feat: consistent esm syntax err (#90)

build.js

@@ -19,4 +19,7 @@ const minified = MINIFY && terser.minify(jsSourceProcessed, {
   }
 });
 
+if (minified.error)
+  throw minified.error;
+
 fs.writeFileSync('./dist/lexer.mjs', minified ? minified.code : jsSourceProcessed);

include-wasm/cjs-module-lexer.h

@@ -117,8 +117,6 @@ bool ru () {
   return true;
 }
 
-bool parse (uint32_t point);
-
 void _addExport (const uint16_t* start, const uint16_t* end) {
   Slice* export = (Slice*)(analysis_head);
   analysis_head = analysis_head + sizeof(Slice);
@@ -163,7 +161,7 @@ void (*addExport)(const uint16_t*, const uint16_t*) = &_addExport;
 void (*addReexport)(const uint16_t*, const uint16_t*) = &_addReexport;
 void (*addUnsafeGetter)(const uint16_t*, const uint16_t*) = &_addUnsafeGetter;
 void (*clearReexports)() = &_clearReexports;
-bool parseCJS (uint16_t* source, uint32_t sourceLen, void (*addExport)(const uint16_t* start, const uint16_t* end), void (*addReexport)(const uint16_t* start, const uint16_t* end), void (*addUnsafeGetter)(const uint16_t*, const uint16_t*), void (*clearReexports)());
+uint32_t parseCJS (uint16_t* source, uint32_t sourceLen, void (*addExport)(const uint16_t* start, const uint16_t* end), void (*addReexport)(const uint16_t* start, const uint16_t* end), void (*addUnsafeGetter)(const uint16_t*, const uint16_t*), void (*clearReexports)());
 
 enum RequireType {
   Import,
@@ -237,4 +235,4 @@ void nextChar (uint16_t ch);
 void nextCharSurrogate (uint16_t ch);
 uint16_t readChar ();
 
-void syntaxError ();
+void syntaxError (uint32_t code);

lexer.js

@@ -1149,6 +1149,10 @@ function codePointAtLast (bPos) {
   return ch;
 }
 
+function esmSyntaxErr (msg) {
+  return Object.assign(new Error(msg), { code: 'ERR_LEXER_ESM_SYNTAX' });
+}
+
 function throwIfImportStatement () {
   const startPos = pos;
   pos += 6;
@@ -1160,7 +1164,7 @@ function throwIfImportStatement () {
       return;
     // import.meta
     case 46/*.*/:
-      throw new Error('Unexpected import.meta in CJS module.');
+      throw esmSyntaxErr('Unexpected import.meta in CJS module.');
 
     default:
       // no space after "import" -> not an import keyword
@@ -1176,7 +1180,7 @@ function throwIfImportStatement () {
         return;
       }
       // import statements are a syntax error in CommonJS
-      throw new Error('Unexpected import statement in CJS module.');
+      throw esmSyntaxErr('Unexpected import statement in CJS module.');
   }
 }
 
@@ -1186,7 +1190,7 @@ function throwIfExportStatement () {
   const ch = commentWhitespace();
   if (pos === curPos && !isPunctuator(ch))
     return;
-  throw new Error('Unexpected export statement in CJS module.');
+  throw esmSyntaxErr('Unexpected export statement in CJS module.');
 }
 
 function commentWhitespace () {

package.json

@@ -13,10 +13,10 @@
   "types": "lexer.d.ts",
   "scripts": {
     "test-js": "mocha -b -u tdd test/*.js",
-    "test-wasm": "WASM=1 mocha -b -u tdd test/*.js",
+    "test-wasm": "cross-env WASM=1 mocha -b -u tdd test/*.js",
     "test": "npm run test-wasm && npm run test-js",
     "bench": "node --expose-gc bench/index.mjs",
-    "build": "node build.js && babel dist/lexer.mjs | terser -o dist/lexer.js",
+    "build": "node build.js ; babel dist/lexer.mjs -o dist/lexer.js ; terser dist/lexer.js -o dist/lexer.js",
     "build-wasm": "make lib/lexer.wasm && node build.js",
     "prepublishOnly": "make && npm run build",
     "footprint": "npm run build && cat dist/lexer.js | gzip -9f | wc -c"
@@ -27,6 +27,7 @@
     "@babel/cli": "^7.5.5",
     "@babel/core": "^7.5.5",
     "@babel/plugin-transform-modules-commonjs": "^7.5.0",
+    "cross-env": "^7.0.3",
     "kleur": "^2.0.2",
     "mocha": "^9.1.3",
     "terser": "^4.1.4"

src/lexer.c

@@ -30,7 +30,7 @@ uint16_t* lastReexportEnd;
 // -> source
 // -> analysis starts after source
 uint32_t parse_error;
-bool has_error = false;
+uint32_t error = 0;
 uint32_t sourceLen;
 
 uint16_t templateStack_[STACK_DEPTH];
@@ -45,7 +45,7 @@ void (*addUnsafeGetter)(const uint16_t*, const uint16_t*);
 void (*clearReexports)();
 
 // Note: parsing is based on the _assumption_ that the source is already valid
-bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const uint16_t*, const uint16_t*), void (*_addReexport)(const uint16_t*, const uint16_t*), void (*_addUnsafeGetter)(const uint16_t*, const uint16_t*), void (*_clearReexports)()) {
+uint32_t parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const uint16_t*, const uint16_t*), void (*_addReexport)(const uint16_t*, const uint16_t*), void (*_addUnsafeGetter)(const uint16_t*, const uint16_t*), void (*_clearReexports)()) {
   source = _source;
   sourceLen = _sourceLen;
   if (_addExport)
@@ -61,7 +61,7 @@ bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const
   lastTokenPos = (uint16_t*)EMPTY_CHAR;
   lastSlashWasDivision = false;
   parse_error = 0;
-  has_error = false;
+  error = 0;
   templateStack = &templateStack_[0];
   openTokenPosStack = &openTokenPosStack_[0];
   starExportStack = &starExportStack_[0];
@@ -74,7 +74,7 @@ bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const
   // Handle #!
   if (*source == '#' && *(source + 1) == '!') {
     if (sourceLen == 2)
-      return true;
+      return 0;
     pos += 2;
     while (pos++ < end) {
       ch = *pos;
@@ -155,7 +155,7 @@ bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const
         break;
       case ')':
         if (openTokenDepth == 0)
-          return syntaxError(), false;
+          return syntaxError(8), error;
         openTokenDepth--;
         break;
       case '{':
@@ -165,14 +165,14 @@ bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const
         break;
       case '}':
         if (openTokenDepth == 0)
-          return syntaxError(), false;
+          return syntaxError(2), false;
         if (openTokenDepth-- == templateDepth) {
           templateDepth = templateStack[--templateStackDepth];
           templateString();
         }
         else {
           if (templateDepth != UINT16_MAX && openTokenDepth < templateDepth)
-            return syntaxError(), false;
+            return syntaxError(3), error;
         }
         break;
       case '<':
@@ -219,18 +219,18 @@ bool parseCJS (uint16_t* _source, uint32_t _sourceLen, void (*_addExport)(const
       }
       case '`':
         if (templateDepth == UINT16_MAX - 1)
-          return syntaxError(), false;
+          return syntaxError(4), error;
         templateString();
         break;
     }
     lastTokenPos = pos;
   }
 
-  if (templateDepth != UINT16_MAX || openTokenDepth || has_error)
-    return false;
+  if (templateDepth != UINT16_MAX || openTokenDepth || error)
+    return error;
 
   // success
-  return true;
+  return 0;
 }
 
 void tryBacktrackAddStarExportBinding (uint16_t* bPos) {
@@ -1154,7 +1154,7 @@ void throwIfImportStatement () {
       return;
     // import.meta
     case '.':
-      syntaxError();
+      syntaxError(5);
       return;
 
     default:
@@ -1171,7 +1171,7 @@ void throwIfImportStatement () {
         return;
       }
       // import statements are a syntax error in CommonJS
-      syntaxError();
+      syntaxError(6);
   }
 }
 
@@ -1181,7 +1181,7 @@ void throwIfExportStatement () {
   uint16_t ch = commentWhitespace();
   if (pos == curPos && !isPunctuator(ch))
     return;
-  syntaxError();
+  syntaxError(7);
 }
 
 uint16_t commentWhitespace () {
@@ -1218,7 +1218,7 @@ void templateString () {
     if (ch == '\\')
       pos++;
   }
-  syntaxError();
+  syntaxError(8);
 }
 
 void blockComment () {
@@ -1253,7 +1253,7 @@ void stringLiteral (uint16_t quote) {
     else if (isBr(ch))
       break;
   }
-  syntaxError();
+  syntaxError(9);
 }
 
 uint16_t regexCharacterClass () {
@@ -1266,7 +1266,7 @@ uint16_t regexCharacterClass () {
     else if (ch == '\n' || ch == '\r')
       break;
   }
-  syntaxError();
+  syntaxError(10);
   return '\0';
 }
 
@@ -1282,7 +1282,7 @@ void regularExpression () {
     else if (ch == '\n' || ch == '\r')
       break;
   }
-  syntaxError();
+  syntaxError(11);
 }
 
 uint16_t readToWsOrPunctuator (uint16_t ch) {
@@ -1503,14 +1503,14 @@ bool isExpressionTerminator (uint16_t* curPos) {
   return false;
 }
 
-void bail (uint32_t error) {
-  has_error = true;
-  parse_error = error;
+void bail (uint32_t err) {
+  error = 1;
+  parse_error = err;
   pos = end + 1;
 }
 
-void syntaxError () {
-  has_error = true;
+void syntaxError (uint32_t code) {
+  if (error == 0) error = code;
   parse_error = pos - source;
   pos = end + 1;
 }

src/lexer.js

@@ -16,8 +16,15 @@ export function parse (source, name = '@') {
   const addr = wasm.sa(len);
   (isLE ? copyLE : copyBE)(source, new Uint16Array(wasm.memory.buffer, addr, len));
 
-  if (!wasm.parseCJS(addr, source.length, 0, 0, 0))
-    throw Object.assign(new Error(`Parse error ${name}${wasm.e()}:${source.slice(0, wasm.e()).split('\n').length}:${wasm.e() - source.lastIndexOf('\n', wasm.e() - 1)}`), { idx: wasm.e() });
+  const err_code = wasm.parseCJS(addr, source.length, 0, 0, 0);
+
+  if (err_code) {
+    const err = new Error(`Parse error ${name}${wasm.e()}:${source.slice(0, wasm.e()).split('\n').length}:${wasm.e() - source.lastIndexOf('\n', wasm.e() - 1)}`);
+    Object.assign(err, { idx: wasm.e() });
+    if (err_code === 5 || err_code === 6 || err_code === 7)
+      Object.assign(err, { code: 'ERR_LEXER_ESM_SYNTAX' });
+    throw err;
+  }
 
   let exports = new Set(), reexports = new Set(), unsafeGetters = new Set();