Fix opencc-jieba npm config resolution (#1197)

* Fix opencc-jieba npm config resolution

* Add Windows prebuild for opencc-jieba npm package

* Declare  as manual

* Set version number for npm opencc-jieba to 1.3.1

Author: frankslin

node/README.md

@@ -189,18 +189,24 @@ Some OpenCC configurations use jieba segmentation. Install the optional
 npm install opencc opencc-jieba
 ```
 
-When `opencc-jieba` is installed, the JavaScript API can load its configs
-automatically:
+When `opencc-jieba` is installed, the JavaScript API and npm `opencc` CLI can
+load its configs automatically:
 
 ```js
 import OpenCC from 'opencc';
 
-const converter = new OpenCC('s2twp_jieba.json');
+const converter = new OpenCC('s2twp_jieba');
 console.log(converter.convertSync('软件鼠标'));
 ```
 
-The npm `opencc` CLI does not support plugin-backed segmentation. Use the
-JavaScript API or the native OpenCC CLI for those workflows.
+The npm CLI also accepts the same config names for text conversion:
+
+```bash
+opencc -c s2twp_jieba
+```
+
+Use the native OpenCC CLI for diagnostic modes such as `--inspect` and
+`--segmentation`.
 
 ## Related npm Packages
 

node/README.zh.md

@@ -185,17 +185,23 @@ CLI 的相對配置路徑會從目前工作目錄解析。在 JavaScript API 中
 npm install opencc opencc-jieba
 ```
 
-安裝 `opencc-jieba` 後，JavaScript API 可以自動載入其中的配置：
+安裝 `opencc-jieba` 後，JavaScript API 與 npm `opencc` CLI 可以自動
+載入其中的配置：
 
 ```js
 import OpenCC from 'opencc';
 
-const converter = new OpenCC('s2twp_jieba.json');
+const converter = new OpenCC('s2twp_jieba');
 console.log(converter.convertSync('软件鼠标'));
 ```
 
-npm `opencc` CLI 不支援插件分詞。這類工作流程請使用 JavaScript API
-或原生 OpenCC CLI。
+npm CLI 也可使用相同配置名稱進行文字轉換：
+
+```bash
+opencc -c s2twp_jieba
+```
+
+`--inspect`、`--segmentation` 等診斷模式仍請使用原生 OpenCC CLI。
 
 ## 相關 npm 套件
 
@@ -231,4 +237,3 @@ npm test
 
 若要為發布建置預編譯原生 addon，請參考
 [`node/PUBLISHING.md`](./PUBLISHING.md)。
-

node/cli.js

@@ -24,6 +24,16 @@ const BUILT_IN_CONFIG_NAMES = new Set(BUILT_IN_CONFIGS.map(([name]) => name));
 const BUILT_IN_CONFIG_STEMS = new Set(
   BUILT_IN_CONFIGS.map(([name]) => name.replace(/\.json$/, ''))
 );
+const OPTIONAL_JIEBA_CONFIGS = new Set([
+  's2hk_jieba.json',
+  's2t_jieba.json',
+  's2tw_jieba.json',
+  's2twp_jieba.json',
+  'tw2sp_jieba.json',
+]);
+const OPTIONAL_JIEBA_CONFIG_STEMS = new Set(
+  Array.from(OPTIONAL_JIEBA_CONFIGS).map((name) => name.replace(/\.json$/, ''))
+);
 
 function printHelp() {
   console.log(`Open Chinese Convert (OpenCC) npm Command Line Tool
@@ -41,7 +51,6 @@ Options:
 Unsupported in the npm CLI:
   --inspect            Use the native OpenCC CLI for inspection output.
   --segmentation       Use the native OpenCC CLI for segmentation output.
-  plugins              Plugin-backed segmentation is not supported by this npm CLI.
 
 Built-in Configurations:
 ${BUILT_IN_CONFIGS.map(([name, description]) => `  ${name.padEnd(11)} ${description}`).join('\n')}
@@ -129,14 +138,18 @@ function writeOutput(outputFileName, text) {
 }
 
 function resolveConfigPath(config) {
-  if (BUILT_IN_CONFIG_NAMES.has(config) || path.isAbsolute(config)) {
+  if (BUILT_IN_CONFIG_NAMES.has(config) || OPTIONAL_JIEBA_CONFIGS.has(config) || path.isAbsolute(config)) {
     return config;
   }
 
   if (!config.endsWith('.json') && BUILT_IN_CONFIG_STEMS.has(config)) {
     return config + '.json';
   }
 
+  if (!config.endsWith('.json') && OPTIONAL_JIEBA_CONFIG_STEMS.has(config)) {
+    return config + '.json';
+  }
+
   return path.resolve(process.cwd(), config);
 }
 

node/opencc.js

@@ -44,19 +44,34 @@ const getAssetsPath = function (bindingPath) {
   return bindingDir;
 };
 
+function requireOptionalPeer(packageName) {
+  const searchPaths = [
+    process.cwd(),
+    path.join(__dirname, '..'),
+    __dirname,
+  ];
+  for (const searchPath of searchPaths) {
+    try {
+      return require(require.resolve(packageName, { paths: [searchPath] }));
+    } catch (e) {
+      // Try the next search path.
+    }
+  }
+  return null;
+}
+
 // Detect optional opencc-jieba plugin package.
 // When installed, jieba-based configs (e.g. s2twp_jieba.json) become available.
-let jiebaInfo = null;
-try {
-  jiebaInfo = require('opencc-jieba');
-} catch (e) {
-  // opencc-jieba not installed — jieba configs won't be available.
-}
+const jiebaInfo = requireOptionalPeer('opencc-jieba');
 
 const assetsPath = getAssetsPath(bindingPath);
+const isAbsolutePath = function (filePath) {
+  return path.isAbsolute(filePath) || /^[A-Za-z]:[\\/]/.test(filePath);
+};
+
 const getConfigPath = function (config) {
   let configPath = config;
-  if (config[0] !== '/' && config[1] !== ':') {
+  if (!isAbsolutePath(config)) {
     // Resolve relative path
     configPath = path.join(assetsPath, config);
   }
@@ -112,6 +127,29 @@ function patchConfigPaths(config, jieba, mainAssetsDir) {
   }
 }
 
+function resolveJiebaConfigPath(config, jieba) {
+  if (!jieba || isAbsolutePath(config)) {
+    return null;
+  }
+  if (typeof jieba.resolveConfigPath === 'function') {
+    return jieba.resolveConfigPath(config);
+  }
+  const candidates = [config];
+  if (!config.endsWith('.json')) {
+    candidates.push(config + '.json');
+  }
+  for (const candidate of candidates) {
+    if (candidate.includes('/') || candidate.includes('\\')) {
+      continue;
+    }
+    const configPath = path.join(jieba.dataDir, candidate);
+    if (fs.existsSync(configPath)) {
+      return configPath;
+    }
+  }
+  return null;
+}
+
 /**
  * OpenCC Node.js API
  *
@@ -127,17 +165,15 @@ const OpenCC = module.exports = function (config) {
   // When opencc-jieba is installed, check if the requested config is a jieba
   // config. If so, load its JSON, patch all paths to absolute, and pass the
   // patched JSON string directly to the C++ layer via NewFromString.
-  if (jiebaInfo && config[0] !== '/' && config[1] !== ':') {
-    const jiebaConfigPath = path.join(jiebaInfo.dataDir, config);
-    if (fs.existsSync(jiebaConfigPath)) {
-      const raw = JSON.parse(fs.readFileSync(jiebaConfigPath, 'utf-8'));
-      patchConfigPaths(raw, jiebaInfo, assetsPath);
-      this.handler = new binding.Opencc(
-        JSON.stringify(raw),
-        jiebaInfo.dataDir + '/'
-      );
-      return;
-    }
+  const jiebaConfigPath = resolveJiebaConfigPath(config, jiebaInfo);
+  if (jiebaConfigPath) {
+    const raw = JSON.parse(fs.readFileSync(jiebaConfigPath, 'utf-8'));
+    patchConfigPaths(raw, jiebaInfo, assetsPath);
+    this.handler = new binding.Opencc(
+      JSON.stringify(raw),
+      jiebaInfo.dataDir + '/'
+    );
+    return;
   }
 
   config = getConfigPath(config);

node/test.js

@@ -11,6 +11,29 @@ const { prepareArtifacts } = require('../scripts/prepare-node-prebuild-artifacts
 
 const cases = JSON.parse(fs.readFileSync('test/testcases/testcases.json', 'utf-8')).cases || [];
 
+function createLocalInstalledShape() {
+  const rootDir = path.resolve(__dirname, '..');
+  const jiebaPackageDir = path.join(rootDir, 'plugins', 'jieba', 'node');
+  const libName = os.platform() === 'win32' ? 'opencc-jieba.dll'
+    : os.platform() === 'darwin' ? 'libopencc-jieba.dylib'
+      : 'libopencc-jieba.so';
+  const requiredFiles = [
+    path.join(jiebaPackageDir, 'index.js'),
+    path.join(jiebaPackageDir, 'data', 's2twp_jieba.json'),
+    path.join(jiebaPackageDir, 'data', 'jieba_dict', 'jieba_merged.ocd2'),
+    path.join(jiebaPackageDir, 'prebuilds', `${os.platform()}-${os.arch()}`, libName),
+  ];
+  if (!requiredFiles.every((file) => fs.existsSync(file))) {
+    return null;
+  }
+  const root = fs.mkdtempSync(path.join(os.tmpdir(), 'opencc-node-jieba-'));
+  const nodeModulesDir = path.join(root, 'node_modules');
+  fs.mkdirSync(nodeModulesDir, { recursive: true });
+  fs.symlinkSync(rootDir, path.join(nodeModulesDir, 'opencc'), 'dir');
+  fs.symlinkSync(jiebaPackageDir, path.join(nodeModulesDir, 'opencc-jieba'), 'dir');
+  return root;
+}
+
 const testSync = function (tc, cfg, expected, done) {
   const opencc = new OpenCC(cfg + '.json');
   const converted = opencc.convertSync(tc.input);
@@ -157,7 +180,6 @@ describe('npm CLI', function () {
     assert.match(result.stdout, /Unsupported in the npm CLI:/);
     assert.match(result.stdout, /--inspect/);
     assert.match(result.stdout, /--segmentation/);
-    assert.match(result.stdout, /plugins/);
   });
 
   it('prints version', function () {
@@ -194,6 +216,44 @@ describe('npm CLI', function () {
   });
 });
 
+describe('Optional opencc-jieba package integration', function () {
+  it('loads jieba configs by mode name in the JavaScript API', function () {
+    const installRoot = createLocalInstalledShape();
+    if (!installRoot) this.skip();
+
+    const script = [
+      "const OpenCC = require('opencc');",
+      "const converter = new OpenCC('s2twp_jieba');",
+      "process.stdout.write(converter.convertSync('云计算'));",
+    ].join('');
+    const result = childProcess.spawnSync(process.execPath, ['-e', script], {
+      cwd: installRoot,
+      env: { ...process.env },
+      encoding: 'utf8',
+    });
+    assert.equal(result.status, 0, result.stderr);
+    assert.equal(result.stdout, '雲端計算');
+  });
+
+  it('loads jieba configs by mode name in the npm CLI', function () {
+    const installRoot = createLocalInstalledShape();
+    if (!installRoot) this.skip();
+
+    const result = childProcess.spawnSync(process.execPath, [
+      path.join(installRoot, 'node_modules', 'opencc', 'node', 'cli.js'),
+      '-c',
+      's2twp_jieba',
+    ], {
+      cwd: installRoot,
+      env: { ...process.env },
+      input: '云计算',
+      encoding: 'utf8',
+    });
+    assert.equal(result.status, 0, result.stderr);
+    assert.equal(result.stdout, '雲端計算');
+  });
+});
+
 describe('Node prebuild assets', function () {
   it('collects only runtime json and ocd2 assets', function () {
     const root = fs.mkdtempSync(path.join(os.tmpdir(), 'opencc-prebuild-assets-'));

plugins/jieba/BUILD.bazel

@@ -51,6 +51,22 @@ cc_binary(
     ],
 )
 
+genrule(
+    name = "opencc_jieba_windows_zig",
+    srcs = [
+        "//deps/marisa-0.3.1:all_files",
+        "//plugins/jieba/deps/cppjieba:all_files",
+        "//src:all_files",
+        "include/JiebaSegmentation.hpp",
+        "src/JiebaSegmentation.cpp",
+        "src/JiebaSegmentationPlugin.cpp",
+    ],
+    outs = ["windows-x64/opencc-jieba.dll"],
+    cmd = "bash $(location //scripts:build_jieba_windows_zig) $@",
+    tags = ["manual"],
+    tools = ["//scripts:build_jieba_windows_zig"],
+)
+
 genrule(
     name = "jieba_merged_dict",
     srcs = [

plugins/jieba/deps/cppjieba/BUILD.bazel

@@ -2,6 +2,11 @@ load("@rules_cc//cc:defs.bzl", "cc_library")
 
 package(default_visibility = ["//visibility:public"])
 
+filegroup(
+    name = "all_files",
+    srcs = glob(["**/*"]),
+)
+
 exports_files([
     "dict/jieba.dict.utf8",
     "dict/user.dict.utf8",

plugins/jieba/node/build.js

@@ -10,6 +10,7 @@ process.chdir(rootDir);
 let platform = os.platform();
 let arch = os.arch();
 let bazelFlags = '-c opt';
+let libTarget = '//plugins/jieba:opencc-jieba';
 
 // E.g., node build.js linux-arm64
 if (process.argv[2]) {
@@ -20,6 +21,8 @@ if (process.argv[2]) {
     if (platform === 'linux') {
       const configArch = arch === 'x64' ? 'x86_64' : arch;
       bazelFlags += ` --config=linux-${configArch}-remote`;
+    } else if (platform === 'win32' && arch === 'x64') {
+      libTarget = '//plugins/jieba:opencc_jieba_windows_zig';
     }
   }
 }
@@ -29,14 +32,14 @@ bazelFlags += ' --remote_download_toplevel';
 
 console.log(`Building native addon and dictionary via Bazel for ${platform}-${arch}...`);
 try {
-  execSync(`bazel build ${bazelFlags} //plugins/jieba:opencc-jieba //plugins/jieba:jieba_merged_dict`, { stdio: 'inherit' });
+  execSync(`bazel build ${bazelFlags} ${libTarget} //plugins/jieba:jieba_merged_dict`, { stdio: 'inherit' });
 } catch (err) {
   console.error('Bazel build failed.', err);
   process.exit(1);
 }
 
 // Dynamically resolve exact output paths using cquery
-const libPathCmd = `bazel cquery ${bazelFlags} --output=files //plugins/jieba:opencc-jieba`;
+const libPathCmd = `bazel cquery ${bazelFlags} --output=files ${libTarget}`;
 const libSrcRaw = execSync(libPathCmd, { encoding: 'utf8', stdio: ['pipe', 'pipe', 'ignore'] }).trim().split('\n').pop();
 const libSrc = path.join(rootDir, libSrcRaw);
 

plugins/jieba/node/index.js

@@ -10,11 +10,38 @@ const libName = platform === 'win32' ? 'opencc-jieba.dll'
   : platform === 'darwin' ? 'libopencc-jieba.dylib'
   : 'libopencc-jieba.so';
 
+const dataDir = path.join(__dirname, 'data');
+const configDir = dataDir;
+
+function resolveConfigPath(config) {
+  if (typeof config !== 'string' || path.isAbsolute(config)) {
+    return null;
+  }
+  const candidates = [config];
+  if (!config.endsWith('.json')) {
+    candidates.push(`${config}.json`);
+  }
+  for (const candidate of candidates) {
+    if (candidate.includes('/') || candidate.includes('\\')) {
+      continue;
+    }
+    const configPath = path.join(configDir, candidate);
+    if (require('fs').existsSync(configPath)) {
+      return configPath;
+    }
+  }
+  return null;
+}
+
 module.exports = {
   /** Directory containing the prebuilt libopencc-jieba shared library. */
   pluginDir: prebuildDir,
   /** Absolute path to the platform-specific plugin shared library. */
   pluginLibrary: path.join(prebuildDir, libName),
   /** Root data directory (config files and jieba_dict/ live here). */
-  dataDir: path.join(__dirname, 'data'),
+  dataDir,
+  /** Directory containing jieba-backed OpenCC configuration JSON files. */
+  configDir,
+  /** Resolve a jieba-backed config name, with or without the .json suffix. */
+  resolveConfigPath,
 };