metin2-client/src/codepage/codepage.cpp

#include "codepage.h"

#include <algorithm>
#include <cstdint>
#include <memory>
#include <mutex>

namespace mt_codepage {
namespace {

constexpr uint16_t kNone = 0xFFFF;

struct DbcsRow {
	uint8_t lead;
	uint8_t trail_lo;
	uint8_t trail_hi;
	uint32_t offset;  // into the Dbcs array, trail_lo first
};

// A WCTABLE entry the first-match inverse of the byte tables gets wrong.
struct EncodeOverride {
	uint16_t wc;
	uint16_t mb;
	uint8_t kind;  // 0 exact (round-trips), 1 best fit only, 2 not encodable
};

struct CodePageTable {
	unsigned cp;
	uint16_t mb_default;  // Unicode char for an unmapped byte / pair (CPINFO)
	const uint16_t* single;
	const uint8_t* leads;  // lo, hi pairs
	size_t lead_count;
	const DbcsRow* rows;
	size_t row_count;
	const uint16_t* dbcs;
	const EncodeOverride* overrides;
	size_t override_count;
};

#include "codepage_tables.inc"

constexpr size_t kTableCount = sizeof(kTables) / sizeof(kTables[0]);

const CodePageTable* find(unsigned cp)
{
	for (const CodePageTable& t : kTables)
		if (t.cp == cp) return &t;
	return nullptr;
}

bool lead(const CodePageTable& t, unsigned char b)
{
	for (size_t i = 0; i < t.lead_count; ++i)
		if (b >= t.leads[i * 2] && b <= t.leads[i * 2 + 1]) return true;
	return false;
}

uint16_t pair(const CodePageTable& t, unsigned char l, unsigned char trail)
{
	for (size_t i = 0; i < t.row_count; ++i) {
		const DbcsRow& r = t.rows[i];
		if (r.lead != l) continue;
		if (trail < r.trail_lo || trail > r.trail_hi) return kNone;
		return t.dbcs[r.offset + (trail - r.trail_lo)];
	}
	return kNone;
}

// Exact encoder (BMP code point -> bytes, kNone when unmapped), built on first use per code page.
struct Encoder {
	std::once_flag once;
	std::unique_ptr<uint16_t[]> exact;
};
Encoder g_encoders[kTableCount];

const uint16_t* exact_encoder(const CodePageTable& t)
{
	Encoder& e = g_encoders[&t - kTables];
	std::call_once(e.once, [&] {
		e.exact.reset(new uint16_t[0x10000]);
		std::fill_n(e.exact.get(), 0x10000, kNone);
		uint16_t* map = e.exact.get();
		// First match in ascending byte order, as the generator assumed.
		for (unsigned b = 0; b < 256; ++b)
			if (t.single[b] != kNone && map[t.single[b]] == kNone) map[t.single[b]] = uint16_t(b);
		for (size_t i = 0; i < t.row_count; ++i) {
			const DbcsRow& r = t.rows[i];
			for (unsigned trail = r.trail_lo; trail <= r.trail_hi; ++trail) {
				const uint16_t wc = t.dbcs[r.offset + (trail - r.trail_lo)];
				if (wc != kNone && map[wc] == kNone) map[wc] = uint16_t(r.lead << 8 | trail);
			}
		}
		for (size_t i = 0; i < t.override_count; ++i) {
			const EncodeOverride& o = t.overrides[i];
			if (o.kind == 0) map[o.wc] = o.mb;
			else if (o.kind == 2) map[o.wc] = kNone;
		}
	});
	return e.exact.get();
}

uint16_t best_fit(const CodePageTable& t, char32_t c)
{
	const EncodeOverride* end = t.overrides + t.override_count;
	const EncodeOverride* it = std::lower_bound(t.overrides, end, c,
		[](const EncodeOverride& o, char32_t v) { return o.wc < v; });
	return it != end && it->wc == c && it->kind == 1 ? it->mb : kNone;
}

}  // namespace

bool supported(unsigned cp)
{
	return find(cp) != nullptr;
}

bool is_lead_byte(unsigned cp, unsigned char b)
{
	const CodePageTable* t = find(cp);
	return t && lead(*t, b);
}

bool decode(unsigned cp, const unsigned char* p, size_t n, std::u32string& out)
{
	const CodePageTable* t = find(cp);
	if (!t) return false;
	bool ok = true;
	for (size_t i = 0; i < n;) {
		const unsigned char b = p[i];
		uint16_t wc = kNone;
		if (lead(*t, b) && i + 1 < n && p[i + 1] != 0) {
			wc = pair(*t, b, p[i + 1]);
			i += 2;
		} else {
			wc = t->single[b];
			++i;
		}
		if (wc == kNone) {
			wc = t->mb_default;
			ok = false;
		}
		out.push_back(wc);
	}
	return ok;
}

void encode(unsigned cp, std::u32string_view in, std::string& out, bool allow_best_fit, bool* used_default)
{
	const CodePageTable* t = find(cp);
	const uint16_t* exact = t ? exact_encoder(*t) : nullptr;
	for (char32_t c : in) {
		uint16_t mb = kNone;
		if (exact && c < 0x10000) {
			mb = exact[c];
			if (mb == kNone && allow_best_fit) mb = best_fit(*t, c);
		}
		if (mb == kNone) {
			out.push_back('?');
			if (used_default) *used_default = true;
		} else if (mb > 0xFF) {
			out.push_back(char(mb >> 8));
			out.push_back(char(mb & 0xFF));
		} else {
			out.push_back(char(mb));
		}
	}
}

}  // namespace mt_codepage