diff options
1 files changed, 212 insertions, 0 deletions
| -rw-r--r-- | spasm/src/asm/assembler.cpp | 212 |
diff --git a/spasm/src/asm/assembler.cpp b/spasm/src/asm/assembler.cpp new file mode 100644 index 0000000..1f69b1a --- /dev/null +++ b/spasm/src/asm/assembler.cpp @@ -0,0 +1,212 @@ +#include <cassert> + +#include "assembler.hpp" +#include "bytecode.hpp" +#include "symbol.hpp" +#include "tokenizer.hpp" + +namespace SpasmImpl +{ +namespace ASM +{ +Assembler::Assembler(Lexer::Tokenizer& tokenizer, Bytecode_Stream& bytecode) + : _tokenizer(&tokenizer), _bytecode(&bytecode) +{ +} + +int get_arg_size(const Lexer::Token args[]) +{ + int size = 0; + for (int i = 0; i < 3; ++i) + { + switch (args[i].type()) + { + case Lexer::Token::Integer: + case Lexer::Token::XInteger: + { + if (args[i].value_int() > 0xffffffff) + { + size = std::max(size, 3); + } + else if (args[i].value_int() > 0xffff) + { + size = std::max(size, 2); + } + else if (args[i].value_int() > 0xff) + { + size = std::max(size, 1); + } + else if (args[i].value_int() < -0xffffffffLL) + { + size = std::max(size, 3); + } + else if (args[i].value_int() < -0xffff) + { + size = std::max(size, 2); + } + else if (args[i].value_int() < -0xff) + { + size = std::max(size, 1); + } + break; + } + case Lexer::Token::FloatingPoint: + { + size = std::max(size, 3); + break; + } + case Lexer::Token::StringValue: + { + const auto length = args[i].value_str().size(); + if (length > 0xffffffff) + { + size = std::max(size, 3); + } + else if (length > 0xffff) + { + size = std::max(size, 2); + } + else if (length > 0xff) + { + size = std::max(size, 1); + } + break; + } + default: + break; + } + } + return size; +} + +void Assembler::assemble() +{ + Lexer::Token token = _tokenizer->next_token(); + + while (token.type() != Lexer::Token::EndInput) + { + const auto type = token.type(); + if (type >= Lexer::Token::_NotOpCodeBegin) + { + switch (token.type()) + { + case Lexer::Token::Ident: + assemble_identifier(token); + break; + case Lexer::Token::Label: + token = _tokenizer->next_token(); + assert(token.type() == Lexer::Token::Ident); + _symbols.define(token.value_str(), _bytecode->size()); + break; + default: + break; + } + } + else + { + Lexer::Token args[3]; + if (type >= Lexer::Token::_OneArgBegin) + { + args[0] = _tokenizer->next_token(); + } + if (type >= Lexer::Token::_TwoArgBegin) + { + args[1] = _tokenizer->next_token(); + } + if (type >= Lexer::Token::_ThreeArgBegin) + { + args[2] = _tokenizer->next_token(); + } + const auto size = get_arg_size(args); + _bytecode->push_opcode( + (Bytecode_Stream::Opcode_t)((size << 6) | token.type())); + const auto arg_size = 1 << size; + + if (args[0].type() != Lexer::Token::NotUsed) + { + if (type == Lexer::Token::Call || type == Lexer::Token::Jump) + { + assert(args[0].type() == Lexer::Token::Ident); + assemble_identifier(args[0]); + } + else + { + assert(args[0].type() == Lexer::Token::Integer); + _bytecode->push_integer(args[0].value_int(), arg_size); + } + } + if (args[1].type() != Lexer::Token::NotUsed) + { + if (type == Lexer::Token::JumpF || type == Lexer::Token::JumpT) + { + assert(args[1].type() == Lexer::Token::Ident); + assemble_identifier(args[1]); + } + else if (type == Lexer::Token::String) + { + assert(args[1].type() == Lexer::Token::StringValue); + const auto& s = args[1].value_str(); + _bytecode->push_string(s.data(), s.size(), arg_size); + } + else + { + if (args[1].type() == Lexer::Token::Integer) + { + _bytecode->push_integer(args[1].value_int(), arg_size); + } + else if (args[1].type() == Lexer::Token::FloatingPoint) + { + _bytecode->push_double(args[1].value_double()); + } + else + { + assert(false && "expected a number type"); + } + } + } + if (args[2].type() != Lexer::Token::NotUsed) + { + assert(args[2].type() == Lexer::Token::Integer); + _bytecode->push_integer(args[2].value_int(), arg_size); + } + } + + token = _tokenizer->next_token(); + } + + Symbol_Table::const_iterator i = _symbols.begin(); + while (i != _symbols.end()) + { + backpatch(i->second); + ++i; + } +} + +void Assembler::backpatch(const Symbol* symbol) +{ + Symbol::Positions_t::const_iterator i = symbol->positions_begin(); + while (i != symbol->positions_end()) + { + _bytecode->set_location(*i, symbol->definition()); + ++i; + } +} + +void Assembler::assemble_identifier(const Lexer::Token& token) +{ + Symbol* symbol = _symbols.insert(token.value_str(), _bytecode->size()); + _bytecode->push_location(symbol->definition()); +} + +bool compile(std::istream& istr, Bytecode_Stream& bytecode) +{ + Lexer::Tokenizer tokenizer(istr); + Assembler assembler(tokenizer, bytecode); + + assembler.assemble(); + + return true; +} + +} // namespace ASM +} // namespace SpasmImpl |
