UserScript/tests/parser_test.cpp

#include "UserScript/parser.h"
#include <catch2/catch_test_macros.hpp>
#include <ranges>
#include <iostream>
#include <fstream>
#include <sstream>
#include <random>
#include <algorithm>

TEST_CASE("Can parse") {
	
	std::string code = "/salad (/potato) 12 13 \"hello\" ident\n"
	                   "/salad 12 13 \"hello\" ident\n"
	                   "if(/test)\n"
	                   "    /nice\n"
	                   "endif";
	std::vector<scripting::script_error> errors;
	auto lexed = scripting::ast::lex(code, errors);
	auto parsed = scripting::ast::parse(lexed, errors);
	
	if(not errors.empty()) {
		for(auto& line : errors) {
			std::cout << line.message << "\n    at line " << line.location->line_number << ":" << line.location->column_number << "\n";
			std::cout << "    " << *line.location->line_contents << "\n";
			std::cout << "    " << std::string(line.location->column_number - 1, ' ') << "^\n";
		}
	}
	
	auto& block =  parsed;
	REQUIRE(block.contents.size() == 3);
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::command_expression>>(block.contents.front().contents));
	auto& cmd1 = std::get<std::unique_ptr<scripting::ast::command_expression>>(block.contents.front().contents);
	REQUIRE(cmd1->name.value == "salad");
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::command_expression>>(std::span(block.contents).subspan(1).front().contents));
	auto& cmd2 = std::get<std::unique_ptr<scripting::ast::command_expression>>(std::span(block.contents).subspan(1).front().contents);
	REQUIRE(cmd2->name.value == "salad");
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::conditional>>(std::span(block.contents).subspan(2).front().contents));
	auto& conditional = std::get<std::unique_ptr<scripting::ast::conditional>>(std::span(block.contents).subspan(2).front().contents);
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::paren_expression>>(conditional->condition->contents));
	auto& paren = std::get<std::unique_ptr<scripting::ast::paren_expression>>(conditional->condition->contents)->content;
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::command_expression>>(paren));
	auto& condition = std::get<std::unique_ptr<scripting::ast::command_expression>>(paren);
	REQUIRE(condition->name.value == "test");
}

TEST_CASE("Can parse 2") {
	
	std::string code = "/salad (/potato) 12+13*16/(/potato)+myvar \"hello\" ident\n"
	                   "/salad 12 13 \"hello\" ident\n"
	                   "if !(/test)\n"
	                   "    /nice\n"
	                   "endif";
	std::vector<scripting::script_error> errors;
	auto lexed = scripting::ast::lex(code, errors);
	auto parsed = scripting::ast::parse(lexed, errors);
	
	if(not errors.empty()) {
		for(auto& line : errors) {
			std::cout << line.message << "\n    at line " << line.location->line_number << ":" << line.location->column_number << "\n";
			std::cout << "    " << *line.location->line_contents << "\n";
			std::cout << "    " << std::string(line.location->column_number - 1, ' ') << "^\n";
		}
	}
	
	auto& block =  parsed;
	REQUIRE(block.contents.size() == 3);
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::command_expression>>(block.contents.front().contents));
	auto& cmd1 = std::get<std::unique_ptr<scripting::ast::command_expression>>(block.contents.front().contents);
	REQUIRE(cmd1->name.value == "salad");
	REQUIRE(cmd1->arguments.size() == 4);
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::command_expression>>(std::span(block.contents).subspan(1).front().contents));
	auto& cmd2 = std::get<std::unique_ptr<scripting::ast::command_expression>>(std::span(block.contents).subspan(1).front().contents);
	REQUIRE(cmd2->name.value == "salad");
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::conditional>>(std::span(block.contents).subspan(2).front().contents));
	auto& conditional = std::get<std::unique_ptr<scripting::ast::conditional>>(std::span(block.contents).subspan(2).front().contents);
	REQUIRE(std::holds_alternative<std::unique_ptr<scripting::ast::unary_algebraic_expression>>(conditional->condition->contents));
}

template<auto seed_template = -1>
constexpr auto runner = [](){
	std::vector<std::string> sources = {
		"../tests/scripts/001.script",
		"../tests/scripts/002.script",
		"../tests/scripts/003.script",
		"../tests/scripts/004.script",
		"../tests/scripts/005.script",
		"../tests/scripts/006.script",
		"../tests/scripts/007.script",
		"../tests/scripts/008.script",
	};
	
	auto seed = seed_template == -1 ? std::random_device{}() : seed_template;
	
	std::cout << "TEST \"Try to crash the parser 1\" with seed " << seed << std::endl;
	
	std::mt19937_64 rand(seed);
	
	auto mod = [&](std::string tmp) -> std::string {
		if(tmp.empty()) return tmp;
		auto alter_idx = rand()%tmp.size();
		switch(rand()%3) {
			case 0:{
				tmp.erase(alter_idx);
			}break;
			case 1:{
				tmp[alter_idx] = rand() % 256;
			}break;
			case 2:{
				tmp.insert(alter_idx, 1, char(rand() % 256));
			}break;
		}
		return tmp;
	};
	
	auto codes = sources | std::ranges::views::transform([](std::string file){
		std::ifstream file_str{file};
		std::stringstream read;
		read << file_str.rdbuf();
		return read.str();
	});
	
	std::vector<std::string> vec;
	std::copy(codes.begin(), codes.end(), std::back_inserter(vec));
	
	size_t count = 0;
	size_t error_cnt = 0;
	size_t success_cnt = 0;
	constexpr size_t max_count = 8000000;
	
	auto begin = std::chrono::high_resolution_clock::now();
	while(count < max_count) {
		std::cout << 100.0*double(count)/max_count <<"%"<< std::endl;
		for(auto& code : vec) {
			std::vector<scripting::script_error> errors;
			auto lexed = scripting::ast::lex(code, errors);
			auto parsed = scripting::ast::parse(lexed, errors);
			if(errors.empty()) success_cnt++;
			else error_cnt++;
			count++;
		}
		
		auto limit = std::min<size_t>(vec.size(), 5000) ;
		for(size_t idx = 0; idx < limit; ++idx) {
			vec.push_back(mod(vec[idx]));
		}
		
		std::transform(vec.begin(), vec.end(), vec.begin(), mod);
		
		std::shuffle(vec.begin(), vec.end(), rand);
		if(vec.size()>30000) vec.resize(30000);
	}
	auto end = std::chrono::high_resolution_clock::now();
	
	std::cout
	    << "Successes: " << success_cnt << "\n"
	    << "Failures: " << error_cnt << "\n"
		<< "Ratio: " << double(success_cnt)/double(success_cnt+error_cnt) << "\n"
        << "Total time: " << std::chrono::duration_cast<std::chrono::microseconds>(end-begin).count() << "µs\n"
	    << "Time per iteration: " << (std::chrono::duration_cast<std::chrono::nanoseconds>(end-begin)/(error_cnt+success_cnt)).count() << "ns\n";
};

TEST_CASE("Try to crash the parser (known seeds)") {
	runner<4138740281>();
	runner<1547293717>();
	runner<1759257947>();
	runner<2909912711>();
	runner<1236548620>();
}

TEST_CASE("Try to crash the parser (new seeds)") {
	runner<>();
}