Lua 5.4 support

luaparse.js

@@ -106,16 +106,32 @@
  highMask | 0x80 | ((codepoint >> 6) & 0x3f),
  highMask | 0x80 | ( codepoint & 0x3f)
  );
- } else /* istanbul ignore else */ if (codepoint < 0x110000) {
+ } else if (codepoint < 0x200000) {
  return String.fromCharCode(
  highMask | 0xf0 | (codepoint >> 18) ,
  highMask | 0x80 | ((codepoint >> 12) & 0x3f),
  highMask | 0x80 | ((codepoint >> 6) & 0x3f),
  highMask | 0x80 | ( codepoint & 0x3f)
  );
+ } else if (codepoint < 0x4000000) {
+ return String.fromCharCode(
+ highMask | 0xf8 | (codepoint >> 24) ,
+ highMask | 0x80 | ((codepoint >> 18) & 0x3f),
+ highMask | 0x80 | ((codepoint >> 12) & 0x3f),
+ highMask | 0x80 | ((codepoint >> 6) & 0x3f),
+ highMask | 0x80 | ( codepoint & 0x3f)
+ );
+ } else /* istanbul ignore else */ if (codepoint <= 0x7fffffff) {
+ return String.fromCharCode(
+ highMask | 0xfc | (codepoint >> 30) ,
+ highMask | 0x80 | ((codepoint >> 24) & 0x3f),
+ highMask | 0x80 | ((codepoint >> 18) & 0x3f),
+ highMask | 0x80 | ((codepoint >> 12) & 0x3f),
+ highMask | 0x80 | ((codepoint >> 6) & 0x3f),
+ highMask | 0x80 | ( codepoint & 0x3f)
+ );
  } else {
- // TODO: Lua 5.4 allows up to six-byte sequences, as in UTF-8:1993
- return null;
+ throw new Error('Should not happen');
  }
  }
 
@@ -218,6 +234,7 @@
  , gotoJumpInLocalScope: '<goto %1> jumps into the scope of local \'%2\''
  , cannotUseVararg: 'cannot use \'...\' outside a vararg function near \'%1\''
  , invalidCodeUnit: 'code unit U+%1 is not allowed in the current encoding mode'
+ , unknownAttribute: 'unknown attribute \'%1\''
  };
 
  // ### Abstract Syntax Tree
@@ -363,13 +380,28 @@
  };
  }
 
+ , attribute: function(name) {
+ return {
+ type: 'Attribute'
+ , name: name
+ };
+ }
+
  , identifier: function(name) {
  return {
  type: 'Identifier'
  , name: name
  };
  }
 
+ , identifierWithAttribute: function(name, attribute) {
+ return {
+ type: 'IdentifierWithAttribute'
+ , name: name
+ , attribute: attribute
+ };
+ }
+
  , literal: function(type, value, raw) {
  type = (type === StringLiteral) ? 'StringLiteral'
  : (type === NumericLiteral) ? 'NumericLiteral'
@@ -1144,7 +1176,7 @@
 
  while (isHexDigit(input.charCodeAt(index))) {
  ++index;
- if (index - escStart > 6)
+ if (index - escStart > (features.relaxedUTF8 ? 8 : 6))
  raise(null, errors.tooLargeCodepoint, '\\' + input.slice(sequenceStart, index));
  }
 
@@ -1159,7 +1191,7 @@
  var codepoint = parseInt(input.slice(escStart, index - 1) || '0', 16);
  var frag = '\\' + input.slice(sequenceStart, index);
 
- if (codepoint > 0x10ffff) {
+ if (codepoint > (features.relaxedUTF8 ? 0x7fffffff : 0x10ffff)) {
  raise(null, errors.tooLargeCodepoint, frag);
  }
 
@@ -1596,6 +1628,17 @@
  return false;
  };
 
+ FullFlowContext.prototype.findLabel = function (name) {
+ var i = this.scopes.length;
+ while (i --> 0) {
+ if (Object.prototype.hasOwnProperty.call(this.scopes[i].labels, name))
+ return this.scopes[i].labels[name];
+ if (!features.noLabelShadowing)
+ return null;
+ }
+ return null;
+ };
+
  FullFlowContext.prototype.pushScope = function (isLoop) {
  var scope = {
  labels: {},
@@ -1637,9 +1680,10 @@
 
  FullFlowContext.prototype.addLabel = function (name, token) {
  var scope = this.currentScope();
+ var definedLabel = this.findLabel(name);
 
- if (Object.prototype.hasOwnProperty.call(scope.labels, name)) {
- raise(token, errors.labelAlreadyDefined, name, scope.labels[name].line);
+ if (definedLabel !== null) {
+ raise(token, errors.labelAlreadyDefined, name, definedLabel.line);
  } else {
  var newGotos = [];
 
@@ -2062,16 +2106,31 @@
 
  function parseLocalStatement(flowContext) {
  var name
+ , attribute
+ , marker
  , declToken = previousToken;
 
  if (Identifier === token.type) {
  var variables = []
  , init = [];
 
  do {
+ if (trackLocations) marker = createLocationMarker();
+
  name = parseIdentifier();
 
- variables.push(name);
+ attribute = null;
+ if (features.attributes) {
+ attribute = parseAttribute();
+ }
+
+ if (attribute !== null) {
+ if (trackLocations) pushLocation(marker);
+ variables.push(finishNode(ast.identifierWithAttribute(name, attribute)));
+ } else {
+ variables.push(name);
+ }
+
  flowContext.addLocal(name.name, declToken);
  } while (consume(','));
 
@@ -2210,6 +2269,21 @@
  return finishNode(ast.identifier(identifier));
  }
 
+ function parseAttribute() {
+ markLocation();
+ if (consume('<')) {
+ if (Identifier !== token.type) raiseUnexpectedToken('<name>', token);
+ var identifier = token.value;
+ if (!features.attributes[identifier])
+ raise(token, errors.unknownAttribute, identifier);
+ next();
+ expect('>');
+ return finishNode(ast.attribute(identifier));
+ }
+ if (trackLocations) locations.pop();
+ return null;
+ }
+
  // Parse the functions parameters and body block. The name should already
  // have been parsed and passed to this declaration function. By separating
  // this we allow for anonymous functions in expressions.
@@ -2647,6 +2721,20 @@
  integerDivision: true,
  relaxedBreak: true
  },
+ '5.4': {
+ labels: true,
+ emptyStatement: true,
+ hexEscapes: true,
+ skipWhitespaceEscape: true,
+ strictEscapes: true,
+ unicodeEscapes: true,
+ bitwiseOperators: true,
+ integerDivision: true,
+ relaxedBreak: true,
+ noLabelShadowing: true,
+ attributes: { 'const': true, 'close': true },
+ relaxedUTF8: true
+ },
  'LuaJIT': {
  // XXX: LuaJIT language features may depend on compilation options; may need to
  // rethink how to handle this. Specifically, there is a LUAJIT_ENABLE_LUA52COMPAT

test/runner.js

@@ -267,7 +267,12 @@
  visitKey('base');
  visitKey('argument');
  break;
+ case 'IdentifierWithAttribute':
+ visitKey('name');
+ visitKey('attribute');
+ break;
  case 'Identifier':
+ case 'Attribute':
  case 'NumericLiteral':
  case 'BooleanLiteral':
  case 'StringLiteral':
@@ -515,9 +520,19 @@
  , [ '"\\u{1f4a9}"'
  , '"\\u{000001f4a9}"'
  , '"\\xf0\\x9f\\x92\\xa9"'
- , '"\ud83d\udca9"'
- , "'\ud83d\udca9'"
- , '[[\ud83d\udca9]]'
+ , '[[\xf0\x9f\x92\xa9]]'
+ ]
+ , [ '"\\u{db80}"'
+ , '"\\xed\\xae\\x80"'
+ , '[[\xed\xae\x80]]'
+ ]
+ , [ '"\\u{1ffffff}"'
+ , '"\\xf9\\xbf\\xbf\\xbf\\xbf"'
+ , '[[\xf9\xbf\xbf\xbf\xbf]]'
+ ]
+ , [ '"\\u{7fffffff}"'
+ , '"\\xfd\\xbf\\xbf\\xbf\\xbf\\xbf"'
+ , '[[\xfd\xbf\xbf\xbf\xbf\xbf]]'
  ]
  , [ '"\\\\a"'
  , '[==[\\a]==]'
@@ -546,11 +561,13 @@
 
  var list = testcases[i];
  var left = luaparse.parse('return ' + list[0],
- { "luaVersion": "5.3" }).body[0].arguments[0];
+ { "luaVersion": "5.4",
+ "encodingMode": "pseudo-latin1" }).body[0].arguments[0];
 
  for (var j = 1; j < list.length; ++j) {
  var right = luaparse.parse('return ' + list[j],
- { "luaVersion": "5.3" }).body[0].arguments[0];
+ { "luaVersion": "5.4",
+ "encodingMode": "pseudo-latin1" }).body[0].arguments[0];
 
  this.equal(left.value, right.value, symbolicControlChars(left.raw) + ' == ' + symbolicControlChars(right.raw));
  left = right;

test/scaffolding/for

@@ -56,3 +56,5 @@ for a = p, q do do return end end
 // { "luaVersion": "5.2" }
 for a in b do break break end
 for a = p, q do break break end
+// { "luaVersion": "5.4" }
+for a <const> in b do end -- FAIL

test/scaffolding/labels

@@ -23,6 +23,9 @@ do goto x local a ::x:: ::y:: ::z:: end
 do do ::x:: end do goto x end end -- FAIL
 do do goto x end do ::x:: end end -- FAIL
 do do goto x end do end ::x:: end
+::x:: do ::x:: end
+// { "luaVersion": "5.4" }
+::x:: do ::x:: end -- FAIL
 // { "luaVersion": "LuaJIT" }
 ::foo::
 goto "foo"

test/scaffolding/local

@@ -55,3 +55,10 @@ local a = { b = { z = 1 } }
 local Дождь = {} -- FAIL
 // { "extendedIdentifiers": true }
 local Дождь = {}
+// { "luaVersion": "5.4" }
+local x <close> = {}
+local x <const> = {}
+local x <0> = {} -- FAIL
+local x <+> = {} -- FAIL
+local x <xyzzy> = {} -- FAIL
+local x = {}

test/spec/for.js

@@ -3878,6 +3878,17 @@
  "scope": true,
  "luaVersion": "5.2"
  }
+ },
+ {
+ "source": "for a <const> in b do end",
+ "result": "[1:6] 'in' expected near '<'",
+ "options": {
+ "comments": true,
+ "locations": true,
+ "ranges": true,
+ "scope": true,
+ "luaVersion": "5.4"
+ }
  }
  ];
 }));