mtgodot-poc/extension/tests/net_text_codec_test.cpp

// net_text_codec_test — §1.10 fixed-length wire text: the client keeps UTF-8
// internally, while the Chinese 40250 wire uses GB2312-compatible bytes.
#include "../src/net/text_codec.h"

#include <cstdio>
#include <cstring>
#include <string>

static int g_fail = 0;
#define CHECK(c, msg)                               \
	do {                                            \
		if (!(c)) {                                 \
			std::fprintf(stderr, "FAIL: %s\n", msg); \
			++g_fail;                               \
		}                                          \
	} while (0)

// U+6C49 "汉": 3 bytes in UTF-8, 2 bytes in GB2312 (BA BA).
static std::string cjk(int n) {
	std::string s;
	for (int i = 0; i < n; ++i) {
		s += "\xE6\xB1\x89";
	}
	return s;
}

int main() {
	using mtnet::from_wire_str;
	using mtnet::to_wire;

	// CHARACTER_NAME_MAX_LEN is a 24-byte wire capacity. Twelve Chinese
	// characters fit; measuring the UTF-8 input would incorrectly allow eight.
	{
		const std::string in = cjk(12);
		CHECK(in.size() == 36, "12 CJK == 36 UTF-8 bytes");
		const std::string w = to_wire(in, 24);
		CHECK(w.size() == 24, "12 CJK == 24 GB2312 bytes");
		CHECK(from_wire_str(w.data(), w.size()) == in, "12 CJK round-trip through GB2312");
		CHECK(mtnet::wire_text_fits(in, 24), "12 CJK fits 24-byte name field");
	}

	// Thirteen characters need 26 GB2312 bytes. Conversion truncates only at a
	// legacy character boundary, while input validation rejects the overlong name.
	{
		const std::string in = cjk(13);
		CHECK(in.size() == 39, "13 CJK == 39 UTF-8 bytes");
		const std::string w = to_wire(in, 24);
		CHECK(w.size() == 24, "13 CJK is capped at 24 wire bytes");
		CHECK(from_wire_str(w.data(), w.size()) == cjk(12), "cap keeps 12 whole CJK chars");
		CHECK(!mtnet::wire_text_fits(in, 24), "13 CJK rejected by 24-byte name field");
	}

	// The preflight used by M2Client must use the selected protocol's rule.
	// A 32-byte 40250 shop-sign field therefore accepts sixteen GB2312 CJK
	// characters even though the UTF-8 input occupies 48 bytes.  The generic
	// transport intentionally keeps its own byte contract.
	{
		CHECK(mtnet::protocol_text_fits(cjk(16), 32, true),
				"classic preflight measures the encoded 32-byte sign");
		CHECK(!mtnet::protocol_text_fits(cjk(17), 32, true),
				"classic preflight rejects a 34-byte sign");
		CHECK(!mtnet::protocol_text_fits(cjk(16), 32, false),
				"generic preflight keeps its UTF-8 byte contract");
	}

	// Invalid UTF-8 is not leaked to the legacy wire and is rejected by the
	// strict validation helper.
	{
		std::string in = cjk(12);
		in += '\xE6'; // dangling UTF-8 lead byte
		CHECK(to_wire(in, 24).size() == 24, "invalid suffix cannot overflow cap");
		CHECK(!mtnet::wire_text_fits(in, 24), "invalid UTF-8 rejected");
	}

	// ASCII remains one byte per character and round-trips unchanged.
	{
		const std::string in = "Warrior";
		CHECK(to_wire(in, 24) == in, "ascii under cap unchanged");
		CHECK(to_wire(std::string(40, 'x'), 24).size() == 24, "ascii over cap -> 24");
		CHECK(from_wire_str(in.data(), in.size()) == in, "ascii round-trip");
	}

	// The user-visible UTF-8 text is not limited to the test character 汉.  A
	// real Chinese name must round-trip through the 40250 GB2312 boundary too.
	{
		const std::string in = "龙驹简体中文";
		const std::string wire = to_wire(in, 24);
		CHECK(wire.size() == 12, "Chinese name uses GB2312 bytes, not UTF-8 bytes");
		CHECK(from_wire_str(wire.data(), wire.size()) == in,
				"Chinese name round-trips through the 40250 wire code page");
	}
	{
		const std::string in = "龙驹简体中文免费版";
		const std::string wire = to_wire(in, 24);
		CHECK(in.size() == 27, "target Chinese name is 27 UTF-8 bytes");
		CHECK(wire.size() == 18 && mtnet::wire_text_fits(in, 24),
				"target Chinese name fits the 24-byte GB2312 field");
		CHECK(from_wire_str(wire.data(), wire.size()) == in,
				"target Chinese name round-trips through the 40250 wire code page");
	}

	// Decode legacy bytes up to NUL.
	{
		char buf[25] = {};
		const std::string legacy = "\xBA\xBA\xBA\xBA\xBA\xBA";
		std::memcpy(buf, legacy.data(), legacy.size());
		CHECK(from_wire_str(buf, sizeof(buf) - 1) == cjk(3), "decodes GB2312 up to NUL");
	}
	{
		char buf[25] = {};
		const std::string legacy = to_wire(cjk(12), 24);
		std::memcpy(buf, legacy.data(), legacy.size());
		buf[24] = '\xBA'; // dangling lead byte at the field boundary
		CHECK(from_wire_str(buf, 25) == cjk(12), "dangling legacy lead byte is ignored");
	}
	{
		// Full 24-byte field, no NUL terminator: read all twelve characters.
		char buf[24];
		const std::string legacy = to_wire(cjk(12), 24);
		std::memcpy(buf, legacy.data(), legacy.size());
		CHECK(from_wire_str(buf, 24) == cjk(12), "unterminated full field read whole");
	}

	if (g_fail) {
		std::fprintf(stderr, "%d checks FAILED\n", g_fail);
		return 1;
	}
	std::puts("net_text_codec_test: OK");
	return 0;
}