Merge pull request #31 from rafalgrodzinski/29-match-line-of-tokens-through-a-function

Match line of tokens through a function
This commit is contained in:
Rafał
2025-07-31 14:02:57 +09:00
committed by GitHub
12 changed files with 769 additions and 186 deletions

View File

@@ -24,15 +24,20 @@ i u32 <- 0, rep text[i] != 0:
add $1, $0 add $1, $0
;*/ ;*/
normAdd fun: num1 sint32, num2 sint32 -> sint32 /*normAdd fun: num1 sint32, num2 sint32 -> sint32
ret num1 + num2 ret num1 + num2
; ;*/
rawAdd raw<"=r,r,r">: num1 sint32, num2 sint32 -> sint32 rawAdd raw<"=r,r,r">: num1 sint32, num2 sint32 -> sint32
add $1, $2 add $1, $2
mov $0, $1 mov $0, $1
; ;
/*rawAdd raw: num1 sint32, num2 sint32 -> sint32
add $1, $2
mov $0, $1
;*/
/*printChar raw /*printChar raw
.global REGISTER .global REGISTER
.text .text
@@ -44,10 +49,12 @@ rawAdd raw<"=r,r,r">: num1 sint32, num2 sint32 -> sint32
;*/ ;*/
main fun -> sint32 main fun -> sint32
//printChar() /*a sint32 <- 4
b sint32 <- 5
res1 sint32 <- normAdd(4, 5) res sint32 <- rawAdd(4, 5)*/
res2 sint32 <- rawAdd(4, 5) putchar('@')
putchar('\n')
ret 0 ret 0
; ;

View File

@@ -482,7 +482,7 @@ void Logger::print(shared_ptr<Error> error) {
switch (error->getKind()) { switch (error->getKind()) {
case ErrorKind::LEXER_ERROR: { case ErrorKind::LEXER_ERROR: {
string lexme = error->getLexme() ? *(error->getLexme()) : ""; string lexme = error->getLexme() ? *(error->getLexme()) : "";
message = format("Unexpected token \"{}\" at line: {}, column: {}", lexme, error->getLine() + 1, error->getColumn() + 1); message = format("At line {}, column {}: Unexpected token \"{}\"", error->getLine() + 1, error->getColumn() + 1, lexme);
break; break;
} }
case ErrorKind::PARSER_ERROR: { case ErrorKind::PARSER_ERROR: {
@@ -492,13 +492,13 @@ void Logger::print(shared_ptr<Error> error) {
if (expectedTokenKind) { if (expectedTokenKind) {
message = format( message = format(
"Expected token {} but instead found {} at line: {}, column: {}", "At line {}, column {}: Expected token {} but found {} instead",
toString(*expectedTokenKind), toString(token), token->getLine() + 1, token->getColumn() + 1 token->getLine() + 1, token->getColumn() + 1, toString(*expectedTokenKind), toString(token)
); );
} else { } else {
message = format( message = format(
"Unexpected token \"{}\" found at line: {}, column: {}", "At line {}, column {}: Unexpected token \"{}\" found",
toString(token), token->getLine() + 1, token->getColumn() + 1 token->getLine() + 1, token->getColumn() + 1, toString(token)
); );
} }
if (errorMessage) if (errorMessage)
@@ -507,7 +507,7 @@ void Logger::print(shared_ptr<Error> error) {
} }
case ErrorKind::BUILDER_ERROR: { case ErrorKind::BUILDER_ERROR: {
string errorMessage = error->getMessage() ? *(error->getMessage()) : ""; string errorMessage = error->getMessage() ? *(error->getMessage()) : "";
message = format("Error at line {}, column {}: {}", error->getLine(), error->getColumn(), errorMessage); message = format("At line {}, column {}: {}", error->getLine(), error->getColumn(), errorMessage);
break; break;
} }
} }

View File

@@ -0,0 +1,44 @@
#include "Parsee.h"
Parsee Parsee::tokenParsee(TokenKind tokenKind, bool isRequired, bool shouldReturn) {
Parsee parsee;
parsee.kind = ParseeKind::TOKEN;
parsee.tokenKind = tokenKind;
parsee.isRequired = isRequired;
parsee.shouldReturn = shouldReturn;
return parsee;
}
Parsee Parsee::valueTypeParsee(bool isRequired) {
Parsee parsee;
parsee.kind = ParseeKind::VALUE_TYPE;
parsee.isRequired = isRequired;
parsee.shouldReturn = true;
return parsee;
}
Parsee Parsee::expressionParsee(bool isRequired) {
Parsee parsee;
parsee.kind = ParseeKind::EXPRESSION;
parsee.isRequired = isRequired;
parsee.shouldReturn = true;
return parsee;
}
Parsee::Parsee() { }
ParseeKind Parsee::getKind() {
return kind;
}
TokenKind Parsee::getTokenKind() {
return tokenKind;
}
bool Parsee::getIsRequired() {
return isRequired;
}
bool Parsee::getShouldReturn() {
return shouldReturn;
}

View File

@@ -0,0 +1,33 @@
#ifndef PARSEE_H
#define PARSEE_H
#include <memory>
enum class TokenKind;
enum class ParseeKind {
TOKEN,
VALUE_TYPE,
EXPRESSION
};
class Parsee {
private:
ParseeKind kind;
TokenKind tokenKind;
bool isRequired;
bool shouldReturn;
Parsee();
public:
static Parsee tokenParsee(TokenKind tokenKind, bool isRequired, bool shouldReturn);
static Parsee valueTypeParsee(bool isRequired);
static Parsee expressionParsee(bool isRequired);
ParseeKind getKind();
TokenKind getTokenKind();
bool getIsRequired();
bool getShouldReturn();
};
#endif

View File

@@ -0,0 +1,20 @@
#include "ParseeGroup.h"
#include "Parsee.h"
ParseeGroup::ParseeGroup(vector<Parsee> parsees, optional<ParseeGroup> repeatedGroup):
parsees(parsees) {
if (repeatedGroup) {
this->repeatedGroup = *repeatedGroup;
} else {
this->repeatedGroup = {};
}
}
vector<Parsee> ParseeGroup::getParsees() {
return parsees;
}
optional<reference_wrapper<ParseeGroup>> ParseeGroup::getRepeatedGroup() {
return repeatedGroup;
}

View File

@@ -0,0 +1,22 @@
#ifndef PARSEE_GROUP_H
#define PARSEE_GROUP_H
#include <vector>
#include <optional>
class Parsee;
using namespace std;
class ParseeGroup {
private:
vector<Parsee> parsees;
optional<reference_wrapper<ParseeGroup>> repeatedGroup;
public:
ParseeGroup(vector<Parsee> parsees, optional<ParseeGroup> repeatedGroup);
vector<Parsee> getParsees();
optional<reference_wrapper<ParseeGroup>> getRepeatedGroup();
};
#endif

View File

@@ -0,0 +1,51 @@
#include "ParseeResult.h"
#include "Lexer/Token.h"
#include "Parser/ValueType.h"
ParseeResult ParseeResult::tokenResult(shared_ptr<Token> token) {
ParseeResult parseeResult;
parseeResult.kind = ParseeResultKind::TOKEN;
parseeResult.token = token;
parseeResult.tokensCount = 1;
return parseeResult;
}
ParseeResult ParseeResult::valueTypeResult(shared_ptr<ValueType> valueType, int tokensCount) {
ParseeResult parseeResult;
parseeResult.kind = ParseeResultKind::VALUE_TYPE;
parseeResult.valueType = valueType;
parseeResult.tokensCount = tokensCount;
return parseeResult;
}
ParseeResult ParseeResult::expressionResult(shared_ptr<Expression> expression, int tokensCount) {
ParseeResult result;
result.kind = ParseeResultKind::EXPRESSION;
result.expression = expression;
result.tokensCount = tokensCount;
return result;
}
ParseeResult::ParseeResult() { }
ParseeResultKind ParseeResult::getKind() {
return kind;
}
shared_ptr<Token> ParseeResult::getToken() {
return token;
}
shared_ptr<ValueType> ParseeResult::getValueType() {
return valueType;
}
shared_ptr<Expression> ParseeResult::getExpression() {
return expression;
}
int ParseeResult::getTokensCount() {
return tokensCount;
}

View File

@@ -0,0 +1,39 @@
#ifndef PARSEE_RESULT_H
#define PARSEE_RESULT_H
#include <memory>
class Token;
class ValueType;
class Expression;
using namespace std;
enum class ParseeResultKind {
TOKEN,
VALUE_TYPE,
EXPRESSION
};
class ParseeResult {
private:
ParseeResultKind kind;
shared_ptr<Token> token;
shared_ptr<ValueType> valueType;
shared_ptr<Expression> expression;
int tokensCount;
ParseeResult();
public:
static ParseeResult tokenResult(shared_ptr<Token> token);
static ParseeResult valueTypeResult(shared_ptr<ValueType> valueType, int tokensCount);
static ParseeResult expressionResult(shared_ptr<Expression> expression, int tokensCount);
ParseeResultKind getKind();
shared_ptr<Token> getToken();
shared_ptr<ValueType> getValueType();
shared_ptr<Expression> getExpression();
int getTokensCount();
};
#endif

View File

@@ -0,0 +1,30 @@
#include "ParseeResultsGroup.h"
#include "ParseeResult.h"
ParseeResultsGroup ParseeResultsGroup::success(vector<ParseeResult> results) {
ParseeResultsGroup resultsGroup;
resultsGroup.kind = ParseeResultsGroupKind::SUCCESS;
resultsGroup.results = results;
return resultsGroup;
}
ParseeResultsGroup ParseeResultsGroup::noMatch() {
ParseeResultsGroup resultsGroup;
resultsGroup.kind = ParseeResultsGroupKind::NO_MATCH;
return resultsGroup;
}
ParseeResultsGroup ParseeResultsGroup::failure() {
ParseeResultsGroup resultsGroup;
resultsGroup.kind = ParseeResultsGroupKind::FAILURE;
return resultsGroup;
}
ParseeResultsGroupKind ParseeResultsGroup::getKind() {
return kind;
}
vector<ParseeResult> ParseeResultsGroup::getResults() {
return results;
}

View File

@@ -0,0 +1,30 @@
#ifndef PARSEE_RESULTS_GROUP_H
#define PARSEE_RESULTS_GROUP_H
#include <vector>
class ParseeResult;
using namespace std;
enum class ParseeResultsGroupKind {
SUCCESS,
NO_MATCH,
FAILURE
};
class ParseeResultsGroup {
private:
ParseeResultsGroupKind kind;
vector<ParseeResult> results;
public:
static ParseeResultsGroup success(vector<ParseeResult> results);
static ParseeResultsGroup noMatch();
static ParseeResultsGroup failure();
ParseeResultsGroupKind getKind();
vector<ParseeResult> getResults();
};
#endif

View File

@@ -25,6 +25,11 @@
#include "Parser/Statement/StatementBlock.h" #include "Parser/Statement/StatementBlock.h"
#include "Parser/Statement/StatementRepeat.h" #include "Parser/Statement/StatementRepeat.h"
#include "Parsee/Parsee.h"
#include "Parsee/ParseeGroup.h"
#include "Parsee/ParseeResult.h"
#include "Parsee/ParseeResultsGroup.h"
Parser::Parser(vector<shared_ptr<Token>> tokens) : Parser::Parser(vector<shared_ptr<Token>> tokens) :
tokens(tokens) { } tokens(tokens) { }
@@ -107,112 +112,207 @@ shared_ptr<Statement> Parser::nextInBlockStatement() {
} }
shared_ptr<Statement> Parser::matchStatementMetaExternFunction() { shared_ptr<Statement> Parser::matchStatementMetaExternFunction() {
if (!tryMatchingTokenKinds({TokenKind::M_EXTERN, TokenKind::IDENTIFIER, TokenKind::FUNCTION}, true, false)) ParseeResultsGroup resultsGroup;
return nullptr;
string name; string identifier;
vector<pair<string, shared_ptr<ValueType>>> arguments; vector<pair<string, shared_ptr<ValueType>>> arguments;
shared_ptr<ValueType> returnType = ValueType::NONE; shared_ptr<ValueType> returnType = ValueType::NONE;
currentIndex++; // skip meta // identifier
shared_ptr<Token> identifierToken = tokens.at(currentIndex++); resultsGroup = parseeResultsGroupForParseeGroup(
currentIndex++; // skip fun ParseeGroup(
{
Parsee::tokenParsee(TokenKind::M_EXTERN, true, false),
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::tokenParsee(TokenKind::FUNCTION, true, false)
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
identifier = resultsGroup.getResults().at(0).getToken()->getLexme();
break;
case ParseeResultsGroupKind::NO_MATCH:
case ParseeResultsGroupKind::FAILURE:
return nullptr;
}
// arguments // arguments
if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) { resultsGroup = parseeResultsGroupForParseeGroup(
do { ParseeGroup(
tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line {
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) { Parsee::tokenParsee(TokenKind::COLON, true, false),
markError({}, "Expected function argument"); Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
return nullptr; Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::valueTypeParsee(true)
},
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::COMMA, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::valueTypeParsee(true)
},
{}
)
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
for (int i=0; i<resultsGroup.getResults().size(); i+=2) {
pair<string, shared_ptr<ValueType>> arg;
arg.first = resultsGroup.getResults().at(i).getToken()->getLexme();
arg.second = resultsGroup.getResults().at(i+1).getValueType();
arguments.push_back(arg);
} }
shared_ptr<Token> identifierToken = tokens.at(currentIndex++); break;
//shared_ptr<Token> argumentTypeToken = tokens.at(currentIndex++); case ParseeResultsGroupKind::NO_MATCH:
shared_ptr<ValueType> argumentType = matchValueType(); break;
if (argumentType == nullptr) { case ParseeResultsGroupKind::FAILURE:
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
//shared_ptr<Token> returnTypeToken = tokens.at(currentIndex);
returnType = matchValueType();
if (returnType == nullptr) {
markError(TokenKind::TYPE, {});
return nullptr; return nullptr;
}
} }
return make_shared<StatementMetaExternFunction>(identifierToken->getLexme(), arguments, returnType); // return type
resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::RIGHT_ARROW, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::valueTypeParsee(true)
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
returnType = resultsGroup.getResults().at(0).getValueType();
break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
return nullptr;
}
return make_shared<StatementMetaExternFunction>(identifier, arguments, returnType);
} }
shared_ptr<Statement> Parser::matchStatementVariable() { shared_ptr<Statement> Parser::matchStatementVariable() {
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) ParseeResultsGroup resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::valueTypeParsee(true),
Parsee::tokenParsee(TokenKind::LEFT_ARROW, true, false),
Parsee::expressionParsee(true)
},
{}
)
);
if (resultsGroup.getKind() != ParseeResultsGroupKind::SUCCESS)
return nullptr; return nullptr;
shared_ptr<Token> identifierToken = tokens.at(currentIndex++); string identifier = resultsGroup.getResults().at(0).getToken()->getLexme();
shared_ptr<ValueType> valueType = matchValueType(); shared_ptr<ValueType> valueType = resultsGroup.getResults().at(1).getValueType();
shared_ptr<Expression> expression = resultsGroup.getResults().at(2).getExpression();
// Expect left arrow return make_shared<StatementVariable>(identifier, valueType, expression);
if (!tryMatchingTokenKinds({TokenKind::LEFT_ARROW}, true, true)) {
markError(TokenKind::LEFT_ARROW, {});
return nullptr;
}
shared_ptr<Expression> expression = nextExpression();
if (expression == nullptr)
return nullptr;
return make_shared<StatementVariable>(identifierToken->getLexme(), valueType, expression);
} }
shared_ptr<Statement> Parser::matchStatementFunction() { shared_ptr<Statement> Parser::matchStatementFunction() {
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::FUNCTION}, true, false)) bool hasError = false;
return nullptr; ParseeResultsGroup resultsGroup;
string name; string name;
vector<pair<string, shared_ptr<ValueType>>> arguments; vector<pair<string, shared_ptr<ValueType>>> arguments;
shared_ptr<ValueType> returnType = ValueType::NONE; shared_ptr<ValueType> returnType = ValueType::NONE;
shared_ptr<Statement> statementBlock; shared_ptr<Statement> statementBlock;
// name // identifier
name = tokens.at(currentIndex++)->getLexme(); resultsGroup = parseeResultsGroupForParseeGroup(
currentIndex++; // skip fun ParseeGroup(
{
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::tokenParsee(TokenKind::FUNCTION, true, false)
},
{}
)
);
// arguments switch (resultsGroup.getKind()) {
if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) { case ParseeResultsGroupKind::SUCCESS:
do { name = resultsGroup.getResults().at(0).getToken()->getLexme();
tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line break;
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) { case ParseeResultsGroupKind::NO_MATCH:
markError({}, "Expected function argument"); return nullptr;
return nullptr; case ParseeResultsGroupKind::FAILURE:
} hasError = true;
shared_ptr<Token> identifierToken = tokens.at(currentIndex++); break;
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 // arguments
if (tryMatchingTokenKinds({TokenKind::RIGHT_ARROW}, true, true)) { if (!hasError) {
tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::COLON, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::valueTypeParsee(true)
},
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::COMMA, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::valueTypeParsee(true)
},
{}
)
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
for (int i=0; i<resultsGroup.getResults().size(); i+=2) {
pair<string, shared_ptr<ValueType>> arg;
arg.first = resultsGroup.getResults().at(i).getToken()->getLexme();
arg.second = resultsGroup.getResults().at(i+1).getValueType();
arguments.push_back(arg);
}
break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
}
}
returnType = matchValueType(); // return type
if (returnType == nullptr) { if (!hasError) {
markError(TokenKind::TYPE, {}); resultsGroup = parseeResultsGroupForParseeGroup(
return nullptr; ParseeGroup(
{
Parsee::tokenParsee(TokenKind::RIGHT_ARROW, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::valueTypeParsee(true)
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
returnType = resultsGroup.getResults().at(0).getValueType();
break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
} }
} }
@@ -223,10 +323,11 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
} }
// block // block
statementBlock = matchStatementBlock({TokenKind::SEMICOLON}); statementBlock = matchStatementBlock({TokenKind::SEMICOLON, TokenKind::END});
if (statementBlock == nullptr) if (statementBlock == nullptr)
return nullptr; return nullptr;
// closing semicolon
if(!tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true)) { if(!tryMatchingTokenKinds({TokenKind::SEMICOLON}, false, true)) {
markError(TokenKind::SEMICOLON, {}); markError(TokenKind::SEMICOLON, {});
return nullptr; return nullptr;
@@ -236,8 +337,8 @@ shared_ptr<Statement> Parser::matchStatementFunction() {
} }
shared_ptr<Statement> Parser::matchStatementRawFunction() { shared_ptr<Statement> Parser::matchStatementRawFunction() {
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::RAW_FUNCTION}, true, false)) bool hasError = false;
return nullptr; ParseeResultsGroup resultsGroup;
string name; string name;
string constraints; string constraints;
@@ -245,50 +346,118 @@ shared_ptr<Statement> Parser::matchStatementRawFunction() {
shared_ptr<ValueType> returnType = ValueType::NONE; shared_ptr<ValueType> returnType = ValueType::NONE;
string rawSource; string rawSource;
// name // identifier
name = tokens.at(currentIndex++)->getLexme(); resultsGroup = parseeResultsGroupForParseeGroup(
currentIndex++; // skip raw ParseeGroup(
{
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
Parsee::tokenParsee(TokenKind::RAW_FUNCTION, true, false)
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
name = resultsGroup.getResults().at(0).getToken()->getLexme();
break;
case ParseeResultsGroupKind::NO_MATCH:
return nullptr;
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
}
// constraints // constraints
if (!hasError) {
resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::LESS, true, false),
Parsee::tokenParsee(TokenKind::STRING, true, true),
Parsee::tokenParsee(TokenKind::GREATER, true, false)
},
{}
)
);
if (tryMatchingTokenKinds({TokenKind::LESS}, true, true)) { switch (resultsGroup.getKind()) {
if (tokens.at(currentIndex)->isOfKind({TokenKind::STRING})) { case ParseeResultsGroupKind::SUCCESS:
constraints = tokens.at(currentIndex++)->getLexme(); constraints = resultsGroup.getResults().at(0).getToken()->getLexme();
// remove enclosing quotes // remove enclosing quotes
if (constraints.length() >= 2) if (constraints.length() >= 2)
constraints = constraints.substr(1, constraints.length() - 2); constraints = constraints.substr(1, constraints.length() - 2);
break;
case ParseeResultsGroupKind::NO_MATCH:
return nullptr;
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
} }
if (!tryMatchingTokenKinds({TokenKind::GREATER}, true, true))
markError({TokenKind::GREATER}, {});
} }
// arguments // arguments
if (tryMatchingTokenKinds({TokenKind::COLON}, true, true)) { if (!hasError) {
do { resultsGroup = parseeResultsGroupForParseeGroup(
tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line ParseeGroup(
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER, TokenKind::TYPE}, true, false)) { {
markError({}, "Expected function argument"); Parsee::tokenParsee(TokenKind::COLON, true, false),
return nullptr; Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
} Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
shared_ptr<Token> identifierToken = tokens.at(currentIndex++); Parsee::valueTypeParsee(true)
shared_ptr<ValueType> argumentType = matchValueType(); },
if (argumentType == nullptr) { ParseeGroup(
markError(TokenKind::TYPE, {}); {
return nullptr; Parsee::tokenParsee(TokenKind::COMMA, true, false),
} Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
arguments.push_back(pair<string, shared_ptr<ValueType>>(identifierToken->getLexme(), argumentType)); Parsee::valueTypeParsee(true)
} while (tryMatchingTokenKinds({TokenKind::COMMA}, true, true)); },
{}
)
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
for (int i=0; i<resultsGroup.getResults().size(); i+=2) {
pair<string, shared_ptr<ValueType>> arg;
arg.first = resultsGroup.getResults().at(i).getToken()->getLexme();
arg.second = resultsGroup.getResults().at(i+1).getValueType();
arguments.push_back(arg);
}
break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
}
} }
// return type // return type
if (tryMatchingTokenKinds({TokenKind::RIGHT_ARROW}, true, true)) { if (!hasError) {
tryMatchingTokenKinds({TokenKind::NEW_LINE}, true, true); // skip new line resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::RIGHT_ARROW, true, false),
Parsee::tokenParsee(TokenKind::NEW_LINE, false, false),
Parsee::valueTypeParsee(true)
},
{}
)
);
returnType = matchValueType(); switch (resultsGroup.getKind()) {
if (returnType == nullptr) { case ParseeResultsGroupKind::SUCCESS:
markError(TokenKind::TYPE, {}); returnType = resultsGroup.getResults().at(0).getValueType();
return nullptr; break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
hasError = true;
break;
} }
} }
@@ -337,44 +506,90 @@ shared_ptr<Statement> Parser::matchStatementBlock(vector<TokenKind> terminalToke
shared_ptr<Statement> Parser::matchStatementAssignment() { shared_ptr<Statement> Parser::matchStatementAssignment() {
int startIndex = currentIndex; int startIndex = currentIndex;
ParseeResultsGroup resultsGroup;
if (!tryMatchingTokenKinds({TokenKind::IDENTIFIER}, true, false)) string identifier;
return nullptr;
shared_ptr<Token> identifierToken = tokens.at(currentIndex++);
shared_ptr<Expression> indexExpression; shared_ptr<Expression> indexExpression;
shared_ptr<Expression> expression;
if (tryMatchingTokenKinds({TokenKind::LEFT_SQUARE_BRACKET}, true, true)) { // identifier
indexExpression = nextExpression(); resultsGroup = parseeResultsGroupForParseeGroup(
if (indexExpression == nullptr) ParseeGroup(
return nullptr; {
Parsee::tokenParsee(TokenKind::IDENTIFIER, true, true),
},
{}
)
);
if (!tryMatchingTokenKinds({TokenKind::RIGHT_SQUARE_BRACKET}, true, true)) { if (resultsGroup.getKind() != ParseeResultsGroupKind::SUCCESS)
markError(TokenKind::RIGHT_SQUARE_BRACKET, {});
return nullptr;
}
}
// assignment requires left arrow, otherwise abort
if (!tryMatchingTokenKinds({TokenKind::LEFT_ARROW}, true, true)) {
currentIndex = startIndex;
return nullptr;
}
shared_ptr<Expression> expression = nextExpression();
if (expression == nullptr)
return nullptr; return nullptr;
return make_shared<StatementAssignment>(identifierToken->getLexme(), indexExpression, expression); identifier = resultsGroup.getResults().at(0).getToken()->getLexme();
// index expression
resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::LEFT_SQUARE_BRACKET, true, false),
Parsee::expressionParsee(true),
Parsee::tokenParsee(TokenKind::RIGHT_SQUARE_BRACKET, true, false),
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
indexExpression = resultsGroup.getResults().at(0).getExpression();
break;
case ParseeResultsGroupKind::NO_MATCH:
break;
case ParseeResultsGroupKind::FAILURE:
return nullptr;
}
// expression
resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::LEFT_ARROW, true, false),
Parsee::expressionParsee(true)
},
{}
)
);
switch (resultsGroup.getKind()) {
case ParseeResultsGroupKind::SUCCESS:
expression = resultsGroup.getResults().at(0).getExpression();
break;
case ParseeResultsGroupKind::NO_MATCH:
currentIndex = startIndex;
return nullptr;
case ParseeResultsGroupKind::FAILURE:
return nullptr;
}
return make_shared<StatementAssignment>(identifier, indexExpression, expression);
} }
shared_ptr<Statement> Parser::matchStatementReturn() { shared_ptr<Statement> Parser::matchStatementReturn() {
if (!tryMatchingTokenKinds({TokenKind::RETURN}, true, true)) ParseeResultsGroup resultsGroup = parseeResultsGroupForParseeGroup(
ParseeGroup(
{
Parsee::tokenParsee(TokenKind::RETURN, true, false),
Parsee::expressionParsee(false)
},
{}
)
);
if (resultsGroup.getKind() != ParseeResultsGroupKind::SUCCESS)
return nullptr; return nullptr;
shared_ptr<Expression> expression = nextExpression(); shared_ptr<Expression> expression = !resultsGroup.getResults().empty() ? resultsGroup.getResults().at(0).getExpression() : nullptr;
if (expression == nullptr)
return nullptr;
return make_shared<StatementReturn>(expression); return make_shared<StatementReturn>(expression);
} }
@@ -471,7 +686,6 @@ shared_ptr<Expression> Parser::nextExpression() {
if (expression != nullptr || errors.size() > errorsCount) if (expression != nullptr || errors.size() > errorsCount)
return expression; return expression;
markError({}, {});
return nullptr; return nullptr;
} }
@@ -707,6 +921,7 @@ shared_ptr<Expression> Parser::matchExpressionBinary(shared_ptr<Expression> left
} }
if (right == nullptr) { if (right == nullptr) {
markError({}, "Expected expression");
return nullptr; return nullptr;
} else { } else {
return make_shared<ExpressionBinary>(token, left, right); return make_shared<ExpressionBinary>(token, left, right);
@@ -737,44 +952,130 @@ shared_ptr<Expression> Parser::matchExpressionBlock(vector<TokenKind> terminalTo
return make_shared<ExpressionBlock>(statements); return make_shared<ExpressionBlock>(statements);
} }
shared_ptr<ValueType> Parser::matchValueType() { ParseeResultsGroup Parser::parseeResultsGroupForParseeGroup(ParseeGroup group) {
if (!tryMatchingTokenKinds({TokenKind::TYPE}, true, false)) int errorsCount = errors.size();
return nullptr; int startIndex = currentIndex;
shared_ptr<Token> typeToken = tokens.at(currentIndex++); vector<ParseeResult> results;
shared_ptr<ValueType> subType; bool mustFulfill = false;
int valueArg = 0;
if (tryMatchingTokenKinds({TokenKind::LESS}, true, true)) { for (Parsee &parsee : group.getParsees()) {
if (!tryMatchingTokenKinds({TokenKind::TYPE}, true, false)) { optional<ParseeResult> result;
markError(TokenKind::TYPE, {}); switch (parsee.getKind()) {
return nullptr; case ParseeKind::TOKEN:
} result = tokenParseeResult(currentIndex, parsee.getTokenKind());
subType = matchValueType(); break;
if (subType == nullptr) case ParseeKind::VALUE_TYPE:
return subType; result = valueTypeParseeResult(currentIndex);
break;
if (tryMatchingTokenKinds({TokenKind::COMMA}, true, true)) { case ParseeKind::EXPRESSION:
if (!tryMatchingTokenKinds({TokenKind::INTEGER_DEC, TokenKind::INTEGER_HEX, TokenKind::INTEGER_BIN, TokenKind::INTEGER_CHAR}, false, false)) { result = expressionParseeResult(currentIndex);
markError({}, "Expected integer literal"); break;
return nullptr;
}
shared_ptr<Expression> expressionValue = matchExpressionLiteral();
if (expressionValue == nullptr) {
markError({}, "Expected integer literal");
return nullptr;
}
valueArg = dynamic_pointer_cast<ExpressionLiteral>(expressionValue)->getSint32Value();
} }
// generated an error?
if (errors.size() > errorsCount)
return ParseeResultsGroup::failure();
if (!tryMatchingTokenKinds({TokenKind::GREATER}, true, true)) { // if doesn't match on optional group
markError(TokenKind::GREATER, {}); if (!result && parsee.getIsRequired() && !mustFulfill) {
return nullptr; currentIndex = startIndex;
//return vector<ParseeResult>();
return ParseeResultsGroup::noMatch();
} }
// return matching token?
if (result && parsee.getShouldReturn())
results.push_back(*result);
// decide if we're decoding the expected sequence
if (!parsee.getIsRequired() && currentIndex > startIndex)
mustFulfill = true;
// invalid sequence detected?
if (!result && parsee.getIsRequired() && mustFulfill) {
markError(parsee.getTokenKind(), {});
//return {};
return ParseeResultsGroup::failure();
}
// got to the next token if we got a match
if (result)
currentIndex += (*result).getTokensCount();
} }
return ValueType::valueTypeForToken(typeToken, subType, valueArg); if (group.getRepeatedGroup()) {
ParseeResultsGroup subResultsGroup;
do {
subResultsGroup = parseeResultsGroupForParseeGroup(*group.getRepeatedGroup());
if (subResultsGroup.getKind() == ParseeResultsGroupKind::FAILURE)
return ParseeResultsGroup::failure();
for (ParseeResult &subResult : subResultsGroup.getResults())
results.push_back(subResult);
} while (subResultsGroup.getKind() == ParseeResultsGroupKind::SUCCESS);
}
return ParseeResultsGroup::success(results);
}
optional<ParseeResult> Parser::tokenParseeResult(int index, TokenKind tokenKind) {
shared_ptr<Token> token = tokens.at(index);
if (token->isOfKind({tokenKind}))
return ParseeResult::tokenResult(token);
return {};
}
optional<ParseeResult> Parser::valueTypeParseeResult(int index) {
int startIndex = index;
if (!tokens.at(index)->isOfKind({TokenKind::TYPE}))
return {};
shared_ptr<Token> typeToken = tokens.at(index++);
shared_ptr<ValueType> subType;
int typeArg = 0;
if (tokens.at(index)->isOfKind({TokenKind::LESS})) {
index++;
optional<ParseeResult> subResult = valueTypeParseeResult(index);
if (!subResult)
return {};
subType = (*subResult).getValueType();
index += (*subResult).getTokensCount();
if (tokens.at(index)->isOfKind({TokenKind::COMMA})) {
index++;
if (!tokens.at(index)->isOfKind({TokenKind::INTEGER_DEC, TokenKind::INTEGER_HEX, TokenKind::INTEGER_BIN, TokenKind::INTEGER_CHAR}))
return {};
int storedIndex = currentIndex;
currentIndex = index;
shared_ptr<Expression> expressionValue = matchExpressionLiteral();
typeArg = dynamic_pointer_cast<ExpressionLiteral>(expressionValue)->getSint32Value();
currentIndex = storedIndex;
index++;
}
if (!tokens.at(index)->isOfKind({TokenKind::GREATER}))
return {};
index++;
}
shared_ptr<ValueType> valueType = ValueType::valueTypeForToken(typeToken, subType, typeArg);
return ParseeResult::valueTypeResult(valueType, index - startIndex);
}
optional<ParseeResult> Parser::expressionParseeResult(int index) {
int startIndex = currentIndex;
int errorsCount = errors.size();
shared_ptr<Expression> expression = nextExpression();
if (errors.size() > errorsCount)
return {};
int tokensCount = currentIndex - startIndex;
currentIndex = startIndex;
return ParseeResult::expressionResult(expression, tokensCount);
} }
bool Parser::tryMatchingTokenKinds(vector<TokenKind> kinds, bool shouldMatchAll, bool shouldAdvance) { bool Parser::tryMatchingTokenKinds(vector<TokenKind> kinds, bool shouldMatchAll, bool shouldAdvance) {

View File

@@ -12,6 +12,10 @@ class ValueType;
class Expression; class Expression;
class Statement; class Statement;
class ParseeGroup;
class ParseeResult;
class ParseeResultsGroup;
using namespace std; using namespace std;
class Parser { class Parser {
@@ -50,8 +54,10 @@ private:
shared_ptr<Expression> matchExpressionBinary(shared_ptr<Expression> left); shared_ptr<Expression> matchExpressionBinary(shared_ptr<Expression> left);
shared_ptr<Expression> matchExpressionBlock(vector<TokenKind> terminalTokenKinds); shared_ptr<Expression> matchExpressionBlock(vector<TokenKind> terminalTokenKinds);
shared_ptr<ValueType> matchValueType(); ParseeResultsGroup parseeResultsGroupForParseeGroup(ParseeGroup group);
optional<ParseeResult> tokenParseeResult(int index, TokenKind tokenKind);
optional<ParseeResult> valueTypeParseeResult(int index);
optional<ParseeResult> expressionParseeResult(int index);
bool tryMatchingTokenKinds(vector<TokenKind> kinds, bool shouldMatchAll, bool shouldAdvance); bool tryMatchingTokenKinds(vector<TokenKind> kinds, bool shouldMatchAll, bool shouldAdvance);
void markError(optional<TokenKind> expectedTokenKind, optional<string> message); void markError(optional<TokenKind> expectedTokenKind, optional<string> message);