Initial work on expression parsing.
All checks were successful
continuous-integration/drone/push Build is passing

This commit is contained in:
2020-11-01 18:43:35 +01:00
parent e30bfe1686
commit c20a1089a9
7 changed files with 556 additions and 115 deletions

View File

@@ -11,6 +11,8 @@
#define EXPECT_TOKEN(token, kind) \
if (token->GetKind() != LexTokenKind::kind) { \
LogError(Diagnostics::DiagnosticType::UnexpectedToken, token->GetSpan()); \
} else { \
PROGRESS_TOKEN(token); \
}
namespace MalachScript::Parser {
@@ -20,18 +22,20 @@ namespace MalachScript::Parser {
std::vector<const ParsedStatement*> statements;
statements.reserve(32);
size_t current = 0;
const auto* currentToken = _firstToken;
while (true) {
while (_currentToken->GetKind() == LexTokenKind::Whitespace) {
_currentToken = _currentToken->GetNext().get();
while (currentToken->GetKind() == LexTokenKind::Whitespace) {
currentToken = currentToken->GetNext().get();
}
if (_currentToken->GetKind() == LexTokenKind::EndOfFile) {
if (currentToken->GetKind() == LexTokenKind::EndOfFile) {
break;
}
const ParsedStatement* statement;
auto result = ParseClass(statement) || ParseFunc(statement) || ParseNamespace(statement);
auto result = ParseClass(statement, currentToken) || ParseFunc(statement, currentToken) ||
ParseNamespace(statement, currentToken);
if (!result) {
// TODO: Log error
PROGRESS_TOKEN(_currentToken);
PROGRESS_TOKEN(currentToken);
continue;
}
statements.push_back(statement);
@@ -44,8 +48,8 @@ namespace MalachScript::Parser {
}
return new ParsedScriptStatement(TextSpan(0, end), statements);
}
bool Parser::ParseClass(const ParsedStatement*& out) {
const auto* current = _currentToken;
bool Parser::ParseClass(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
auto start = current->GetSpan().GetStart();
bool lookingForClass = true;
while (lookingForClass) {
@@ -62,7 +66,7 @@ namespace MalachScript::Parser {
// After class keyword, an identifier should always follow, if it doesn't, log an error.
Identifier identifier;
if (!ParseIdentifier(identifier, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
PROGRESS_TOKEN(current);
std::vector<Identifier> inherits;
@@ -77,14 +81,14 @@ namespace MalachScript::Parser {
case LexTokenKind::ColonSymbol: {
PROGRESS_TOKEN(current);
Identifier id;
if (!ParseIdentifier(id, _currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
if (!ParseIdentifier(id, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
inherits.push_back(id);
while (current->GetKind() == LexTokenKind::CommaSymbol) {
PROGRESS_TOKEN(current);
if (!ParseIdentifier(id, _currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
if (!ParseIdentifier(id, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
inherits.push_back(id);
PROGRESS_TOKEN(current);
@@ -105,123 +109,124 @@ namespace MalachScript::Parser {
}
const ParsedStatement* statement = nullptr;
// TODO: Sort by complexity
if (!ParseVirtProp(statement, current) && !ParseFunc(statement) && !ParseVar(statement) &&
!ParseFuncDef(statement)) {
if (!ParseVirtProp(statement, current) && !ParseFunc(statement, current) &&
!ParseVar(statement, current) && !ParseFuncDef(statement, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
break;
} else {
body.push_back(statement);
}
body.push_back(statement);
}
break;
}
default: throw;
}
out = new ParsedClassStatement(TextSpan(start, current->GetSpan().GetEnd()), identifier, inherits, body);
_currentToken = current;
currentToken = current;
return true;
}
bool Parser::ParseTypeDef(const ParsedStatement*& out) {
if (_currentToken->GetKind() != LexTokenKind::TypedefKeyword) {
bool Parser::ParseTypeDef(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
if (current->GetKind() != LexTokenKind::TypedefKeyword) {
return false;
}
auto start = _currentToken->GetSpan().GetStart();
PROGRESS_TOKEN(_currentToken);
auto start = current->GetSpan().GetStart();
PROGRESS_TOKEN(current);
Identifier defineFrom;
if (!ParsePrimType(defineFrom, _currentToken) && !ParseIdentifier(defineFrom, _currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
if (!ParsePrimType(defineFrom, current) && !ParseIdentifier(defineFrom, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
PROGRESS_TOKEN(_currentToken);
PROGRESS_TOKEN(current);
Identifier defineTo;
if (!ParseIdentifier(defineTo, _currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
if (!ParseIdentifier(defineTo, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
PROGRESS_TOKEN(_currentToken);
EXPECT_TOKEN(_currentToken, SemicolonSymbol);
PROGRESS_TOKEN(_currentToken);
out = new ParsedTypeDefStatement(TextSpan(start, _currentToken->GetSpan().GetEnd()), defineTo, defineFrom);
PROGRESS_TOKEN(current);
EXPECT_TOKEN(current, SemicolonSymbol);
out = new ParsedTypeDefStatement(TextSpan(start, current->GetSpan().GetEnd()), defineTo, defineFrom);
return true;
}
bool Parser::ParseNamespace(const ParsedStatement*& out) {
if (_currentToken->GetKind() != LexTokenKind::NamespaceKeyword) {
bool Parser::ParseNamespace(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
if (current->GetKind() != LexTokenKind::NamespaceKeyword) {
return false;
}
auto start = _currentToken->GetSpan().GetStart();
PROGRESS_TOKEN(_currentToken);
auto start = current->GetSpan().GetStart();
PROGRESS_TOKEN(current);
Identifier identifier;
if (!ParseIdentifier(identifier, _currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, _currentToken->GetSpan());
if (!ParseIdentifier(identifier, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
const auto* script = ParseScript();
auto end = _currentToken->GetSpan().GetEnd();
PROGRESS_TOKEN(_currentToken);
auto end = current->GetSpan().GetEnd();
PROGRESS_TOKEN(current);
out = new ParsedNamespaceStatement(TextSpan(start, end), identifier, script);
currentToken = current;
return true;
}
bool Parser::ParseFunc(const ParsedStatement*& out) {
auto start = _currentToken->GetSpan().GetStart();
const auto* token = _currentToken;
bool Parser::ParseFunc(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
auto start = current->GetSpan().GetStart();
bool isShared = false;
bool isExternal = false;
bool modifiers = true;
while (modifiers) {
switch (token->GetKind()) {
switch (current->GetKind()) {
case LexTokenKind::SharedKeyword:
isShared = true;
PROGRESS_TOKEN(token);
PROGRESS_TOKEN(current);
continue;
case LexTokenKind::ExternalKeyword:
isExternal = true;
PROGRESS_TOKEN(token);
PROGRESS_TOKEN(current);
continue;
default: modifiers = false; break;
}
}
AccessModifier accessModifier = AccessModifier::Public;
if (token->GetKind() == LexTokenKind::PrivateKeyword) {
if (current->GetKind() == LexTokenKind::PrivateKeyword) {
accessModifier = AccessModifier::Private;
PROGRESS_TOKEN(token);
} else if (token->GetKind() == LexTokenKind::ProtectedKeyword) {
PROGRESS_TOKEN(current);
} else if (current->GetKind() == LexTokenKind::ProtectedKeyword) {
accessModifier = AccessModifier::Protected;
PROGRESS_TOKEN(token);
PROGRESS_TOKEN(current);
}
const ParsedStatement* typeStatement = nullptr;
bool returnsReference = false;
if (token->GetKind() == LexTokenKind::TildeSymbol) {
if (current->GetKind() == LexTokenKind::TildeSymbol) {
// TODO: Handle destructor
throw std::logic_error("not implemented");
} else if (ParseType(typeStatement, token)) {
if (token->GetKind() == LexTokenKind::AmpersandSymbol) {
} else if (ParseType(typeStatement, current)) {
if (current->GetKind() == LexTokenKind::AmpersandSymbol) {
returnsReference = true;
PROGRESS_TOKEN(token);
PROGRESS_TOKEN(current);
}
}
Identifier identifier;
if (!ParseIdentifier(identifier, token)) {
if (!ParseIdentifier(identifier, current)) {
return false;
}
PROGRESS_TOKEN(token);
PROGRESS_TOKEN(current);
const ParsedStatement* paramList = nullptr;
if (!ParseParamList(paramList, token)) {
if (!ParseParamList(paramList, current)) {
return false;
}
_currentToken = token;
bool isConst = false;
if (_currentToken->GetKind() == LexTokenKind::ConstKeyword) {
if (current->GetKind() == LexTokenKind::ConstKeyword) {
isConst = true;
PROGRESS_TOKEN(_currentToken);
PROGRESS_TOKEN(current);
}
FuncAttr funcAttr = FuncAttr::None;
ParseFuncAttr(funcAttr, _currentToken);
ParseFuncAttr(funcAttr, current);
const ParsedStatement* statblock = nullptr;
if (_currentToken->GetKind() != LexTokenKind::SemicolonSymbol) {
if (current->GetKind() != LexTokenKind::SemicolonSymbol) {
// TODO: Parse stat block.
throw std::logic_error("not implemented");
}
out = new ParsedFuncStatement(TextSpan(start, _currentToken->GetSpan().GetEnd()), isShared, isExternal,
out = new ParsedFuncStatement(TextSpan(start, current->GetSpan().GetEnd()), isShared, isExternal,
accessModifier, typeStatement, returnsReference, identifier, paramList, isConst,
funcAttr, statblock);
currentToken = current;
return true;
}
@@ -355,7 +360,7 @@ namespace MalachScript::Parser {
const ParsedStatement* typeStatement = nullptr;
TypeMod typeMod = TypeMod::None;
Identifier identifier;
const ParsedExpression* defaultExpression = nullptr;
const ParsedStatement* defaultExpression = nullptr;
if (!ParseType(typeStatement, currentToken)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, currentToken->GetSpan());
@@ -410,8 +415,8 @@ namespace MalachScript::Parser {
}
bool Parser::ParseVirtProp([[maybe_unused]] const ParsedStatement*& out, const LexToken*& currentToken) {
AccessModifier access = AccessModifier::Public;
const auto* current = currentToken;
AccessModifier access = AccessModifier::Public;
if (current->GetKind() == LexTokenKind::PrivateKeyword) {
access = AccessModifier::Private;
PROGRESS_TOKEN(current);
@@ -456,11 +461,9 @@ namespace MalachScript::Parser {
}
ParseFuncAttr(getAttr, current);
if (current->GetKind() != LexTokenKind::SemicolonSymbol) {
// TODO: Parse stat block.
// if (ParseStatBlock(getStatement, current)){
//
// }
this->LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
if (!ParseStatBlock(getStatement, current)) {
this->LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
}
PROGRESS_TOKEN(current);
@@ -477,11 +480,9 @@ namespace MalachScript::Parser {
}
ParseFuncAttr(setAttr, current);
if (current->GetKind() != LexTokenKind::SemicolonSymbol) {
// TODO: Parse stat block.
// if (ParseStatBlock(setStatement, current)){
//
// }
this->LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
if (!ParseStatBlock(setStatement, current)) {
this->LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
}
PROGRESS_TOKEN(current);
@@ -507,8 +508,108 @@ namespace MalachScript::Parser {
return true;
}
bool Parser::ParseVar([[maybe_unused]] const ParsedStatement*& out) { return false; }
bool Parser::ParseFuncDef([[maybe_unused]] const ParsedStatement*& out) { return false; }
bool Parser::ParseIfStatement([[maybe_unused]] const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
if (current->GetKind() != LexTokenKind::IfKeyword) {
return false;
}
PROGRESS_TOKEN(current);
EXPECT_TOKEN(current, OpenParenthesisSymbol);
const ParsedStatement* condition = nullptr;
if (!ParseAssign(condition, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
return false;
}
EXPECT_TOKEN(current, CloseParenthesisSymbol);
const ParsedStatement* body = nullptr;
if (!ParseStatement(body, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
return false;
}
const ParsedStatement* elseStatement = nullptr;
if (current->GetKind() == LexTokenKind::ElseKeyword) {
PROGRESS_TOKEN(current);
if (!ParseStatement(elseStatement, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
}
out = new ParsedIfStatement(TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()),
condition, body, elseStatement);
currentToken = current;
return true;
}
bool Parser::ParseStatement(const ParsedStatement*& out, const LexToken*& currentToken) {
// TODO: All the other statements.
return ParseIfStatement(out, currentToken);
}
bool Parser::ParseVar([[maybe_unused]] const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
AccessModifier access = AccessModifier::Public;
if (current->GetKind() == LexTokenKind::PrivateKeyword) {
access = AccessModifier::Private;
PROGRESS_TOKEN(current);
} else if (current->GetKind() == LexTokenKind::ProtectedKeyword) {
access = AccessModifier::Protected;
PROGRESS_TOKEN(current);
}
const ParsedStatement* typeStatement = nullptr;
if (!ParseType(typeStatement, current)) {
return false;
}
Identifier identifier;
if (!ParseIdentifier(identifier, current)) {
delete typeStatement;
return false;
}
PROGRESS_TOKEN(current);
// TODO: Default values
// TODO: Creating multiple vars in a single line (int a, b, c)
if (current->GetKind() == LexTokenKind::SemicolonSymbol) {
PROGRESS_TOKEN(current);
} else {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
out = new ParsedVarStatement(TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), access,
typeStatement, identifier);
currentToken = current;
return true;
}
bool Parser::ParseStatBlock(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
if (current->GetKind() != LexTokenKind::OpenCurlyParenthesisSymbol) {
return false;
}
std::vector<const ParsedStatement*> statements;
while (true) {
if (current->GetKind() == LexTokenKind::CloseCurlyParenthesisSymbol) {
break;
}
const ParsedStatement* stat = nullptr;
if (ParseVar(stat, current) || ParseStatement(stat, current)) {
statements.push_back(stat);
} else {
break;
}
}
if (current->GetKind() == LexTokenKind::CloseCurlyParenthesisSymbol) {
PROGRESS_TOKEN(current);
} else {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
}
out = new ParsedStatBlockStatement(TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()),
statements);
currentToken = current;
return true;
}
bool Parser::ParseFuncDef([[maybe_unused]] const ParsedStatement*& out,
[[maybe_unused]] const LexToken*& currentToken) {
return false;
}
bool Parser::ParsePrimType(Identifier& out, const LexToken*& token) {
switch (token->GetKind()) {
case LexTokenKind::VoidKeyword: out = PrimitiveTypes::VoidName(); return true;
@@ -549,4 +650,107 @@ namespace MalachScript::Parser {
default: typeMod = TypeMod::RefInOut; return true;
}
}
}
bool Parser::ParseAssign(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
const ParsedStatement* leftHand = nullptr;
if (!ParseTernary(leftHand, current)) {
return false;
}
AssignmentOperator op;
if (!ParseAssignOp(op, current)) {
out = leftHand;
currentToken = current;
return true;
}
PROGRESS_TOKEN(current);
const ParsedStatement* rightHand = nullptr;
if (!ParseAssign(rightHand, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
out = leftHand;
currentToken = current;
return true;
}
out = new ParsedBinaryStatement<AssignmentOperator>(
TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), leftHand, op, rightHand);
currentToken = current;
return true;
}
bool Parser::ParseTernary(const ParsedStatement*& out, const LexToken*& currentToken) {
// TODO: implement ternary.
return ParseExpr(out, currentToken);
}
bool Parser::ParseExpr(const ParsedStatement*& out, const LexToken*& currentToken) {
const auto* current = currentToken;
const ParsedStatement* leftHand = nullptr;
if (!ParseExprTerm(leftHand, current)) {
return false;
}
MathOperator mathOp;
if (ParseMathOp(mathOp, current)) {
PROGRESS_TOKEN(current);
const ParsedStatement* rightHand = nullptr;
if (!ParseExprTerm(rightHand, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
out = leftHand;
currentToken = current;
return true;
}
out = new ParsedBinaryStatement<MathOperator>(
TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), leftHand, mathOp, rightHand);
return true;
}
ComparisonOperator compOp;
if (ParseCompOp(compOp, current)) {
PROGRESS_TOKEN(current);
const ParsedStatement* rightHand = nullptr;
if (!ParseExprTerm(rightHand, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
out = leftHand;
currentToken = current;
return true;
}
out = new ParsedBinaryStatement<ComparisonOperator>(
TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), leftHand, compOp, rightHand);
return true;
}
LogicOperator logicOp;
if (ParseLogicOp(logicOp, current)) {
PROGRESS_TOKEN(current);
const ParsedStatement* rightHand = nullptr;
if (!ParseExprTerm(rightHand, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
out = leftHand;
currentToken = current;
return true;
}
out = new ParsedBinaryStatement<LogicOperator>(
TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), leftHand, logicOp,
rightHand);
return true;
}
BitOperator bitOp;
if (ParseBitOp(bitOp, current)) {
PROGRESS_TOKEN(current);
const ParsedStatement* rightHand = nullptr;
if (!ParseExprTerm(rightHand, current)) {
LogError(Diagnostics::DiagnosticType::UnexpectedToken, current->GetSpan());
out = leftHand;
currentToken = current;
return true;
}
out = new ParsedBinaryStatement<BitOperator>(
TextSpan(currentToken->GetSpan().GetStart(), current->GetSpan().GetEnd()), leftHand, bitOp, rightHand);
return true;
}
out = leftHand;
currentToken = current;
return true;
}
bool Parser::ParseExprTerm([[maybe_unused]] const ParsedStatement*& out,
[[maybe_unused]] const LexToken*& currentToken) {
return false;
}
}