fix: perf and static calls

Author: ThaUnknown

.dockerignore

@@ -0,0 +1,10 @@
+node_modules
+dist
+.git
+.gitignore
+.gitmodules
+.gitattributes
+.vscode
+test
+*.md
+LICENSE

Dockerfile

@@ -0,0 +1,11 @@
+FROM emscripten/emsdk:latest AS build
+
+WORKDIR /src
+COPY build-config.mjs build.mjs ./
+COPY src/ src/
+COPY include/ include/
+
+RUN mkdir -p dist && node build.mjs
+
+FROM alpine
+COPY --from=build /src/dist/ /out/

build-config.mjs

@@ -0,0 +1,53 @@
+import path from 'node:path'
+import { fileURLToPath } from 'node:url'
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url))
+const inc = path.resolve(__dirname, 'include')
+const src = path.resolve(__dirname, 'src')
+
+export const includeDirs = [inc]
+
+export const sources = [
+  path.resolve(inc, 'anitomy/anitomy.cpp'),
+  path.resolve(inc, 'anitomy/element.cpp'),
+  path.resolve(inc, 'anitomy/keyword.cpp'),
+  path.resolve(inc, 'anitomy/parser.cpp'),
+  path.resolve(inc, 'anitomy/parser_helper.cpp'),
+  path.resolve(inc, 'anitomy/parser_number.cpp'),
+  path.resolve(inc, 'anitomy/string.cpp'),
+  path.resolve(inc, 'anitomy/token.cpp'),
+  path.resolve(inc, 'anitomy/tokenizer.cpp'),
+  path.resolve(src, 'anitomyscript.cpp'),
+]
+
+export const outputFile = path.resolve(__dirname, 'dist', 'anitomyscript.js')
+
+export const baseFlags = [
+  '-std=c++17',
+  '-fno-exceptions',
+  '-Wall',
+  '-Wpedantic',
+  '-s', 'EXPORT_NAME="anitomyscript"',
+  '-s', 'WASM=1',
+  '-s', 'ENVIRONMENT=web,node',
+  '-s', 'FILESYSTEM=0',
+  '-s', 'MODULARIZE=1',
+  '-s', 'EXPORT_ES6=1',
+  '-s', 'DYNAMIC_EXECUTION=0',
+  '-s', 'TEXTDECODER=2',
+  '-s', 'SUPPORT_LONGJMP=0',
+  '-s', 'EMBIND_AOT=1',
+  '-s', 'INITIAL_MEMORY=524288',
+  '-s', 'ALLOW_MEMORY_GROWTH=1',
+  '-lembind',
+]
+
+export const releaseFlags = [
+  '-O3',
+]
+
+export const debugFlags = [
+  '-O0',
+  '-g',
+  '-v',
+]

build.mjs

@@ -0,0 +1,26 @@
+import { spawn } from 'node:child_process'
+import { baseFlags, releaseFlags, includeDirs, sources, outputFile } from './build-config.mjs'
+
+const includeArgs = includeDirs.flatMap(d => ['-I', d])
+
+const spawnArgs = [
+  ...baseFlags,
+  ...releaseFlags,
+  ...includeArgs,
+  ...sources,
+  '-o', outputFile,
+]
+
+const emcc = process.platform === 'win32' ? 'emcc.bat' : 'emcc'
+
+console.log('Starting Docker build with:', spawnArgs.join(' '))
+const s = spawn(emcc, spawnArgs, { stdio: 'inherit' })
+
+s.on('error', (e) => {
+  console.error('Failed to start emcc:', e)
+  process.exit(1)
+})
+
+s.on('close', (code) => {
+  process.exit(code ?? 0)
+})

dist/anitomyscript.js

@@ -1,66 +1,32 @@
 import gulp from 'gulp'
 import { spawn } from 'child_process'
 import path from 'path'
-import fs from 'fs'
+import { baseFlags, releaseFlags, debugFlags, includeDirs, sources, outputFile } from './build-config.mjs'
 
-function build (cb) {
+function getEmcc () {
   const emscriptenPath = process.env.EMSCRIPTEN || process.env.EMSCRIPTEN_ROOT
-
-  if (!emscriptenPath || !fs.existsSync(path.resolve(emscriptenPath))) {
-    return cb('Unable to find emscripten root. Use the env variable EMSCRIPTEN or EMSCRIPTEN_ROOT to set it manually.')
+  const emcc = process.platform === 'win32' ? 'emcc.bat' : 'emcc'
+  if (emscriptenPath) {
+    return path.join(emscriptenPath, emcc)
   }
+  return emcc
+}
 
-  const out = path.resolve('./', 'dist', 'anitomyscript.js')
+function build (cb) {
   const isRelease = process.env.NODE_ENV === 'prod'
-
-  let spawnArgs = [
-    '--memory-init-file', '0',
-    '-std=c++14',
-    '-fPIC', '-fno-exceptions', '-Wall', '-Wextra', '-Wpedantic', '-Werror',
-    '--bind',
-    '-v',
-    '-I', path.resolve('./include'),
-    '-s', 'EXPORT_NAME="anitomyscript"',
-    '-s', 'WASM=1',
-    '-s', 'ENVIRONMENT=web,node',
-    '-s', 'FILESYSTEM=0',
-    '-s', 'MODULARIZE=1',
-    '-s', 'EXPORT_ES6=1',
-    '-s', 'AUTO_JS_LIBRARIES=0',
-    '-s', 'AUTO_NATIVE_LIBRARIES=0',
-    '-s', 'HTML5_SUPPORT_DEFERRING_USER_SENSITIVE_REQUESTS=0',
-    '-s', 'USE_SDL=0',
-    '-s', 'INITIAL_MEMORY=5308416',
-    '-s', 'ALLOW_MEMORY_GROWTH=1',
-    '--no-heap-copy',
-    path.resolve('./include/anitomy/anitomy.cpp'),
-    path.resolve('./include/anitomy/element.cpp'),
-    path.resolve('./include/anitomy/keyword.cpp'),
-    path.resolve('./include/anitomy/parser.cpp'),
-    path.resolve('./include/anitomy/parser_helper.cpp'),
-    path.resolve('./include/anitomy/parser_number.cpp'),
-    path.resolve('./include/anitomy/string.cpp'),
-    path.resolve('./include/anitomy/token.cpp'),
-    path.resolve('./include/anitomy/tokenizer.cpp'),
-    path.resolve('./src/anitomyscript.cpp'),
-    '-o', out
+  const includeArgs = includeDirs.flatMap(d => ['-I', d])
+
+  const spawnArgs = [
+    ...baseFlags,
+    ...(isRelease ? releaseFlags : debugFlags),
+    ...includeArgs,
+    ...sources,
+    '-o', outputFile,
   ]
 
-  if (isRelease) {
-    spawnArgs = spawnArgs.concat([
-      '-O3',
-      '-s', 'USE_CLOSURE_COMPILER=1',
-      '-s', 'IGNORE_CLOSURE_COMPILER_ERRORS=1',
-      '--closure', '1',
-      '--llvm-lto', '3'
-    ])
-  }
-
-  console.log(`Starting ${isRelease ? 'release' : 'debug'} build with emcc args`, spawnArgs)
-  const emccExec = process.platform === 'win32' ? 'emcc.bat' : 'emcc'
-  const s = spawn(path.join(emscriptenPath, emccExec), spawnArgs, {
-    cwd: './dist'
-  })
+  console.log(`Starting ${isRelease ? 'release' : 'debug'} build`)
+  const emcc = getEmcc()
+  const s = spawn(emcc, spawnArgs)
 
   s.stdout.on('data', (data) => {
     console.log(data.toString())

dist/anitomyscript.wasm

@@ -1,78 +1,47 @@
 import AnitomyNative from './dist/anitomyscript.js'
-let anitomyModule
 
-export default async function (file) {
-  if (!Array.isArray(file) && typeof file !== 'string') {
-    return new Error('Input must be either an Array or a string')
+const FIELDS = [
+  'anime_season', 'season_prefix', 'anime_title',
+  'anime_type', 'anime_year', 'audio_term',
+  'device_compatibility', 'episode_number', 'episode_number_alt',
+  'episode_prefix', 'episode_title', 'file_checksum',
+  'file_extension', 'file_name', 'language',
+  'other', 'release_group', 'release_information',
+  'release_version', 'source', 'subtitles',
+  'video_resolution', 'video_term', 'volume_number',
+  'volume_prefix', 'unknown',
+]
+
+let modPromise, pairs
+
+export default async function parse (input) {
+  if (!Array.isArray(input) && typeof input !== 'string') {
+    throw new Error('Input must be either an Array or a string')
   }
 
-  if (anitomyModule) {
-    return parse(file)
-  }
-
-  anitomyModule = await AnitomyNative()
-  return parse(file)
-}
-
-async function parse (file) {
-  const vector = mapArray(file)
-  const result = mapVector(anitomyModule.parseMultiple(vector))
-  vector.delete()
-  return result.map((each) => elements(each))
-}
-
-function elements (elements) {
-  const returnObj = {
-    anime_season: elementEntry(elements, anitomyModule.ElementCategory.kElementAnimeSeason),
-    season_prefix: elementEntry(elements, anitomyModule.ElementCategory.kElementAnimeSeasonPrefix),
-    anime_title: elementEntry(elements, anitomyModule.ElementCategory.kElementAnimeTitle),
-    anime_type: elementEntry(elements, anitomyModule.ElementCategory.kElementAnimeType),
-    anime_year: elementEntry(elements, anitomyModule.ElementCategory.kElementAnimeYear),
-    audio_term: elementEntry(elements, anitomyModule.ElementCategory.kElementAudioTerm),
-    device_compatibility: elementEntry(elements, anitomyModule.ElementCategory.kElementDeviceCompatibility),
-    episode_number: elementEntry(elements, anitomyModule.ElementCategory.kElementEpisodeNumber),
-    episode_prefix: elementEntry(elements, anitomyModule.ElementCategory.kElementEpisodePrefix),
-    episode_title: elementEntry(elements, anitomyModule.ElementCategory.kElementEpisodeTitle),
-    file_checksum: elementEntry(elements, anitomyModule.ElementCategory.kElementFileChecksum),
-    file_extension: elementEntry(elements, anitomyModule.ElementCategory.kElementFileExtension),
-    file_name: elementEntry(elements, anitomyModule.ElementCategory.kElementFileName),
-    language: elementEntry(elements, anitomyModule.ElementCategory.kElementLanguage),
-    other: elementEntry(elements, anitomyModule.ElementCategory.kElementOther),
-    release_group: elementEntry(elements, anitomyModule.ElementCategory.kElementReleaseGroup),
-    release_information: elementEntry(elements, anitomyModule.ElementCategory.kElementReleaseInformation),
-    release_version: elementEntry(elements, anitomyModule.ElementCategory.kElementReleaseVersion),
-    source: elementEntry(elements, anitomyModule.ElementCategory.kElementSource),
-    subtitles: elementEntry(elements, anitomyModule.ElementCategory.kElementSubtitles),
-    video_resolution: elementEntry(elements, anitomyModule.ElementCategory.kElementVideoResolution),
-    video_term: elementEntry(elements, anitomyModule.ElementCategory.kElementVideoTerm),
-    volume_number: elementEntry(elements, anitomyModule.ElementCategory.kElementVolumeNumber),
-    volume_prefix: elementEntry(elements, anitomyModule.ElementCategory.kElementVolumePrefix),
-    unknown: elementEntry(elements, anitomyModule.ElementCategory.kElementUnknown)
-  }
-  elements.delete()
-  return returnObj
-}
-
-function elementEntry (elements, key) {
-  return mapVector(elements.get_all(key))
-}
-
-function mapArray (array) {
-  const vector = new anitomyModule.StringVector()
-  array.forEach((element, index) => {
-    if (typeof element !== 'string') {
-      throw new Error(`Element at index ${index} is not a string`)
+  const mod = await (modPromise ??= AnitomyNative())
+  pairs ??= FIELDS.map((name, i) => [mod.ElementCategory.values[i], name])
+  const files = typeof input === 'string' ? [input] : input
+  const anit = new mod.Anitomy()
+  const results = []
+
+  try {
+    for (let i = 0; i < files.length; i++) {
+      const file = files[i]
+      if (typeof file !== 'string') throw new Error(`Element at index ${i} is not a string`)
+      anit.parse(file)
+      const obj = {}
+      for (const [cat, field] of pairs) {
+        const vec = anit.get_all(cat)
+        const length = vec.size()
+        if (!length) continue
+        obj[field] = Array.from({ length }, (_, j) => vec.get(j))
+      }
+      results.push(obj)
     }
-    vector.push_back(element)
-  })
-  return vector
-}
-
-function mapVector (vector) {
-  const array = []
-  for (let index = 0; index < vector.size(); index++) {
-    array.push(vector.get(index))
+  } finally {
+    anit.delete()
   }
-  vector.delete()
-  return array
+
+  return results
 }

gulpfile.js

@@ -7,7 +7,8 @@
   "scripts": {
     "test": "mocha test/*.spec.js",
     "build": "cross-env NODE_ENV=prod gulp",
-    "build:dev": "cross-env NODE_ENV=dev gulp"
+    "build:dev": "cross-env NODE_ENV=dev gulp",
+    "build:docker": "docker build -t anitomyscript-builder -f Dockerfile . && docker create --name tmp-extract anitomyscript-builder && docker cp tmp-extract:/out/. dist/ && docker rm tmp-extract"
   },
   "files": [
     "index.js",