Implement string literal module export names

Implements https://github.com/tc39/ecma262/pull/2154, which allows
module export names to be string literals.

Semantics highlights:
  - It is a SyntaxError for string literal export names to have unpaired
    UTF16 surrogates.
  - It is a SyntaxError for string literal export names to be used as
    the local name without being followed by a 'from' clause. For example,
    `export { "foo" }` and `export { "foo" as "bar" }` are errors, but
    `export { "foo" } from "./module.js"` is allowed.

The remaining failing test262 test is wrong:
https://github.com/tc39/test262/issues/2866

Bug: v8:10964
Change-Id: Ib3e06e1ee6b3f1b60ed7f24e21902e17ddfc0351
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2482335
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Marja Hölttä <marja@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70692}

src/common/message-template.h

@@ -395,6 +395,8 @@ namespace internal {
     "Async functions can only be declared at the top level or inside a "       \
     "block.")                                                                  \
   T(IllegalBreak, "Illegal break statement")                                   \
+  T(ModuleExportNameWithoutFromClause,                                         \
+    "String literal module export names must be followed by a 'from' clause")  \
   T(NoIterationStatement,                                                      \
     "Illegal continue statement: no surrounding iteration statement")          \
   T(IllegalContinue,                                                           \
@@ -419,6 +421,8 @@ namespace internal {
     "Invalid left-hand side expression in postfix operation")                  \
   T(InvalidLhsInPrefixOp,                                                      \
     "Invalid left-hand side expression in prefix operation")                   \
+  T(InvalidModuleExportName,                                                   \
+    "Invalid module export name: contains unpaired surrogate")                 \
   T(InvalidRegExpFlags, "Invalid flags supplied to RegExp constructor '%'")    \
   T(InvalidOrUnexpectedToken, "Invalid or unexpected token")                   \
   T(InvalidPrivateBrand, "Object must be an instance of class %")              \

src/parsing/parser.cc

@@ -27,6 +27,7 @@
 #include "src/runtime/runtime.h"
 #include "src/strings/char-predicates-inl.h"
 #include "src/strings/string-stream.h"
+#include "src/strings/unicode-inl.h"
 #include "src/tracing/trace-event.h"
 #include "src/zone/zone-list-inl.h"
 
@@ -1071,7 +1072,8 @@ const AstRawString* Parser::ParseModuleSpecifier() {
 }
 
 ZoneChunkList<Parser::ExportClauseData>* Parser::ParseExportClause(
-    Scanner::Location* reserved_loc) {
+    Scanner::Location* reserved_loc,
+    Scanner::Location* string_literal_local_name_loc) {
   // ExportClause :
   //   '{' '}'
   //   '{' ExportsList '}'
@@ -1084,6 +1086,12 @@ ZoneChunkList<Parser::ExportClauseData>* Parser::ParseExportClause(
   // ExportSpecifier :
   //   IdentifierName
   //   IdentifierName 'as' IdentifierName
+  //   IdentifierName 'as' ModuleExportName
+  //   ModuleExportName
+  //   ModuleExportName 'as' ModuleExportName
+  //
+  // ModuleExportName :
+  //   StringLiteral
   ZoneChunkList<ExportClauseData>* export_data =
       zone()->New<ZoneChunkList<ExportClauseData>>(zone());
 
@@ -1091,23 +1099,27 @@ ZoneChunkList<Parser::ExportClauseData>* Parser::ParseExportClause(
 
   Token::Value name_tok;
   while ((name_tok = peek()) != Token::RBRACE) {
-    // Keep track of the first reserved word encountered in case our
-    // caller needs to report an error.
-    if (!reserved_loc->IsValid() &&
+    const AstRawString* local_name = ParseExportSpecifierName();
+    if (!string_literal_local_name_loc->IsValid() &&
+        name_tok == Token::STRING) {
+      // Keep track of the first string literal local name exported for error
+      // reporting. These must be followed by a 'from' clause.
+      *string_literal_local_name_loc = scanner()->location();
+    } else if (!reserved_loc->IsValid() &&
                !Token::IsValidIdentifier(name_tok, LanguageMode::kStrict, false,
                                          flags().is_module())) {
+      // Keep track of the first reserved word encountered in case our
+      // caller needs to report an error.
       *reserved_loc = scanner()->location();
     }
-    const AstRawString* local_name = ParsePropertyName();
-    const AstRawString* export_name = nullptr;
+    const AstRawString* export_name;
     Scanner::Location location = scanner()->location();
     if (CheckContextualKeyword(ast_value_factory()->as_string())) {
-      export_name = ParsePropertyName();
+      export_name = ParseExportSpecifierName();
       // Set the location to the whole "a as b" string, so that it makes sense
       // both for errors due to "a" and for errors due to "b".
       location.end_pos = scanner()->location().end_pos;
-    }
-    if (export_name == nullptr) {
+    } else {
       export_name = local_name;
     }
     export_data->push_back({export_name, local_name, location});
@@ -1122,6 +1134,31 @@ ZoneChunkList<Parser::ExportClauseData>* Parser::ParseExportClause(
   return export_data;
 }
 
+const AstRawString* Parser::ParseExportSpecifierName() {
+  Token::Value next = Next();
+
+  // IdentifierName
+  if (V8_LIKELY(Token::IsPropertyName(next))) {
+    return GetSymbol();
+  }
+
+  // ModuleExportName
+  if (next == Token::STRING) {
+    const AstRawString* export_name = GetSymbol();
+    if (V8_LIKELY(export_name->is_one_byte())) return export_name;
+    if (!unibrow::Utf16::HasUnpairedSurrogate(
+            reinterpret_cast<const uint16_t*>(export_name->raw_data()),
+            export_name->length())) {
+      return export_name;
+    }
+    ReportMessage(MessageTemplate::kInvalidModuleExportName);
+    return EmptyIdentifierString();
+  }
+
+  ReportUnexpectedToken(next);
+  return EmptyIdentifierString();
+}
+
 ZonePtrList<const Parser::NamedImport>* Parser::ParseNamedImports(int pos) {
   // NamedImports :
   //   '{' '}'
@@ -1135,12 +1172,13 @@ ZonePtrList<const Parser::NamedImport>* Parser::ParseNamedImports(int pos) {
   // ImportSpecifier :
   //   BindingIdentifier
   //   IdentifierName 'as' BindingIdentifier
+  //   ModuleExportName 'as' BindingIdentifier
 
   Expect(Token::LBRACE);
 
   auto result = zone()->New<ZonePtrList<const NamedImport>>(1, zone());
   while (peek() != Token::RBRACE) {
-    const AstRawString* import_name = ParsePropertyName();
+    const AstRawString* import_name = ParseExportSpecifierName();
     const AstRawString* local_name = import_name;
     Scanner::Location location = scanner()->location();
     // In the presence of 'as', the left-side of the 'as' can
@@ -1450,9 +1488,14 @@ void Parser::ParseExportStar() {
   //   export * as x from "...";
   // ~>
   //   import * as .x from "..."; export {.x as x};
+  //
+  // Note that the desugared internal namespace export name (.x above) will
+  // never conflict with a string literal export name, as literal string export
+  // names in local name positions (i.e. left of 'as' or in a clause without
+  // 'as') are disallowed without a following 'from' clause.
 
   ExpectContextualKeyword(ast_value_factory()->as_string());
-  const AstRawString* export_name = ParsePropertyName();
+  const AstRawString* export_name = ParseExportSpecifierName();
   Scanner::Location export_name_loc = scanner()->location();
   const AstRawString* local_name = NextInternalNamespaceExportName();
   Scanner::Location local_name_loc = Scanner::Location::invalid();
@@ -1478,12 +1521,18 @@ Statement* Parser::ParseExportDeclaration() {
   //    'export' '*' 'as' IdentifierName 'from' ModuleSpecifier ';'
   //    'export' '*' 'as' IdentifierName 'from' ModuleSpecifier
   //        [no LineTerminator here] AssertClause ';'
+  //    'export' '*' 'as' ModuleExportName 'from' ModuleSpecifier ';'
+  //    'export' '*' 'as' ModuleExportName 'from' ModuleSpecifier ';'
+  //        [no LineTerminator here] AssertClause ';'
   //    'export' ExportClause ('from' ModuleSpecifier)? ';'
   //    'export' ExportClause ('from' ModuleSpecifier [no LineTerminator here]
   //        AssertClause)? ';'
   //    'export' VariableStatement
   //    'export' Declaration
   //    'export' 'default' ... (handled in ParseExportDefault)
+  //
+  // ModuleExportName :
+  //   StringLiteral
 
   Expect(Token::EXPORT);
   Statement* result = nullptr;
@@ -1510,8 +1559,10 @@ Statement* Parser::ParseExportDeclaration() {
       // encountered, and then throw a SyntaxError if we are in the
       // non-FromClause case.
       Scanner::Location reserved_loc = Scanner::Location::invalid();
+      Scanner::Location string_literal_local_name_loc =
+          Scanner::Location::invalid();
       ZoneChunkList<ExportClauseData>* export_data =
-          ParseExportClause(&reserved_loc);
+          ParseExportClause(&reserved_loc, &string_literal_local_name_loc);
       if (CheckContextualKeyword(ast_value_factory()->from_string())) {
         Scanner::Location specifier_loc = scanner()->peek_location();
         const AstRawString* module_specifier = ParseModuleSpecifier();
@@ -1533,6 +1584,10 @@ Statement* Parser::ParseExportDeclaration() {
           // No FromClause, so reserved words are invalid in ExportClause.
           ReportMessageAt(reserved_loc, MessageTemplate::kUnexpectedReserved);
           return nullptr;
+        } else if (string_literal_local_name_loc.IsValid()) {
+          ReportMessageAt(string_literal_local_name_loc,
+                          MessageTemplate::kModuleExportNameWithoutFromClause);
+          return nullptr;
         }
 
         ExpectSemicolon();

src/parsing/parser.h

@@ -269,7 +269,8 @@ class V8_EXPORT_PRIVATE Parser : public NON_EXPORTED_BASE(ParserBase<Parser>) {
     Scanner::Location location;
   };
   ZoneChunkList<ExportClauseData>* ParseExportClause(
-      Scanner::Location* reserved_loc);
+      Scanner::Location* reserved_loc,
+      Scanner::Location* string_literal_local_name_loc);
   struct NamedImport : public ZoneObject {
     const AstRawString* import_name;
     const AstRawString* local_name;
@@ -280,6 +281,7 @@ class V8_EXPORT_PRIVATE Parser : public NON_EXPORTED_BASE(ParserBase<Parser>) {
           local_name(local_name),
           location(location) {}
   };
+  const AstRawString* ParseExportSpecifierName();
   ZonePtrList<const NamedImport>* ParseNamedImports(int pos);
   using ImportAssertions =
       ZoneMap<const AstRawString*,

src/strings/unicode-inl.h

@@ -59,6 +59,25 @@ int Mapping<T, s>::CalculateValue(uchar c, uchar n, uchar* result) {
 }
 #endif  // !V8_INTL_SUPPORT
 
+bool Utf16::HasUnpairedSurrogate(const uint16_t* code_units, size_t length) {
+  for (size_t i = 0; i < length; ++i) {
+    const int code_unit = code_units[i];
+    if (IsLeadSurrogate(code_unit)) {
+      // The current code unit is a leading surrogate. Check if it is followed
+      // by a trailing surrogate.
+      if (i == length - 1) return true;
+      if (!IsTrailSurrogate(code_units[i + 1])) return true;
+      // Skip the paired trailing surrogate.
+      ++i;
+    } else if (IsTrailSurrogate(code_unit)) {
+      // All paired trailing surrogates are skipped above, so this branch is
+      // only for those that are unpaired.
+      return true;
+    }
+  }
+  return false;
+}
+
 // Decodes UTF-8 bytes incrementally, allowing the decoding of bytes as they
 // stream in. This **must** be followed by a call to ValueOfIncrementalFinish
 // when the stream is complete, to ensure incomplete sequences are handled.

src/strings/unicode.h

@@ -128,6 +128,8 @@ class Utf16 {
   static inline uint16_t TrailSurrogate(uint32_t char_code) {
     return 0xdc00 + (char_code & 0x3ff);
   }
+  static inline bool HasUnpairedSurrogate(const uint16_t* code_units,
+                                          size_t length);
 };
 
 class Latin1 {

test/cctest/interpreter/bytecode_expectations/PrivateAccessorAccess.golden

@@ -84,7 +84,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   48 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   53 S> */ B(Wide), B(LdaSmi), I16(270),
+  /*   53 S> */ B(Wide), B(LdaSmi), I16(272),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),
@@ -115,7 +115,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   41 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   46 S> */ B(Wide), B(LdaSmi), I16(269),
+  /*   46 S> */ B(Wide), B(LdaSmi), I16(271),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),
@@ -146,7 +146,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   48 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   53 S> */ B(Wide), B(LdaSmi), I16(270),
+  /*   53 S> */ B(Wide), B(LdaSmi), I16(272),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),
@@ -177,7 +177,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   41 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   46 S> */ B(Wide), B(LdaSmi), I16(269),
+  /*   46 S> */ B(Wide), B(LdaSmi), I16(271),
                 B(Star), R(4),
                 B(LdaConstant), U8(0),
                 B(Star), R(5),

test/cctest/interpreter/bytecode_expectations/PrivateMethodAccess.golden

@@ -57,7 +57,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   44 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   49 S> */ B(Wide), B(LdaSmi), I16(268),
+  /*   49 S> */ B(Wide), B(LdaSmi), I16(270),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),
@@ -89,7 +89,7 @@ bytecodes: [
                 B(Mov), R(this), R(0),
                 B(Mov), R(context), R(2),
   /*   44 E> */ B(CallRuntime), U16(Runtime::kAddPrivateBrand), R(0), U8(3),
-  /*   49 S> */ B(Wide), B(LdaSmi), I16(268),
+  /*   49 S> */ B(Wide), B(LdaSmi), I16(270),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),

test/cctest/interpreter/bytecode_expectations/StaticPrivateMethodAccess.golden

@@ -25,7 +25,7 @@ bytecodes: [
                 B(TestReferenceEqual), R(this),
                 B(Mov), R(this), R(1),
                 B(JumpIfTrue), U8(18),
-                B(Wide), B(LdaSmi), I16(266),
+                B(Wide), B(LdaSmi), I16(268),
                 B(Star), R(2),
                 B(LdaConstant), U8(0),
                 B(Star), R(3),
@@ -56,7 +56,7 @@ frame size: 2
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   56 S> */ B(Wide), B(LdaSmi), I16(268),
+  /*   56 S> */ B(Wide), B(LdaSmi), I16(270),
                 B(Star), R(0),
                 B(LdaConstant), U8(0),
                 B(Star), R(1),
@@ -83,7 +83,7 @@ frame size: 2
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   56 S> */ B(Wide), B(LdaSmi), I16(268),
+  /*   56 S> */ B(Wide), B(LdaSmi), I16(270),
                 B(Star), R(0),
                 B(LdaConstant), U8(0),
                 B(Star), R(1),
@@ -122,7 +122,7 @@ bytecodes: [
   /*   94 E> */ B(TestReferenceEqual), R(this),
                 B(Mov), R(this), R(0),
                 B(JumpIfTrue), U8(18),
-                B(Wide), B(LdaSmi), I16(266),
+                B(Wide), B(LdaSmi), I16(268),
                 B(Star), R(2),
                 B(LdaConstant), U8(0),
                 B(Star), R(3),
@@ -144,7 +144,7 @@ bytecodes: [
   /*  109 E> */ B(TestReferenceEqual), R(this),
                 B(Mov), R(this), R(1),
                 B(JumpIfTrue), U8(18),
-                B(Wide), B(LdaSmi), I16(267),
+                B(Wide), B(LdaSmi), I16(269),
                 B(Star), R(3),
                 B(LdaConstant), U8(0),
                 B(Star), R(4),
@@ -159,7 +159,7 @@ bytecodes: [
   /*  133 E> */ B(TestReferenceEqual), R(this),
                 B(Mov), R(this), R(0),
                 B(JumpIfTrue), U8(18),
-                B(Wide), B(LdaSmi), I16(266),
+                B(Wide), B(LdaSmi), I16(268),
                 B(Star), R(2),
                 B(LdaConstant), U8(0),
                 B(Star), R(3),
@@ -189,7 +189,7 @@ frame size: 2
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   60 S> */ B(Wide), B(LdaSmi), I16(270),
+  /*   60 S> */ B(Wide), B(LdaSmi), I16(272),
                 B(Star), R(0),
                 B(LdaConstant), U8(0),
                 B(Star), R(1),
@@ -215,7 +215,7 @@ frame size: 2
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   53 S> */ B(Wide), B(LdaSmi), I16(269),
+  /*   53 S> */ B(Wide), B(LdaSmi), I16(271),
                 B(Star), R(0),
                 B(LdaConstant), U8(0),
                 B(Star), R(1),
@@ -241,7 +241,7 @@ frame size: 2
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   60 S> */ B(Wide), B(LdaSmi), I16(270),
+  /*   60 S> */ B(Wide), B(LdaSmi), I16(272),
                 B(Star), R(0),
                 B(LdaConstant), U8(0),
                 B(Star), R(1),
@@ -267,7 +267,7 @@ frame size: 3
 parameter count: 1
 bytecode array length: 16
 bytecodes: [
-  /*   46 S> */ B(Wide), B(LdaSmi), I16(269),
+  /*   46 S> */ B(Wide), B(LdaSmi), I16(271),
                 B(Star), R(1),
                 B(LdaConstant), U8(0),
                 B(Star), R(2),

test/test262/test262.status

@@ -649,16 +649,6 @@
   'built-ins/TypedArray/prototype/item/returns-undefined-for-holes-in-sparse-arrays': [FAIL],
   'built-ins/TypedArray/prototype/item/returns-undefined-for-out-of-range-index': [FAIL],
 
-  # http://crbug/v8/10964
-  'language/module-code/export-expname-binding-string': [FAIL],
-  'language/module-code/export-expname-from-binding-string': [FAIL],
-  'language/module-code/export-expname-from-star': [FAIL],
-  'language/module-code/export-expname-from-star-string': [FAIL],
-  'language/module-code/export-expname-from-string': [FAIL],
-  'language/module-code/export-expname-from-string-binding': [FAIL],
-  'language/module-code/export-expname-from-string-string': [FAIL],
-  'language/module-code/export-expname-import-string-binding': [FAIL],
-
   # http://crbug/v8/11039
   'intl402/Locale/reject-duplicate-variants-in-tlang': [FAIL],