[Maintenance] Introduce proper Shift-JIS and JIS X 0208 types

Much more semantic than that twobyte_t abomination.

Part of P0212, funded by GhostRiderCog, Lmocinemod, and LeyDud.

decomp.hpp

@@ -3,6 +3,8 @@
  * Declarations to help decompiling the seemingly impossible
  */
 
+#define DECOMP_HPP
+
 // Flag comparisons
 // ----------------
 // When used inside a conditional expression like

shiftjis.hpp

@@ -0,0 +1,43 @@
+// 2-byte JIS X 0208 codepoint.
+typedef uint16_t jis_t;
+
+// Multibyte Shift-JIS string. Use as replacement for `unsigned char` in
+// situations where the contents are assumed to be Shift-JIS.
+typedef uint8_t shiftjis_t;
+typedef int8_t sshiftjis_t; // MODDERS: Delete
+
+// 2-byte/fullwidth Shift-JIS codepoint.
+typedef uint16_t shiftjis_kanji_t;
+typedef int8_t sshiftjis_kanji_t; // MODDERS: Delete
+
+// ZUN bloat: Byte-swapped 2-byte/fullwidth Shift-JIS codepoint for bytewise
+// access.
+typedef uint16_t shiftjis_kanji_swapped_t;
+typedef int16_t sshiftjis_kanji_swapped_t; // MODDERS: Delete
+
+#define kanji_swap(kanji) static_cast<shiftjis_kanji_swapped_t>( \
+	(static_cast<shiftjis_kanji_t>(kanji) << 8) | \
+	(static_cast<shiftjis_kanji_t>(kanji) >> 8) \
+)
+
+#ifdef DECOMP_HPP
+	// Way too often, the games store fullwidth Shift-JIS codepoints in two
+	// 8-bit variables rather than a single 16-bit one.
+	typedef StupidBytewiseWrapperAround<shiftjis_kanji_t> ShiftJISKanji;
+#endif
+
+// Good luck deriving this from the StupidBytewiseWrapperAround template.
+template <unsigned int Length> union ShiftJISKanjiBuffer {
+	shiftjis_kanji_t t[Length];
+	int8_t byte[sizeof(uint16_t) * Length];
+	uint8_t ubyte[sizeof(uint16_t) * Length];
+};
+
+inline jis_t shiftjis_to_jis(shiftjis_kanji_t shiftjis) {
+	return (0x1F21 +
+		(((shiftjis >> 8) - (shiftjis >> 15) < 0x9E)
+			? ((((shiftjis >> 8) - (shiftjis >> 15) - 0x40) & 0x1FF))
+			: ((shiftjis >> 8) - (shiftjis >> 15) + 0x62)
+		) + ((shiftjis & 0x3F) << 9)
+	);
+}

th01/core/initexit.cpp

@@ -6,6 +6,7 @@
 #include "pc98.h"
 #include "planar.h"
 #include "master.hpp"
+#include "shiftjis.hpp"
 extern "C" {
 #include "th01/hardware/graph.h"
 #include "th01/hardware/palette.h"

th01/end/end.cpp

@@ -4,8 +4,8 @@
 #include "platform.h"
 #include "pc98.h"
 #include "planar.h"
-#include "twobyte.h"
 #include "master.hpp"
+#include "shiftjis.hpp"
 #include "th01/common.h"
 #include "th01/rank.h"
 #include "th01/resident.hpp"
@@ -511,7 +511,7 @@ void verdict_title_calculate_and_render(void)
 
 void verdict_animate_and_regist(void)
 {
-	const char* RANKS[RANK_COUNT] = {
+	const shiftjis_t* RANKS[RANK_COUNT] = {
 		" EASY ",
 		"NORMAL",
 		" HARD ",

th01/end/type.hpp

@@ -3,10 +3,10 @@ static const uint4_t COL_TYPE = 15;
 // Types [len] half- (ank) or full-width (kanji) characters of [str] onto the
 // given position in VRAM, with a frame delay between each character.
 void pascal graph_type_ank_n(
-	screen_x_t left, vram_y_t top, int len, const char *str
+	screen_x_t left, vram_y_t top, int len, const sshiftjis_t *str
 );
 void pascal graph_type_kanji_n(
-	screen_x_t left, vram_y_t top, int len, const char *str
+	screen_x_t left, vram_y_t top, int len, const sshiftjis_kanji_t *str
 );
 
 #define graph_type_ank(left, top, str) { \

th01/formats/scoredat.hpp

@@ -17,25 +17,26 @@
 int8_t scoredat_name_byte_encode(int8_t byte);
 int8_t scoredat_name_byte_decode(int8_t byte);
 
+#if 0
 // On-disk structure of the REYHI*.DAT files.
 // For reference, never actually used by the game itself
 struct scoredat_t {
 	// Not null-terminated!
 	char magic[sizeof(SCOREDAT_MAGIC) - 1];
 
-	// Exclusively used to store full-width Shift-JIS code points.
 	// Not null-terminated.
-	int16_t name[SCOREDAT_PLACES][SCOREDAT_NAME_KANJI];
+	shiftjis_kanji_t name[SCOREDAT_PLACES][SCOREDAT_NAME_KANJI];
 
 	uint32_t score[SCOREDAT_PLACES];
 	int16_t stage[SCOREDAT_PLACES];
-	twobyte_t route[SCOREDAT_PLACES];
+	shiftjis_kanji_t route[SCOREDAT_PLACES];
 };
+#endif
 
 extern int8_t* scoredat_names; // Yeah, technically a scoredat_name_t.
 extern int16_t* scoredat_stages;
 extern int32_t* scoredat_score;
-extern int8_t* scoredat_routes; // Yeah, technically a twobyte_t.
+extern int8_t* scoredat_routes; // Yeah, technically a shiftjis_kanji_t.
 
 // Byte-wise access to [scoredat_routes].
 inline int8_t& scoredat_route_byte(int place, int byte) {

th01/fuuin_02.cpp

@@ -14,8 +14,8 @@
 #include "planar.h"
 #include "decomp.hpp"
 #include "pc98kbd.h"
-#include "twobyte.h"
 #include "master.hpp"
+#include "shiftjis.hpp"
 #include "th01/rank.h"
 extern "C" {
 #include "th01/hardware/graph.h"

th01/fuuin_04.cpp

@@ -9,6 +9,7 @@
 #include "pc98.h"
 #include "planar.h"
 #include "master.hpp"
+#include "shiftjis.hpp"
 extern "C" {
 #include "th01/hardware/graph.h"
 #include "th01/hardware/grppsafx.h"
@@ -102,7 +103,7 @@ void pascal grp_palette_white_in(unsigned int frames)
 #pragma option -O-
 
 void pascal graph_type_ank_n(
-	screen_x_t left, vram_y_t top, int len, const char *str
+	screen_x_t left, vram_y_t top, int len, const sshiftjis_t *str
 )
 {
 	for(int i = 0; i < len; i++) {
@@ -114,7 +115,7 @@ void pascal graph_type_ank_n(
 }
 
 void pascal graph_type_kanji_n(
-	screen_x_t left, vram_y_t top, int len, const char *str
+	screen_x_t left, vram_y_t top, int len, const sshiftjis_kanji_t *str
 )
 {
 	for(int i = 0; i < len; i++) {

th01/hardware/graph.cpp

@@ -3,9 +3,11 @@
 #include <mbctype.h>
 #include <mbstring.h>
 #include "platform.h"
+#include "decomp.hpp"
 #include "pc98.h"
 #include "planar.h"
 #include "master.hpp"
+#include "shiftjis.hpp"
 #include "th01/v_colors.hpp"
 #include "th01/math/clamp.hpp"
 #include "th01/hardware/egc.h"
@@ -729,13 +731,13 @@ inline pixel_t fx_spacing_from(int col_and_fx) {
 	return ((col_and_fx / 0x40) % 8);
 }
 
-pixel_t text_extent_fx(int col_and_fx, const unsigned char *str)
+pixel_t text_extent_fx(int col_and_fx, const shiftjis_t *str)
 {
 	register pixel_t ret = 0;
 	register pixel_t spacing = fx_spacing_from(col_and_fx);
 	while(*str) {
 		if(_ismbblead(str[0])) {
-			uint16_t codepoint = ((char)str[0] << 8) + str[0];
+			shiftjis_kanji_t codepoint = ((char)str[0] << 8) + str[0];
 			str++;
 			str++;
 			if(codepoint < 0x8540) {
@@ -757,11 +759,11 @@ pixel_t text_extent_fx(int col_and_fx, const unsigned char *str)
 #include "th01/hardware/grppsafx.cpp"
 
 void graph_putsa_fx(
-	screen_x_t left, vram_y_t top, int16_t col_and_fx, const unsigned char *str
+	screen_x_t left, vram_y_t top, int16_t col_and_fx, const shiftjis_t *str
 )
 {
 	register screen_x_t x = left;
-	uint16_t codepoint;
+	jis_t codepoint;
 	dots_t(GLYPH_FULL_W) glyph_row;
 	dots8_t far *vram;
 	int fullwidth;

th01/hardware/grp2xscs.cpp

@@ -4,6 +4,7 @@
 #include "platform.h"
 #include "pc98.h"
 #include "planar.h"
+#include "shiftjis.hpp"
 extern "C" {
 #include "th01/hardware/graph.h"
 #include "th01/hardware/grppsafx.h"

th01/hardware/grp2xscs.hpp

@@ -14,8 +14,8 @@ void graph_2xscale_byterect_1_to_0_slow(
 // (Which are the only things that this 2× scale is performed on anyway.)
 
 // Blits [str] to the given glyph [row] with the given color and effects, for
-// a later 2× scale of the
-inline void graph_glyphrow_put(int row, int col_and_fx, const char *str) {
+// a later 2× nearest-neighbor scale of the same row.
+inline void graph_glyphrow_put(int row, int col_and_fx, const shiftjis_t *str) {
 	graph_putsa_fx(0, (GLYPH_H * row), (FX_CLEAR_BG | col_and_fx), str);
 }
 

th01/hardware/grppffx.cpp

@@ -2,12 +2,17 @@
 #include <stdarg.h>
 #include "platform.h"
 #include "pc98.h"
+#include "shiftjis.hpp"
 extern "C" {
 #include "th01/hardware/grppsafx.h"
 }
 
 void graph_printf_fx(
-	screen_x_t left, vram_y_t top, int16_t col_and_fx, const char *fmt, ...
+	screen_x_t left,
+	vram_y_t top,
+	int16_t col_and_fx,
+	const shiftjis_t *fmt,
+	...
 )
 {
 	char str[256];

th01/hardware/grppfnfx.cpp

@@ -10,7 +10,7 @@ void graph_putfwnum_fx(
 	int digit;
 	int digit_prev;
 	unsigned long divisor = 1;
-	const char* FULLWIDTH_NUMERAL[10] = FULLWIDTH_NUMERALS;
+	const shiftjis_t* FULLWIDTH_NUMERAL[10] = FULLWIDTH_NUMERALS;
 
 	for(divisor_i = 0; divisor_i < digits; divisor_i++) {
 		divisor *= 10;

th01/hardware/grppsafx.h

@@ -44,19 +44,20 @@
 	// Puts the given [str] onto the graphics RAM at the given position,
 	// with the given graphics color and effect.
 	void DEFCONV graph_putsa_fx(
-		screen_x_t left,
-		vram_y_t top,
-		int16_t col_and_fx,
-		const unsigned char *str
+		screen_x_t left, vram_y_t top, int16_t col_and_fx, const shiftjis_t *str
 	);
 #endif
 
 #if (GAME == 1)
 	// Variadic version of graph_putsa_fx().
 	void graph_printf_fx(
-		screen_x_t left, vram_y_t top, int16_t col_and_fx, const char *fmt, ...
+		screen_x_t left,
+		vram_y_t top,
+		int16_t col_and_fx,
+		const shiftjis_t *fmt,
+		...
 	);
 
 	// Calculates the width of [str], displayed with the given [fx].
-	int text_extent_fx(int fx, const unsigned char *str);
+	int text_extent_fx(int fx, const shiftjis_t *str);
 #endif

th01/hardware/tram_x16.cpp

@@ -1,4 +1,4 @@
-void TRAMCursor::putkanji(uint16_t jis_kanji, int atrb)
+void TRAMCursor::putkanji(jis_t jis_kanji, int atrb)
 {
 	// Yes, this is a correct way of writing any fullwidth character to TRAM.
 	// More research and proof forthcoming...
@@ -10,7 +10,7 @@ void TRAMCursor::putkanji(uint16_t jis_kanji, int atrb)
 	p++;
 }
 
-void TRAMCursor::putkanji_for_5_rows(uint16_t jis_kanji, int atrb)
+void TRAMCursor::putkanji_for_5_rows(jis_t jis_kanji, int atrb)
 {
 	for(int y = 0; y < 5; y++) {
 		for(int x = 0; x < (RES_X / GLYPH_FULL_W); x++) {
@@ -21,7 +21,7 @@ void TRAMCursor::putkanji_for_5_rows(uint16_t jis_kanji, int atrb)
 
 // Heavily inspired by the INT 18h, AH=14h sample program from the PC-9801
 // Programmers' Bible, on p. 121.
-void pascal tram_x16_kanji_center_reverse(uint16_t jis_kanji)
+void pascal tram_x16_kanji_center_reverse(jis_t jis_kanji)
 {
 	utram_kanji_amount_t x;
 	upixel_t glyph_y;
@@ -56,7 +56,7 @@ void pascal tram_x16_kanji_center_reverse(uint16_t jis_kanji)
 }
 
 // Where was this function above? :P
-void int18h_14h(REGS& in, pc98_glyph_t& glyph, uint16_t jis)
+void int18h_14h(REGS& in, pc98_glyph_t& glyph, jis_t jis)
 {
 	REGS out;
 	in.w.bx = FP_SEG(&glyph);

th01/hardware/tram_x16.hpp

@@ -1,12 +1,3 @@
-inline uint16_t shiftjis_to_jis(uint16_t shiftjis) {
-	return (0x1F21 +
-		(((shiftjis >> 8) - (shiftjis >> 15) < 0x9E)
-			? ((((shiftjis >> 8) - (shiftjis >> 15) - 0x40) & 0x1FF))
-			: ((shiftjis >> 8) - (shiftjis >> 15) + 0x62)
-		) + ((shiftjis & 0x3F) << 9)
-	);
-}
-
 // Font ROM glyph retrieval
 // ------------------------
 
@@ -26,7 +17,7 @@ struct pc98_glyph_kanji_t : public pc98_glyph_t {
 	DotRect<dots_t(GLYPH_FULL_W), GLYPH_H> dots;
 };
 
-void int18h_14h(REGS& in, pc98_glyph_t& glyph, uint16_t jis);
+void int18h_14h(REGS& in, pc98_glyph_t& glyph, jis_t jis);
 
 inline void fontrom_get(REGS& in, pc98_glyph_ank_8x16_t& glyph, char ank) {
 	int18h_14h(in, glyph, (0x8000 + ank));
@@ -35,23 +26,23 @@ inline void fontrom_get(REGS& in, pc98_glyph_ank_8x16_t& glyph, char ank) {
 
 class TRAMCursor {
 	struct {
-		uint16_t left;
-		uint16_t right;
+		jis_t left;
+		jis_t right;
 	} near* p;
 
 public:
 	// Writes the given fullwidth JIS code point with the given attribute to
 	// the left and right cells at the current cursor position, then advances
 	// the cursor. Halfwidth code points will also be written to both cells.
-	void putkanji(uint16_t jis_kanji, int atrb);
+	void putkanji(jis_t jis_kanji, int atrb);
 
 	// Calls putkanji() for the next 5 TRAM rows.
-	void putkanji_for_5_rows(uint16_t jis_kanji, int atrb);
+	void putkanji_for_5_rows(jis_t jis_kanji, int atrb);
 
 	// This is always called at the (0-based) line 21, and therefore always
 	// ends up writing into the second TRAM page. Luckily, that page is used,
 	// and no code cares about it...
-	void putkanji_until_end(uint16_t jis_kanji, int atrb) {
+	void putkanji_until_end(jis_t jis_kanji, int atrb) {
 		putkanji_for_5_rows(jis_kanji, atrb);
 	}
 
@@ -68,7 +59,7 @@ class TRAMCursor {
 
 // Fills text RAM with black and renders a transparent [jis_kanji] at its
 // center.
-void pascal tram_x16_kanji_center_reverse(uint16_t jis_kanji);
+void pascal tram_x16_kanji_center_reverse(jis_t jis_kanji);
 
 // Shows the red "STAGE [stage_num]" letters.
 void pascal tram_x16_stage(unsigned int stage_num);