Huge refactoring of builtin features into an external plugin

This commit is contained in:
WerWolv
2021-01-22 18:01:39 +01:00
parent 9bc569bf9a
commit 104000fbc4
30 changed files with 1095 additions and 842 deletions

View File

@@ -76,7 +76,7 @@ namespace hex {
/* Pattern Language Functions */
void ContentRegistry::PatternLanguageFunctions::add(std::string_view name, u32 parameterCount, const std::function<hex::lang::ASTNode*(hex::lang::LogConsole&, std::vector<hex::lang::ASTNode*>)> &func) {
void ContentRegistry::PatternLanguageFunctions::add(std::string_view name, u32 parameterCount, const std::function<hex::lang::ASTNode*(hex::lang::Evaluator&, std::vector<hex::lang::ASTNode*>)> &func) {
getEntries()[name.data()] = Function{ parameterCount, func };
}

View File

@@ -1,182 +1,10 @@
#include <hex/lang/evaluator.hpp>
#include <hex.hpp>
#include <hex/api/content_registry.hpp>
namespace hex::lang {
#define LITERAL_COMPARE(literal, cond) std::visit([&, this](auto &&literal) { return (cond) != 0; }, literal)
void Evaluator::registerBuiltinFunctions() {
/* findSequence */
ContentRegistry::PatternLanguageFunctions::add("findSequence", ContentRegistry::PatternLanguageFunctions::MoreParametersThan | 1, [this](auto &console, auto params) {
auto& occurrenceIndex = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
std::vector<u8> sequence;
for (u32 i = 1; i < params.size(); i++) {
sequence.push_back(std::visit([&](auto &&value) -> u8 {
if (value <= 0xFF)
return value;
else
console.abortEvaluation("sequence bytes need to fit into 1 byte");
}, asType<ASTNodeIntegerLiteral>(params[i])->getValue()));
}
std::vector<u8> bytes(sequence.size(), 0x00);
u32 occurrences = 0;
for (u64 offset = 0; offset < this->m_provider->getSize() - sequence.size(); offset++) {
this->m_provider->read(offset, bytes.data(), bytes.size());
if (bytes == sequence) {
if (LITERAL_COMPARE(occurrenceIndex, occurrences < occurrenceIndex)) {
occurrences++;
continue;
}
return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, offset });
}
}
console.abortEvaluation("failed to find sequence");
});
/* assert */
ContentRegistry::PatternLanguageFunctions::add("readUnsigned", 2, [this](auto &console, auto params) {
auto address = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
auto size = asType<ASTNodeIntegerLiteral>(params[1])->getValue();
if (LITERAL_COMPARE(address, address >= this->m_provider->getActualSize()))
console.abortEvaluation("address out of range");
return std::visit([&, this](auto &&address, auto &&size) {
if (size <= 0 || size > 16)
console.abortEvaluation("invalid read size");
u8 value[(u8)size];
this->m_provider->read(address, value, size);
switch ((u8)size) {
case 1: return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned8Bit, hex::changeEndianess(*reinterpret_cast<u8*>(value), 1, this->getCurrentEndian()) });
case 2: return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned16Bit, hex::changeEndianess(*reinterpret_cast<u16*>(value), 2, this->getCurrentEndian()) });
case 4: return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned32Bit, hex::changeEndianess(*reinterpret_cast<u32*>(value), 4, this->getCurrentEndian()) });
case 8: return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, hex::changeEndianess(*reinterpret_cast<u64*>(value), 8, this->getCurrentEndian()) });
case 16: return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned128Bit, hex::changeEndianess(*reinterpret_cast<u128*>(value), 16, this->getCurrentEndian()) });
default: console.abortEvaluation("invalid read size");
}
}, address, size);
});
ContentRegistry::PatternLanguageFunctions::add("readSigned", 2, [this](auto &console, auto params) {
auto address = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
auto size = asType<ASTNodeIntegerLiteral>(params[1])->getValue();
if (LITERAL_COMPARE(address, address >= this->m_provider->getActualSize()))
console.abortEvaluation("address out of range");
return std::visit([&, this](auto &&address, auto &&size) {
if (size <= 0 || size > 16)
console.abortEvaluation("invalid read size");
u8 value[(u8)size];
this->m_provider->read(address, value, size);
switch ((u8)size) {
case 1: return new ASTNodeIntegerLiteral({ Token::ValueType::Signed8Bit, hex::changeEndianess(*reinterpret_cast<s8*>(value), 1, this->getCurrentEndian()) });
case 2: return new ASTNodeIntegerLiteral({ Token::ValueType::Signed16Bit, hex::changeEndianess(*reinterpret_cast<s16*>(value), 2, this->getCurrentEndian()) });
case 4: return new ASTNodeIntegerLiteral({ Token::ValueType::Signed32Bit, hex::changeEndianess(*reinterpret_cast<s32*>(value), 4, this->getCurrentEndian()) });
case 8: return new ASTNodeIntegerLiteral({ Token::ValueType::Signed64Bit, hex::changeEndianess(*reinterpret_cast<s64*>(value), 8, this->getCurrentEndian()) });
case 16: return new ASTNodeIntegerLiteral({ Token::ValueType::Signed128Bit, hex::changeEndianess(*reinterpret_cast<s128*>(value), 16, this->getCurrentEndian()) });
default: console.abortEvaluation("invalid read size");
}
}, address, size);
});
ContentRegistry::PatternLanguageFunctions::add("assert", 2, [this](auto &console, auto params) {
auto condition = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
auto message = asType<ASTNodeStringLiteral>(params[1])->getString();
if (LITERAL_COMPARE(condition, condition == 0))
console.abortEvaluation(hex::format("assert failed \"%s\"", message.data()));
return nullptr;
});
ContentRegistry::PatternLanguageFunctions::add("warnAssert", 2, [this](auto console, auto params) {
auto condition = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
auto message = asType<ASTNodeStringLiteral>(params[1])->getString();
if (LITERAL_COMPARE(condition, condition == 0))
console.log(LogConsole::Level::Warning, hex::format("assert failed \"%s\"", message.data()));
return nullptr;
});
ContentRegistry::PatternLanguageFunctions::add("print", ContentRegistry::PatternLanguageFunctions::MoreParametersThan | 0, [](auto &console, auto params) {
std::string message;
for (auto& param : params) {
if (auto integerLiteral = dynamic_cast<ASTNodeIntegerLiteral*>(param); integerLiteral != nullptr) {
switch (integerLiteral->getType()) {
case Token::ValueType::Character: message += std::get<s8>(integerLiteral->getValue()); break;
case Token::ValueType::Unsigned8Bit: message += std::to_string(std::get<u8>(integerLiteral->getValue())); break;
case Token::ValueType::Signed8Bit: message += std::to_string(std::get<s8>(integerLiteral->getValue())); break;
case Token::ValueType::Unsigned16Bit: message += std::to_string(std::get<u16>(integerLiteral->getValue())); break;
case Token::ValueType::Signed16Bit: message += std::to_string(std::get<s16>(integerLiteral->getValue())); break;
case Token::ValueType::Unsigned32Bit: message += std::to_string(std::get<u32>(integerLiteral->getValue())); break;
case Token::ValueType::Signed32Bit: message += std::to_string(std::get<s32>(integerLiteral->getValue())); break;
case Token::ValueType::Unsigned64Bit: message += std::to_string(std::get<u64>(integerLiteral->getValue())); break;
case Token::ValueType::Signed64Bit: message += std::to_string(std::get<s64>(integerLiteral->getValue())); break;
case Token::ValueType::Unsigned128Bit: message += hex::to_string(std::get<u128>(integerLiteral->getValue())); break;
case Token::ValueType::Signed128Bit: message += hex::to_string(std::get<s128>(integerLiteral->getValue())); break;
case Token::ValueType::Float: message += std::to_string(std::get<float>(integerLiteral->getValue())); break;
case Token::ValueType::Double: message += std::to_string(std::get<double>(integerLiteral->getValue())); break;
case Token::ValueType::Boolean: message += std::get<s32>(integerLiteral->getValue()) ? "true" : "false"; break;
case Token::ValueType::CustomType: message += "< Custom Type >"; break;
}
}
else if (auto stringLiteral = dynamic_cast<ASTNodeStringLiteral*>(param); stringLiteral != nullptr)
message += stringLiteral->getString();
}
console.log(LogConsole::Level::Info, message);
return nullptr;
});
ContentRegistry::PatternLanguageFunctions::add("addressof", 1, [this](auto &console, auto params) -> ASTNode* {
auto name = asType<ASTNodeStringLiteral>(params[0])->getString();
std::vector<std::string> path = splitString(name, ".");
auto pattern = this->patternFromName(path);
return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, u64(pattern->getOffset()) });
});
ContentRegistry::PatternLanguageFunctions::add("sizeof", 1, [this](auto &console, auto params) -> ASTNode* {
auto name = asType<ASTNodeStringLiteral>(params[0])->getString();
std::vector<std::string> path = splitString(name, ".");
auto pattern = this->patternFromName(path);
return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, u64(pattern->getSize()) });
});
ContentRegistry::PatternLanguageFunctions::add("nextAfter", 1, [this](auto &console, auto params) -> ASTNode* {
auto name = asType<ASTNodeStringLiteral>(params[0])->getString();
std::vector<std::string> path = splitString(name, ".");
auto pattern = this->patternFromName(path);
return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, u64(pattern->getOffset() + pattern->getSize()) });
});
ContentRegistry::PatternLanguageFunctions::add("alignTo", 2, [this](auto &console, auto params) -> ASTNode* {
auto alignment = asType<ASTNodeIntegerLiteral>(params[0])->getValue();
auto value = asType<ASTNodeIntegerLiteral>(params[1])->getValue();
auto result = std::visit([](auto &&alignment, auto &&value) {
u64 remainder = u64(value) % u64(alignment);
return remainder != 0 ? u64(value) + (u64(alignment) - remainder) : u64(value);
}, alignment, value);
return new ASTNodeIntegerLiteral({ Token::ValueType::Unsigned64Bit, u64(result) });
});
}
}

View File

@@ -13,8 +13,6 @@ namespace hex::lang {
Evaluator::Evaluator(prv::Provider* &provider, std::endian defaultDataEndian)
: m_provider(provider), m_defaultDataEndian(defaultDataEndian) {
this->registerBuiltinFunctions();
}
ASTNodeIntegerLiteral* Evaluator::evaluateScopeResolution(ASTNodeScopeResolution *node) {
@@ -157,7 +155,7 @@ namespace hex::lang {
this->getConsole().abortEvaluation(hex::format("invalid number of parameters for function '%s'. Expected %d", node->getFunctionName().data(), function.parameterCount));
}
return function.func(this->getConsole(), evaluatedParams);
return function.func(*this, evaluatedParams);
}
#define FLOAT_BIT_OPERATION(name) \

View File

@@ -420,7 +420,7 @@ namespace hex::lang {
if (MATCHES(sequence(VALUETYPE_PADDING, SEPARATOR_SQUAREBRACKETOPEN)))
member = parsePadding();
else if (MATCHES((optional(KEYWORD_BE), optional(KEYWORD_LE)) && variant(IDENTIFIER, VALUETYPE_ANY) && sequence(IDENTIFIER, SEPARATOR_SQUAREBRACKETOPEN) && none(sequence(SEPARATOR_SQUAREBRACKETOPEN))))
else if (MATCHES((optional(KEYWORD_BE), optional(KEYWORD_LE)) && variant(IDENTIFIER, VALUETYPE_ANY) && sequence(IDENTIFIER, SEPARATOR_SQUAREBRACKETOPEN) && sequence<Not>(SEPARATOR_SQUAREBRACKETOPEN)))
member = parseMemberArrayVariable();
else if (MATCHES((optional(KEYWORD_BE), optional(KEYWORD_LE)) && variant(IDENTIFIER, VALUETYPE_ANY) && sequence(IDENTIFIER)))
member = parseMemberVariable();

View File

@@ -0,0 +1,120 @@
#include <hex/lang/pattern_language.hpp>
#include <hex/providers/provider.hpp>
#include <hex/lang/preprocessor.hpp>
#include <hex/lang/lexer.hpp>
#include <hex/lang/parser.hpp>
#include <hex/lang/validator.hpp>
#include <hex/lang/evaluator.hpp>
#include <hex/lang/pattern_data.hpp>
#include <unistd.h>
namespace hex::lang {
PatternLanguage::PatternLanguage(prv::Provider *provider) : m_provider(provider) {
this->m_preprocessor = new Preprocessor();
this->m_lexer = new Lexer();
this->m_parser = new Parser();
this->m_validator = new Validator();
this->m_evaluator = new Evaluator(provider);
}
PatternLanguage::~PatternLanguage() {
delete this->m_preprocessor;
delete this->m_lexer;
delete this->m_parser;
delete this->m_validator;
delete this->m_evaluator;
}
std::optional<std::vector<PatternData*>> PatternLanguage::executeString(std::string_view string) {
this->m_currError.reset();
this->m_evaluator->getConsole().clear();
hex::lang::Preprocessor preprocessor;
std::endian defaultEndian;
preprocessor.addPragmaHandler("endian", [&defaultEndian](std::string value) {
if (value == "big") {
defaultEndian = std::endian::big;
return true;
} else if (value == "little") {
defaultEndian = std::endian::little;
return true;
} else if (value == "native") {
defaultEndian = std::endian::native;
return true;
} else
return false;
});
preprocessor.addDefaultPragmaHandlers();
auto preprocessedCode = preprocessor.preprocess(string.data());
if (!preprocessedCode.has_value()) {
this->m_currError = preprocessor.getError();
return { };
}
hex::lang::Lexer lexer;
auto tokens = lexer.lex(preprocessedCode.value());
if (!tokens.has_value()) {
this->m_currError = lexer.getError();
return { };
}
hex::lang::Parser parser;
auto ast = parser.parse(tokens.value());
if (!ast.has_value()) {
this->m_currError = parser.getError();
return { };
}
SCOPE_EXIT( for(auto &node : ast.value()) delete node; );
hex::lang::Validator validator;
auto validatorResult = validator.validate(ast.value());
if (!validatorResult) {
this->m_currError = validator.getError();
return { };
}
auto provider = SharedData::currentProvider;
hex::lang::Evaluator evaluator(provider, defaultEndian);
auto patternData = evaluator.evaluate(ast.value());
if (!patternData.has_value())
return { };
return patternData.value();
}
std::optional<std::vector<PatternData*>> PatternLanguage::executeFile(std::string_view path) {
FILE *file = fopen(path.data(), "r");
if (file == nullptr)
return { };
fseek(file, 0, SEEK_END);
size_t size = ftell(file);
rewind(file);
std::string code(size + 1, 0x00);
fread(code.data(), size, 1, file);
fclose(file);
return this->executeString(code);
}
std::vector<std::pair<LogConsole::Level, std::string>> PatternLanguage::getConsoleLog() {
return this->m_evaluator->getConsole().getLog();
}
std::optional<std::pair<u32, std::string>> PatternLanguage::getError() {
return this->m_currError;
}
}