native function validator for your consideration

tc39 · Jul 26, 2020 · 405d896 · 405d896
1 parent 4e7f377
commit 405d896
Show file tree

Hide file tree

Showing 2 changed files with 238 additions and 48 deletions.
diff --git a/harness/nativeFunctionMatcher.js b/harness/nativeFunctionMatcher.js
@@ -3,20 +3,198 @@
 /*---
 description: Assert _NativeFunction_ Syntax
 info: |
-    This regex makes a best-effort determination that the tested string matches
-    the NativeFunction grammar production without requiring a correct tokeniser.
-
     NativeFunction :
       function _NativeFunctionAccessor_ opt _IdentifierName_ opt ( _FormalParameters_ ) { [ native code ] }
     NativeFunctionAccessor :
       get
       set
 defines:
-  - NATIVE_FUNCTION_RE
   - assertToStringOrNativeFunction
   - assertNativeFunction
+  - validateNativeFunctionSource
 ---*/
-const NATIVE_FUNCTION_RE = /\bfunction\b((get|set)\b)?[\s\S]*\([\s\S]*\)[\s\S]*\{[\s\S]*\[[\s\S]*\bnative\b[\s\S]+\bcode\b[\s\S]*\][\s\S]*\}/;
+
+const validateNativeFunctionSource = function(source) {
+  const UnicodeIDStart = /\p{ID_Start}/u;
+  const UnicodeIDContinue = /\p{ID_Continue}/u;
+  const isNewline = (c) => /[\u000A\u000D\u2028\u2029]/u.test(c);
+  const isWhitespace = (c) => /[\u0009\u000B\u000C\u0020\u00A0\uFEFF]|\p{Space_Separator}/u.test(c);
+
+  let i = 0;
+
+  const eatWhitespace = () => {
+    while (i < source.length) {
+      const c = source[i];
+      if (isWhitespace(c) || isNewline(c)) {
+        i += 1;
+        continue;
+      }
+
+      if (c === '/') {
+        if (source[i + 1] === '/') {
+          while (i < source.length) {
+            if (isNewline(source[i])) {
+              break;
+            }
+            i += 1;
+          }
+          continue;
+        }
+        if (source[i + 1] === '*') {
+          const end = source.indexOf('*/', i);
+          if (end === -1) {
+            throw new SyntaxError();
+          }
+          i = end + '*/'.length;
+          continue;
+        }
+      }
+
+      break;
+    }
+  };
+
+  const getIdentifier = () => {
+    eatWhitespace();
+
+    const start = i;
+    let end = i;
+    switch (source[end]) {
+      case '_':
+      case '$':
+        end += 1;
+        break;
+      default:
+        if (UnicodeIDStart.test(source[end])) {
+          end += 1;
+          break;
+        }
+        return null;
+    }
+    while (end < source.length) {
+      const c = source[end];
+      switch (c) {
+        case '_':
+        case '$':
+          end += 1;
+          break;
+        default:
+          if (UnicodeIDContinue.test(c)) {
+            end += 1;
+            break;
+          }
+          return source.slice(start, end);
+      }
+    }
+    return source.slice(start, end);
+  };
+
+  const test = (s) => {
+    eatWhitespace();
+
+    if (/\w/.test(s)) {
+      return getIdentifier() === s;
+    }
+    return source.slice(i, i + s.length) === s;
+  };
+
+  const eat = (s) => {
+    if (test(s)) {
+      i += s.length;
+      return true;
+    }
+    return false;
+  };
+
+  const eatIdentifier = () => {
+    const n = getIdentifier();
+    if (n !== null) {
+      i += n.length;
+      return true;
+    }
+    return false;
+  };
+
+  const expect = (s) => {
+    if (!eat(s)) {
+      throw new SyntaxError();
+    }
+  };
+
+  const eatString = () => {
+    if (source[i] === '\'' || source[i] === '"') {
+      const match = source[i];
+      i += 1;
+      while (i < source.length) {
+        if (source[i] === match && source[i - 1] !== '\\') {
+          return;
+        }
+        if (isNewline(source[i])) {
+          throw new SyntaxError();
+        }
+        i += 1;
+      }
+      throw new SyntaxError();
+    }
+  };
+
+  // "Stumble" through source text until matching character is found.
+  // Assumes ECMAScript syntax keeps `[]` and `()` balanced.
+  const stumbleUntil = (c) => {
+    const match = {
+      ']': '[',
+      ')': '(',
+    }[c];
+    let nesting = 1;
+    while (i < source.length) {
+      eatWhitespace();
+      eatString(); // Strings may contain unbalanced characters.
+      if (source[i] === match) {
+        nesting += 1;
+      } else if (source[i] === c) {
+        nesting -= 1;
+      }
+      i += 1;
+      if (nesting === 0) {
+        return;
+      }
+    }
+    throw new SyntaxError();
+  };
+
+  // function
+  expect('function');
+
+  // NativeFunctionAccessor
+  eat('get') || eat('set');
+
+  // PropertyName
+  if (eatIdentifier()) {
+  } else if (eat('[')) {
+    stumbleUntil(']');
+  }
+
+  // ( FormalParameters )
+  expect('(');
+  stumbleUntil(')');
+
+  // {
+  expect('{');
+
+  // [native code]
+  expect('[');
+  expect('native');
+  expect('code');
+  expect(']');
+
+  // }
+  expect('}');
+
+  eatWhitespace();
+  if (i !== source.length) {
+    throw new SyntaxError();
+  }
+};
 
 const assertToStringOrNativeFunction = function(fn, expected) {
   const actual = "" + fn;
@@ -29,8 +207,9 @@ const assertToStringOrNativeFunction = function(fn, expected) {
 
 const assertNativeFunction = function(fn, special) {
   const actual = "" + fn;
-  assert(
-    NATIVE_FUNCTION_RE.test(actual),
-    "Conforms to NativeFunction Syntax: '" + actual + "'." + (special ? "(" + special + ")" : "")
-  );
+  try {
+    validateNativeFunctionSource(actual);
+  } catch (unused) {
+    $ERROR("Conforms to NativeFunction Syntax: '" + actual + "'." + (special ? "(" + special + ")" : ""))
+  }
 };
diff --git a/test/harness/nativeFunctionMatcher.js b/test/harness/nativeFunctionMatcher.js
@@ -8,42 +8,53 @@ description: >
 includes: [nativeFunctionMatcher.js]
 ---*/
 
-if (!NATIVE_FUNCTION_RE.test('function(){[native code]}')) {
-  $ERROR('expected string to pass: "function(){[native code]}"');
-}
-
-if (!NATIVE_FUNCTION_RE.test('function(){ [native code] }')) {
-  $ERROR('expected string to pass: "function(){ [native code] }"');
-}
-
-if (!NATIVE_FUNCTION_RE.test('function ( ) { [ native code ] }')) {
-  $ERROR('expected string to pass: "function ( ) { [ native code ] }"');
-}
-
-if (!NATIVE_FUNCTION_RE.test('function a(){ [native code] }')) {
-  $ERROR('expected string to pass: "function a(){ [native code] }"');
-}
-
-if (!NATIVE_FUNCTION_RE.test('function a(){ /* } */ [native code] }')) {
-  $ERROR('expected string to pass: "function a(){ /* } */ [native code] }"');
-}
-
-if (NATIVE_FUNCTION_RE.test('')) {
-  $ERROR('expected string to fail: ""');
-}
-
-if (NATIVE_FUNCTION_RE.test('native code')) {
-  $ERROR('expected string to fail: "native code"');
-}
-
-if (NATIVE_FUNCTION_RE.test('function(){}')) {
-  $ERROR('expected string to fail: "function(){}"');
-}
-
-if (NATIVE_FUNCTION_RE.test('function(){ "native code" }')) {
-  $ERROR('expected string to fail: "function(){ "native code" }"');
-}
-
-if (NATIVE_FUNCTION_RE.test('function(){ [] native code }')) {
-  $ERROR('expected string to fail: "function(){ [] native code }"');
-}
+[
+  'function(){[native code]}',
+  'function(){ [native code] }',
+  'function ( ) { [ native code ] }',
+  'function a(){ [native code] }',
+  'function a(){ /* } */ [native code] }',
+  `function a() {
+    // test
+    [native code]
+    /* test */
+  }`,
+  'function(a, b = function() { []; }) { [native code] }',
+  'function [Symbol.xyz]() { [native code] }',
+  'function [x[y][z[d]()]]() { [native code] }',
+  'function ["]"] () { [native code] }',
+  'function [\']\'] () { [native code] }',
+  '/* test */ function() { [native code] }',
+  'function() { [native code] } /* test */',
+  'function() { [native code] } // test',
+].forEach((s) => {
+  try {
+    validateNativeFunctionSource(s);
+  } catch (unused) {
+    $ERROR(`"${s}" should pass`);
+  }
+});
+
+[
+  'native code',
+  'function() {}',
+  'function(){ "native code" }',
+  'function(){ [] native code }',
+  'function()) { [native code] }',
+  'function(() { [native code] }',
+  'function []] () { [native code] }',
+  'function [[] () { [native code] }',
+  'function ["]] () { [native code] }',
+  'function [\']] () { [native code] }',
+  'function() { [native code] /* }',
+  '// function() { [native code] }',
+].forEach((s) => {
+  let fail = false;
+  try {
+    validateNativeFunctionSource(s);
+    fail = true;
+  } catch (unused) {}
+  if (fail) {
+    $ERROR(`"${s}" should fail`);
+  }
+});