add optimizer v2

Author: Uzlopak

benchmark/bench-thread.js

@@ -3,7 +3,7 @@
 const { workerData: benchmark, parentPort } = require('worker_threads')
 
 const Benchmark = require('benchmark')
-Benchmark.options.minSamples = 500
+Benchmark.options.minSamples = 100
 
 const suite = Benchmark.Suite()
 

index.js

@@ -11,6 +11,7 @@ const Serializer = require('./lib/serializer')
 const Validator = require('./lib/validator')
 const RefResolver = require('./lib/ref-resolver')
 const Location = require('./lib/location')
+const optimize = require('./lib/optimize')
 
 let largeArraySize = 2e4
 let largeArrayMechanism = 'default'
@@ -27,6 +28,18 @@ const validLargeArrayMechanisms = [
   'json-stringify'
 ]
 
+const serializerFns = `
+const {
+  asString,
+  asInteger,
+  asNumber,
+  asBoolean,
+  asDateTime,
+  asDate,
+  asTime,
+} = serializer
+`
+
 const addComma = '!addComma && (addComma = true) || (json += \',\')'
 
 function isValidSchema (schema, name) {
@@ -119,21 +132,8 @@ function build (schema, options) {
   const location = new Location(schema, context.rootSchemaId)
   const code = buildValue(context, location, 'input')
 
-  let contextFunctionCode
-
-  // If we have only the invocation of the 'anonymous0' function, we would
-  // basically just wrap the 'anonymous0' function in the 'main' function and
-  // and the overhead of the intermediate variable 'json'. We can avoid the
-  // wrapping and the unnecessary memory allocation by aliasing 'anonymous0' to
-  // 'main'
-  if (code === 'json += anonymous0(input)') {
-    contextFunctionCode = `
-    ${context.functions.join('\n')}
-    const main = anonymous0
-    return main
-    `
-  } else {
-    contextFunctionCode = `
+  let contextFunctionCode = `
+    ${serializerFns}
     function main (input) {
       let json = ''
       ${code}
@@ -142,7 +142,8 @@ function build (schema, options) {
     ${context.functions.join('\n')}
     return main
     `
-  }
+
+  contextFunctionCode = optimize(contextFunctionCode)
 
   const serializer = new Serializer(options)
   const validator = new Validator(options.ajv)
@@ -263,7 +264,7 @@ function buildExtraObjectPropertiesSerializer (context, location) {
       code += `
         if (/${propertyKey.replace(/\\*\//g, '\\/')}/.test(key)) {
           ${addComma}
-          json += serializer.asString(key) + ':'
+          json += asString(key) + ':'
           ${buildValue(context, propertyLocation, 'value')}
           continue
         }
@@ -278,13 +279,13 @@ function buildExtraObjectPropertiesSerializer (context, location) {
     if (additionalPropertiesSchema === true) {
       code += `
         ${addComma}
-        json += serializer.asString(key) + ':' + JSON.stringify(value)
+        json += asString(key) + ':' + JSON.stringify(value)
       `
     } else {
       const propertyLocation = location.getPropertyLocation('additionalProperties')
       code += `
         ${addComma}
-        json += serializer.asString(key) + ':'
+        json += asString(key) + ':'
         ${buildValue(context, propertyLocation, 'value')}
       `
     }
@@ -743,21 +744,21 @@ function buildSingleTypeSerializer (context, location, input) {
       return 'json += \'null\''
     case 'string': {
       if (schema.format === 'date-time') {
-        return `json += serializer.asDateTime(${input})`
+        return `json += asDateTime(${input})`
       } else if (schema.format === 'date') {
-        return `json += serializer.asDate(${input})`
+        return `json += asDate(${input})`
       } else if (schema.format === 'time') {
-        return `json += serializer.asTime(${input})`
+        return `json += asTime(${input})`
       } else {
-        return `json += serializer.asString(${input})`
+        return `json += asString(${input})`
       }
     }
     case 'integer':
-      return `json += serializer.asInteger(${input})`
+      return `json += asInteger(${input})`
     case 'number':
-      return `json += serializer.asNumber(${input})`
+      return `json += asNumber(${input})`
     case 'boolean':
-      return `json += serializer.asBoolean(${input})`
+      return `json += asBoolean(${input})`
     case 'object': {
       const funcName = buildObject(context, location)
       return `json += ${funcName}(${input})`

lib/optimize.js

@@ -0,0 +1,89 @@
+'use strict'
+
+const returnFnRE = /^\s+return ([.a-zA-Z0-9]+)\(\w+\)$/
+const fnRE = /^\s*function\s+/
+const fnNameRE = /^\s+function ([a-zA-Z0-9_]+) \(input\) {$/
+const jsonConcatRE = /^\s*json\s*\+=/
+const letJsonRE = /^\s*let json =/
+const returnJsonRE = /^\s*return json\s*$/
+const returnEmptyStringRE = /^\s*return '' \+/
+const closingCurlyBracketRE = /^\s*}\s*$/
+/**
+ * @param {Array<string>} code
+ * @returns {Array<string>}
+ */
+function optimize (raw) {
+  const code = raw.split('\n')
+  /**
+     * @type {Array<string>}
+     */
+  const dedupedLevel1 = []
+
+  for (let i = 0; i < code.length; i++) {
+    if (i > 0 && jsonConcatRE.test(code[i]) && jsonConcatRE.test(code[i - 1])) {
+      const mergedEntry = code[i - 1] + ' +' + code[i].substring(code[i].indexOf('json +=') + 7)
+      dedupedLevel1.pop() // Remove the previous entry
+      dedupedLevel1.push(mergedEntry)
+    } else {
+      dedupedLevel1.push(code[i])
+    }
+  }
+
+  /**
+     * @type {Array<string>}
+     */
+  const dedupedLevel2 = []
+  for (let i = 0; i < dedupedLevel1.length; i++) {
+    if (i > 0 && jsonConcatRE.test(dedupedLevel1[i]) && letJsonRE.test(dedupedLevel1[i - 1])) {
+      const mergedEntry = dedupedLevel1[i - 1] + ' +' + dedupedLevel1[i].substring(dedupedLevel1[i].indexOf('json +=') + 7)
+      dedupedLevel2.pop() // Remove the previous entry
+      dedupedLevel2.push(mergedEntry)
+    } else {
+      dedupedLevel2.push(dedupedLevel1[i])
+    }
+  }
+
+  /**
+     * @type {Array<string>}
+     */
+  const dedupedLevel3 = []
+  for (let i = 0; i < dedupedLevel2.length; i++) {
+    if (i > 0 && returnJsonRE.test(dedupedLevel2[i]) && letJsonRE.test(dedupedLevel2[i - 1])) {
+      const mergedEntry = dedupedLevel2[i].slice(0, dedupedLevel2[i].indexOf('return') + 6) + dedupedLevel2[i - 1].substring(dedupedLevel2[i - 1].indexOf('let json =') + 10)
+      dedupedLevel3.pop() // Remove the previous entry
+      dedupedLevel3.push(mergedEntry)
+    } else {
+      dedupedLevel3.push(dedupedLevel2[i])
+    }
+  }
+
+  /**
+     * @type {Array<string>}
+     */
+  for (let i = 0; i < dedupedLevel3.length; i++) {
+    if (returnEmptyStringRE.test(dedupedLevel3[i])) {
+      dedupedLevel3[i] = dedupedLevel3[i].replace('return \'\' +', 'return')
+    }
+  }
+
+  const dedupedLevel4 = []
+  for (let i = 0; i < dedupedLevel3.length; i++) {
+    if (
+      fnRE.test(dedupedLevel3[i]) &&
+      returnFnRE.test(dedupedLevel3[i + 1]) &&
+      closingCurlyBracketRE.test(dedupedLevel3[i + 2])
+    ) {
+      const serializerFnName = dedupedLevel3[i + 1].match(returnFnRE)[1]
+      const fnName = dedupedLevel3[i].match(fnNameRE)[1]
+      const whitespace = dedupedLevel3[i].slice(0, dedupedLevel3[i].indexOf('f'))
+      dedupedLevel4[i] = `${whitespace}const ${fnName} = ${serializerFnName}`
+      i += 2
+    } else {
+      dedupedLevel4.push(dedupedLevel3[i])
+    }
+  }
+
+  return dedupedLevel4.join('\n')
+}
+
+module.exports = optimize