String literals

.gitignore

@@ -1,12 +1,9 @@
-# ignore files without extensions
-*
-!*.*
-# brb build artifiacts
-*.o
-
 .DS_Store
 .vscode/settings.json
 
 # project build artifacts
 *.dSYM
 build/
+
+# brb build artifiacts
+*.o

samples/test.brc

@@ -0,0 +1,7 @@
+@extern putchar fun: character sint32 -> sint32
+
+main fun -> sint32
+    text data<sint32> <- "Hello string!\n"
+
+    ret 0
+;

src/Lexer/Lexer.cpp

@@ -250,6 +250,10 @@ shared_ptr<Token> Lexer::nextToken() {
     if (token != nullptr)
         return token;
 
+    token = matchString();
+    if (token != nullptr)
+        return token;
+
     // type
     token = matchType();
     if (token != nullptr)
@@ -405,6 +409,27 @@ shared_ptr<Token> Lexer::matchReal() {
     return token;
 }
 
+shared_ptr<Token> Lexer::matchString() {
+    int nextIndex = currentIndex;
+
+    if (currentIndex >= source.size() || source.at(nextIndex) != '\"')
+        return nullptr;
+
+    bool isClosing = false;
+    do {
+        nextIndex++;
+        isClosing = source.at(nextIndex) == '\"' && source.at(nextIndex - 1) != '\\';
+    } while (nextIndex < source.length() && !isClosing);
+
+    if (!isClosing)
+        return nullptr;
+
+    string lexme = source.substr(currentIndex, nextIndex - currentIndex + 1);
+    shared_ptr<Token> token = make_shared<Token>(TokenKind::STRING, lexme, currentLine, currentColumn);
+    advanceWithToken(token);
+    return token;
+}
+
 shared_ptr<Token> Lexer::matchIdentifier() {
     int nextIndex = currentIndex;
 

src/Lexer/Lexer.h

@@ -25,6 +25,7 @@ private:
     shared_ptr<Token> matchIntegerBin();
     shared_ptr<Token> matchIntegerChar();
     shared_ptr<Token> matchReal();
+    shared_ptr<Token> matchString();
     shared_ptr<Token> matchType();
     shared_ptr<Token> matchIdentifier();
     shared_ptr<Token> matchEnd();

src/Lexer/Token.cpp

@@ -41,7 +41,8 @@ vector<TokenKind> Token::tokensLiteral = {
     TokenKind::INTEGER_HEX,
     TokenKind::INTEGER_BIN,
     TokenKind::INTEGER_CHAR,
-    TokenKind::REAL
+    TokenKind::REAL,
+    TokenKind::STRING
 };
 
 Token::Token(TokenKind kind, string lexme, int line, int column):

src/Lexer/Token.h

@@ -41,6 +41,7 @@ enum class TokenKind {
     INTEGER_BIN,
     INTEGER_CHAR,
     REAL,
+    STRING,
     IDENTIFIER,
     TYPE,
 

src/Logger.cpp

@@ -84,6 +84,8 @@ string Logger::toString(shared_ptr<Token> token) {
             return "INT_CHAR(" + token->getLexme() + ")";
         case TokenKind::REAL:
             return "REAL(" + token->getLexme() + ")";
+        case TokenKind::STRING:
+            return "STRING(" + token->getLexme() + ")";
         case TokenKind::IDENTIFIER:
             return "ID(" + token->getLexme() + ")";
         case TokenKind::TYPE:
@@ -164,6 +166,8 @@ string Logger::toString(TokenKind tokenKind) {
             return "LITERAL(INTEGER)";
         case TokenKind::REAL:
             return "LITERAL(REAL)";
+        case TokenKind::STRING:
+            return "LITERAL(STRING)";
         case TokenKind::IDENTIFIER:
             return "LITERAL(ID)";
         case TokenKind::TYPE:

src/Parser/Expression/ExpressionArrayLiteral.cpp

@@ -1,8 +1,34 @@
 #include "ExpressionArrayLiteral.h"
 
+#include "Lexer/Token.h"
+#include "Parser/Expression/ExpressionLiteral.h"
+
 ExpressionArrayLiteral::ExpressionArrayLiteral(vector<shared_ptr<Expression>> expressions):
 Expression(ExpressionKind::ARRAY_LITERAL, nullptr), expressions(expressions) { }
 
+shared_ptr<ExpressionArrayLiteral> ExpressionArrayLiteral::expressionArrayLiteralForExpressions(vector<shared_ptr<Expression>> expressions) {
+    return make_shared<ExpressionArrayLiteral>(expressions);
+}
+
+shared_ptr<ExpressionArrayLiteral> ExpressionArrayLiteral::expressionArrayLiteralForTokenString(shared_ptr<Token> tokenString) {
+    if (tokenString->getKind() != TokenKind::STRING)
+        return nullptr;
+
+    vector<shared_ptr<Expression>> expressions;
+    string stringValue = tokenString->getLexme();
+    for (int i=1; i<stringValue.length()-1; i++) {
+        string lexme = stringValue.substr(i, 1);
+        if (stringValue[i] == '\\') {
+            lexme = stringValue.substr(i, 2);
+            i++;
+        }
+        shared_ptr<Token> token = make_shared<Token>(TokenKind::INTEGER_CHAR, lexme, tokenString->getLine(), tokenString->getColumn() + i);
+        shared_ptr<ExpressionLiteral> expression = ExpressionLiteral::expressionLiteralForToken(token);
+        expressions.push_back(expression);
+    }
+    return make_shared<ExpressionArrayLiteral>(expressions);
+}
+
 vector<shared_ptr<Expression>> ExpressionArrayLiteral::getExpressions() {
     return expressions;
 } 

src/Parser/Expression/ExpressionArrayLiteral.h

@@ -9,6 +9,9 @@ private:
 
 public:
     ExpressionArrayLiteral(vector<shared_ptr<Expression>> expressions);
+    //ExpressionArrayLiteral(shared_ptr<Token> tokenString);
+    static shared_ptr<ExpressionArrayLiteral> expressionArrayLiteralForExpressions(vector<shared_ptr<Expression>> expressions);
+    static shared_ptr<ExpressionArrayLiteral> expressionArrayLiteralForTokenString(shared_ptr<Token> tokenString);
     vector<shared_ptr<Expression>> getExpressions();
 };
 

src/Parser/Expression/ExpressionLiteral.cpp

@@ -1,8 +1,26 @@
 #include "ExpressionLiteral.h"
 
+#include "Utils.h"
 #include "Lexer/Token.h"
 #include "Parser/ValueType.h"
 
+shared_ptr<ExpressionLiteral> ExpressionLiteral::expressionLiteralForToken(shared_ptr<Token> token) {
+    switch (token->getKind()) {
+        case TokenKind::INTEGER_CHAR: {
+            string charString = token->getLexme();
+            optional<int> charValue = Utils::charStringToInt(charString);
+            if (!charValue)
+                return nullptr;
+            shared_ptr<ExpressionLiteral> expression = make_shared<ExpressionLiteral>();
+            expression->valueType = ValueType::valueTypeForToken(token, nullptr, 0);
+            expression->sint32Value = *charValue;
+            return expression;
+        }
+        default:
+            return nullptr;
+    }
+}
+
 ExpressionLiteral::ExpressionLiteral():
 Expression(ExpressionKind::LITERAL, nullptr) { }
 

src/Parser/Expression/ExpressionLiteral.h

@@ -10,6 +10,8 @@ private:
     float real32Value;
 
 public:
+    static shared_ptr<ExpressionLiteral> expressionLiteralForToken(shared_ptr<Token> token);
+    
     ExpressionLiteral(shared_ptr<Token> token);
     ExpressionLiteral();
     bool getBoolValue();

src/Parser/Parser.cpp

@@ -421,11 +421,11 @@ shared_ptr<Expression> Parser::matchPrimary() {
     if (expression != nullptr)
         return expression;
 
-    expression = matchExpressionLiteral();
+    expression = matchExpressionArrayLiteral();
         if (expression != nullptr)
             return expression;
 
-    expression = matchExpressionArrayLiteral();
+    expression = matchExpressionLiteral();
     if (expression != nullptr)
         return expression;
 
@@ -466,9 +466,9 @@ shared_ptr<Expression> Parser::matchExpressionLiteral() {
 }
 
 shared_ptr<Expression> Parser::matchExpressionArrayLiteral() {
-    if (!tryMatchingTokenKinds({TokenKind::LEFT_SQUARE_BRACKET}, true, true))
+    if (tryMatchingTokenKinds({TokenKind::STRING}, true, false)) {
-        return nullptr;
+        return ExpressionArrayLiteral::expressionArrayLiteralForTokenString(tokens.at(currentIndex++));
-
+    } else if (tryMatchingTokenKinds({TokenKind::LEFT_SQUARE_BRACKET}, true, true)) {
         vector<shared_ptr<Expression>> expressions;
         if (!tryMatchingTokenKinds({TokenKind::RIGHT_SQUARE_BRACKET}, true, true)) {
             do {
@@ -485,6 +485,9 @@ shared_ptr<Expression> Parser::matchExpressionArrayLiteral() {
 
 
         return make_shared<ExpressionArrayLiteral>(expressions);
+    }
+
+    return nullptr;
 }
 
 shared_ptr<Expression> Parser::matchExpressionVariable() {

src/Utils.cpp

@@ -0,0 +1,34 @@
+#include "Utils.h"
+
+optional<int> Utils::charStringToInt(string charString) {
+    switch (charString.length()) {
+        case 1:
+            return charString[0];
+        case 3:
+            return charString[1];
+        case 4:
+            charString[0] = charString[1];
+            charString[1] = charString[2];
+        case 2:
+            if (charString[0] != '\\')
+                return {};
+            switch (charString[1]) {
+                case 'b':
+                    return '\b';
+                case 'n':
+                    return '\n';
+                case 't':
+                    return '\t';
+                case '\\':
+                    return '\\';
+                case '\'':
+                    return '\'';
+                case '\"':
+                    return '\"';
+                default:
+                    return {};
+            }
+        default:
+            return {};
+    }
+}

src/Utils.h

@@ -0,0 +1,13 @@
+#ifndef UTILS_H
+#define UTILS_H
+
+#include <iostream>
+
+using namespace std;
+
+class Utils {
+public:
+    static optional<int> charStringToInt(string charString);
+};
+
+#endif