',
+			'<!-- 111 -->222',
+			'<![CDATA[ >>]>>]]>zzz',
+			'<?xml version="3.11"?><a/>',
+		]
+	) {
+	
+		let n = new XMLNode (s)
+		
+		n.trim ()
+		
+		console.log (n)		
 	
+	}
 
-	console.log (xs.stringify (data))
+}
+
+function test_011_iterator (fn) {
+
+	const xml = fs.readFileSync (
+		'test/' + fn
+		, 'utf-8'
+	)
+	
+console.log (xml)
+
+	for (const i of new XMLIterator (xml)) {
+
+console.log ([i, i.detach ()])
+
+	}
+
+}
 
+function test_012_parser (fn) {
+
+	const xml = fs.readFileSync (
+		'test/' + fn
+		, 'utf-8'
+	)
+	
+//console.log (new XMLParser ().process (xml).detach ())
+
+const parser = new XMLParser  ({})
+const doc = parser.process (xml)
+
+for (const element of doc.detach ().children) {
+  console.log (element.attributes)
 }
 
+}
 
 async function main () {
 
 //	await test_001_lexer_sync ('E05a.xml')
-//	await test_001_lexer_sync ('not-sa01.xml')
+//await test_001_lexer_sync ('not-sa01.xml')
 //	await test_001_lexer_sync ('not-sa02.xml')
 //	await test_001_lexer_sync ('param_types.xml')
 //	await test_002_lexer_stream ('E05a.xml')
@@ -444,10 +445,17 @@ async function main () {
 //	await test_003_emitter_sync ('soap.xml')
 //	await test_004_schemata ()
 //	await test_005_schemata ()
-	await test_006_schemata ()
+//	await test_006_schemata ()
 //	await test_007_wsdl ()
-	await test_008_schemata ()
+//	await test_008_schemata ()
 
+//	test_010_node ()
+//	test_011_iterator ('param_types.xml')
+//	test_011_iterator ('20040.wsdl')
+
+	test_012_parser ('param_types.xml')
+//	test_012_parser ('20040.wsdl')
+	
 }
 
 main ()