From a53568bf5de7a854c0bb4432d317ae976d45b5c2 Mon Sep 17 00:00:00 2001
From: Zoe Roux <zoe.roux@sdg.moe>
Date: Wed, 17 Feb 2021 17:10:35 +0100
Subject: [PATCH] Finishing the WlaDx label's parser

---
 CMakeLists.txt                               |  2 +
 sources/Debugger/CPU/CPUDebug.cpp            |  7 +-
 sources/Debugger/CPU/SymbolLoaders/WlaDx.cpp | 72 +++++++++++++++++---
 sources/Debugger/CPU/SymbolLoaders/WlaDx.hpp | 19 +++++-
 sources/Utility/Utility.cpp                  |  2 +
 sources/Utility/Utility.hpp                  | 13 ++--
 6 files changed, 94 insertions(+), 21 deletions(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index e576844..034550a 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -7,6 +7,8 @@ project(ComSquare)
 # show compilation warnings
 add_compile_options(-W -Wall -Wextra -Wshadow)
 
+include_directories(sources)
+
 # make unit tests
 add_executable(unit_tests
         tests/CPU/testAddressingMode.cpp
diff --git a/sources/Debugger/CPU/CPUDebug.cpp b/sources/Debugger/CPU/CPUDebug.cpp
index cd8cb22..61e848c 100644
--- a/sources/Debugger/CPU/CPUDebug.cpp
+++ b/sources/Debugger/CPU/CPUDebug.cpp
@@ -3,8 +3,9 @@
 //
 
 #include "CPUDebug.hpp"
-#include "../../Utility/Utility.hpp"
-#include "../../Exceptions/InvalidOpcode.hpp"
+#include "Utility/Utility.hpp"
+#include "Exceptions/InvalidOpcode.hpp"
+#include "SymbolLoaders/WlaDx.hpp"
 #include <QtEvents>
 #include <QPainter>
 #include <iostream>
@@ -316,7 +317,7 @@ namespace ComSquare::Debugger
 		std::ifstream sym(symbolPath);
 
 		if (sym) {
-			std::vector<Label> symLabels = WlaDx.parse(sym);
+			std::vector<Label> symLabels = WlaDx::parse(sym);
 			labels.insert(labels.end(),
 			              std::make_move_iterator(symLabels.begin()),
 			              std::make_move_iterator(symLabels.end()));
diff --git a/sources/Debugger/CPU/SymbolLoaders/WlaDx.cpp b/sources/Debugger/CPU/SymbolLoaders/WlaDx.cpp
index c38f05a..ab439c0 100644
--- a/sources/Debugger/CPU/SymbolLoaders/WlaDx.cpp
+++ b/sources/Debugger/CPU/SymbolLoaders/WlaDx.cpp
@@ -3,23 +3,79 @@
 //
 
 #include <regex>
+#include "Utility/Utility.hpp"
 #include "WlaDx.hpp"
 
 namespace ComSquare::Debugger
 {
-	std::vector<Label> WlaDx::parse(std::fstream symbolFile)
+	std::vector<Label> WlaDx::parse(std::ifstream &symbolFile)
 	{
-		std::vector<Label> labels;
+		std::vector<Label> ret;
+		std::map<std::string, Label> labels;
 		std::string line;
+		std::smatch match;
+		std::regex re(R"(\[(\S+)\])");
 
-		while (std::getline(symbolFile, line)) {
-			line = line.substr(0, line.rfind(';'));
-			std::smatch match;
-			std::regex re("\\[\\S+\\]");
+		while (symbolFile) {
+			if (line.empty()) {
+				std::getline(symbolFile, line);
+				line = WlaDx::_cleanLine(line);
+			}
+			if (!std::regex_match(line, match, re))
+				continue;
 
-			if (!std::regex_search(line, match, re)
+			if (match[1] == "labels")
+				line = WlaDx::_parseLabels(labels, symbolFile);
+			else if (match[1] == "definitions")
+				line = WlaDx::_parseDefinitions(labels, symbolFile);
 		}
-		return labels;
+		std::transform(labels.begin(), labels.end(), std::back_inserter(ret), [](auto &i){return i.second;});
+		return ret;
+	}
+
+	std::string WlaDx::_parseLabels(std::map<std::string, Label> &labels, std::ifstream &f)
+	{
+		std::string line;
+		std::smatch match;
+		std::regex re(R"(([0-9a-fA-F]{2}):([0-9a-fA-F]{4}) (.*))");
+
+		while (std::getline(f, line)) {
+			line = WlaDx::_cleanLine(line);
+			if (!std::regex_match(line, match, re))
+				return line;
+			labels[match[3]].name = match[3];
+			labels[match[3]].address = std::stoi(match[1], nullptr, 16) << 16
+			                           | std::stoi(match[2], nullptr, 16);
+		}
+		return "";
+	}
+
+	std::string WlaDx::_parseDefinitions(std::map<std::string, Label> &labels, std::ifstream &f)
+	{
+		std::string line;
+		std::smatch match;
+		std::regex re(R"(([0-9a-fA-F]{8}) (.*))");
+		std::string name;
+
+		while (std::getline(f, line)) {
+			line = WlaDx::_cleanLine(line);
+			if (!std::regex_match(line, match, re))
+				return line;
+			// TODO support other definitions.
+			if (!match[2].str().starts_with("_sizeof_"))
+				continue;
+			name = match[2].str().substr(std::string("_sizeof_").size());
+			labels[name].size = std::stoi(match[1], nullptr, 16);
+		}
+		return "";
+	}
+
+	std::string WlaDx::_cleanLine(std::string line)
+	{
+		line = line.substr(0, line.rfind(';'));
+		line.erase(0, line.find_first_not_of(Utility::WHITESPACES));
+		line.erase(line.find_last_not_of(Utility::WHITESPACES) + 1);
+		return line;
 	}
 }
 
diff --git a/sources/Debugger/CPU/SymbolLoaders/WlaDx.hpp b/sources/Debugger/CPU/SymbolLoaders/WlaDx.hpp
index b779341..f8ead36 100644
--- a/sources/Debugger/CPU/SymbolLoaders/WlaDx.hpp
+++ b/sources/Debugger/CPU/SymbolLoaders/WlaDx.hpp
@@ -12,6 +12,23 @@ namespace ComSquare::Debugger
 	//! @brief Class to parse WLA-DX symbol files.
 	class WlaDx {
 	public:
-		static std::vector<Label> parse(std::fstream symbolFile);
+		//! @brief Parse the whole file
+		//! @param symbolFile The file to parse (opened but not seeked)
+		//! @return The list of symbol found.
+		static std::vector<Label> parse(std::ifstream &symbolFile);
+		//! @brief Parse labels name & addresses (the [labels] section)
+		//! @param labels An list of labels. If one is already defined, the address will be set.
+		//! @param symbolFile The file to parse, opened and seeked to the line just after the [labels] line.
+		//! @return The line after the label section (the first line that doesn't match the section's regex).
+		static std::string _parseLabels(std::map<std::string, Label> &labels, std::ifstream &symbolFile);
+		//! @brief Parse definitions (and size of labels) (the [definitions] section)
+		//! @param labels An list of labels. If one is already defined, the size will be set.
+		//! @param symbolFile The file to parse, opened and seeked to the line just after the [definitions] line.
+		//! @return The line after the definitions section (the first line that doesn't match the section's regex).
+		static std::string _parseDefinitions(std::map<std::string, Label> &labels, std::ifstream &symbolFile);
+		//! @brief Remove comments, indentation & trailing spaces from a line.
+		//! @param line The line to clean
+		//! @return The line cleaned up.
+		static std::string _cleanLine(std::string line);
 	};
 }
\ No newline at end of file
diff --git a/sources/Utility/Utility.cpp b/sources/Utility/Utility.cpp
index 66500b7..c7a37b0 100644
--- a/sources/Utility/Utility.cpp
+++ b/sources/Utility/Utility.cpp
@@ -55,4 +55,6 @@ namespace ComSquare::Utility
 	{
 		return std::bitset<24>(i).to_string();
 	}
+
+	const std::string WHITESPACES = " \t\n\r\f\v";
 }
\ No newline at end of file
diff --git a/sources/Utility/Utility.hpp b/sources/Utility/Utility.hpp
index 78b1ced..94b8268 100644
--- a/sources/Utility/Utility.hpp
+++ b/sources/Utility/Utility.hpp
@@ -2,13 +2,12 @@
 // Created by anonymus-raccoon on 2/16/20.
 //
 
-#ifndef COMSQUARE_UTILITY_HPP
-#define COMSQUARE_UTILITY_HPP
+#pragma once
 
 #include <string>
 #include <ios>
 #include <sstream>
-#include "../Models/Int24.hpp"
+#include "Models/Int24.hpp"
 
 namespace ComSquare::Utility
 {
@@ -19,16 +18,12 @@ namespace ComSquare::Utility
 	};
 
 	std::string to_hex(uint8_t i, HexString prefix = AsmPrefix);
-
 	std::string to_hex(uint16_t i, HexString prefix = AsmPrefix);
-
 	std::string to_hex(uint24_t i, HexString prefix = AsmPrefix);
 
 	std::string to_binary(uint8_t i);
-
 	std::string to_binary(uint16_t i);
-
 	std::string to_binary(uint24_t i);
-}
 
-#endif //COMSQUARE_UTILITY_HPP
+	extern const std::string WHITESPACES;
+}
\ No newline at end of file