LibJS: Handle hex and unicode escape sequences in string literals

Introduces the following syntax:

'\x55'
'\u26a0'
'\u{1f41e}'

5 files changed, 118 insertions, 9 deletions

diff --git a/Libraries/LibJS/Parser.cpp b/Libraries/LibJS/Parser.cpp
index bafb72454a..3b1343079f 100644
--- a/Libraries/LibJS/Parser.cpp
+++ b/Libraries/LibJS/Parser.cpp
@@ -405,7 +405,7 @@ NonnullRefPtr<Expression> Parser::parse_primary_expression()
     case TokenType::BoolLiteral:
         return create_ast_node<BooleanLiteral>(consume().bool_value());
     case TokenType::StringLiteral:
-        return create_ast_node<StringLiteral>(consume().string_value());
+        return parse_string_literal(consume());
     case TokenType::NullLiteral:
         consume();
         return create_ast_node<NullLiteral>();
@@ -494,7 +494,7 @@ NonnullRefPtr<ObjectExpression> Parser::parse_object_expression()
             property_value = create_ast_node<Identifier>(identifier);
             need_colon = false;
         } else if (match(TokenType::StringLiteral)) {
-            property_key = create_ast_node<StringLiteral>(consume(TokenType::StringLiteral).string_value());
+            property_key = parse_string_literal(consume());
         } else if (match(TokenType::NumericLiteral)) {
             property_key = create_ast_node<StringLiteral>(consume(TokenType::NumericLiteral).value());
         } else if (match(TokenType::BracketOpen)) {
@@ -559,6 +559,28 @@ NonnullRefPtr<ArrayExpression> Parser::parse_array_expression()
     return create_ast_node<ArrayExpression>(move(elements));
 }
 
+NonnullRefPtr<StringLiteral> Parser::parse_string_literal(Token token)
+{
+    auto status = Token::StringValueStatus::Ok;
+    auto string = token.string_value(status);
+    if (status != Token::StringValueStatus::Ok) {
+        String message;
+        if (status == Token::StringValueStatus::MalformedHexEscape || status == Token::StringValueStatus::MalformedUnicodeEscape) {
+            auto type = status == Token::StringValueStatus::MalformedUnicodeEscape ? "unicode" : "hexadecimal";
+            message = String::format("Malformed %s escape sequence", type);
+        } else if (status == Token::StringValueStatus::UnicodeEscapeOverflow) {
+            message = "Unicode codepoint must not be greater than 0x10ffff in escape sequence";
+        }
+
+        syntax_error(
+            message,
+            m_parser_state.m_current_token.line_number(),
+            m_parser_state.m_current_token.line_column()
+        );
+    }
+    return create_ast_node<StringLiteral>(string);
+}
+
 NonnullRefPtr<TemplateLiteral> Parser::parse_template_literal(bool is_tagged)
 {
     consume(TokenType::TemplateLiteralStart);
@@ -579,7 +601,7 @@ NonnullRefPtr<TemplateLiteral> Parser::parse_template_literal(bool is_tagged)
     while (!match(TokenType::TemplateLiteralEnd) && !match(TokenType::UnterminatedTemplateLiteral)) {
         if (match(TokenType::TemplateLiteralString)) {
             auto token = consume();
-            expressions.append(create_ast_node<StringLiteral>(token.string_value()));
+            expressions.append(parse_string_literal(token));
             if (is_tagged)
                 raw_strings.append(create_ast_node<StringLiteral>(token.value()));
         } else if (match(TokenType::TemplateLiteralExprStart)) {
diff --git a/Libraries/LibJS/Parser.h b/Libraries/LibJS/Parser.h
index 91f90953ea..1f5d2f4fec 100644
--- a/Libraries/LibJS/Parser.h
+++ b/Libraries/LibJS/Parser.h
@@ -70,6 +70,7 @@ public:
     NonnullRefPtr<Expression> parse_unary_prefixed_expression();
     NonnullRefPtr<ObjectExpression> parse_object_expression();
     NonnullRefPtr<ArrayExpression> parse_array_expression();
+    NonnullRefPtr<StringLiteral> parse_string_literal(Token token);
     NonnullRefPtr<TemplateLiteral> parse_template_literal(bool is_tagged);
     NonnullRefPtr<Expression> parse_secondary_expression(NonnullRefPtr<Expression>, int min_precedence, Associativity associate = Associativity::Right);
     NonnullRefPtr<CallExpression> parse_call_expression(NonnullRefPtr<Expression>);
diff --git a/Libraries/LibJS/Tests/string-escapes.js b/Libraries/LibJS/Tests/string-escapes.js
new file mode 100644
index 0000000000..ad9f47ea12
--- /dev/null
+++ b/Libraries/LibJS/Tests/string-escapes.js
@@ -0,0 +1,17 @@
+load("test-common.js")
+
+try {
+    assert("\x55" === "U");
+    assert("\X55" === "X55");
+    assert(`\x55` === "U");
+    assert(`\X55` === "X55");
+
+    assert("\u26a0" === "⚠");
+    assert(`\u26a0` === "⚠");
+    assert("\u{1f41e}" === "🐞");
+    assert(`\u{1f41e}` === "🐞");
+    
+    console.log("PASS");
+} catch (e) {
+    console.log("FAIL: " + e);
+}
diff --git a/Libraries/LibJS/Token.cpp b/Libraries/LibJS/Token.cpp
index ffb95908a5..cc3a1ee0eb 100644
--- a/Libraries/LibJS/Token.cpp
+++ b/Libraries/LibJS/Token.cpp
@@ -27,6 +27,7 @@
 #include "Token.h"
 #include <AK/Assertions.h>
 #include <AK/StringBuilder.h>
+#include <AK/Utf32View.h>
 #include <ctype.h>
 
 namespace JS {
@@ -72,13 +73,26 @@ double Token::double_value() const
     return strtod(value_string.characters(), nullptr);
 }
 
-String Token::string_value() const
+static u32 hex2int(char x)
+{
+    ASSERT(isxdigit(x));
+    if (x >= '0' && x <= '9')
+        return x - '0';
+    return 10u + (tolower(x) - 'a');
+}
+
+String Token::string_value(StringValueStatus& status) const
 {
     ASSERT(type() == TokenType::StringLiteral || type() == TokenType::TemplateLiteralString);
     auto is_template = type() == TokenType::TemplateLiteralString;
 
     auto offset = type() == TokenType::TemplateLiteralString ? 0 : 1;
 
+    auto encoding_failure = [&status](StringValueStatus parse_status) -> String {
+        status = parse_status;
+        return {};
+    };
+
     StringBuilder builder;
     for (size_t i = offset; i < m_value.length() - offset; ++i) {
         if (m_value[i] == '\\' && i + 1 < m_value.length() - offset) {
@@ -114,14 +128,62 @@ String Token::string_value() const
             case '\\':
                 builder.append('\\');
                 break;
+            case 'x': {
+                if (i + 2 >= m_value.length() - offset)
+                    return encoding_failure(StringValueStatus::MalformedHexEscape);
+
+                auto digit1 = m_value[++i];
+                auto digit2 = m_value[++i];
+                if (!isxdigit(digit1) || !isxdigit(digit2))
+                    return encoding_failure(StringValueStatus::MalformedHexEscape);
+                builder.append(static_cast<char>(hex2int(digit1) * 16 + hex2int(digit2)));
+                break;
+            }
+            case 'u': {
+                if (i + 1 >= m_value.length() - offset)
+                    return encoding_failure(StringValueStatus::MalformedUnicodeEscape);
+                u32 code_point = m_value[++i];
+
+                if (code_point == '{') {
+                    code_point = 0;
+                    do {
+                        if (i + 1 >= m_value.length() - offset)
+                            return encoding_failure(StringValueStatus::MalformedUnicodeEscape);
+
+                        auto ch = m_value[++i];
+                        if (!isxdigit(ch))
+                            return encoding_failure(StringValueStatus::MalformedUnicodeEscape);
+
+                        auto new_code_point = (code_point << 4u) | hex2int(ch);
+                        if (new_code_point < code_point)
+                            return encoding_failure(StringValueStatus::UnicodeEscapeOverflow);
+                        code_point = new_code_point;
+                    } while (m_value[i + 1] != '}');
+                    ++i;
+                } else {
+                    if (i + 3 >= m_value.length() - offset || !isxdigit(code_point))
+                        return encoding_failure(StringValueStatus::MalformedUnicodeEscape);
+
+                    code_point = hex2int(code_point);
+                    for (int j = 0; j < 3; ++j) {
+                        auto ch = m_value[++i];
+                        if (!isxdigit(ch))
+                            return encoding_failure(StringValueStatus::MalformedUnicodeEscape);
+                        code_point = (code_point << 4u) | hex2int(ch);
+                    }
+                }
+
+                builder.append({ &code_point, 1 });
+                break;
+            }
             default:
                 if (is_template && (m_value[i] == '$' || m_value[i] == '`')) {
                     builder.append(m_value[i]);
-                } else {
-                    // FIXME: Also parse octal, hex and unicode sequences
-                    // should anything else generate a syntax error?
-                    builder.append(m_value[i]);
+                    break;
                 }
+
+                // FIXME: Also parse octal. Should anything else generate a syntax error?
+                builder.append(m_value[i]);
             }
         } else {
             builder.append(m_value[i]);
diff --git a/Libraries/LibJS/Token.h b/Libraries/LibJS/Token.h
index c66d3c5431..884a249f27 100644
--- a/Libraries/LibJS/Token.h
+++ b/Libraries/LibJS/Token.h
@@ -172,9 +172,16 @@ public:
     size_t line_number() const { return m_line_number; }
     size_t line_column() const { return m_line_column; }
     double double_value() const;
-    String string_value() const;
     bool bool_value() const;
 
+    enum class StringValueStatus {
+        Ok,
+        MalformedHexEscape,
+        MalformedUnicodeEscape,
+        UnicodeEscapeOverflow,
+    };
+    String string_value(StringValueStatus& status) const;
+
     bool is_identifier_name() const;
 
 private: