cbor-x 0.9.4 → 1.0.0

package/decode.js

@@ -7,7 +7,10 @@ let srcEnd
 let position = 0
 let alreadySet
 const EMPTY_ARRAY = []
-const RECORD_TAG_ID = 0x69
+const LEGACY_RECORD_INLINE_ID = 105
+const RECORD_DEFINITIONS_ID = 0xdffe
+const RECORD_INLINE_ID = 0xdfff // temporary first-come first-serve tag // proposed tag: 0x7265 // 're'
+const BUNDLED_STRINGS_ID = 0xdff9
 const PACKED_TABLE_TAG_ID = 51
 const PACKED_REFERENCE_TAG_ID = 6
 const STOP_CODE = {}
@@ -18,13 +21,13 @@ let currentStructures
 let srcString
 let srcStringStart = 0
 let srcStringEnd = 0
+let bundledStrings
 let referenceMap
 let currentExtensions = []
 let currentExtensionRanges = []
 let packedValues
 let dataView
 let restoreMapsAsObject
-let sharedValues
 let defaultOptions = {
 	useRecords: false,
 	mapsAsObjects: true
@@ -34,13 +37,70 @@ let sequentialMode = false
 export class Decoder {
 	constructor(options) {
 		if (options) {
+			if ((options.keyMap || options._keyMap) && !options.useRecords) {
+				options.useRecords = false
+				options.mapsAsObjects = true
+			}
 			if (options.useRecords === false && options.mapsAsObjects === undefined)
 				options.mapsAsObjects = true
-			if (options.getStructures && !options.structures)
+			if (options.getStructures)
+				options.getShared = options.getStructures
+			if (options.getShared && !options.structures)
 				(options.structures = []).uninitialized = true // this is what we use to denote an uninitialized structures
+			if (options.keyMap) {
+				this.mapKey = new Map()
+				for (let [k,v] of Object.entries(options.keyMap)) this.mapKey.set(v,k)
+			}
 		}
 		Object.assign(this, options)
 	}
+	/*
+	decodeKey(key) {
+		return this.keyMap
+			? Object.keys(this.keyMap)[Object.values(this.keyMap).indexOf(key)] || key
+			: key
+	}
+	*/
+	decodeKey(key) {
+		return this.keyMap ? this.mapKey.get(key) || key : key
+	}
+	
+	encodeKey(key) {
+		return this.keyMap && this.keyMap.hasOwnProperty(key) ? this.keyMap[key] : key
+	}
+
+	encodeKeys(rec) {
+		if (!this._keyMap) return rec
+		let map = new Map()
+		for (let [k,v] of Object.entries(rec)) map.set((this._keyMap.hasOwnProperty(k) ? this._keyMap[k] : k), v)
+		return map
+	}
+
+	decodeKeys(map) {
+		if (!this._keyMap || map.constructor.name != 'Map') return map
+		if (!this._mapKey) {
+			this._mapKey = new Map()
+			for (let [k,v] of Object.entries(this._keyMap)) this._mapKey.set(v,k)
+		}
+		let res = {}
+		//map.forEach((v,k) => res[Object.keys(this._keyMap)[Object.values(this._keyMap).indexOf(k)] || k] = v)
+		map.forEach((v,k) => res[this._mapKey.has(k) ? this._mapKey.get(k) : k] =  v)
+		return res
+	}
+	
+	mapDecode(source, end) {
+	
+		let res = this.decode(source)
+		if (this._keyMap) { 
+			//Experiemntal support for Optimised KeyMap  decoding 
+			switch (res.constructor.name) {
+				case 'Array': return res.map(r => this.decodeKeys(r))
+				//case 'Map': return this.decodeKeys(res)
+			}
+		}
+		return res
+	}
+
 	decode(source, end) {
 		if (src) {
 			// re-entrant execution, save the state and restore it after we do this decode
@@ -55,6 +115,7 @@ export class Decoder {
 		srcStringEnd = 0
 		srcString = null
 		strings = EMPTY_ARRAY
+		bundledStrings = null
 		src = source
 		// this provides cached access to the data view for a buffer if it is getting reused, which is a recommend
 		// technique for getting data from a database where it can be copied into an existing buffer instead of creating
@@ -128,6 +189,9 @@ export function getPosition() {
 export function checkedRead() {
 	try {
 		let result = read()
+		if (bundledStrings) // bundled strings to skip past
+			position = bundledStrings.postBundlePosition
+
 		if (position == srcEnd) {
 			// finished reading this source, cleanup references
 			currentStructures = null
@@ -202,6 +266,7 @@ export function read() {
 				switch(majorType) {
 					case 2: // byte string
 					case 3: // text string
+						throw new Error('Indefinite length not supported for byte or text strings')
 					case 4: // array
 						let array = []
 						let value, i = 0
@@ -213,8 +278,8 @@ export function read() {
 						let key
 						if (currentDecoder.mapsAsObjects) {
 							let object = {}
-							while ((key = readKey()) != STOP_CODE)
-								object[key] = read()
+							if (currentDecoder.keyMap) while((key = read()) != STOP_CODE) object[currentDecoder.decodeKey(key)] = read()
+							else while ((key = read()) != STOP_CODE) object[key] = read()
 							return object
 						} else {
 							if (restoreMapsAsObject) {
@@ -222,8 +287,8 @@ export function read() {
 								restoreMapsAsObject = false
 							}
 							let map = new Map()
-							while ((key = read()) != STOP_CODE)
-								map.set(key, read())
+							if (currentDecoder.keyMap) while((key = read()) != STOP_CODE) map.set(currentDecoder.decodeKey(key), read())
+							else while ((key = read()) != STOP_CODE) map.set(key, read())
 							return map
 						}
 					case 7:
@@ -255,16 +320,15 @@ export function read() {
 			return readFixedString(token)
 		case 4: // array
 			let array = new Array(token)
-			for (let i = 0; i < token; i++) {
-				array[i] = read()
-			}
+		  //if (currentDecoder.keyMap) for (let i = 0; i < token; i++) array[i] = currentDecoder.decodeKey(read())	
+			//else 
+			for (let i = 0; i < token; i++) array[i] = read()
 			return array
 		case 5: // map
 			if (currentDecoder.mapsAsObjects) {
 				let object = {}
-				for (let i = 0; i < token; i++) {
-					object[readKey()] = read()
-				}
+				if (currentDecoder.keyMap) for (let i = 0; i < token; i++) object[currentDecoder.decodeKey(read())] = read()
+				else for (let i = 0; i < token; i++) object[read()] = read()
 				return object
 			} else {
 				if (restoreMapsAsObject) {
@@ -272,63 +336,66 @@ export function read() {
 					restoreMapsAsObject = false
 				}
 				let map = new Map()
-				for (let i = 0; i < token; i++) {
-					map.set(read(), read())
-				}
+				if (currentDecoder.keyMap) for (let i = 0; i < token; i++) map.set(currentDecoder.decodeKey(read()),read())
+				else for (let i = 0; i < token; i++) map.set(read(), read())
 				return map
 			}
 		case 6: // extension
-			if ((token >> 8) == RECORD_TAG_ID) { // record structures
-				let structure = currentStructures[token & 0xff]
+			if (token >= BUNDLED_STRINGS_ID) {
+				let structure = currentStructures[token & 0x1fff] // check record structures first
+				// At some point we may provide an option for dynamic tag assignment with a range like token >= 8 && (token < 16 || (token > 0x80 && token < 0xc0) || (token > 0x130 && token < 0x4000))
 				if (structure) {
-					if (!structure.read)
-						structure.read = createStructureReader(structure)
+					if (!structure.read) structure.read = createStructureReader(structure)
 					return structure.read()
-				} else if (currentDecoder.getStructures) {
-					let updatedStructures = saveState(() => {
-						// save the state in case getStructures modifies our buffer
-						src = null
-						return currentDecoder.getStructures()
-					})
-					if (currentStructures === true)
-						currentDecoder.structures = currentStructures = updatedStructures
-					else
-						currentStructures.splice.apply(currentStructures, [0, updatedStructures.length].concat(updatedStructures))
-					structure = currentStructures[token & 0xff]
-					if (structure) {
-						if (!structure.read)
-							structure.read = createStructureReader(structure)
-						return structure.read()
-					} else
-						return token
-				} else
-					return token
-			} else {
-				let extension = currentExtensions[token]
-				if (extension) {
-					if (extension.handlesRead)
-						return extension(read)
-					else
-						return extension(read())
-				} else {
-					let input = read()
-					for (let i = 0; i < currentExtensionRanges.length; i++) {
-						let value = currentExtensionRanges[i](token, input)
-						if (value !== undefined)
-							return value
+				}
+				if (token < 0x10000) {
+					if (token == RECORD_INLINE_ID) // we do a special check for this so that we can keep the currentExtensions as densely stored array (v8 stores arrays densely under about 3000 elements)
+						return recordDefinition(read())
+					else if (token == RECORD_DEFINITIONS_ID) {
+						let length = readJustLength()
+						let id = read()
+						for (let i = 2; i < length; i++) {
+							recordDefinition([id++, read()])
+						}
+						return read()
+					} else if (token == BUNDLED_STRINGS_ID) {
+						return readBundleExt()
+					}
+					if (currentDecoder.getShared) {
+						loadShared()
+						structure = currentStructures[token & 0x1fff]
+						if (structure) {
+							if (!structure.read)
+								structure.read = createStructureReader(structure)
+							return structure.read()
+						}
 					}
-					return new Tag(input)
 				}
 			}
+			let extension = currentExtensions[token]
+			if (extension) {
+				if (extension.handlesRead)
+					return extension(read)
+				else
+					return extension(read())
+			} else {
+				let input = read()
+				for (let i = 0; i < currentExtensionRanges.length; i++) {
+					let value = currentExtensionRanges[i](token, input)
+					if (value !== undefined)
+						return value
+				}
+				return new Tag(input, token)
+			}
 		case 7: // fixed value
 			switch (token) {
 				case 0x14: return false
 				case 0x15: return true
 				case 0x16: return null
 				case 0x17: return; // undefined
-				case 0x1f: 
+				case 0x1f:
 				default:
-					let packedValue = packedValues[token]
+					let packedValue = (packedValues || getPackedValues())[token]
 					if (packedValue !== undefined)
 						return packedValue
 					throw new Error('Unknown token ' + token)
@@ -344,41 +411,47 @@ export function read() {
 }
 const validName = /^[a-zA-Z_$][a-zA-Z\d_$]*$/
 function createStructureReader(structure) {
-	let l = structure.length
 	function readObject() {
+		// get the array size from the header
+		let length = src[position++]
+		//let majorType = token >> 5
+		length = length & 0x1f
+		if (length > 0x17) {
+			switch (length) {
+				case 0x18:
+					length = src[position++]
+					break
+				case 0x19:
+					length = dataView.getUint16(position)
+					position += 2
+					break
+				case 0x1a:
+					length = dataView.getUint32(position)
+					position += 4
+					break
+				default:
+					throw new Error('Expected array header, but got ' + src[position - 1])
+			}
+		}
 		// This initial function is quick to instantiate, but runs slower. After several iterations pay the cost to build the faster function
-		if (readObject.count++ > 2) {
-			this.read = (new Function('a', 'r', 'return function(){a();return {' + structure.map(key => validName.test(key) ? key + ':r()' : ('[' + JSON.stringify(key) + ']:r()')).join(',') + '}}'))(readArrayHeader, read)
-			return this.read()
+		if (this.objectLiteralSize === length) // we have a fast object literal reader, use it (assuming it is the right length)
+			return this.objectLiteral(read)
+		if (this.count++ == 3) { // create a fast reader
+			this.objectLiteralSize = length
+			this.objectLiteral = currentDecoder.keyMap 
+				? new Function('r', 'return {' + this.map(k => currentDecoder.decodeKey(k)).map(k => validName.test(k) ? k + ':r()' : ('[' + JSON.stringify(k) + ']:r()')).join(',') + '}')
+				: new Function('r', 'return {' + this.map(key => validName.test(key) ? key + ':r()' : ('[' + JSON.stringify(key) + ']:r()')).join(',') + '}')
+			return this.objectLiteral(read)
 		}
-		readArrayHeader(l)
 		let object = {}
-		for (let i = 0; i < l; i++) {
-			let key = structure[i]
-			object[key] = read()
-		}
+		if (currentDecoder.keyMap) for (let i = 0; i < length; i++) object[currentDecoder.decodeKey(this[i])] = read()
+		else for (let i = 0; i < length; i++) object[this[i]] = read()
 		return object
 	}
-	readObject.count = 0
+	structure.count = 0
 	return readObject
 }
 
-function readArrayHeader(expectedLength) {
-	// consume the array header, TODO: check expected length
-	let token = src[position++]
-	//let majorType = token >> 5
-	token = token & 0x1f
-	if (token > 0x17) {
-		switch (token) {
-			case 0x18: position++
-				break
-			case 0x19: position += 2
-				break
-			case 0x1a: position += 4
-		}
-	}
-}
-
 let readFixedString = readStringJS
 let readString8 = readStringJS
 let readString16 = readStringJS
@@ -395,6 +468,8 @@ export function setExtractor(extractStrings) {
 		return function readString(length) {
 			let string = strings[stringPosition++]
 			if (string == null) {
+				if (bundledStrings)
+					return readStringJS(length)
 				let extraction = extractStrings(position, srcEnd, length, src)
 				if (typeof extraction == 'string') {
 					string = extraction
@@ -721,12 +796,13 @@ function readKey() {
 }
 
 export class Tag {
-	constructor(value) {
+	constructor(value, tag) {
 		this.value = value
+		this.tag = tag
 	}
 }
 
-let glbl = typeof window == 'object' ? window : global
+let glbl = typeof self == 'object' ? self : global
 
 currentExtensions[0] = (dateString) => {
 	// string date extension
@@ -740,38 +816,62 @@ currentExtensions[1] = (epochSec) => {
 
 currentExtensions[2] = (buffer) => {
 	// bigint extension
-	return new DataView(buffer.buffer, buffer.byteOffset, buffer.byteLength).getBigUint64(0)
+	let value = BigInt(0)
+	for (let i = 0, l = buffer.byteLength; i < l; i++) {
+		value = BigInt(buffer[i]) + value << BigInt(8)
+	}
+	return value
 }
 
 currentExtensions[3] = (buffer) => {
 	// negative bigint extension
-	return BigInt(-1) - (new DataView(buffer.buffer, buffer.byteOffset, buffer.byteLength).getBigUint64(0))
+	return BigInt(-1) - currentExtensions[2](buffer)
+}
+currentExtensions[4] = (fraction) => {
+	// best to reparse to maintain accuracy
+	return +(fraction[1] + 'e' + fraction[0])
+}
+
+currentExtensions[5] = (fraction) => {
+	// probably not sufficiently accurate
+	return fraction[1] * Math.exp(fraction[0] * Math.log(2))
 }
 
 // the registration of the record definition extension (tag 105)
-const recordDefinition = () => {
-	let definition = read()
-	let structure = definition[0]
-	let id = definition[1]
-	currentStructures[id & 0xff] = structure
+const recordDefinition = (definition) => {
+	let id = definition[0] - 0xe000
+	let structure = definition[1]
+	currentStructures[id] = structure
 	structure.read = createStructureReader(structure)
 	let object = {}
-	for (let i = 2,l = definition.length; i < l; i++) {
-		let key = structure[i - 2]
-		object[key] = definition[i]
-	}
+	if (currentDecoder.keyMap) for (let i = 2,l = definition.length; i < l; i++) {
+			let key = currentDecoder.decodeKey(structure[i - 2])
+			object[key] = definition[i]
+		}
+	else for (let i = 2,l = definition.length; i < l; i++) {
+			let key = structure[i - 2]
+			object[key] = definition[i]
+		}
 	return object
 }
-
-recordDefinition.handlesRead = true
-currentExtensions[RECORD_TAG_ID] = recordDefinition
+currentExtensions[LEGACY_RECORD_INLINE_ID] = recordDefinition
+currentExtensions[14] = (value) => {
+	if (bundledStrings)
+		return bundledStrings[0].slice(bundledStrings.position0, bundledStrings.position0 += value)
+	return new Tag(value, 14)
+}
+currentExtensions[15] = (value) => {
+	if (bundledStrings)
+		return bundledStrings[1].slice(bundledStrings.position1, bundledStrings.position1 += value)
+	return new Tag(value, 15)
+}
 
 currentExtensions[27] = (data) => { // http://cbor.schmorp.de/generic-object
 	return (glbl[data[0]] || Error)(data[1], data[2])
 }
 const packedTable = (read) => {
 	if (src[position++] != 0x84)
-		throw new Error('Packed values structure must be followed by 4 element array')
+		throw new Error('Packed values structure must be followed by a 4 element array')
 	let newPackedValues = read() // packed values
 	packedValues = packedValues ? newPackedValues.concat(packedValues.slice(newPackedValues.length)) : newPackedValues
 	packedValues.prefixes = read()
@@ -782,15 +882,36 @@ packedTable.handlesRead = true
 currentExtensions[51] = packedTable
 
 currentExtensions[PACKED_REFERENCE_TAG_ID] = (data) => { // packed reference
+	if (!packedValues) {
+		if (currentDecoder.getShared)
+			loadShared()
+		else
+			return new Tag(data, PACKED_REFERENCE_TAG_ID)
+	}
 	if (typeof data == 'number')
 		return packedValues[16 + (data >= 0 ? 2 * data : (-2 * data - 1))]
 	throw new Error('No support for non-integer packed references yet')
 }
+currentExtensions[25] = (id) => {
+	return stringRefs[id]
+}
+currentExtensions[256] = (read) => {
+	stringRefs = []
+	try {
+		return read()
+	} finally {
+		stringRefs = null
+	}
+}
+currentExtensions[256].handlesRead = true
 
-currentExtensions[40009] = (id) => {
-	// id extension (for structured clones)
-	if (!referenceMap)
+currentExtensions[28] = (read) => { 
+	// shareable http://cbor.schmorp.de/value-sharing (for structured clones)
+	if (!referenceMap) {
 		referenceMap = new Map()
+		referenceMap.id = 0
+	}
+	let id = referenceMap.id++
 	let token = src[position]
 	let target
 	// TODO: handle Maps, Sets, and other types that can cycle; this is complicated, because you potentially need to read
@@ -808,9 +929,10 @@ currentExtensions[40009] = (id) => {
 	refEntry.target = targetProperties // the placeholder wasn't used, replace with the deserialized one
 	return targetProperties // no cycle, can just use the returned read object
 }
+currentExtensions[28].handlesRead = true
 
-currentExtensions[40010] = (id) => {
-	// pointer extension (for structured clones)
+currentExtensions[29] = (id) => {
+	// sharedref http://cbor.schmorp.de/value-sharing (for structured clones)
 	let refEntry = referenceMap.get(id)
 	refEntry.used = true
 	return refEntry.target
@@ -833,19 +955,36 @@ function combine(a, b) {
 		return a.concat(b)
 	return Object.assign({}, a, b)
 }
+function getPackedValues() {
+	if (!packedValues) {
+		if (currentDecoder.getShared)
+			loadShared()
+		else
+			throw new Error('No packed values available')
+	}
+	return packedValues
+}
+const SHARED_DATA_TAG_ID = 0x53687264 // ascii 'Shrd'
 currentExtensionRanges.push((tag, input) => {
 	if (tag >= 225 && tag <= 255)
-		return combine(packedTable.prefixes[tag - 224], input)
+		return combine(getPackedValues().prefixes[tag - 224], input)
 	if (tag >= 28704 && tag <= 32767)
-		return combine(packedTable.prefixes[tag - 28672], input)
+		return combine(getPackedValues().prefixes[tag - 28672], input)
 	if (tag >= 1879052288 && tag <= 2147483647)
-		return combine(packedTable.prefixes[tag - 1879048192], input)
+		return combine(getPackedValues().prefixes[tag - 1879048192], input)
 	if (tag >= 216 && tag <= 223)
-		return combine(input, packedTable.suffixes[tag - 216])
+		return combine(input, getPackedValues().suffixes[tag - 216])
 	if (tag >= 27647 && tag <= 28671)
-		return combine(input, packedTable.suffixes[tag - 27639])
+		return combine(input, getPackedValues().suffixes[tag - 27639])
 	if (tag >= 1811940352 && tag <= 1879048191)
-		return combine(input, packedTable.suffixes[tag - 1811939328])
+		return combine(input, getPackedValues().suffixes[tag - 1811939328])
+	if (tag == SHARED_DATA_TAG_ID) {// we do a special check for this so that we can keep the currentExtensions as densely stored array (v8 stores arrays densely under about 3000 elements)
+		return {
+			packedValues: packedValues,
+			structures: currentStructures.slice(0),
+			version: input,
+		}
+	}
 })
 
 export const typedArrays = ['Uint8', 'Uint8Clamped', 'Uint16', 'Uint32', 'BigUint64','Int8', 'Int16', 'Int32', 'BigInt64', 'Float32', 'Float64'].map(type => type + 'Array')
@@ -862,6 +1001,61 @@ function registerTypedArray(typedArrayName, tag) {
 	}
 }
 
+function readBundleExt() {
+	let length = readJustLength()
+	let bundlePosition = position + read()
+	for (let i = 2; i < length; i++) {
+		// skip past bundles that were already read
+		let bundleLength = readJustLength() // this will increment position, so must add to position afterwards
+		position += bundleLength
+	}
+	let dataPosition = position
+	position = bundlePosition
+	bundledStrings = [readStringJS(readJustLength()), readStringJS(readJustLength())]
+	bundledStrings.position0 = 0
+	bundledStrings.position1 = 0
+	bundledStrings.postBundlePosition = position
+	position = dataPosition
+	return read()
+}
+
+function readJustLength() {
+	let token = src[position++] & 0x1f
+	if (token > 0x17) {
+		switch (token) {
+			case 0x18:
+				token = src[position++]
+				break
+			case 0x19:
+				token = dataView.getUint16(position)
+				position += 2
+				break
+			case 0x1a:
+				token = dataView.getUint32(position)
+				position += 4
+				break
+		}
+	}
+	return token
+}
+
+function loadShared() {
+	if (currentDecoder.getShared) {
+		let sharedData = saveState(() => {
+			// save the state in case getShared modifies our buffer
+			src = null
+			return currentDecoder.getShared()
+		}) || {}
+		let updatedStructures = sharedData.structures || []
+		currentDecoder.sharedVersion = sharedData.version
+		packedValues = currentDecoder.sharedValues = sharedData.packedValues
+		if (currentStructures === true)
+			currentDecoder.structures = currentStructures = updatedStructures
+		else
+			currentStructures.splice.apply(currentStructures, [0, updatedStructures.length].concat(updatedStructures))
+	}
+}
+
 function saveState(callback) {
 	let savedSrcEnd = srcEnd
 	let savedPosition = position
@@ -871,6 +1065,8 @@ function saveState(callback) {
 	let savedSrcString = srcString
 	let savedStrings = strings
 	let savedReferenceMap = referenceMap
+	let savedBundledStrings = bundledStrings
+
 	// TODO: We may need to revisit this if we do more external calls to user code (since it could be slow)
 	let savedSrc = new Uint8Array(src.slice(0, srcEnd)) // we copy the data in case it changes while external data is processed
 	let savedStructures = currentStructures
@@ -885,6 +1081,7 @@ function saveState(callback) {
 	srcString = savedSrcString
 	strings = savedStrings
 	referenceMap = savedReferenceMap
+	bundledStrings = savedBundledStrings
 	src = savedSrc
 	sequentialMode = savedSequentialMode
 	currentStructures = savedStructures