Better statement errors parsing

src/Error.cpp

@@ -1,7 +1,14 @@
 #include "Error.h"
 
 Error::Error(int line, int column, string lexme) :
-line(line), column(column), lexme(lexme) { }
+kind(ErrorKind::LEXER_ERROR), line(line), column(column), lexme(lexme) { }
+
+Error::Error(shared_ptr<Token> actualToken, optional<TokenKind> expectedTokenKind, optional<string> message) :
+kind(ErrorKind::PARSER_ERROR), actualToken(actualToken), expectedTokenKind(expectedTokenKind), message(message) { }
+
+ErrorKind Error::getKind() {
+    return kind;
+}
 
 int Error::getLine() {
     return line;
@@ -14,3 +21,15 @@ int Error::getColumn() {
 string Error::getLexme() {
     return lexme;
 }
+
+shared_ptr<Token> Error::getActualToken() {
+    return actualToken;
+}
+
+optional<TokenKind> Error::getExpectedTokenKind() {
+    return expectedTokenKind;
+}
+
+optional<string> Error::getMessage() {
+    return message;
+}

src/Error.h

@@ -3,19 +3,41 @@
 
 #include <iostream>
 
+class Token;
+enum class TokenKind;
+
 using namespace std;
 
+enum class ErrorKind {
+    LEXER_ERROR,
+    PARSER_ERROR
+};
+
 class Error {
 private:
+    ErrorKind kind;
+
     int line;
     int column;
     string lexme;
 
+    shared_ptr<Token> actualToken;
+    optional<TokenKind> expectedTokenKind;
+    optional<string> message;
+
 public:
     Error(int line, int column, string lexme);
+    Error(shared_ptr<Token> actualToken, optional<TokenKind> expectedTokenKind, optional<string> message);
+
+    ErrorKind getKind();
+
     int getLine();
     int getColumn();
     string getLexme();
+
+    shared_ptr<Token> getActualToken();
+    optional<TokenKind> getExpectedTokenKind();
+    optional<string> getMessage();
 };
 
 #endif

src/Lexer/Lexer.h

@@ -15,7 +15,6 @@ private:
     int currentIndex;
     int currentLine;
     int currentColumn;
-
     vector<shared_ptr<Error>> errors;
 
     shared_ptr<Token> nextToken();

src/Logger.cpp

@@ -102,6 +102,81 @@ string Logger::toString(shared_ptr<Token> token) {
     }
 }
 
+string Logger::toString(TokenKind tokenKind) {
+        switch (tokenKind) {
+        case TokenKind::PLUS:
+            return "+";
+        case TokenKind::MINUS:
+            return "-";
+        case TokenKind::STAR:
+            return "*";
+        case TokenKind::SLASH:
+            return "/";
+        case TokenKind::PERCENT:
+            return "%";
+        
+        case TokenKind::EQUAL:
+            return "=";
+        case TokenKind::NOT_EQUAL:
+            return "≠";
+        case TokenKind::LESS:
+            return "<";
+        case TokenKind::LESS_EQUAL:
+            return "≤";
+        case TokenKind::GREATER:
+            return ">";
+        case TokenKind::GREATER_EQUAL:
+            return "≥";
+
+        case TokenKind::LEFT_PAREN:
+            return "(";
+        case TokenKind::RIGHT_PAREN:
+            return ")";
+        case TokenKind::COMMA:
+            return ",";
+        case TokenKind::COLON:
+            return ":";
+        case TokenKind::SEMICOLON:
+            return ";";
+        case TokenKind::LEFT_ARROW:
+            return "←";
+        case TokenKind::RIGHT_ARROW:
+            return "→";
+
+        case TokenKind::BOOL:
+            return "LITERAL(BOOLEAN)";
+        case TokenKind::INTEGER_DEC:
+        case TokenKind::INTEGER_HEX:
+        case TokenKind::INTEGER_BIN:
+            return "LITERAL(INTEGER)";
+        case TokenKind::REAL:
+            return "LITERAL(REAL)";
+        case TokenKind::IDENTIFIER:
+            return "LITERAL(ID)";
+        case TokenKind::TYPE:
+            return "TYPE";
+
+        case TokenKind::IF:
+            return "IF";
+        case TokenKind::ELSE:
+            return "ELSE";
+        case TokenKind::FUNCTION:
+            return "FUN";
+        case TokenKind::RETURN:
+            return "RET";
+        case TokenKind::REPEAT:
+            return "REP";
+
+        case TokenKind::M_EXTERN:
+            return "@EXTERN";
+
+        case TokenKind::NEW_LINE:
+            return "↲";
+        case TokenKind::END:
+            return "END";
+    }
+}
+
 string Logger::toString(shared_ptr<Statement> statement) {
     switch (statement->getKind()) {
         case StatementKind::META_EXTERN_FUNCTION:
@@ -333,5 +408,30 @@ void Logger::print(vector<shared_ptr<Statement>> statements) {
 }
 
 void Logger::print(shared_ptr<Error> error) {
-    cout << format("Unexpected token \"{}\" at line: {}, column: {}\n", error->getLexme(), error->getLine() + 1, error->getColumn() + 1);
+    string message;
+    switch (error->getKind()) {
+        case ErrorKind::LEXER_ERROR:
+            message = format("Unexpected token \"{}\" at line: {}, column: {}", error->getLexme(), error->getLine() + 1, error->getColumn() + 1);
+            break;
+        case ErrorKind::PARSER_ERROR:
+            shared_ptr<Token> token = error->getActualToken();
+            optional<TokenKind> expectedTokenKind = error->getExpectedTokenKind();
+            optional<string> errorMessage = error->getMessage();
+
+            if (expectedTokenKind) {
+                message = format(
+                    "Expected token {} but instead found \"{}\" at line: {}, column: {}",
+                    toString(*expectedTokenKind), token->getLexme(), token->getLine() + 1, token->getColumn() + 1
+                );
+            } else {
+                message = format(
+                    "Unexpected token \"{}\" found at line: {}, column: {}",
+                    token->getLexme(), token->getLine() + 1, token->getColumn() + 1
+                );
+            }
+            if (errorMessage)
+                message += format(". {}", *errorMessage);
+            break;
+    }
+    cout << message << endl;
 }

src/Logger.h

@@ -4,6 +4,7 @@
 #include <vector>
 
 class Token;
+enum class TokenKind;
 class Statement;
 class StatementMetaExternFunction;
 class StatementVariable;
@@ -32,6 +33,7 @@ using namespace std;
 class Logger {
 private:
     static string toString(shared_ptr<Token> token);
+    static string toString(TokenKind tokenKind);
 
     static string toString(shared_ptr<Statement> statement);
     static string toString(shared_ptr<StatementMetaExternFunction> statement);

src/Parser/Parser.cpp

@@ -1,5 +1,8 @@
 #include "Parser.h"
 
+#include "Error.h"
+#include "Logger.h"
+
 #include "Parser/Expression/ExpressionGrouping.h"
 #include "Parser/Expression/ExpressionLiteral.h"
 #include "Parser/Expression/ExpressionVariable.h"
@@ -27,20 +30,21 @@ vector<shared_ptr<Statement>> Parser::getStatements() {
 
     while (!tryMatchingTokenKinds({TokenKind::END}, true, false)) {
         shared_ptr<Statement> statement = nextStatement();
-        // Abort parsing if we got an error
+        if (statement != nullptr) {
-        if (!statement->isValid()) {
-            //cerr << statement->toString(0);
-            exit(1);
-        }
             statements.push_back(statement);
 
             // Expect new line after statement
-        if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true)) {
+            if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true))
-            cerr << "Expected new line" << endl;
+                markError(TokenKind::NEW_LINE, {});
-            exit(1);
         }
     }
 
+    if (!errors.empty()) {
+        for (shared_ptr<Error> &error : errors)
+            Logger::print(error);
+        exit(1);
+    }
+
     return statements;
 }
 
@@ -49,20 +53,22 @@ vector<shared_ptr<Statement>> Parser::getStatements() {
 //
 shared_ptr<Statement> Parser::nextStatement() {
     shared_ptr<Statement> statement;
+    int errorsCount = errors.size();
 
     statement = matchStatementFunction();
-    if (statement != nullptr)
+    if (statement != nullptr || errors.size() > errorsCount)
         return statement;
 
     statement = matchStatementVariable();
-    if (statement != nullptr)
+    if (statement != nullptr || errors.size() > errorsCount)
         return statement;
 
     statement = matchStatementMetaExternFunction();
-    if (statement != nullptr)
+    if (statement != nullptr || errors.size() > errorsCount)
         return statement;
 
-    return matchStatementInvalid("Unexpected token");
+    markError({}, {});
+    return nullptr;
 }
 
 shared_ptr<Statement> Parser::nextInBlockStatement() {
@@ -88,7 +94,8 @@ shared_ptr<Statement> Parser::nextInBlockStatement() {
     if (statement != nullptr)
         return statement;
 
-    return matchStatementInvalid("Unexpected token");
+    markError({}, {});
+    return nullptr;
 }
 
 shared_ptr<Statement> Parser::matchStatementMetaExternFunction() {
@@ -107,13 +114,17 @@ shared_ptr<Statement> Parser::matchStatementMetaExternFunction() {
     if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) {
         do {
             tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line
-            if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false))
+            if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) {
-                return matchStatementInvalid("Expected function argument");
+                markError({}, "Expected function argument");
+                return nullptr;
+            }
             shared_ptr<Token> identifierToken = tokens.at(currentIndex++);
             shared_ptr<Token> typeToken = tokens.at(currentIndex++);
             optional<ValueType> argumentType = valueTypeForToken(typeToken);
-            if (!argumentType)
+            if (!argumentType) {
-                return matchStatementInvalid("Invalid argument type");
+                markError(TokenKind::TYPE, {});
+                return nullptr;
+            }
             
             arguments.push_back(pair<string, ValueType>(identifierToken->getLexme(), *argumentType));
         } while (tryMatchingTokenKinds({TokenKind::COMMA}, true, true));
@@ -125,8 +136,10 @@ shared_ptr<Statement> Parser::matchStatementMetaExternFunction() {
 
         shared_ptr<Token> typeToken = tokens.at(currentIndex);
         optional<ValueType> type = valueTypeForToken(typeToken);
-        if (!type)
+        if (!type) {
-            return matchStatementInvalid("Expected return type");
+            markError(TokenKind::TYPE, {});
+            return nullptr;
+        }
         returnType = *type;
 
         currentIndex++; // type
@@ -149,18 +162,23 @@ shared_ptr<Statement> Parser::matchStatementVariable() {
         valueType = ValueType::SINT32;
     else if (valueTypeToken->getLexme().compare("real32") == 0)
         valueType = ValueType::REAL32;
-    else
+    else {
-        return matchStatementInvalid("Invalid type");
+        markError(TokenKind::TYPE, {});
+        return nullptr;
+    }
 
     currentIndex++; // type
 
     // Expect left arrow
-    if (!tryMatchingTokenKinds({TokenKind::LEFT_ARROW}, true, true))
+    if (!tryMatchingTokenKinds({TokenKind::LEFT_ARROW}, true, true)) {
-        return matchStatementInvalid("Expected left arrow");
+        markError(TokenKind::LEFT_ARROW, {});
+        return nullptr;
+    }
 
     shared_ptr<Expression> expression = nextExpression();
-    if (expression == nullptr || !expression->isValid())
+    if (expression == nullptr || !expression->isValid()) {
-        return matchStatementInvalid("Invalid expression");
+        return nullptr;
+    }
 
     return make_shared<StatementVariable>(identifierToken->getLexme(), valueType, expression);
 }
@@ -182,13 +200,17 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
     if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) {
         do {
             tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line
-            if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false))
+            if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) {
-                return matchStatementInvalid("Expected function argument");
+                markError({}, "Expected function argument");
+                return nullptr;
+            }
             shared_ptr<Token> identifierToken = tokens.at(currentIndex++);
             shared_ptr<Token> typeToken = tokens.at(currentIndex++);
             optional<ValueType> argumentType = valueTypeForToken(typeToken);
-            if (!argumentType)
+            if (!argumentType) {
-                return matchStatementInvalid("Invalid argument type");
+                markError(TokenKind::TYPE, {});
+                return nullptr;
+            }
             
             arguments.push_back(pair<string, ValueType>(identifierToken->getLexme(), *argumentType));
         } while (tryMatchingTokenKinds({TokenKind::COMMA}, true, true));
@@ -200,24 +222,31 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
 
         shared_ptr<Token> typeToken = tokens.at(currentIndex);
         optional<ValueType> type = valueTypeForToken(typeToken);
-        if (!type)
+        if (!type) {
-            return matchStatementInvalid("Expected return type");
+            markError(TokenKind::TYPE, {});
+            return nullptr;
+        }
         returnType = *type;
 
         currentIndex++; // type
     }
 
     // consume new line
-    if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true))
+    if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true)) {
-        return matchStatementInvalid("Expected new line after function declaration");
+        markError(TokenKind::NEW_LINE, {});
+        return nullptr;
+    }
 
     // block
     statementBlock = matchStatementBlock({TokenKind::SEMICOLON});
-    if (statementBlock == nullptr || !statementBlock->isValid())
+    if (statementBlock == nullptr)
-        return statementBlock ?: matchStatementInvalid();
+        return nullptr;
 
-    if(!tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true))
+    if(!tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true)) {
-        return matchStatementInvalid("Expected a \";\" after a function declaration");
+        markError(TokenKind::SEMICOLON, {});
+        return nullptr;
+    }
+        //return matchStatementInvalid("Expected a \";\" after a function declaration");
 
     return make_shared<StatementFunction>(name, arguments, returnType, dynamic_pointer_cast<StatementBlock>(statementBlock));
 }
@@ -227,16 +256,18 @@ shared_ptr<Statement> Parser::matchStatementBlock(vector<TokenKind> terminalToke
 
     while (!tryMatchingTokenKinds(terminalTokenKinds, false, false)) {
         shared_ptr<Statement> statement = nextInBlockStatement();
-        if (statement == nullptr || !statement->isValid())
+        if (statement == nullptr)
-            return statement ?: matchStatementInvalid("Expected statement");
+            return nullptr;
         statements.push_back(statement);
 
         if (tryMatchingTokenKinds(terminalTokenKinds, false, false))
             break;
 
         // except new line
-        if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true))
+        if (!tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true)) {
-            return matchStatementInvalid("Expected new line");
+            markError(TokenKind::NEW_LINE, {});
+            return nullptr;
+        }
     }
 
     return make_shared<StatementBlock>(statements);
@@ -250,8 +281,8 @@ shared_ptr<Statement> Parser::matchStatementAssignment() {
     currentIndex++; // arrow
 
     shared_ptr<Expression> expression = nextExpression();
-    if (expression == nullptr || !expression->isValid())
+    if (expression == nullptr)
-        return matchStatementInvalid("Expected expression");
+        return nullptr;
 
     return make_shared<StatementAssignment>(identifierToken->getLexme(), expression);
 }
@@ -261,8 +292,8 @@ shared_ptr<Statement> Parser::matchStatementReturn() {
         return nullptr;
 
     shared_ptr<Expression> expression = nextExpression();
-    if (expression != nullptr && !expression->isValid())
+    if (expression == nullptr)
-        return matchStatementInvalid("Expected expression");
+        return nullptr;
     
     return make_shared<StatementReturn>(expression);
 }
@@ -285,33 +316,35 @@ shared_ptr<Statement> Parser::matchStatementRepeat() {
 
     if (!tryMatchingTokenKinds({TokenKind::COLON}, false, true)) {
         // got initial, expect comma
-        if (initStatement != nullptr && !tryMatchingTokenKinds({TokenKind::COMMA}, true, true))
+        if (initStatement != nullptr && !tryMatchingTokenKinds({TokenKind::COMMA}, true, true)) {
-            return matchStatementInvalid("Expected comma after initial statement");
+            markError(TokenKind::COMMA, {});
+            return nullptr;
+        }
 
         // optional new line
         tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true);
 
         // pre condition
         preConditionExpression = nextExpression();
-        if (preConditionExpression != nullptr && !preConditionExpression->isValid())
-            return matchStatementInvalid("Expected pre-condition expression");
 
         if (!tryMatchingTokenKinds({TokenKind::COLON}, true, true)) {
             // got pre-condition, expect comma
-            if (!tryMatchingTokenKinds({TokenKind::COMMA}, true, true))
+            if (!tryMatchingTokenKinds({TokenKind::COMMA}, true, true)) {
-                return matchStatementInvalid("Expected comma after pre-condition statement");
+                markError(TokenKind::COMMA, {});
+                return nullptr;
+            }
 
             // optional new line
             tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true);
 
             // post condition
             postConditionExpression = nextExpression();
-            if (postConditionExpression == nullptr || !postConditionExpression->isValid())
-                return matchStatementInvalid("Expected post-condition expression");
             
             // expect colon
-            if (!tryMatchingTokenKinds({TokenKind::COLON}, true, true))
+            if (!tryMatchingTokenKinds({TokenKind::COLON}, true, true)) {
-                return matchStatementInvalid("Expected \":\"");
+                markError(TokenKind::COLON, {});
+                return nullptr;
+            }
         }
     }
 
@@ -323,8 +356,8 @@ shared_ptr<Statement> Parser::matchStatementRepeat() {
     else
         bodyBlockStatement = matchStatementBlock({TokenKind::NEW_LINE});
 
-    if (bodyBlockStatement == nullptr || !bodyBlockStatement->isValid())
+    if (bodyBlockStatement == nullptr)
-        return bodyBlockStatement ?: matchStatementInvalid("Expected block statement");
+        return nullptr;
     
     tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true);
 
@@ -342,10 +375,6 @@ shared_ptr<Statement> Parser::matchStatementExpression() {
     return make_shared<StatementExpression>(expression);
 }
 
-shared_ptr<StatementInvalid> Parser::matchStatementInvalid(string message) {
-    return make_shared<StatementInvalid>(tokens.at(currentIndex), message);
-}
-
 //
 // Expression
 //
@@ -632,3 +661,19 @@ optional<ValueType> Parser::valueTypeForToken(shared_ptr<Token> token) {
 
     return {};
 }
+
+void Parser::markError(optional<TokenKind> expectedTokenKind, optional<string> message) {
+    shared_ptr<Token> actualToken = tokens.at(currentIndex);
+
+    // Try reaching the next safe token
+    vector<TokenKind> safeKinds = {TokenKind::END};
+    if (!actualToken->isOfKind({TokenKind::NEW_LINE}))
+        safeKinds.push_back(TokenKind::NEW_LINE);
+    if (!actualToken->isOfKind({TokenKind::SEMICOLON}))
+        safeKinds.push_back(TokenKind::SEMICOLON);
+
+    while (!tryMatchingTokenKinds(safeKinds, false, true))
+        currentIndex++;
+
+    errors.push_back(make_shared<Error>(actualToken, expectedTokenKind, message));
+}

src/Parser/Parser.h

@@ -2,8 +2,11 @@
 #define PARSER_H
 
 #include <vector>
+#include "Types.h"
 
-#include "Lexer/Token.h"
+class Token;
+enum class TokenKind;
+class Error;
 
 class Expression;
 class ExpressionInvalid;
@@ -17,6 +20,7 @@ class Parser {
 private:
     vector<shared_ptr<Token>> tokens;
     int currentIndex = 0;
+    vector<shared_ptr<Error>> errors;
 
     shared_ptr<Statement> nextStatement();
     shared_ptr<Statement> nextInBlockStatement();
@@ -30,7 +34,6 @@ private:
     shared_ptr<Statement> matchStatementReturn();
     shared_ptr<Statement> matchStatementRepeat();
     shared_ptr<Statement> matchStatementExpression();
-    shared_ptr<StatementInvalid> matchStatementInvalid(string message = "");
 
     shared_ptr<Expression> nextExpression();
     shared_ptr<Expression> matchEquality(); // =, !=
@@ -51,6 +54,8 @@ private:
     bool tryMatchingTokenKinds(vector<TokenKind> kinds, bool shouldMatchAll, bool shouldAdvance);
     optional<ValueType> valueTypeForToken(shared_ptr<Token> token);
 
+    void markError(optional<TokenKind> expectedTokenKind, optional<string> message);
+
 public:
     Parser(vector<shared_ptr<Token>> tokens);
     vector<shared_ptr<Statement>> getStatements();