Merge pull request #30 from rafalgrodzinski/28-support-for-embedded-assembly

Support for embedded assembly

.vscode/launch.json

@@ -6,7 +6,7 @@
             "type": "lldb-dap",
             "request": "launch",
             "program": "${command:cmake.launchTargetPath}",
-            "args": ["-v", "${workspaceFolder}/samples/hello.brc"],
+            "args": ["-v", "${workspaceFolder}/samples/test.brc"],
             "cwd": "${workspaceFolder}",
             "internalConsoleOptions": "openOnSessionStart",
         }

samples/test.brc

@@ -1,5 +1,7 @@
 @extern putchar fun: character sint32 -> sint32
 
+// ./build/brb samples/test.brc -S -x86-asm-syntax=intel
+
 /*
 User type
     name data<u8, 32>
@@ -18,8 +20,34 @@ i u32 <- 0, rep text[i] != 0:
 
 // text data<u8> <- "Hello world!"
 
+/*addStuff asm<"+r, r">: num1 u32, num2 u32 -> u32
+    add $1, $0
+;*/
+
+normAdd fun: num1 sint32, num2 sint32 -> sint32
+    ret num1 + num2
+;
+
+rawAdd raw<"=r,r,r">: num1 sint32, num2 sint32 -> sint32
+    add $1, $2
+    mov $0, $1
+;
+
+/*printChar raw
+    .global REGISTER
+    .text
+    .REGISTER:
+    .byte "Hello", 0xa0
+    .long RegisterTable
+    //push 0x21
+
+;*/
+
 main fun -> sint32
-    text data<sint32> <- "Hello string!\n"
+    //printChar()
+
+    res1 sint32 <- normAdd(4, 5)
+    res2 sint32 <- rawAdd(4, 5)
 
     ret 0
 ;

src/Compiler/ModuleBuilder.cpp

@@ -14,6 +14,7 @@
 #include "Parser/Expression/ExpressionBlock.h"
 
 #include "Parser/Statement/StatementFunction.h"
+#include "Parser/Statement/StatementRawFunction.h"
 #include "Parser/Statement/StatementVariable.h"
 #include "Parser/Statement/StatementAssignment.h"
 #include "Parser/Statement/StatementReturn.h"
@@ -40,6 +41,12 @@ shared_ptr<llvm::Module> ModuleBuilder::getModule() {
     for (shared_ptr<Statement> &statement : statements)
         buildStatement(statement);
 
+    // verify module
+    string errorMessage;
+    llvm::raw_string_ostream llvmErrorMessage(errorMessage);
+    if (llvm::verifyModule(*module, &llvmErrorMessage))
+        markError(0, 0, errorMessage);
+
     if (!errors.empty()) {
         for (shared_ptr<Error> &error : errors)
             Logger::print(error);
@@ -52,7 +59,10 @@ shared_ptr<llvm::Module> ModuleBuilder::getModule() {
 void ModuleBuilder::buildStatement(shared_ptr<Statement> statement) {
     switch (statement->getKind()) {
         case StatementKind::FUNCTION:
-            buildFunctionDeclaration(dynamic_pointer_cast<StatementFunction>(statement));
+            buildFunction(dynamic_pointer_cast<StatementFunction>(statement));
+            break;
+        case StatementKind::RAW_FUNCTION:
+            buildRawFunction(dynamic_pointer_cast<StatementRawFunction>(statement));
             break;
         case StatementKind::VARIABLE:
             buildVarDeclaration(dynamic_pointer_cast<StatementVariable>(statement));
@@ -80,15 +90,15 @@ void ModuleBuilder::buildStatement(shared_ptr<Statement> statement) {
     }
 }
 
-void ModuleBuilder::buildFunctionDeclaration(shared_ptr<StatementFunction> statement) {
+void ModuleBuilder::buildFunction(shared_ptr<StatementFunction> statement) {
-    // get argument types
+    // function types
-    vector<llvm::Type *> types;
+    llvm::Type *returnType = typeForValueType(statement->getReturnValueType());
-    for (pair<string, shared_ptr<ValueType>> &arg : statement->getArguments()) {
+    vector<llvm::Type *> argTypes;
-        types.push_back(typeForValueType(arg.second));
+    for (pair<string, shared_ptr<ValueType>> &arg : statement->getArguments())
-    }
+        argTypes.push_back(typeForValueType(arg.second));
 
     // build function declaration
-    llvm::FunctionType *funType = llvm::FunctionType::get(typeForValueType(statement->getReturnValueType()), types, false);
+    llvm::FunctionType *funType = llvm::FunctionType::get(returnType, argTypes, false);
     llvm::Function *fun = llvm::Function::Create(funType, llvm::GlobalValue::ExternalLinkage, statement->getName(), module.get());
     if (!setFun(statement->getName(), fun))
         return;
@@ -103,7 +113,7 @@ void ModuleBuilder::buildFunctionDeclaration(shared_ptr<StatementFunction> state
     int i=0;
     for (auto &arg : fun->args()) {
         string name = statement->getArguments()[i].first;
-        llvm::Type *type = types[i];
+        llvm::Type *type = argTypes[i];
         arg.setName(name);
 
         llvm::AllocaInst *alloca = builder->CreateAlloca(type, nullptr, name);
@@ -126,6 +136,20 @@ void ModuleBuilder::buildFunctionDeclaration(shared_ptr<StatementFunction> state
         markError(0, 0, errorMessage);
 }
 
+void ModuleBuilder::buildRawFunction(shared_ptr<StatementRawFunction> statement) {
+    // function types
+    llvm::Type *returnType = typeForValueType(statement->getReturnValueType());
+    vector<llvm::Type *> argTypes;
+    for (pair<string, shared_ptr<ValueType>> &arg : statement->getArguments())
+        argTypes.push_back(typeForValueType(arg.second));
+
+    // build function declaration & body
+    llvm::FunctionType *funType = llvm::FunctionType::get(returnType, argTypes, false);
+    llvm::InlineAsm *rawFun = llvm::InlineAsm::get(funType, statement->getRawSource(), statement->getConstraints(), true, false, llvm::InlineAsm::AsmDialect::AD_Intel);
+    if (!setRawFun(statement->getName(), rawFun))
+        return;
+}
+
 void ModuleBuilder::buildVarDeclaration(shared_ptr<StatementVariable> statement) {
     if (statement->getValueType()->getKind() == ValueTypeKind::DATA) {
         vector<llvm::Value *> values = valuesForExpression(statement->getExpression());
@@ -481,15 +505,28 @@ llvm::Value *ModuleBuilder::valueForVar(shared_ptr<ExpressionVariable> expressio
 
 llvm::Value *ModuleBuilder::valueForCall(shared_ptr<ExpressionCall> expression) {
     llvm::Function *fun = getFun(expression->getName());
-    if (fun == nullptr)
+    if (fun != nullptr) {
-        return nullptr;
+        llvm::FunctionType *funType = fun->getFunctionType();
-    llvm::FunctionType *funType = fun->getFunctionType();
+        vector<llvm::Value*> argValues;
-    vector<llvm::Value*> argValues;
+        for (shared_ptr<Expression> &argumentExpression : expression->getArgumentExpressions()) {
-    for (shared_ptr<Expression> &argumentExpression : expression->getArgumentExpressions()) {
+            llvm::Value *argValue = valueForExpression(argumentExpression);
-        llvm::Value *argValue = valueForExpression(argumentExpression);
+            argValues.push_back(argValue);
-        argValues.push_back(argValue);
+        }
+        return builder->CreateCall(funType, fun, llvm::ArrayRef(argValues));
     }
-    return builder->CreateCall(funType, fun, llvm::ArrayRef(argValues));
+
+    llvm::InlineAsm *rawFun = getRawFun(expression->getName());
+    if (rawFun != nullptr) {
+        vector<llvm::Value *>argValues;
+        for (shared_ptr<Expression> &argumentExpression : expression->getArgumentExpressions()) {
+            llvm::Value *argValue = valueForExpression(argumentExpression);
+            argValues.push_back(argValue);
+        }
+        return builder->CreateCall(rawFun, llvm::ArrayRef(argValues));
+    }
+
+    markError(0, 0, format("Function \"{}\" not defined in scope", expression->getName()));
+    return nullptr;
 }
 
 bool ModuleBuilder::setAlloca(string name, llvm::AllocaInst *alloca) {
@@ -518,7 +555,7 @@ llvm::AllocaInst* ModuleBuilder::getAlloca(string name) {
 
 bool ModuleBuilder::setFun(string name, llvm::Function *fun) {
     if (scopes.top().funMap[name] != nullptr) {
-        markError(0, 0, format("Function \"{}\" already defined", name));
+        markError(0, 0, format("Function \"{}\" already defined in scope", name));
         return false;
     }
 
@@ -536,7 +573,29 @@ llvm::Function* ModuleBuilder::getFun(string name) {
         scopes.pop();
     }
 
-    markError(0, 0, format("Function \"{}\" not defined in scope", name));
+    return nullptr;
+}
+
+bool ModuleBuilder::setRawFun(string name, llvm::InlineAsm *rawFun) {
+    if (scopes.top().rawFunMap[name] != nullptr) {
+        markError(0, 0, format("Raw function \"{}\" already defined in scope", name));
+        return false;
+    }
+    
+    scopes.top().rawFunMap[name] = rawFun;
+    return true;
+}
+
+llvm::InlineAsm *ModuleBuilder::getRawFun(string name) {
+    stack<Scope> scopes = this->scopes;
+
+    while (!scopes.empty()) {
+        llvm::InlineAsm *rawFun = scopes.top().rawFunMap[name];
+        if (rawFun != nullptr)
+            return rawFun;
+        scopes.pop();
+    }
+
     return nullptr;
 }
 

src/Compiler/ModuleBuilder.h

@@ -10,6 +10,7 @@
 #include <llvm/IR/Constants.h>
 #include <llvm/Support/raw_ostream.h>
 #include <llvm/IR/Verifier.h>
+#include <llvm/IR/InlineAsm.h>
 
 class Error;
 class ValueType;
@@ -26,6 +27,7 @@ enum class ExpressionBinaryOperation;
 
 class Statement;
 class StatementFunction;
+class StatementRawFunction;
 class StatementVariable;
 class StatementAssignment;
 class StatementReturn;
@@ -39,6 +41,7 @@ using namespace std;
 typedef struct {
     map<string, llvm::AllocaInst*> allocaMap;
     map<string, llvm::Function*> funMap;
+    map<string, llvm::InlineAsm*> rawFunMap;
 } Scope;
 
 class ModuleBuilder {
@@ -60,7 +63,8 @@ private:
     stack<Scope> scopes;
 
     void buildStatement(shared_ptr<Statement> statement);
-    void buildFunctionDeclaration(shared_ptr<StatementFunction> statement);
+    void buildFunction(shared_ptr<StatementFunction> statement);
+    void buildRawFunction(shared_ptr<StatementRawFunction> statement);
     void buildVarDeclaration(shared_ptr<StatementVariable> statement);
     void buildAssignment(shared_ptr<StatementAssignment> statement);
     void buildBlock(shared_ptr<StatementBlock> statement);
@@ -88,6 +92,9 @@ private:
     bool setFun(string name, llvm::Function *fun);
     llvm::Function *getFun(string name);
 
+    bool setRawFun(string name, llvm::InlineAsm *rawFun);
+    llvm::InlineAsm *getRawFun(string name);
+
     llvm::Type *typeForValueType(shared_ptr<ValueType> valueType, int count = 0);
 
     void markError(int line, int column, string message);

src/Lexer/Lexer.cpp

@@ -12,6 +12,8 @@ vector<shared_ptr<Token>> Lexer::getTokens() {
     currentIndex = 0;
     currentLine = 0;
     currentColumn = 0;
+    foundRawSourceStart = false;
+    isParsingRawSource = false;
 
     tokens.clear();
     errors.clear();
@@ -117,6 +119,11 @@ shared_ptr<Token> Lexer::nextToken() {
             return nextToken(); // gets rid of remaining white spaces without repeating the code
     }
 
+    // raw source
+    token = matchRawSourceLine();
+    if (token != nullptr)
+        return token;
+
     // structural
     token = match(TokenKind::LEFT_PAREN, "(", false);
     if (token != nullptr)
@@ -213,6 +220,12 @@ shared_ptr<Token> Lexer::nextToken() {
     if (token != nullptr)
         return token;
 
+    token = match(TokenKind::RAW_FUNCTION, "raw", true);
+    if (token != nullptr) {
+        foundRawSourceStart = true;
+        return token;
+    }
+    
     token = match(TokenKind::RETURN, "ret", true);
     if (token != nullptr)
         return token;
@@ -271,8 +284,10 @@ shared_ptr<Token> Lexer::nextToken() {
 
     // new line
     token = match(TokenKind::NEW_LINE, "\n", false);
-    if (token != nullptr)
+    if (token != nullptr) {
+        tryStartingRawSourceParsing();
         return token;
+    }
     
     // eof
     token = matchEnd();
@@ -430,21 +445,6 @@ shared_ptr<Token> Lexer::matchString() {
     return token;
 }
 
-shared_ptr<Token> Lexer::matchIdentifier() {
-    int nextIndex = currentIndex;
-
-    while (nextIndex < source.length() && isIdentifier(nextIndex))
-        nextIndex++;
-
-    if (nextIndex == currentIndex || !isSeparator(nextIndex))
-        return nullptr;
-
-    string lexme = source.substr(currentIndex, nextIndex - currentIndex);
-    shared_ptr<Token> token = make_shared<Token>(TokenKind::IDENTIFIER, lexme, currentLine, currentColumn);
-    advanceWithToken(token);
-    return token;
-}
-
 shared_ptr<Token> Lexer::matchType() {
     int nextIndex = currentIndex;
 
@@ -463,6 +463,52 @@ shared_ptr<Token> Lexer::matchType() {
     return token;
 }
 
+shared_ptr<Token> Lexer::matchIdentifier() {
+    int nextIndex = currentIndex;
+
+    while (nextIndex < source.length() && isIdentifier(nextIndex))
+        nextIndex++;
+
+    if (nextIndex == currentIndex || !isSeparator(nextIndex))
+        return nullptr;
+
+    string lexme = source.substr(currentIndex, nextIndex - currentIndex);
+    shared_ptr<Token> token = make_shared<Token>(TokenKind::IDENTIFIER, lexme, currentLine, currentColumn);
+    advanceWithToken(token);
+    return token;
+}
+
+void Lexer::tryStartingRawSourceParsing() {
+    if (!foundRawSourceStart)
+        return;
+
+    if (!tokens.at(tokens.size() - 1)->isOfKind({TokenKind::COLON, TokenKind::COMMA, TokenKind::RIGHT_ARROW})) {
+        foundRawSourceStart = false;
+        isParsingRawSource = true;
+    }
+}
+
+shared_ptr<Token> Lexer::matchRawSourceLine() {
+    int nextIndex = currentIndex;
+
+    if (!isParsingRawSource)
+        return nullptr;
+
+    if (source.at(nextIndex) == ';') {
+        isParsingRawSource = false;
+        return nullptr;
+    }
+
+    while (source.at(nextIndex) != '\n')
+        nextIndex++;
+
+    string lexme = source.substr(currentIndex, nextIndex - currentIndex);
+    shared_ptr<Token> token =  make_shared<Token>(TokenKind::RAW_SOURCE_LINE, lexme, currentLine, currentColumn);
+    advanceWithToken(token);
+    currentIndex++; // skip newline
+    return token;
+}
+
 shared_ptr<Token> Lexer::matchEnd() {
     if (currentIndex >= source.length())
         return make_shared<Token>(TokenKind::END, "", currentLine, currentColumn);
@@ -530,11 +576,15 @@ bool Lexer::isSeparator(int index) {
 }
 
 void Lexer::advanceWithToken(shared_ptr<Token> token) {
-    if (token->getKind() == TokenKind::NEW_LINE) {
+    switch (token->getKind()) {
-        currentLine++;
+        case TokenKind::NEW_LINE:
-        currentColumn = 0;
+        case TokenKind::RAW_SOURCE_LINE:
-    } else {
+            currentLine++;
-        currentColumn += token->getLexme().length();
+            currentColumn = 0;
+            break;
+        default:
+            currentColumn += token->getLexme().length();
+            break;
     }
     currentIndex += token->getLexme().length();
 }

src/Lexer/Lexer.h

@@ -17,6 +17,8 @@ private:
     int currentColumn;
     vector<shared_ptr<Token>> tokens;
     vector<shared_ptr<Error>> errors;
+    bool foundRawSourceStart;
+    bool isParsingRawSource;
 
     shared_ptr<Token> nextToken();
     shared_ptr<Token> match(TokenKind kind, string lexme, bool needsSeparator);
@@ -28,6 +30,8 @@ private:
     shared_ptr<Token> matchString();
     shared_ptr<Token> matchType();
     shared_ptr<Token> matchIdentifier();
+    void tryStartingRawSourceParsing();
+    shared_ptr<Token> matchRawSourceLine();
     shared_ptr<Token> matchEnd();
 
     bool isWhiteSpace(int index);

src/Lexer/Token.h

@@ -30,6 +30,8 @@ enum class TokenKind {
     RIGHT_ARROW,
 
     FUNCTION,
+    RAW_FUNCTION,
+    RAW_SOURCE_LINE,
     RETURN,
     REPEAT,
     IF,

src/Logger.cpp

@@ -11,6 +11,7 @@
 #include "Parser/Statement/StatementMetaExternFunction.h"
 #include "Parser/Statement/StatementVariable.h"
 #include "Parser/Statement/StatementFunction.h"
+#include "Parser/Statement/StatementRawFunction.h"
 #include "Parser/Statement/StatementBlock.h"
 #include "Parser/Statement/StatementAssignment.h"
 #include "Parser/Statement/StatementReturn.h"
@@ -97,6 +98,10 @@ string Logger::toString(shared_ptr<Token> token) {
             return "ELSE";
         case TokenKind::FUNCTION:
             return "FUN";
+        case TokenKind::RAW_FUNCTION:
+            return "RAW";
+        case TokenKind::RAW_SOURCE_LINE:
+            return format("RAW_SOURCE_LINE({})", token->getLexme());
         case TokenKind::RETURN:
             return "RET";
         case TokenKind::REPEAT:
@@ -179,6 +184,8 @@ string Logger::toString(TokenKind tokenKind) {
             return "ELSE";
         case TokenKind::FUNCTION:
             return "FUN";
+        case TokenKind::RAW_FUNCTION:
+            return "RAW";
         case TokenKind::RETURN:
             return "RET";
         case TokenKind::REPEAT:
@@ -217,6 +224,8 @@ string Logger::toString(shared_ptr<Statement> statement) {
             return toString(dynamic_pointer_cast<StatementVariable>(statement));
         case StatementKind::FUNCTION:
             return toString(dynamic_pointer_cast<StatementFunction>(statement));
+        case StatementKind::RAW_FUNCTION:
+            return toString(dynamic_pointer_cast<StatementRawFunction>(statement));
         case StatementKind::BLOCK:
             return toString(dynamic_pointer_cast<StatementBlock>(statement));
         case StatementKind::ASSIGNMENT:
@@ -262,6 +271,21 @@ string Logger::toString(shared_ptr<StatementFunction> statement) {
     return text;
 }
 
+string Logger::toString(shared_ptr<StatementRawFunction> statement) {
+    string text;
+
+    string argsString;
+    for (int i = 0; i < statement->getArguments().size(); i++) {
+        auto arg = statement->getArguments().at(i);
+        argsString +=  format("ARG({}, {})", arg.first, toString(arg.second));
+    }
+    text += format("RAW(\"{}\"|{}|{}):\n", statement->getName(), argsString, toString(statement->getReturnValueType()));
+
+    text += statement->getRawSource();
+
+    return text;
+}
+
 string Logger::toString(shared_ptr<StatementBlock> statement) {
     string text;
 
@@ -468,13 +492,13 @@ void Logger::print(shared_ptr<Error> error) {
 
             if (expectedTokenKind) {
                 message = format(
-                    "Expected token {} but instead found \"{}\" at line: {}, column: {}",
+                    "Expected token {} but instead found {} at line: {}, column: {}",
-                    toString(*expectedTokenKind), token->getLexme(), token->getLine() + 1, token->getColumn() + 1
+                    toString(*expectedTokenKind), toString(token), token->getLine() + 1, token->getColumn() + 1
                 );
             } else {
                 message = format(
                     "Unexpected token \"{}\" found at line: {}, column: {}",
-                    token->getLexme(), token->getLine() + 1, token->getColumn() + 1
+                    toString(token), token->getLine() + 1, token->getColumn() + 1
                 );
             }
             if (errorMessage)

src/Logger.h

@@ -11,6 +11,7 @@ class Statement;
 class StatementMetaExternFunction;
 class StatementVariable;
 class StatementFunction;
+class StatementRawFunction;
 class StatementBlock;
 class StatementAssignment;
 class StatementReturn;
@@ -41,6 +42,7 @@ private:
     static string toString(shared_ptr<StatementMetaExternFunction> statement);
     static string toString(shared_ptr<StatementVariable> statement);
     static string toString(shared_ptr<StatementFunction> statement);
+    static string toString(shared_ptr<StatementRawFunction> statement);
     static string toString(shared_ptr<StatementBlock> statement);
     static string toString(shared_ptr<StatementAssignment> statement);
     static string toString(shared_ptr<StatementReturn> statement);

src/Parser/Parser.cpp

@@ -16,6 +16,7 @@
 #include "Parser/Expression/ExpressionBlock.h"
 
 #include "Parser/Statement/StatementFunction.h"
+#include "Parser/Statement/StatementRawFunction.h"
 #include "Parser/Statement/StatementVariable.h"
 #include "Parser/Statement/StatementAssignment.h"
 #include "Parser/Statement/StatementReturn.h"
@@ -61,6 +62,10 @@ shared_ptr<Statement> Parser::nextStatement() {
     if (statement != nullptr || errors.size() > errorsCount)
         return statement;
 
+    statement = matchStatementRawFunction();
+    if (statement != nullptr || errors.size() > errorsCount)
+        return statement;
+
     statement = matchStatementVariable();
     if (statement != nullptr || errors.size() > errorsCount)
         return statement;
@@ -230,6 +235,87 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
     return make_shared<StatementFunction>(name, arguments, returnType, dynamic_pointer_cast<StatementBlock>(statementBlock));
 }
 
+shared_ptr<Statement> Parser::matchStatementRawFunction() {
+    if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::RAW_FUNCTION}, true, false))
+        return nullptr;
+
+    string name;
+    string constraints;
+    vector<pair<string, shared_ptr<ValueType>>> arguments;
+    shared_ptr<ValueType> returnType = ValueType::NONE;
+    string rawSource;
+
+    // name
+    name = tokens.at(currentIndex++)->getLexme();
+    currentIndex++; // skip raw
+
+    // constraints
+
+    if (tryMatchingTokenKinds({TokenKind::LESS}, true, true)) {
+        if (tokens.at(currentIndex)->isOfKind({TokenKind::STRING})) {
+            constraints = tokens.at(currentIndex++)->getLexme();
+            // remove enclosing quotes
+            if (constraints.length() >= 2)
+                constraints = constraints.substr(1, constraints.length() - 2);
+        }
+        if (!tryMatchingTokenKinds({TokenKind::GREATER}, true, true))
+            markError({TokenKind::GREATER}, {});
+    }
+
+    // arguments
+    if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) {
+        do {
+            tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line
+            if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) {
+                markError({}, "Expected function argument");
+                return nullptr;
+            }
+            shared_ptr<Token> identifierToken = tokens.at(currentIndex++);
+            shared_ptr<ValueType> argumentType = matchValueType();
+            if (argumentType == nullptr) {
+                markError(TokenKind::TYPE, {});
+                return nullptr;
+            }
+            
+            arguments.push_back(pair<string, shared_ptr<ValueType>>(identifierToken->getLexme(), argumentType));
+        } while (tryMatchingTokenKinds({TokenKind::COMMA}, true, true));
+    }
+
+    // return type
+    if (tryMatchingTokenKinds({TokenKind::RIGHT_ARROW}, true, true)) {
+        tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line
+
+        returnType = matchValueType();
+        if (returnType == nullptr) {
+            markError(TokenKind::TYPE, {});
+            return nullptr;
+        }
+    }
+
+    // consume new line
+    if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true)) {
+        markError(TokenKind::NEW_LINE, {});
+        return nullptr;
+    }
+
+    // source
+    while (tryMatchingTokenKinds({TokenKind::RAW_SOURCE_LINE}, true, false)) {
+        if (!rawSource.empty())
+            rawSource += "\n";
+        rawSource += tokens.at(currentIndex++)->getLexme();
+
+        // Consume optional new line (for example because of a comment)
+        tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true);
+    }
+
+    if(!tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true)) {
+        markError(TokenKind::SEMICOLON, {});
+        return nullptr;
+    }
+
+    return make_shared<StatementRawFunction>(name, constraints, arguments, returnType, rawSource);
+}
+
 shared_ptr<Statement> Parser::matchStatementBlock(vector<TokenKind> terminalTokenKinds) {
     vector<shared_ptr<Statement>> statements;
 

src/Parser/Parser.h

@@ -26,6 +26,7 @@ private:
     shared_ptr<Statement> matchStatementMetaExternFunction();
     shared_ptr<Statement> matchStatementVariable();
     shared_ptr<Statement> matchStatementFunction();
+    shared_ptr<Statement> matchStatementRawFunction();
 
     shared_ptr<Statement> matchStatementBlock(vector<TokenKind> terminalTokenKinds);
     shared_ptr<Statement> matchStatementAssignment();

src/Parser/Statement/Statement.h

@@ -10,6 +10,7 @@ enum class StatementKind {
     BLOCK,
     RETURN,
     FUNCTION,
+    RAW_FUNCTION,
     VARIABLE,
     ASSIGNMENT,
     REPEAT,

src/Parser/Statement/StatementRawFunction.cpp

@@ -0,0 +1,24 @@
+#include "StatementRawFunction.h"
+
+StatementRawFunction::StatementRawFunction(string name, string constraints, vector<pair<string, shared_ptr<ValueType>>> arguments, shared_ptr<ValueType> returnValueType, string rawSource):
+Statement(StatementKind::RAW_FUNCTION), name(name), constraints(constraints), arguments(arguments), returnValueType(returnValueType), rawSource(rawSource) { }
+
+string StatementRawFunction::getName() {
+    return name;
+}
+
+string StatementRawFunction::getConstraints() {
+    return constraints;
+}
+
+vector<pair<string, shared_ptr<ValueType>>> StatementRawFunction::getArguments() {
+    return arguments;
+}
+
+shared_ptr<ValueType> StatementRawFunction::getReturnValueType() {
+    return returnValueType;
+}
+
+string StatementRawFunction::getRawSource() {
+    return rawSource;
+}

src/Parser/Statement/StatementRawFunction.h

@@ -0,0 +1,20 @@
+#include "Parser/Statement/Statement.h"
+
+class ValueType;
+
+class StatementRawFunction: public Statement {
+private:
+    string name;
+    string constraints;
+    vector<pair<string, shared_ptr<ValueType>>> arguments;
+    shared_ptr<ValueType> returnValueType;
+    string rawSource;
+
+public:
+    StatementRawFunction(string name, string constraints, vector<pair<string, shared_ptr<ValueType>>> arguments, shared_ptr<ValueType> returnValueType, string rawSource);
+    string getName();
+    string getConstraints();
+    vector<pair<string, shared_ptr<ValueType>>> getArguments();
+    shared_ptr<ValueType> getReturnValueType();
+    string getRawSource();
+};