LibJS: Allow Unicode escape sequences in identifiers

For example, "property.br\u{64}wn" should resolve to "property.brown".

To support this behavior, this commit changes the Token class to hold
both the evaluated identifier name and a view into the original source
for the unevaluated name. There are some contexts in which identifiers
are not allowed to contain Unicode escape sequences; for example, export
statements of the form "export {} from foo.js" forbid escapes in the
identifier "from".

The test file is added to .prettierignore because prettier will replace
all escaped Unicode sequences with their unescaped value.

.prettierignore

@@ -1,3 +1,3 @@
 Base/home/anon/Source/js
 Userland/Libraries/LibJS/Tests/eval-aliasing.js
-
+Userland/Libraries/LibJS/Tests/unicode-identifier-escape.js

Userland/Libraries/LibJS/Lexer.cpp

@@ -8,6 +8,7 @@
 #include "Lexer.h"
 #include <AK/CharacterTypes.h>
 #include <AK/Debug.h>
+#include <AK/GenericLexer.h>
 #include <AK/HashMap.h>
 #include <AK/Utf8View.h>
 #include <LibUnicode/CharacterTypes.h>
@@ -350,6 +351,8 @@ u32 Lexer::current_code_point() const
     if (m_position == 0)
         return REPLACEMENT_CHARACTER;
     Utf8View utf_8_view { m_source.substring_view(m_position - 1) };
+    if (utf_8_view.is_empty())
+        return REPLACEMENT_CHARACTER;
     return *utf_8_view.begin();
 }
 
@@ -369,30 +372,60 @@ bool Lexer::is_whitespace() const
     return false;
 }
 
-bool Lexer::is_identifier_start() const
+Optional<u32> Lexer::is_unicode_escape(size_t& identifier_length) const
 {
-    if (!is_unicode_character())
-        return is_ascii_alpha(m_current_char) || m_current_char == '_' || m_current_char == '$';
-    auto code_point = current_code_point();
+    GenericLexer lexer(source().substring_view(m_position - 1));
 
-    static auto id_start_category = Unicode::property_from_string("ID_Start"sv);
-    if (id_start_category.has_value())
-        return Unicode::code_point_has_property(code_point, *id_start_category);
-    return false;
+    if (auto code_point_or_error = lexer.consume_escaped_code_point(); !code_point_or_error.is_error()) {
+        identifier_length = lexer.tell();
+        return code_point_or_error.value();
+    }
+
+    return {};
 }
 
-bool Lexer::is_identifier_middle() const
+Optional<u32> Lexer::is_identifier_start(size_t& identifier_length) const
 {
-    if (!is_unicode_character())
-        return is_identifier_start() || is_ascii_digit(m_current_char);
-    auto code_point = current_code_point();
-    if (code_point == ZERO_WIDTH_NON_JOINER || code_point == ZERO_WIDTH_JOINER)
-        return true;
+    u32 code_point = current_code_point();
+    identifier_length = 1;
+
+    if (code_point == '\\') {
+        if (auto maybe_code_point = is_unicode_escape(identifier_length); maybe_code_point.has_value())
+            code_point = *maybe_code_point;
+        else
+            return {};
+    }
+
+    if (is_ascii_alpha(code_point) || code_point == '_' || code_point == '$')
+        return code_point;
+
+    static auto id_start_category = Unicode::property_from_string("ID_Start"sv);
+    if (id_start_category.has_value() && Unicode::code_point_has_property(code_point, *id_start_category))
+        return code_point;
+
+    return {};
+}
+
+Optional<u32> Lexer::is_identifier_middle(size_t& identifier_length) const
+{
+    u32 code_point = current_code_point();
+    identifier_length = 1;
+
+    if (code_point == '\\') {
+        if (auto maybe_code_point = is_unicode_escape(identifier_length); maybe_code_point.has_value())
+            code_point = *maybe_code_point;
+        else
+            return {};
+    }
+
+    if (is_ascii_alphanumeric(code_point) || (code_point == '$') || (code_point == ZERO_WIDTH_NON_JOINER) || (code_point == ZERO_WIDTH_JOINER))
+        return code_point;
 
     static auto id_continue_category = Unicode::property_from_string("ID_Continue"sv);
-    if (id_continue_category.has_value())
-        return Unicode::code_point_has_property(code_point, *id_continue_category);
-    return false;
+    if (id_continue_category.has_value() && Unicode::code_point_has_property(code_point, *id_continue_category))
+        return code_point;
+
+    return {};
 }
 
 bool Lexer::is_line_comment_start(bool line_has_token_yet) const
@@ -494,6 +527,9 @@ Token Lexer::next()
     // bunch of Invalid* tokens (bad numeric literals, unterminated comments etc.)
     String token_message;
 
+    Optional<FlyString> identifier;
+    size_t identifier_length = 0;
+
     if (m_current_token.type() == TokenType::RegexLiteral && !is_eof() && is_ascii_alpha(m_current_char) && !did_consume_whitespace_or_comments) {
         token_type = TokenType::RegexFlags;
         while (!is_eof() && is_ascii_alpha(m_current_char))
@@ -537,19 +573,26 @@ Token Lexer::next()
             else
                 token_type = TokenType::TemplateLiteralString;
         }
-    } else if (is_identifier_start()) {
+    } else if (auto code_point = is_identifier_start(identifier_length); code_point.has_value()) {
         // identifier or keyword
+        StringBuilder builder;
         do {
-            consume();
-        } while (is_identifier_middle());
+            builder.append_code_point(*code_point);
+            for (size_t i = 0; i < identifier_length; ++i)
+                consume();
 
-        StringView value = m_source.substring_view(value_start - 1, m_position - value_start);
-        auto it = s_keywords.find(value.hash(), [&](auto& entry) { return entry.key == value; });
-        if (it == s_keywords.end()) {
+            code_point = is_identifier_middle(identifier_length);
+        } while (code_point.has_value());
+
+        identifier = builder.build();
+        if (!m_parsed_identifiers.contains_slow(*identifier))
+            m_parsed_identifiers.append(*identifier);
+
+        auto it = s_keywords.find(identifier->hash(), [&](auto& entry) { return entry.key == identifier; });
+        if (it == s_keywords.end())
             token_type = TokenType::Identifier;
-        } else {
+        else
             token_type = it->value;
-        }
     } else if (is_numeric_literal_start()) {
         token_type = TokenType::NumericLiteral;
         bool is_invalid_numeric_literal = false;
@@ -708,15 +751,28 @@ Token Lexer::next()
         }
     }
 
-    m_current_token = Token(
-        token_type,
-        token_message,
-        m_source.substring_view(trivia_start - 1, value_start - trivia_start),
-        m_source.substring_view(value_start - 1, m_position - value_start),
-        m_filename,
-        value_start_line_number,
-        value_start_column_number,
-        m_position);
+    if (identifier.has_value()) {
+        m_current_token = Token(
+            token_type,
+            token_message,
+            m_source.substring_view(trivia_start - 1, value_start - trivia_start),
+            m_source.substring_view(value_start - 1, m_position - value_start),
+            identifier.release_value(),
+            m_filename,
+            value_start_line_number,
+            value_start_column_number,
+            m_position);
+    } else {
+        m_current_token = Token(
+            token_type,
+            token_message,
+            m_source.substring_view(trivia_start - 1, value_start - trivia_start),
+            m_source.substring_view(value_start - 1, m_position - value_start),
+            m_filename,
+            value_start_line_number,
+            value_start_column_number,
+            m_position);
+    }
 
     if constexpr (LEXER_DEBUG) {
         dbgln("------------------------------");

Userland/Libraries/LibJS/Lexer.h

@@ -41,8 +41,9 @@ private:
     bool is_eof() const;
     bool is_line_terminator() const;
     bool is_whitespace() const;
-    bool is_identifier_start() const;
-    bool is_identifier_middle() const;
+    Optional<u32> is_unicode_escape(size_t& identifier_length) const;
+    Optional<u32> is_identifier_start(size_t& identifier_length) const;
+    Optional<u32> is_identifier_middle(size_t& identifier_length) const;
     bool is_line_comment_start(bool line_has_token_yet) const;
     bool is_block_comment_start() const;
     bool is_block_comment_end() const;
@@ -80,6 +81,10 @@ private:
     static HashMap<String, TokenType> s_three_char_tokens;
     static HashMap<String, TokenType> s_two_char_tokens;
     static HashMap<char, TokenType> s_single_char_tokens;
+
+    // Resolved identifiers must be kept alive for the duration of the parsing stage, otherwise
+    // the only references to these strings are deleted by the Token destructor.
+    Vector<FlyString> m_parsed_identifiers;
 };
 
 }

Userland/Libraries/LibJS/Parser.cpp

@@ -210,7 +210,6 @@ constexpr OperatorPrecedenceTable g_operator_precedence;
 
 Parser::ParserState::ParserState(Lexer l, Program::Type program_type)
     : lexer(move(l))
-    , current_token(TokenType::Invalid, {}, {}, {}, {}, 0, 0, 0)
 {
     if (program_type == Program::Type::Module)
         lexer.disallow_html_comments();
@@ -680,7 +679,7 @@ NonnullRefPtr<ClassExpression> Parser::parse_class_expression(bool expect_class_
 
         if (match_property_key()) {
             StringView name;
-            if (!is_generator && m_state.current_token.value() == "static"sv) {
+            if (!is_generator && m_state.current_token.original_value() == "static"sv) {
                 if (match(TokenType::Identifier)) {
                     consume();
                     is_static = true;
@@ -2524,7 +2523,7 @@ NonnullRefPtr<Statement> Parser::parse_for_statement()
 {
     auto rule_start = push_start();
     auto match_for_in_of = [&]() {
-        return match(TokenType::In) || (match(TokenType::Identifier) && m_state.current_token.value() == "of");
+        return match(TokenType::In) || (match(TokenType::Identifier) && m_state.current_token.original_value() == "of");
     };
 
     consume(TokenType::For);
@@ -3019,7 +3018,7 @@ NonnullRefPtr<ImportStatement> Parser::parse_import_statement(Program& program)
     };
 
     auto match_as = [&] {
-        return match(TokenType::Identifier) && m_state.current_token.value() == "as"sv;
+        return match(TokenType::Identifier) && m_state.current_token.original_value() == "as"sv;
     };
 
     bool continue_parsing = true;
@@ -3134,11 +3133,15 @@ NonnullRefPtr<ExportStatement> Parser::parse_export_statement(Program& program)
         syntax_error("Cannot use export statement outside a module");
 
     auto match_as = [&] {
-        return match(TokenType::Identifier) && m_state.current_token.value() == "as"sv;
+        return match(TokenType::Identifier) && m_state.current_token.original_value() == "as"sv;
     };
 
     auto match_from = [&] {
-        return match(TokenType::Identifier) && m_state.current_token.value() == "from"sv;
+        return match(TokenType::Identifier) && m_state.current_token.original_value() == "from"sv;
+    };
+
+    auto match_default = [&] {
+        return match(TokenType::Default) && m_state.current_token.original_value() == "default"sv;
     };
 
     consume(TokenType::Export);
@@ -3158,7 +3161,7 @@ NonnullRefPtr<ExportStatement> Parser::parse_export_statement(Program& program)
 
     RefPtr<ASTNode> expression = {};
 
-    if (match(TokenType::Default)) {
+    if (match_default()) {
         auto default_position = position();
         consume(TokenType::Default);
 

Userland/Libraries/LibJS/Tests/unicode-identifier-escape.js

@@ -0,0 +1,19 @@
+test("basic escapes", () => {
+    var foo = {};
+    foo.brown = 12389;
+
+    expect(foo.brown).toBe(12389);
+    expect(foo.br\u006fwn).toBe(12389);
+    expect(foo.br\u{6f}wn).toBe(12389);
+    expect(foo.\u{62}\u{72}\u{6f}\u{77}\u{6e}).toBe(12389);
+});
+
+test("non-ascii escapes", () => {
+    var foo = {};
+    foo.𝓑𝓻𝓸𝔀𝓷 = 12389;
+
+    expect(foo.𝓑𝓻𝓸𝔀𝓷).toBe(12389);
+    expect(foo.𝓑𝓻\ud835\udcf8𝔀𝓷).toBe(12389);
+    expect(foo.𝓑𝓻\u{1d4f8}𝔀𝓷).toBe(12389);
+    expect(foo.\u{1d4d1}\u{1d4fb}\u{1d4f8}\u{1d500}\u{1d4f7}).toBe(12389);
+});

Userland/Libraries/LibJS/Token.cpp

@@ -56,7 +56,7 @@ double Token::double_value() const
 
     StringBuilder builder;
 
-    for (auto ch : m_value) {
+    for (auto ch : value()) {
         if (ch == '_')
             continue;
         builder.append(ch);
@@ -75,7 +75,7 @@ double Token::double_value() const
             return static_cast<double>(strtoul(value_string.characters() + 2, nullptr, 2));
         } else if (is_ascii_digit(value_string[1])) {
             // also octal, but syntax error in strict mode
-            if (!m_value.contains('8') && !m_value.contains('9'))
+            if (!value().contains('8') && !value().contains('9'))
                 return static_cast<double>(strtoul(value_string.characters() + 1, nullptr, 8));
         }
     }
@@ -95,7 +95,7 @@ String Token::string_value(StringValueStatus& status) const
     VERIFY(type() == TokenType::StringLiteral || type() == TokenType::TemplateLiteralString);
 
     auto is_template = type() == TokenType::TemplateLiteralString;
-    GenericLexer lexer(is_template ? m_value : m_value.substring_view(1, m_value.length() - 2));
+    GenericLexer lexer(is_template ? value() : value().substring_view(1, value().length() - 2));
 
     auto encoding_failure = [&status](StringValueStatus parse_status) -> String {
         status = parse_status;
@@ -195,7 +195,7 @@ String Token::string_value(StringValueStatus& status) const
 bool Token::bool_value() const
 {
     VERIFY(type() == TokenType::BoolLiteral);
-    return m_value == "true";
+    return value() == "true";
 }
 
 bool Token::is_identifier_name() const

Userland/Libraries/LibJS/Token.h

@@ -6,8 +6,10 @@
 
 #pragma once
 
+#include <AK/FlyString.h>
 #include <AK/String.h>
 #include <AK/StringView.h>
+#include <AK/Variant.h>
 
 namespace JS {
 
@@ -172,10 +174,13 @@ enum class TokenCategory {
 
 class Token {
 public:
+    Token() = default;
+
     Token(TokenType type, String message, StringView trivia, StringView value, StringView filename, size_t line_number, size_t line_column, size_t offset)
         : m_type(type)
         , m_message(message)
         , m_trivia(trivia)
+        , m_original_value(value)
         , m_value(value)
         , m_filename(filename)
         , m_line_number(line_number)
@@ -184,6 +189,19 @@ public:
     {
     }
 
+    Token(TokenType type, String message, StringView trivia, StringView original_value, FlyString value, StringView filename, size_t line_number, size_t line_column, size_t offset)
+        : m_type(type)
+        , m_message(message)
+        , m_trivia(trivia)
+        , m_original_value(original_value)
+        , m_value(move(value))
+        , m_filename(filename)
+        , m_line_number(line_number)
+        , m_line_column(line_column)
+        , m_offset(offset)
+    {
+    }
+
     TokenType type() const { return m_type; }
     TokenCategory category() const;
     static TokenCategory category(TokenType);
@@ -192,7 +210,14 @@ public:
 
     const String& message() const { return m_message; }
     const StringView& trivia() const { return m_trivia; }
-    const StringView& value() const { return m_value; }
+    const StringView& original_value() const { return m_original_value; }
+    StringView value() const
+    {
+        return m_value.visit(
+            [](StringView const& view) { return view; },
+            [](FlyString const& identifier) { return identifier.view(); },
+            [](Empty) -> StringView { VERIFY_NOT_REACHED(); });
+    }
     const StringView& filename() const { return m_filename; }
     size_t line_number() const { return m_line_number; }
     size_t line_column() const { return m_line_column; }
@@ -213,14 +238,15 @@ public:
     bool trivia_contains_line_terminator() const;
 
 private:
-    TokenType m_type;
+    TokenType m_type { TokenType::Invalid };
     String m_message;
     StringView m_trivia;
-    StringView m_value;
+    StringView m_original_value;
+    Variant<Empty, StringView, FlyString> m_value { Empty {} };
     StringView m_filename;
-    size_t m_line_number;
-    size_t m_line_column;
-    size_t m_offset;
+    size_t m_line_number { 0 };
+    size_t m_line_column { 0 };
+    size_t m_offset { 0 };
 };
 
 }