apdev-js 0.1.0 → 0.2.0

package/dist/cli.js

@@ -9,88 +9,98 @@ var __dirname = /* @__PURE__ */ getDirname();
 
 // src/cli.ts
 import { readFileSync as readFileSync4 } from "fs";
-import { resolve, dirname, join as join3 } from "path";
-import { fileURLToPath as fileURLToPath2 } from "url";
+import { resolve, dirname as dirname2, join as join4, sep as sep3 } from "path";
+import { fileURLToPath as fileURLToPath3 } from "url";
 import { execFileSync } from "child_process";
 import { Command } from "commander";
 
 // src/check-chars.ts
-import { readFileSync } from "fs";
-import { extname } from "path";
-var EMOJI_RANGES = [
-  [127744, 128511],
-  // Symbols and Pictographs
-  [128512, 128591],
-  // Emoticons
-  [128640, 128767],
-  // Transport and Map Symbols
-  [128896, 129023],
-  // Geometric Shapes Extended
-  [129280, 129535],
-  // Supplemental Symbols and Pictographs
-  [9728, 9983],
-  // Miscellaneous Symbols
-  [9984, 10175]
-  // Dingbats
-];
-var EXTRA_ALLOWED_RANGES = [
-  [128, 255],
-  // Latin-1 Supplement
-  [8192, 8303],
-  // General Punctuation
-  [8448, 8527],
-  // Letterlike Symbols
-  [8592, 8703],
-  // Arrows
-  [8704, 8959],
-  // Mathematical Operators
-  [8960, 9215],
-  // Miscellaneous Technical
-  [9472, 9599],
-  // Box Drawing
-  [9632, 9727],
-  // Geometric Shapes
-  [11008, 11263],
-  // Miscellaneous Symbols and Arrows
-  [65024, 65039]
-  // Variation Selectors
-];
-var ALL_RANGES = [...EMOJI_RANGES, ...EXTRA_ALLOWED_RANGES];
-var DANGEROUS_CODEPOINTS = /* @__PURE__ */ new Map([
-  // Bidi control characters (Trojan Source - CVE-2021-42574)
-  [8234, "LEFT-TO-RIGHT EMBEDDING"],
-  [8235, "RIGHT-TO-LEFT EMBEDDING"],
-  [8236, "POP DIRECTIONAL FORMATTING"],
-  [8237, "LEFT-TO-RIGHT OVERRIDE"],
-  [8238, "RIGHT-TO-LEFT OVERRIDE"],
-  [8294, "LEFT-TO-RIGHT ISOLATE"],
-  [8295, "RIGHT-TO-LEFT ISOLATE"],
-  [8296, "FIRST STRONG ISOLATE"],
-  [8297, "POP DIRECTIONAL ISOLATE"],
-  // Zero-width characters
-  [8203, "ZERO WIDTH SPACE"],
-  [8204, "ZERO WIDTH NON-JOINER"],
-  [8205, "ZERO WIDTH JOINER"],
-  [8206, "LEFT-TO-RIGHT MARK"],
-  [8207, "RIGHT-TO-LEFT MARK"],
-  [8288, "WORD JOINER"]
-]);
-var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
-var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
-function isAllowedChar(c) {
-  const code = c.codePointAt(0);
-  if (code <= 127) {
-    return true;
+import { readFileSync, existsSync } from "fs";
+import { extname, dirname, join, sep } from "path";
+import { fileURLToPath as fileURLToPath2 } from "url";
+function getCharsetsDir() {
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname(fileURLToPath2(import.meta.url));
+  const devPath = join(thisDir, "charsets");
+  if (existsSync(devPath)) {
+    return devPath;
+  }
+  return join(thisDir, "..", "src", "charsets");
+}
+function loadCharset(nameOrPath) {
+  if (nameOrPath.includes(sep) || nameOrPath.includes("/") || nameOrPath.endsWith(".json")) {
+    if (!existsSync(nameOrPath)) {
+      throw new Error(`Charset file not found: ${nameOrPath}`);
+    }
+    return JSON.parse(readFileSync(nameOrPath, "utf-8"));
   }
-  for (const [start, end] of ALL_RANGES) {
-    if (code >= start && code <= end) {
-      return true;
+  const filePath = join(getCharsetsDir(), `${nameOrPath}.json`);
+  if (!existsSync(filePath)) {
+    throw new Error(`Unknown charset: ${nameOrPath}`);
+  }
+  return JSON.parse(readFileSync(filePath, "utf-8"));
+}
+function parseRanges(entries) {
+  return entries.map((e) => [parseInt(e.start, 16), parseInt(e.end, 16)]);
+}
+function parseDangerous(entries) {
+  const map = /* @__PURE__ */ new Map();
+  for (const e of entries) {
+    map.set(parseInt(e.code, 16), e.name);
+  }
+  return map;
+}
+function resolveCharsets(charsetNames, charsetFiles) {
+  const base = loadCharset("base");
+  const rangesSet = /* @__PURE__ */ new Map();
+  const dangerous = parseDangerous(base.dangerous ?? []);
+  function addRanges(entries) {
+    for (const [s, e] of parseRanges(entries)) {
+      rangesSet.set(`${s}-${e}`, [s, e]);
+    }
+  }
+  addRanges(base.emoji_ranges ?? []);
+  addRanges(base.extra_ranges ?? []);
+  for (const name of charsetNames) {
+    const data = loadCharset(name);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
     }
   }
+  for (const path2 of charsetFiles) {
+    const data = loadCharset(path2);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
+    }
+  }
+  const ranges = [...rangesSet.values()].sort((a, b) => a[0] - b[0]);
+  return { ranges, dangerous };
+}
+var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
+var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
+function isInRanges(code, ranges) {
+  if (code <= 127) return true;
+  for (const [start, end] of ranges) {
+    if (code >= start && code <= end) return true;
+  }
   return false;
 }
-function isDangerousChar(c) {
-  return DANGEROUS_CODEPOINTS.has(c.codePointAt(0));
+var _baseRanges = null;
+var _baseDangerous = null;
+function getBaseDefaults() {
+  if (!_baseRanges || !_baseDangerous) {
+    const defaults = resolveCharsets([], []);
+    _baseRanges = defaults.ranges;
+    _baseDangerous = defaults.dangerous;
+  }
+  return { ranges: _baseRanges, dangerous: _baseDangerous };
 }
 function computeCommentMask(content, suffix) {
   if (PYTHON_SUFFIXES.has(suffix)) {
@@ -209,8 +219,13 @@ function computeCommentMaskJs(content) {
   }
   return mask;
 }
-function checkFile(filePath, maxProblems = 5) {
+function checkFile(filePath, maxProblems = 5, extraRanges, dangerousMap) {
   const problems = [];
+  if (!extraRanges || !dangerousMap) {
+    const defaults = getBaseDefaults();
+    extraRanges ??= defaults.ranges;
+    dangerousMap ??= defaults.dangerous;
+  }
   try {
     const content = readFileSync(filePath, "utf-8");
     const suffix = extname(filePath).toLowerCase();
@@ -220,15 +235,15 @@ function checkFile(filePath, maxProblems = 5) {
     for (const char of content) {
       position++;
       const code = char.codePointAt(0);
-      if (isDangerousChar(char)) {
+      if (dangerousMap.has(code)) {
         if (!commentMask.has(offset)) {
-          const name = DANGEROUS_CODEPOINTS.get(code);
+          const name = dangerousMap.get(code);
           const hex = code.toString(16).toUpperCase().padStart(4, "0");
           problems.push(
             `Dangerous character in code at position ${position}: U+${hex} (${name})`
           );
         }
-      } else if (!isAllowedChar(char)) {
+      } else if (!isInRanges(code, extraRanges)) {
         const hex = code.toString(16).toUpperCase().padStart(4, "0");
         problems.push(
           `Illegal character at position ${position}: ${JSON.stringify(char)} (U+${hex})`
@@ -239,15 +254,15 @@ function checkFile(filePath, maxProblems = 5) {
       }
       offset += char.length;
     }
-  } catch {
-    problems.push(`Failed to read file: ${filePath}`);
+  } catch (e) {
+    problems.push(`Failed to read file: ${filePath} (${e})`);
   }
   return problems;
 }
-function checkPaths(paths) {
+function checkPaths(paths, extraRanges, dangerousMap) {
   let hasError = false;
   for (const path2 of paths) {
-    const problems = checkFile(path2);
+    const problems = checkFile(path2, 5, extraRanges, dangerousMap);
     if (problems.length > 0) {
       hasError = true;
       console.log(`
@@ -262,7 +277,7 @@ ${path2} contains illegal characters:`);
 
 // src/check-imports.ts
 import { readFileSync as readFileSync2, readdirSync, statSync } from "fs";
-import { join, relative, sep, extname as extname2, basename } from "path";
+import { join as join2, relative, sep as sep2, extname as extname2, basename } from "path";
 import ts from "typescript";
 var SUPPORTED_EXTENSIONS = /* @__PURE__ */ new Set([
   ".ts",
@@ -282,7 +297,7 @@ var INDEX_BASENAMES = /* @__PURE__ */ new Set([
 ]);
 function fileToModule(filePath, srcDir) {
   const rel = relative(srcDir, filePath);
-  const parts = rel.split(sep);
+  const parts = rel.split(sep2);
   const last = parts[parts.length - 1];
   if (last === "index.ts" || last === "index.js" || last === "index.tsx" || last === "index.jsx") {
     parts.pop();
@@ -355,7 +370,7 @@ function findSourceFiles(dir) {
       if (entry === "node_modules" || entry === "dist" || entry === ".git") {
         continue;
       }
-      const full = join(d, entry);
+      const full = join2(d, entry);
       const stat = statSync(full);
       if (stat.isDirectory()) {
         walk(full);
@@ -472,10 +487,10 @@ Found ${cycles.length} circular import(s):
 
 // src/config.ts
 import { readFileSync as readFileSync3 } from "fs";
-import { join as join2 } from "path";
+import { join as join3 } from "path";
 function loadConfig(projectDir) {
   const dir = projectDir ?? process.cwd();
-  const pkgPath = join2(dir, "package.json");
+  const pkgPath = join3(dir, "package.json");
   let raw;
   try {
     raw = readFileSync3(pkgPath, "utf-8");
@@ -491,9 +506,12 @@ function loadConfig(projectDir) {
 }
 
 // src/cli.ts
+function collect(value, previous) {
+  return [...previous, value];
+}
 function getVersion() {
-  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname(fileURLToPath2(import.meta.url));
-  const pkgPath = join3(thisDir, "..", "package.json");
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname2(fileURLToPath3(import.meta.url));
+  const pkgPath = join4(thisDir, "..", "package.json");
   try {
     const pkg = JSON.parse(readFileSync4(pkgPath, "utf-8"));
     return pkg.version ?? "0.0.0";
@@ -502,15 +520,32 @@ function getVersion() {
   }
 }
 function getReleaseScript() {
-  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname(fileURLToPath2(import.meta.url));
-  return join3(thisDir, "..", "release.sh");
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname2(fileURLToPath3(import.meta.url));
+  return join4(thisDir, "..", "release.sh");
 }
 function buildProgram() {
   const program2 = new Command();
   program2.name("apdev").description("Shared development tools for TypeScript/JavaScript projects").version(getVersion());
-  program2.command("check-chars").description("Validate files contain only allowed characters").argument("<files...>", "Files to check").action((files) => {
+  program2.command("check-chars").description("Validate files contain only allowed characters").argument("<files...>", "Files to check").option("--charset <name>", "Extra charset preset (repeatable)", collect, []).option("--charset-file <path>", "Custom charset JSON file (repeatable)", collect, []).action((files, opts) => {
+    let charsetNames = opts.charset;
+    let charsetFiles = opts.charsetFile;
+    if (charsetNames.length === 0 && charsetFiles.length === 0) {
+      const envVal = process.env.APDEV_EXTRA_CHARS ?? "";
+      if (envVal) {
+        for (const item of envVal.split(",")) {
+          const trimmed = item.trim();
+          if (!trimmed) continue;
+          if (trimmed.includes("/") || trimmed.includes(sep3) || trimmed.endsWith(".json")) {
+            charsetFiles.push(trimmed);
+          } else {
+            charsetNames.push(trimmed);
+          }
+        }
+      }
+    }
+    const { ranges, dangerous } = resolveCharsets(charsetNames, charsetFiles);
     const resolved = files.map((f) => resolve(f));
-    const code = checkPaths(resolved);
+    const code = checkPaths(resolved, ranges, dangerous);
     process.exit(code);
   });
   program2.command("check-imports").description("Detect circular imports in a JS/TS package").option("--package <name>", "Base package name (e.g. mylib). Reads from package.json apdev config if omitted.").option("--src-dir <dir>", "Source directory containing the package (default: src)").action((opts) => {
@@ -526,7 +561,7 @@ function buildProgram() {
     const code = checkCircularImports(resolve(srcDir), basePackage);
     process.exit(code);
   });
-  program2.command("release").description("Interactive release automation (build, tag, GitHub release, npm publish)").option("--yes, -y", "Auto-accept all defaults (silent mode)").argument("[version]", "Version to release (auto-detected from package.json if omitted)").action((version, opts) => {
+  program2.command("release").description("Interactive release automation (build, tag, GitHub release, npm publish)").option("-y, --yes", "Auto-accept all defaults (silent mode)").argument("[version]", "Version to release (auto-detected from package.json if omitted)").action((version, opts) => {
     const script = getReleaseScript();
     try {
       readFileSync4(script);

package/dist/index.cjs

@@ -37,7 +37,10 @@ __export(src_exports, {
   fileToModule: () => fileToModule,
   findCycles: () => findCycles,
   isAllowedChar: () => isAllowedChar,
+  isDangerousChar: () => isDangerousChar,
+  loadCharset: () => loadCharset,
   loadConfig: () => loadConfig,
+  resolveCharsets: () => resolveCharsets,
   version: () => version
 });
 module.exports = __toCommonJS(src_exports);
@@ -49,80 +52,100 @@ var importMetaUrl = /* @__PURE__ */ getImportMetaUrl();
 // src/check-chars.ts
 var import_node_fs = require("fs");
 var import_node_path = require("path");
-var EMOJI_RANGES = [
-  [127744, 128511],
-  // Symbols and Pictographs
-  [128512, 128591],
-  // Emoticons
-  [128640, 128767],
-  // Transport and Map Symbols
-  [128896, 129023],
-  // Geometric Shapes Extended
-  [129280, 129535],
-  // Supplemental Symbols and Pictographs
-  [9728, 9983],
-  // Miscellaneous Symbols
-  [9984, 10175]
-  // Dingbats
-];
-var EXTRA_ALLOWED_RANGES = [
-  [128, 255],
-  // Latin-1 Supplement
-  [8192, 8303],
-  // General Punctuation
-  [8448, 8527],
-  // Letterlike Symbols
-  [8592, 8703],
-  // Arrows
-  [8704, 8959],
-  // Mathematical Operators
-  [8960, 9215],
-  // Miscellaneous Technical
-  [9472, 9599],
-  // Box Drawing
-  [9632, 9727],
-  // Geometric Shapes
-  [11008, 11263],
-  // Miscellaneous Symbols and Arrows
-  [65024, 65039]
-  // Variation Selectors
-];
-var ALL_RANGES = [...EMOJI_RANGES, ...EXTRA_ALLOWED_RANGES];
-var DANGEROUS_CODEPOINTS = /* @__PURE__ */ new Map([
-  // Bidi control characters (Trojan Source - CVE-2021-42574)
-  [8234, "LEFT-TO-RIGHT EMBEDDING"],
-  [8235, "RIGHT-TO-LEFT EMBEDDING"],
-  [8236, "POP DIRECTIONAL FORMATTING"],
-  [8237, "LEFT-TO-RIGHT OVERRIDE"],
-  [8238, "RIGHT-TO-LEFT OVERRIDE"],
-  [8294, "LEFT-TO-RIGHT ISOLATE"],
-  [8295, "RIGHT-TO-LEFT ISOLATE"],
-  [8296, "FIRST STRONG ISOLATE"],
-  [8297, "POP DIRECTIONAL ISOLATE"],
-  // Zero-width characters
-  [8203, "ZERO WIDTH SPACE"],
-  [8204, "ZERO WIDTH NON-JOINER"],
-  [8205, "ZERO WIDTH JOINER"],
-  [8206, "LEFT-TO-RIGHT MARK"],
-  [8207, "RIGHT-TO-LEFT MARK"],
-  [8288, "WORD JOINER"]
-]);
-var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
-var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
-function isAllowedChar(c) {
-  const code = c.codePointAt(0);
-  if (code <= 127) {
-    return true;
+var import_node_url = require("url");
+function getCharsetsDir() {
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : (0, import_node_path.dirname)((0, import_node_url.fileURLToPath)(importMetaUrl));
+  const devPath = (0, import_node_path.join)(thisDir, "charsets");
+  if ((0, import_node_fs.existsSync)(devPath)) {
+    return devPath;
+  }
+  return (0, import_node_path.join)(thisDir, "..", "src", "charsets");
+}
+function loadCharset(nameOrPath) {
+  if (nameOrPath.includes(import_node_path.sep) || nameOrPath.includes("/") || nameOrPath.endsWith(".json")) {
+    if (!(0, import_node_fs.existsSync)(nameOrPath)) {
+      throw new Error(`Charset file not found: ${nameOrPath}`);
+    }
+    return JSON.parse((0, import_node_fs.readFileSync)(nameOrPath, "utf-8"));
+  }
+  const filePath = (0, import_node_path.join)(getCharsetsDir(), `${nameOrPath}.json`);
+  if (!(0, import_node_fs.existsSync)(filePath)) {
+    throw new Error(`Unknown charset: ${nameOrPath}`);
+  }
+  return JSON.parse((0, import_node_fs.readFileSync)(filePath, "utf-8"));
+}
+function parseRanges(entries) {
+  return entries.map((e) => [parseInt(e.start, 16), parseInt(e.end, 16)]);
+}
+function parseDangerous(entries) {
+  const map = /* @__PURE__ */ new Map();
+  for (const e of entries) {
+    map.set(parseInt(e.code, 16), e.name);
+  }
+  return map;
+}
+function resolveCharsets(charsetNames, charsetFiles) {
+  const base = loadCharset("base");
+  const rangesSet = /* @__PURE__ */ new Map();
+  const dangerous = parseDangerous(base.dangerous ?? []);
+  function addRanges(entries) {
+    for (const [s, e] of parseRanges(entries)) {
+      rangesSet.set(`${s}-${e}`, [s, e]);
+    }
   }
-  for (const [start, end] of ALL_RANGES) {
-    if (code >= start && code <= end) {
-      return true;
+  addRanges(base.emoji_ranges ?? []);
+  addRanges(base.extra_ranges ?? []);
+  for (const name of charsetNames) {
+    const data = loadCharset(name);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
+    }
+  }
+  for (const path of charsetFiles) {
+    const data = loadCharset(path);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
     }
   }
+  const ranges = [...rangesSet.values()].sort((a, b) => a[0] - b[0]);
+  return { ranges, dangerous };
+}
+var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
+var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
+function isInRanges(code, ranges) {
+  if (code <= 127) return true;
+  for (const [start, end] of ranges) {
+    if (code >= start && code <= end) return true;
+  }
   return false;
 }
+var _baseRanges = null;
+var _baseDangerous = null;
+function getBaseDefaults() {
+  if (!_baseRanges || !_baseDangerous) {
+    const defaults = resolveCharsets([], []);
+    _baseRanges = defaults.ranges;
+    _baseDangerous = defaults.dangerous;
+  }
+  return { ranges: _baseRanges, dangerous: _baseDangerous };
+}
+function isAllowedChar(c) {
+  const { ranges, dangerous } = getBaseDefaults();
+  const code = c.codePointAt(0);
+  if (dangerous.has(code)) return false;
+  return isInRanges(code, ranges);
+}
 function isDangerousChar(c) {
-  return DANGEROUS_CODEPOINTS.has(c.codePointAt(0));
+  const { dangerous } = getBaseDefaults();
+  return dangerous.has(c.codePointAt(0));
 }
 function computeCommentMask(content, suffix) {
   if (PYTHON_SUFFIXES.has(suffix)) {
@@ -241,8 +264,13 @@ function computeCommentMaskJs(content) {
   }
   return mask;
 }
-function checkFile(filePath, maxProblems = 5) {
+function checkFile(filePath, maxProblems = 5, extraRanges, dangerousMap) {
   const problems = [];
+  if (!extraRanges || !dangerousMap) {
+    const defaults = getBaseDefaults();
+    extraRanges ??= defaults.ranges;
+    dangerousMap ??= defaults.dangerous;
+  }
   try {
     const content = (0, import_node_fs.readFileSync)(filePath, "utf-8");
     const suffix = (0, import_node_path.extname)(filePath).toLowerCase();
@@ -252,15 +280,15 @@ function checkFile(filePath, maxProblems = 5) {
     for (const char of content) {
       position++;
       const code = char.codePointAt(0);
-      if (isDangerousChar(char)) {
+      if (dangerousMap.has(code)) {
         if (!commentMask.has(offset)) {
-          const name = DANGEROUS_CODEPOINTS.get(code);
+          const name = dangerousMap.get(code);
           const hex = code.toString(16).toUpperCase().padStart(4, "0");
           problems.push(
             `Dangerous character in code at position ${position}: U+${hex} (${name})`
           );
         }
-      } else if (!isAllowedChar(char)) {
+      } else if (!isInRanges(code, extraRanges)) {
         const hex = code.toString(16).toUpperCase().padStart(4, "0");
         problems.push(
           `Illegal character at position ${position}: ${JSON.stringify(char)} (U+${hex})`
@@ -271,15 +299,15 @@ function checkFile(filePath, maxProblems = 5) {
       }
       offset += char.length;
     }
-  } catch {
-    problems.push(`Failed to read file: ${filePath}`);
+  } catch (e) {
+    problems.push(`Failed to read file: ${filePath} (${e})`);
   }
   return problems;
 }
-function checkPaths(paths) {
+function checkPaths(paths, extraRanges, dangerousMap) {
   let hasError = false;
   for (const path of paths) {
-    const problems = checkFile(path);
+    const problems = checkFile(path, 5, extraRanges, dangerousMap);
     if (problems.length > 0) {
       hasError = true;
       console.log(`
@@ -525,9 +553,9 @@ function loadConfig(projectDir) {
 // src/index.ts
 var import_node_fs4 = require("fs");
 var import_node_path4 = require("path");
-var import_node_url = require("url");
+var import_node_url2 = require("url");
 function readVersion() {
-  const thisDir = typeof __dirname !== "undefined" ? __dirname : (0, import_node_path4.dirname)((0, import_node_url.fileURLToPath)(importMetaUrl));
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : (0, import_node_path4.dirname)((0, import_node_url2.fileURLToPath)(importMetaUrl));
   try {
     const pkg = JSON.parse((0, import_node_fs4.readFileSync)((0, import_node_path4.join)(thisDir, "..", "package.json"), "utf-8"));
     return pkg.version ?? "0.0.0";
@@ -545,6 +573,9 @@ var version = readVersion();
   fileToModule,
   findCycles,
   isAllowedChar,
+  isDangerousChar,
+  loadCharset,
   loadConfig,
+  resolveCharsets,
   version
 });

package/dist/index.d.cts

@@ -1,14 +1,51 @@
-/** Return true if the character is in the allowed set. */
+/**
+ * Character validation tool.
+ *
+ * Checks that files contain only allowed characters: ASCII, common emoji,
+ * and standard technical symbols (arrows, box-drawing, math operators, etc.).
+ *
+ * Additionally flags dangerous invisible/bidi characters in code regions
+ * (Trojan Source - CVE-2021-42574) while allowing them in comments.
+ */
+interface RangeEntry {
+    start: string;
+    end: string;
+    name: string;
+}
+interface DangerousEntry {
+    code: string;
+    name: string;
+}
+interface CharsetData {
+    name: string;
+    description?: string;
+    emoji_ranges?: RangeEntry[];
+    extra_ranges?: RangeEntry[];
+    dangerous?: DangerousEntry[];
+}
+declare function loadCharset(nameOrPath: string): CharsetData;
+declare function resolveCharsets(charsetNames: string[], charsetFiles: string[]): {
+    ranges: [number, number][];
+    dangerous: Map<number, string>;
+};
+/**
+ * Return true if the character is in the base allowed set.
+ *
+ * Dangerous codepoints (Trojan Source vectors) are excluded even though
+ * they fall within allowed Unicode ranges.
+ */
 declare function isAllowedChar(c: string): boolean;
+/** Return true if the character is a dangerous invisible/bidi codepoint. */
+declare function isDangerousChar(c: string): boolean;
 /**
  * Check a single file for illegal characters.
  * Returns a list of problem descriptions (empty if the file is clean).
  */
-declare function checkFile(filePath: string, maxProblems?: number): string[];
+declare function checkFile(filePath: string, maxProblems?: number, extraRanges?: [number, number][], dangerousMap?: Map<number, string>): string[];
 /**
  * Check multiple files. Returns 0 if all clean, 1 if any have problems.
  */
-declare function checkPaths(paths: string[]): number;
+declare function checkPaths(paths: string[], extraRanges?: [number, number][], dangerousMap?: Map<number, string>): number;
 
 /**
  * Circular import detection tool.
@@ -38,4 +75,4 @@ declare function loadConfig(projectDir?: string): Record<string, unknown>;
 
 declare const version: string;
 
-export { buildDependencyGraph, checkCircularImports, checkFile, checkPaths, fileToModule, findCycles, isAllowedChar, loadConfig, version };
+export { buildDependencyGraph, checkCircularImports, checkFile, checkPaths, fileToModule, findCycles, isAllowedChar, isDangerousChar, loadCharset, loadConfig, resolveCharsets, version };

package/dist/index.d.ts

@@ -1,14 +1,51 @@
-/** Return true if the character is in the allowed set. */
+/**
+ * Character validation tool.
+ *
+ * Checks that files contain only allowed characters: ASCII, common emoji,
+ * and standard technical symbols (arrows, box-drawing, math operators, etc.).
+ *
+ * Additionally flags dangerous invisible/bidi characters in code regions
+ * (Trojan Source - CVE-2021-42574) while allowing them in comments.
+ */
+interface RangeEntry {
+    start: string;
+    end: string;
+    name: string;
+}
+interface DangerousEntry {
+    code: string;
+    name: string;
+}
+interface CharsetData {
+    name: string;
+    description?: string;
+    emoji_ranges?: RangeEntry[];
+    extra_ranges?: RangeEntry[];
+    dangerous?: DangerousEntry[];
+}
+declare function loadCharset(nameOrPath: string): CharsetData;
+declare function resolveCharsets(charsetNames: string[], charsetFiles: string[]): {
+    ranges: [number, number][];
+    dangerous: Map<number, string>;
+};
+/**
+ * Return true if the character is in the base allowed set.
+ *
+ * Dangerous codepoints (Trojan Source vectors) are excluded even though
+ * they fall within allowed Unicode ranges.
+ */
 declare function isAllowedChar(c: string): boolean;
+/** Return true if the character is a dangerous invisible/bidi codepoint. */
+declare function isDangerousChar(c: string): boolean;
 /**
  * Check a single file for illegal characters.
  * Returns a list of problem descriptions (empty if the file is clean).
  */
-declare function checkFile(filePath: string, maxProblems?: number): string[];
+declare function checkFile(filePath: string, maxProblems?: number, extraRanges?: [number, number][], dangerousMap?: Map<number, string>): string[];
 /**
  * Check multiple files. Returns 0 if all clean, 1 if any have problems.
  */
-declare function checkPaths(paths: string[]): number;
+declare function checkPaths(paths: string[], extraRanges?: [number, number][], dangerousMap?: Map<number, string>): number;
 
 /**
  * Circular import detection tool.
@@ -38,4 +75,4 @@ declare function loadConfig(projectDir?: string): Record<string, unknown>;
 
 declare const version: string;
 
-export { buildDependencyGraph, checkCircularImports, checkFile, checkPaths, fileToModule, findCycles, isAllowedChar, loadConfig, version };
+export { buildDependencyGraph, checkCircularImports, checkFile, checkPaths, fileToModule, findCycles, isAllowedChar, isDangerousChar, loadCharset, loadConfig, resolveCharsets, version };

package/dist/index.js

@@ -6,82 +6,102 @@ var getDirname = () => path.dirname(getFilename());
 var __dirname = /* @__PURE__ */ getDirname();
 
 // src/check-chars.ts
-import { readFileSync } from "fs";
-import { extname } from "path";
-var EMOJI_RANGES = [
-  [127744, 128511],
-  // Symbols and Pictographs
-  [128512, 128591],
-  // Emoticons
-  [128640, 128767],
-  // Transport and Map Symbols
-  [128896, 129023],
-  // Geometric Shapes Extended
-  [129280, 129535],
-  // Supplemental Symbols and Pictographs
-  [9728, 9983],
-  // Miscellaneous Symbols
-  [9984, 10175]
-  // Dingbats
-];
-var EXTRA_ALLOWED_RANGES = [
-  [128, 255],
-  // Latin-1 Supplement
-  [8192, 8303],
-  // General Punctuation
-  [8448, 8527],
-  // Letterlike Symbols
-  [8592, 8703],
-  // Arrows
-  [8704, 8959],
-  // Mathematical Operators
-  [8960, 9215],
-  // Miscellaneous Technical
-  [9472, 9599],
-  // Box Drawing
-  [9632, 9727],
-  // Geometric Shapes
-  [11008, 11263],
-  // Miscellaneous Symbols and Arrows
-  [65024, 65039]
-  // Variation Selectors
-];
-var ALL_RANGES = [...EMOJI_RANGES, ...EXTRA_ALLOWED_RANGES];
-var DANGEROUS_CODEPOINTS = /* @__PURE__ */ new Map([
-  // Bidi control characters (Trojan Source - CVE-2021-42574)
-  [8234, "LEFT-TO-RIGHT EMBEDDING"],
-  [8235, "RIGHT-TO-LEFT EMBEDDING"],
-  [8236, "POP DIRECTIONAL FORMATTING"],
-  [8237, "LEFT-TO-RIGHT OVERRIDE"],
-  [8238, "RIGHT-TO-LEFT OVERRIDE"],
-  [8294, "LEFT-TO-RIGHT ISOLATE"],
-  [8295, "RIGHT-TO-LEFT ISOLATE"],
-  [8296, "FIRST STRONG ISOLATE"],
-  [8297, "POP DIRECTIONAL ISOLATE"],
-  // Zero-width characters
-  [8203, "ZERO WIDTH SPACE"],
-  [8204, "ZERO WIDTH NON-JOINER"],
-  [8205, "ZERO WIDTH JOINER"],
-  [8206, "LEFT-TO-RIGHT MARK"],
-  [8207, "RIGHT-TO-LEFT MARK"],
-  [8288, "WORD JOINER"]
-]);
-var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
-var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
-function isAllowedChar(c) {
-  const code = c.codePointAt(0);
-  if (code <= 127) {
-    return true;
+import { readFileSync, existsSync } from "fs";
+import { extname, dirname, join, sep } from "path";
+import { fileURLToPath as fileURLToPath2 } from "url";
+function getCharsetsDir() {
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname(fileURLToPath2(import.meta.url));
+  const devPath = join(thisDir, "charsets");
+  if (existsSync(devPath)) {
+    return devPath;
+  }
+  return join(thisDir, "..", "src", "charsets");
+}
+function loadCharset(nameOrPath) {
+  if (nameOrPath.includes(sep) || nameOrPath.includes("/") || nameOrPath.endsWith(".json")) {
+    if (!existsSync(nameOrPath)) {
+      throw new Error(`Charset file not found: ${nameOrPath}`);
+    }
+    return JSON.parse(readFileSync(nameOrPath, "utf-8"));
+  }
+  const filePath = join(getCharsetsDir(), `${nameOrPath}.json`);
+  if (!existsSync(filePath)) {
+    throw new Error(`Unknown charset: ${nameOrPath}`);
+  }
+  return JSON.parse(readFileSync(filePath, "utf-8"));
+}
+function parseRanges(entries) {
+  return entries.map((e) => [parseInt(e.start, 16), parseInt(e.end, 16)]);
+}
+function parseDangerous(entries) {
+  const map = /* @__PURE__ */ new Map();
+  for (const e of entries) {
+    map.set(parseInt(e.code, 16), e.name);
+  }
+  return map;
+}
+function resolveCharsets(charsetNames, charsetFiles) {
+  const base = loadCharset("base");
+  const rangesSet = /* @__PURE__ */ new Map();
+  const dangerous = parseDangerous(base.dangerous ?? []);
+  function addRanges(entries) {
+    for (const [s, e] of parseRanges(entries)) {
+      rangesSet.set(`${s}-${e}`, [s, e]);
+    }
   }
-  for (const [start, end] of ALL_RANGES) {
-    if (code >= start && code <= end) {
-      return true;
+  addRanges(base.emoji_ranges ?? []);
+  addRanges(base.extra_ranges ?? []);
+  for (const name of charsetNames) {
+    const data = loadCharset(name);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
     }
   }
+  for (const path2 of charsetFiles) {
+    const data = loadCharset(path2);
+    addRanges(data.emoji_ranges ?? []);
+    addRanges(data.extra_ranges ?? []);
+    if (data.dangerous) {
+      for (const [code, dname] of parseDangerous(data.dangerous)) {
+        dangerous.set(code, dname);
+      }
+    }
+  }
+  const ranges = [...rangesSet.values()].sort((a, b) => a[0] - b[0]);
+  return { ranges, dangerous };
+}
+var PYTHON_SUFFIXES = /* @__PURE__ */ new Set([".py"]);
+var JS_SUFFIXES = /* @__PURE__ */ new Set([".js", ".ts", ".tsx", ".jsx", ".mjs", ".cjs"]);
+function isInRanges(code, ranges) {
+  if (code <= 127) return true;
+  for (const [start, end] of ranges) {
+    if (code >= start && code <= end) return true;
+  }
   return false;
 }
+var _baseRanges = null;
+var _baseDangerous = null;
+function getBaseDefaults() {
+  if (!_baseRanges || !_baseDangerous) {
+    const defaults = resolveCharsets([], []);
+    _baseRanges = defaults.ranges;
+    _baseDangerous = defaults.dangerous;
+  }
+  return { ranges: _baseRanges, dangerous: _baseDangerous };
+}
+function isAllowedChar(c) {
+  const { ranges, dangerous } = getBaseDefaults();
+  const code = c.codePointAt(0);
+  if (dangerous.has(code)) return false;
+  return isInRanges(code, ranges);
+}
 function isDangerousChar(c) {
-  return DANGEROUS_CODEPOINTS.has(c.codePointAt(0));
+  const { dangerous } = getBaseDefaults();
+  return dangerous.has(c.codePointAt(0));
 }
 function computeCommentMask(content, suffix) {
   if (PYTHON_SUFFIXES.has(suffix)) {
@@ -200,8 +220,13 @@ function computeCommentMaskJs(content) {
   }
   return mask;
 }
-function checkFile(filePath, maxProblems = 5) {
+function checkFile(filePath, maxProblems = 5, extraRanges, dangerousMap) {
   const problems = [];
+  if (!extraRanges || !dangerousMap) {
+    const defaults = getBaseDefaults();
+    extraRanges ??= defaults.ranges;
+    dangerousMap ??= defaults.dangerous;
+  }
   try {
     const content = readFileSync(filePath, "utf-8");
     const suffix = extname(filePath).toLowerCase();
@@ -211,15 +236,15 @@ function checkFile(filePath, maxProblems = 5) {
     for (const char of content) {
       position++;
       const code = char.codePointAt(0);
-      if (isDangerousChar(char)) {
+      if (dangerousMap.has(code)) {
         if (!commentMask.has(offset)) {
-          const name = DANGEROUS_CODEPOINTS.get(code);
+          const name = dangerousMap.get(code);
           const hex = code.toString(16).toUpperCase().padStart(4, "0");
           problems.push(
             `Dangerous character in code at position ${position}: U+${hex} (${name})`
           );
         }
-      } else if (!isAllowedChar(char)) {
+      } else if (!isInRanges(code, extraRanges)) {
         const hex = code.toString(16).toUpperCase().padStart(4, "0");
         problems.push(
           `Illegal character at position ${position}: ${JSON.stringify(char)} (U+${hex})`
@@ -230,15 +255,15 @@ function checkFile(filePath, maxProblems = 5) {
       }
       offset += char.length;
     }
-  } catch {
-    problems.push(`Failed to read file: ${filePath}`);
+  } catch (e) {
+    problems.push(`Failed to read file: ${filePath} (${e})`);
   }
   return problems;
 }
-function checkPaths(paths) {
+function checkPaths(paths, extraRanges, dangerousMap) {
   let hasError = false;
   for (const path2 of paths) {
-    const problems = checkFile(path2);
+    const problems = checkFile(path2, 5, extraRanges, dangerousMap);
     if (problems.length > 0) {
       hasError = true;
       console.log(`
@@ -253,7 +278,7 @@ ${path2} contains illegal characters:`);
 
 // src/check-imports.ts
 import { readFileSync as readFileSync2, readdirSync, statSync } from "fs";
-import { join, relative, sep, extname as extname2, basename } from "path";
+import { join as join2, relative, sep as sep2, extname as extname2, basename } from "path";
 import ts from "typescript";
 var SUPPORTED_EXTENSIONS = /* @__PURE__ */ new Set([
   ".ts",
@@ -273,7 +298,7 @@ var INDEX_BASENAMES = /* @__PURE__ */ new Set([
 ]);
 function fileToModule(filePath, srcDir) {
   const rel = relative(srcDir, filePath);
-  const parts = rel.split(sep);
+  const parts = rel.split(sep2);
   const last = parts[parts.length - 1];
   if (last === "index.ts" || last === "index.js" || last === "index.tsx" || last === "index.jsx") {
     parts.pop();
@@ -346,7 +371,7 @@ function findSourceFiles(dir) {
       if (entry === "node_modules" || entry === "dist" || entry === ".git") {
         continue;
       }
-      const full = join(d, entry);
+      const full = join2(d, entry);
       const stat = statSync(full);
       if (stat.isDirectory()) {
         walk(full);
@@ -463,10 +488,10 @@ Found ${cycles.length} circular import(s):
 
 // src/config.ts
 import { readFileSync as readFileSync3 } from "fs";
-import { join as join2 } from "path";
+import { join as join3 } from "path";
 function loadConfig(projectDir) {
   const dir = projectDir ?? process.cwd();
-  const pkgPath = join2(dir, "package.json");
+  const pkgPath = join3(dir, "package.json");
   let raw;
   try {
     raw = readFileSync3(pkgPath, "utf-8");
@@ -483,12 +508,12 @@ function loadConfig(projectDir) {
 
 // src/index.ts
 import { readFileSync as readFileSync4 } from "fs";
-import { dirname, join as join3 } from "path";
-import { fileURLToPath as fileURLToPath2 } from "url";
+import { dirname as dirname2, join as join4 } from "path";
+import { fileURLToPath as fileURLToPath3 } from "url";
 function readVersion() {
-  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname(fileURLToPath2(import.meta.url));
+  const thisDir = typeof __dirname !== "undefined" ? __dirname : dirname2(fileURLToPath3(import.meta.url));
   try {
-    const pkg = JSON.parse(readFileSync4(join3(thisDir, "..", "package.json"), "utf-8"));
+    const pkg = JSON.parse(readFileSync4(join4(thisDir, "..", "package.json"), "utf-8"));
     return pkg.version ?? "0.0.0";
   } catch {
     return "0.0.0";
@@ -503,6 +528,9 @@ export {
   fileToModule,
   findCycles,
   isAllowedChar,
+  isDangerousChar,
+  loadCharset,
   loadConfig,
+  resolveCharsets,
   version
 };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "apdev-js",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Shared development tools for TypeScript/JavaScript projects - character validation, circular import detection, and more",
   "type": "module",
   "main": "./dist/index.cjs",
@@ -14,11 +14,12 @@
     }
   },
   "bin": {
-    "apdev": "./dist/cli.js"
+    "apdev-js": "./dist/cli.js"
   },
   "files": [
     "dist",
-    "release.sh"
+    "release.sh",
+    "src/charsets"
   ],
   "scripts": {
     "build": "tsup",
@@ -46,12 +47,20 @@
     "url": "https://github.com/aipartnerup/apdev/issues"
   },
   "dependencies": {
-    "commander": "^13.0.0",
-    "typescript": "^5.7.0"
+    "commander": "^13.0.0"
+  },
+  "peerDependencies": {
+    "typescript": "^5.0.0"
+  },
+  "peerDependenciesMeta": {
+    "typescript": {
+      "optional": true
+    }
   },
   "devDependencies": {
     "@types/node": "^22.0.0",
     "tsup": "^8.0.0",
+    "typescript": "^5.7.0",
     "vitest": "^3.0.0"
   }
 }

package/src/charsets/base.json

@@ -0,0 +1,44 @@
+{
+  "name": "base",
+  "description": "Default: ASCII + emoji + technical symbols",
+  "emoji_ranges": [
+    { "start": "0x1F300", "end": "0x1F5FF", "name": "Symbols and Pictographs" },
+    { "start": "0x1F600", "end": "0x1F64F", "name": "Emoticons" },
+    { "start": "0x1F680", "end": "0x1F6FF", "name": "Transport and Map Symbols" },
+    { "start": "0x1F780", "end": "0x1F7FF", "name": "Geometric Shapes Extended" },
+    { "start": "0x1F900", "end": "0x1F9FF", "name": "Supplemental Symbols and Pictographs" },
+    { "start": "0x2600", "end": "0x26FF", "name": "Miscellaneous Symbols" },
+    { "start": "0x2700", "end": "0x27BF", "name": "Dingbats" }
+  ],
+  "extra_ranges": [
+    { "start": "0x0080", "end": "0x00FF", "name": "Latin-1 Supplement" },
+    { "start": "0x2000", "end": "0x206F", "name": "General Punctuation" },
+    { "start": "0x2100", "end": "0x214F", "name": "Letterlike Symbols" },
+    { "start": "0x2190", "end": "0x21FF", "name": "Arrows" },
+    { "start": "0x2200", "end": "0x22FF", "name": "Mathematical Operators" },
+    { "start": "0x2300", "end": "0x23FF", "name": "Miscellaneous Technical" },
+    { "start": "0x2500", "end": "0x257F", "name": "Box Drawing" },
+    { "start": "0x2580", "end": "0x259F", "name": "Block Elements" },
+    { "start": "0x25A0", "end": "0x25FF", "name": "Geometric Shapes" },
+    { "start": "0x2800", "end": "0x28FF", "name": "Braille Patterns" },
+    { "start": "0x2B00", "end": "0x2BFF", "name": "Miscellaneous Symbols and Arrows" },
+    { "start": "0xFE00", "end": "0xFE0F", "name": "Variation Selectors" }
+  ],
+  "dangerous": [
+    { "code": "0x202A", "name": "LEFT-TO-RIGHT EMBEDDING" },
+    { "code": "0x202B", "name": "RIGHT-TO-LEFT EMBEDDING" },
+    { "code": "0x202C", "name": "POP DIRECTIONAL FORMATTING" },
+    { "code": "0x202D", "name": "LEFT-TO-RIGHT OVERRIDE" },
+    { "code": "0x202E", "name": "RIGHT-TO-LEFT OVERRIDE" },
+    { "code": "0x2066", "name": "LEFT-TO-RIGHT ISOLATE" },
+    { "code": "0x2067", "name": "RIGHT-TO-LEFT ISOLATE" },
+    { "code": "0x2068", "name": "FIRST STRONG ISOLATE" },
+    { "code": "0x2069", "name": "POP DIRECTIONAL ISOLATE" },
+    { "code": "0x200B", "name": "ZERO WIDTH SPACE" },
+    { "code": "0x200C", "name": "ZERO WIDTH NON-JOINER" },
+    { "code": "0x200D", "name": "ZERO WIDTH JOINER" },
+    { "code": "0x200E", "name": "LEFT-TO-RIGHT MARK" },
+    { "code": "0x200F", "name": "RIGHT-TO-LEFT MARK" },
+    { "code": "0x2060", "name": "WORD JOINER" }
+  ]
+}

package/src/charsets/chinese.json

@@ -0,0 +1,9 @@
+{
+  "name": "chinese",
+  "description": "Chinese: CJK Unified Ideographs, CJK Punctuation, Fullwidth Forms",
+  "extra_ranges": [
+    { "start": "0x3000", "end": "0x303F", "name": "CJK Symbols and Punctuation" },
+    { "start": "0x4E00", "end": "0x9FFF", "name": "CJK Unified Ideographs" },
+    { "start": "0xFF00", "end": "0xFFEF", "name": "Fullwidth Forms" }
+  ]
+}

package/src/charsets/japanese.json

@@ -0,0 +1,11 @@
+{
+  "name": "japanese",
+  "description": "Japanese: Hiragana, Katakana, CJK Unified Ideographs, CJK Punctuation, Fullwidth Forms",
+  "extra_ranges": [
+    { "start": "0x3000", "end": "0x303F", "name": "CJK Symbols and Punctuation" },
+    { "start": "0x3040", "end": "0x309F", "name": "Hiragana" },
+    { "start": "0x30A0", "end": "0x30FF", "name": "Katakana" },
+    { "start": "0x4E00", "end": "0x9FFF", "name": "CJK Unified Ideographs" },
+    { "start": "0xFF00", "end": "0xFFEF", "name": "Fullwidth Forms" }
+  ]
+}

package/src/charsets/korean.json

@@ -0,0 +1,11 @@
+{
+  "name": "korean",
+  "description": "Korean: Hangul Jamo, Hangul Syllables, CJK Unified Ideographs, CJK Punctuation, Fullwidth Forms",
+  "extra_ranges": [
+    { "start": "0x1100", "end": "0x11FF", "name": "Hangul Jamo" },
+    { "start": "0x3000", "end": "0x303F", "name": "CJK Symbols and Punctuation" },
+    { "start": "0x4E00", "end": "0x9FFF", "name": "CJK Unified Ideographs" },
+    { "start": "0xAC00", "end": "0xD7AF", "name": "Hangul Syllables" },
+    { "start": "0xFF00", "end": "0xFFEF", "name": "Fullwidth Forms" }
+  ]
+}