Add multicharacter tokens to simple lexer

src/ObjLoading/Parsing/LocalizeFile/LocalizeFileReader.cpp

@@ -29,7 +29,12 @@ void LocalizeFileReader::SetupStreamProxies()
 
 std::vector<LocalizeFileEntry> LocalizeFileReader::ReadLocalizeFile()
 {
-    const auto lexer = std::make_unique<SimpleLexer>(m_stream, SimpleLexer::Config{true, true, false});
+    SimpleLexer::Config lexerConfig;
+    lexerConfig.m_emit_new_line_tokens = true;
+    lexerConfig.m_read_strings = true;
+    lexerConfig.m_read_numbers = false;
+    const auto lexer = std::make_unique<SimpleLexer>(m_stream, std::move(lexerConfig));
+
     const auto parser = std::make_unique<LocalizeFileParser>(lexer.get(), m_language);
 
     if (parser->Parse())

src/ObjLoading/Parsing/Menu/MenuFileReader.cpp

@@ -111,7 +111,12 @@ std::unique_ptr<ParsingResult> MenuFileReader::CreateParsingResult(MenuFileParse
 
 std::unique_ptr<ParsingResult> MenuFileReader::ReadMenuFile()
 {
-    const auto lexer = std::make_unique<SimpleLexer>(m_stream, SimpleLexer::Config{false, true, true});
+    SimpleLexer::Config lexerConfig;
+    lexerConfig.m_emit_new_line_tokens = false;
+    lexerConfig.m_read_strings = true;
+    lexerConfig.m_read_numbers = true;
+    const auto lexer = std::make_unique<SimpleLexer>(m_stream, std::move(lexerConfig));
+
     const auto parser = std::make_unique<MenuFileParser>(lexer.get(), m_feature_level);
 
     if (!parser->Parse())

src/ObjWriting/Menu/AbstractMenuDumper.cpp

@@ -66,7 +66,12 @@ std::vector<std::string> AbstractMenuDumper::CreateScriptTokenList(const char* s
     const std::string scriptString(script);
     std::istringstream stringStream(scriptString);
     ParserSingleInputStream inputStream(stringStream, "MenuScript");
-    SimpleLexer lexer(&inputStream, SimpleLexer::Config{false, true, false});
+
+    SimpleLexer::Config lexerConfig;
+    lexerConfig.m_emit_new_line_tokens = false;
+    lexerConfig.m_read_strings = true;
+    lexerConfig.m_read_numbers = false;
+    SimpleLexer lexer(&inputStream, std::move(lexerConfig));
 
     std::vector<std::string> result;
     auto hasLexerTokens = true;

src/Parser/Parsing/Simple/Matcher/SimpleMatcherFactory.cpp

@@ -5,6 +5,7 @@
 #include "SimpleMatcherKeyword.h"
 #include "SimpleMatcherKeywordIgnoreCase.h"
 #include "SimpleMatcherKeywordPrefix.h"
+#include "SimpleMatcherMultiCharacter.h"
 #include "SimpleMatcherValueType.h"
 
 SimpleMatcherFactory::SimpleMatcherFactory(const IMatcherForLabelSupplier<SimpleParserValue>* labelSupplier)
@@ -57,6 +58,11 @@ MatcherFactoryWrapper<SimpleParserValue> SimpleMatcherFactory::Char(char c) cons
     return MatcherFactoryWrapper<SimpleParserValue>(std::make_unique<SimpleMatcherCharacter>(c));
 }
 
+MatcherFactoryWrapper<SimpleParserValue> SimpleMatcherFactory::MultiChar(int multiCharacterSequenceId) const
+{
+    return MatcherFactoryWrapper<SimpleParserValue>(std::make_unique<SimpleMatcherMultiCharacter>(multiCharacterSequenceId));
+}
+
 MatcherFactoryWrapper<SimpleParserValue> SimpleMatcherFactory::AnyCharBesides(std::vector<char> chars) const
 {
     return MatcherFactoryWrapper<SimpleParserValue>(std::make_unique<SimpleMatcherAnyCharacterBesides>(std::move(chars)));

src/Parser/Parsing/Simple/Matcher/SimpleMatcherFactory.h

@@ -19,5 +19,6 @@ public:
     _NODISCARD MatcherFactoryWrapper<SimpleParserValue> Integer() const;
     _NODISCARD MatcherFactoryWrapper<SimpleParserValue> FloatingPoint() const;
     _NODISCARD MatcherFactoryWrapper<SimpleParserValue> Char(char c) const;
+    _NODISCARD MatcherFactoryWrapper<SimpleParserValue> MultiChar(int multiCharacterSequenceId) const;
     _NODISCARD MatcherFactoryWrapper<SimpleParserValue> AnyCharBesides(std::vector<char> chars) const;
 };

src/Parser/Parsing/Simple/Matcher/SimpleMatcherMultiCharacter.cpp

@@ -0,0 +1,14 @@
+#include "SimpleMatcherMultiCharacter.h"
+
+SimpleMatcherMultiCharacter::SimpleMatcherMultiCharacter(const int multiCharacterSequenceId)
+    : m_multi_character_sequence_id(multiCharacterSequenceId)
+{
+}
+
+MatcherResult<SimpleParserValue> SimpleMatcherMultiCharacter::CanMatch(ILexer<SimpleParserValue>* lexer, const unsigned tokenOffset)
+{
+    const auto& token = lexer->GetToken(tokenOffset);
+    return token.m_type == SimpleParserValueType::MULTI_CHARACTER && token.MultiCharacterValue() == m_multi_character_sequence_id
+               ? MatcherResult<SimpleParserValue>::Match(1)
+               : MatcherResult<SimpleParserValue>::NoMatch();
+}

src/Parser/Parsing/Simple/Matcher/SimpleMatcherMultiCharacter.h

@@ -0,0 +1,15 @@
+#pragma once
+
+#include "Parsing/Simple/SimpleParserValue.h"
+#include "Parsing/Matcher/AbstractMatcher.h"
+
+class SimpleMatcherMultiCharacter final : public AbstractMatcher<SimpleParserValue>
+{
+    int m_multi_character_sequence_id;
+
+protected:
+    MatcherResult<SimpleParserValue> CanMatch(ILexer<SimpleParserValue>* lexer, unsigned tokenOffset) override;
+
+public:
+    explicit SimpleMatcherMultiCharacter(int multiCharacterSequenceId);
+};

src/Parser/Parsing/Simple/SimpleLexer.cpp

@@ -1,17 +1,77 @@
 #include "SimpleLexer.h"
 
+SimpleLexer::Config::MultiCharacterToken::MultiCharacterToken(const int id, std::string value)
+    : m_id(id),
+      m_value(std::move(value))
+{
+}
+
+SimpleLexer::MultiCharacterTokenLookupEntry::MultiCharacterTokenLookupEntry(const int id, std::string value)
+    : m_id(id),
+      m_value(std::move(value))
+{
+}
+
 SimpleLexer::SimpleLexer(IParserLineStream* stream)
     : AbstractLexer(stream),
-    m_config{false, true, true},
+      m_config{false, true, true, {}},
-    m_last_line(1)
+      m_check_for_multi_character_tokens(false),
+      m_last_line(1)
 {
 }
 
 SimpleLexer::SimpleLexer(IParserLineStream* stream, Config config)
     : AbstractLexer(stream),
-    m_config(config),
+      m_config(std::move(config)),
-    m_last_line(1)
+      m_check_for_multi_character_tokens(false),
+      m_last_line(1)
 {
+    for (auto tokenConfig : m_config.m_multi_character_tokens)
+        AddMultiCharacterTokenConfigToLookup(std::move(tokenConfig));
+    m_config.m_multi_character_tokens.clear();
+}
+
+void SimpleLexer::AddMultiCharacterTokenConfigToLookup(Config::MultiCharacterToken tokenConfig)
+{
+    if (tokenConfig.m_value.empty())
+        return;
+
+    m_check_for_multi_character_tokens = true;
+    const auto firstCharacterValue = static_cast<uint8_t>(tokenConfig.m_value[0]);
+
+    if (m_multi_character_token_lookup[firstCharacterValue])
+    {
+        auto* currentEntry = m_multi_character_token_lookup[firstCharacterValue].get();
+        while (currentEntry->m_next)
+            currentEntry = currentEntry->m_next.get();
+
+        currentEntry->m_next = std::make_unique<MultiCharacterTokenLookupEntry>(tokenConfig.m_id, std::move(tokenConfig.m_value));
+    }
+    else
+    {
+        m_multi_character_token_lookup[firstCharacterValue] = std::make_unique<MultiCharacterTokenLookupEntry>(tokenConfig.m_id, std::move(tokenConfig.m_value));
+    }
+}
+
+bool SimpleLexer::ReadMultiCharacterToken(const MultiCharacterTokenLookupEntry* multiTokenLookup)
+{
+    const auto& currentLine = CurrentLine();
+    assert(m_current_line_offset >= 1);
+    assert(multiTokenLookup);
+    assert(!multiTokenLookup->m_value.empty());
+    assert(currentLine.m_line[m_current_line_offset - 1] == multiTokenLookup->m_value[0]);
+
+    const char* linePos = &currentLine.m_line[m_current_line_offset - 1];
+
+    for (const auto c : multiTokenLookup->m_value)
+    {
+        if (!*linePos || *linePos != c)
+            return false;
+        linePos++;
+    }
+
+    m_current_line_offset = m_current_line_offset - 1 + multiTokenLookup->m_value.size();
+    return true;
 }
 
 SimpleParserValue SimpleLexer::GetNextToken()
@@ -32,7 +92,7 @@ SimpleParserValue SimpleLexer::GetNextToken()
         NextChar();
         c = PeekChar();
     }
-    
+
     const auto pos = GetNextCharacterPos();
     if (m_config.m_emit_new_line_tokens && pos.m_line > m_last_line)
     {
@@ -45,9 +105,21 @@ SimpleParserValue SimpleLexer::GetNextToken()
     if (c == EOF)
         return SimpleParserValue::EndOfFile(TokenPos());
 
+    if(m_check_for_multi_character_tokens)
+    {
+        const auto* multiTokenLookup = m_multi_character_token_lookup[static_cast<uint8_t>(c)].get();
+        while(multiTokenLookup)
+        {
+            if(ReadMultiCharacterToken(multiTokenLookup))
+                return SimpleParserValue::MultiCharacter(pos, multiTokenLookup->m_id);
+
+            multiTokenLookup = multiTokenLookup->m_next.get();
+        }
+    }
+
     if (m_config.m_read_strings && c == '\"')
         return SimpleParserValue::String(GetPreviousCharacterPos(), new std::string(ReadString()));
-    
+
     if (m_config.m_read_numbers && (isdigit(c) || c == '.' && isdigit(PeekChar())))
     {
         bool isFloatingPointValue;

src/Parser/Parsing/Simple/SimpleLexer.h

@@ -1,5 +1,9 @@
 #pragma once
 
+#include <cstdint>
+#include <limits>
+#include <memory>
+
 #include "SimpleParserValue.h"
 #include "Parsing/Impl/AbstractLexer.h"
 
@@ -9,15 +13,41 @@ public:
     class Config
     {
     public:
+        class MultiCharacterToken
+        {
+        public:
+            int m_id;
+            std::string m_value;
+
+            MultiCharacterToken(int id, std::string value);
+        };
+
         bool m_emit_new_line_tokens;
         bool m_read_strings;
         bool m_read_numbers;
+        std::vector<MultiCharacterToken> m_multi_character_tokens;
+    };
+
+protected:
+    class MultiCharacterTokenLookupEntry
+    {
+    public:
+        int m_id;
+        std::string m_value;
+        std::unique_ptr<MultiCharacterTokenLookupEntry> m_next;
+
+        MultiCharacterTokenLookupEntry(int id, std::string value);
     };
 
     Config m_config;
+    bool m_check_for_multi_character_tokens;
     int m_last_line;
 
-protected:
+    std::unique_ptr<MultiCharacterTokenLookupEntry> m_multi_character_token_lookup[std::numeric_limits<uint8_t>::max() + 1];
+
+    void AddMultiCharacterTokenConfigToLookup(Config::MultiCharacterToken tokenConfig);
+    bool ReadMultiCharacterToken(const MultiCharacterTokenLookupEntry* multiTokenLookup);
+
     SimpleParserValue GetNextToken() override;
 
 public:

src/Parser/Parsing/Simple/SimpleParserValue.cpp

@@ -27,6 +27,13 @@ SimpleParserValue SimpleParserValue::Character(const TokenPos pos, const char c)
     return pv;
 }
 
+SimpleParserValue SimpleParserValue::MultiCharacter(const TokenPos pos, const int multiCharacterSequenceId)
+{
+    SimpleParserValue pv(pos, SimpleParserValueType::MULTI_CHARACTER);
+    pv.m_value.multi_character_sequence_id = multiCharacterSequenceId;
+    return pv;
+}
+
 SimpleParserValue SimpleParserValue::Integer(const TokenPos pos, const int value)
 {
     SimpleParserValue pv(pos, SimpleParserValueType::INTEGER);
@@ -116,6 +123,12 @@ char SimpleParserValue::CharacterValue() const
     return m_value.char_value;
 }
 
+int SimpleParserValue::MultiCharacterValue() const
+{
+    assert(m_type == SimpleParserValueType::MULTI_CHARACTER);
+    return m_value.multi_character_sequence_id;
+}
+
 int SimpleParserValue::IntegerValue() const
 {
     assert(m_type == SimpleParserValueType::INTEGER);

src/Parser/Parsing/Simple/SimpleParserValue.h

@@ -13,8 +13,9 @@ enum class SimpleParserValueType
     END_OF_FILE,
     NEW_LINE,
 
-    // Single character
+    // Character sequences
     CHARACTER,
+    MULTI_CHARACTER,
 
     // Generic token types
     INTEGER,
@@ -36,6 +37,7 @@ public:
     {
         char char_value;
         int int_value;
+        int multi_character_sequence_id;
         double double_value;
         std::string* string_value;
     } m_value;
@@ -44,6 +46,7 @@ public:
     static SimpleParserValue EndOfFile(TokenPos pos);
     static SimpleParserValue NewLine(TokenPos pos);
     static SimpleParserValue Character(TokenPos pos, char c);
+    static SimpleParserValue MultiCharacter(TokenPos pos, int multiCharacterSequenceId);
     static SimpleParserValue Integer(TokenPos pos, int value);
     static SimpleParserValue FloatingPoint(TokenPos pos, double value);
     static SimpleParserValue String(TokenPos pos, std::string* stringValue);
@@ -63,6 +66,7 @@ public:
     _NODISCARD const TokenPos& GetPos() const override;
 
     _NODISCARD char CharacterValue() const;
+    _NODISCARD int MultiCharacterValue() const;
     _NODISCARD int IntegerValue() const;
     _NODISCARD double FloatingPointValue() const;
     _NODISCARD std::string& StringValue() const;