mtgodot-poc/extension/tests/net_text_codec_test.cpp

// net_text_codec_test — §1.10 fixed-length wire text: to_wire() / from_wire_str()
// must apply a hard BYTE cap (like the stock client's strncpy) while never
// splitting a UTF-8 multi-byte sequence.
#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)

// A CJK code point is 3 bytes in UTF-8 (U+6C49 "汉" = e6 b1 89).
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::to_wire;
	using mtnet::from_wire_str;

	// --- CHARACTER_NAME_MAX_LEN-equivalent cap = 24 bytes ---
	// 8 CJK chars == exactly 24 bytes: accepted whole.
	{
		const std::string in = cjk(8);
		CHECK(in.size() == 24, "8 CJK == 24 bytes");
		const std::string w = to_wire(in, 24);
		CHECK(w.size() == 24 && w == in, "exactly-24 accepted unchanged");
	}

	// 9 CJK chars == 27 bytes: truncated to 24 (8 chars), never 25/26.
	{
		const std::string in = cjk(9);
		CHECK(in.size() == 27, "9 CJK == 27 bytes");
		const std::string w = to_wire(in, 24);
		CHECK(w.size() == 24, "27 -> 24 bytes");
		CHECK(w == cjk(8), "truncated on the code-point boundary (8 chars)");
	}

	// 25-byte input (8 CJK + 1 stray lead byte): a 24-byte cap keeps 8 chars,
	// and even a 25- or 26-byte cap cannot fit the 9th char -> still 24.
	{
		std::string in = cjk(8);
		in += '\xE6'; // dangling lead byte of a 9th char
		CHECK(in.size() == 25, "25-byte input");
		CHECK(to_wire(in, 24).size() == 24, "cap 24 -> 24");
		CHECK(to_wire(in, 25) == cjk(8), "cap 25 -> 24 (no split)");
		CHECK(to_wire(in, 26) == cjk(8), "cap 26 -> 24 (9th char needs 3)");
	}

	// Mid-sequence cap: cutting 2 bytes into a 3-byte char drops the whole char.
	{
		const std::string in = cjk(4); // 12 bytes
		CHECK(to_wire(in, 11).size() == 9, "cap 11 -> 9 (3 whole chars)");
		CHECK(to_wire(in, 10).size() == 9, "cap 10 -> 9");
		CHECK(to_wire(in, 12) == in, "cap 12 -> all 12");
	}

	// ASCII round-trips against the same cap 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");
	}

	// --- from_wire_str: stop at NUL, then guard a dangling partial sequence ---
	{
		char buf[25] = {};
		const std::string src = cjk(3); // 9 bytes
		std::memcpy(buf, src.data(), src.size());
		CHECK(from_wire_str(buf, sizeof(buf) - 1) == src, "reads up to NUL");
	}
	{
		char buf[25];
		std::memset(buf, 0, sizeof(buf));
		const std::string src = cjk(8);
		std::memcpy(buf, src.data(), src.size());
		buf[24] = '\xE6'; // 25th byte: lead of a truncated char, no NUL room
		const std::string got = from_wire_str(buf, 25);
		CHECK(got == cjk(8), "dangling lead byte trimmed on read");
	}
	{
		// Full 24-byte field, no NUL terminator: read all 8 chars.
		char buf[24];
		const std::string src = cjk(8);
		std::memcpy(buf, src.data(), src.size());
		CHECK(from_wire_str(buf, 24) == src, "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;
}