WIP sj4lib

CMakeLists.txt

@@ -31,14 +31,22 @@ target_include_directories(sj4rkcv PUBLIC include/sj4common include/sj4rkcv)
 target_include_directories(sj4rkcv PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
 target_link_libraries(sj4rkcv PUBLIC sj4common)
 
+if(NOT SJ4_GLOBAL)
+	file(GLOB_RECURSE SJ4LIB_SRCS lib/sj4lib/**.c)
+	add_library(sj4lib STATIC ${SJ4LIB_SRCS})
+	target_include_directories(sj4lib PUBLIC include/sj4common include/sj4core include/sj4lib)
+	target_include_directories(sj4lib PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
+	target_link_libraries(sj4lib PUBLIC sj4common sj4core)
+endif()
+
 file(GLOB_RECURSE SJ4MKDIC_SRCS src/sj4mkdic/**.c)
 add_executable(sj4mkdic ${SJ4MKDIC_SRCS})
 target_include_directories(sj4mkdic PRIVATE include/sj4common include/sj4core)
 target_compile_definitions(sj4mkdic PRIVATE UTF8 ENGLISH)
-target_link_libraries(sj4mkdic PUBLIC sj4common)
+target_link_libraries(sj4mkdic PRIVATE sj4common)
 
 if(NOT SJ4_GLOBAL)
 	file(GLOB_RECURSE SJ4TEST_SRCS src/sj4test/**.c)
 	add_executable(sj4test ${SJ4TEST_SRCS})
-	target_link_libraries(sj4test PRIVATE sj4core)
+	target_link_libraries(sj4test PRIVATE sj4lib)
 endif()

include/sj4lib/sj4lib.h

@@ -0,0 +1,35 @@
+#ifndef __SJ4LIB_H__
+#define __SJ4LIB_H__
+
+#include <wchar.h>
+
+/* increase this as needed... */
+#define SJ4BUFSZ (16 * 1024)
+
+typedef union sj4kouhobuffer {
+	unsigned char sjis[SJ4BUFSZ];
+	unsigned char eucjp[SJ4BUFSZ];
+	char	      utf8[SJ4BUFSZ];
+	wchar_t	      utf16[SJ4BUFSZ / sizeof(wchar_t)];
+
+	unsigned char raw[SJ4BUFSZ];
+} Sj4KouhoBuffer;
+
+typedef struct sj4kouho {
+	Sj4KouhoBuffer buffer;
+} Sj4Kouho;
+
+typedef struct sj4lib Sj4Lib;
+
+enum SJ4CHARSET {
+	SJ4EUCJP = 0,
+	SJ4SJIS,
+	SJ4UTF8,
+	SJ4UTF16
+};
+
+Sj4Lib* sj4_open(int, const char*);
+int	sj4_getkan(Sj4Lib*, const void*, int len, Sj4Kouho*);
+void	sj4_close(Sj4Lib*);
+
+#endif

lib/sj4lib/sj.c

@@ -0,0 +1,214 @@
+#include <sj4lib.h>
+#include <sj_kanakan.h>
+
+#include <stdlib.h>
+
+#include <sj_euc2ucs.h>
+#include <sj_ucs2euc.h>
+
+static u_int ucs_to_euc(u_int in) {
+	if(0 <= in && in <= 0xffff) {
+		const T_U2E_BITMAP_INDEX* b = &utf16_to_euc_jp_table[(in >> 8) & 0xff];
+
+		if(b->byType == 2) {
+			b = &utf16_to_euc_jp_table[b->dwBitmapIndex + (in & 0xff)];
+
+			if(b->byType == 3) {
+				return b->dwEucJpCode;
+			}
+		}
+
+		return in;
+	}
+
+	return 0;
+}
+
+static u_int euc_to_ucs(u_int in) {
+	const T_E2U_BITMAP_INDEX* b = NULL;
+
+	if(in <= 0xff) {
+		b = &euc_jp_to_utf16_table[in];
+		if(b->byType != 3) return 0;
+
+		return b->dwUtf16Code;
+	} else if(in <= 0xffff) {
+		b = &euc_jp_to_utf16_table[(in >> 8) & 0xff];
+		if(b->byType != 2) return 0;
+
+		b = &euc_jp_to_utf16_table[b->dwBitmapIndex + ((in >> 0) & 0xff)];
+		if(b->byType != 3) return 0;
+
+		return b->dwUtf16Code;
+	} else {
+		b = &euc_jp_to_utf16_table[(in >> 16) & 0xff];
+		if(b->byType != 2) return 0;
+
+		b = &euc_jp_to_utf16_table[b->dwBitmapIndex + ((in >> 8) & 0xff)];
+		if(b->byType != 2) return 0;
+
+		b = &euc_jp_to_utf16_table[b->dwBitmapIndex + ((in >> 0) & 0xff)];
+		if(b->byType != 3) return 0;
+
+		return b->dwUtf16Code;
+	}
+
+	return 0;
+}
+
+static int from_sjis(u_char* out, const u_char* in, int len) {
+}
+
+static int utf8_later(unsigned char c) {
+	return 0x80 <= c && c < 0xc0;
+}
+
+#define CAST_I32(x) ((int)(x))
+
+static int from_utf8(u_char* out, const u_char* in, int len) {
+	const u_char* o_in  = in;
+	u_char*	      o_out = out;
+
+	while((in - o_in) < len) {
+		u_char c = *in;
+		u_int  n = 0;
+
+		if(c < 0x80) {
+			n = c;
+
+			in += 1;
+		} else if(0xc2 <= c && c < 0xe0) {
+			if(!utf8_later(in[1])) return 0;
+			if((in[0] & 0x1e) == 0) return 0;
+
+			n = CAST_I32(in[0] & 0x1f) << 6;
+			n |= CAST_I32(in[1] & 0x3f);
+
+			in += 2;
+		} else if(0xe0 <= c && c < 0xf0) {
+			if(!utf8_later(in[1]) || !utf8_later(in[2])) return 0;
+			if((in[0] & 0x0f) == 0 && (in[1] & 0x20) == 0) return 0;
+
+			n = CAST_I32(in[0] & 0x0f) << 12;
+			n |= CAST_I32(in[1] & 0x3f) << 6;
+			n |= CAST_I32(in[2] & 0x3f);
+
+			in += 3;
+		} else if(0xf0 <= c && c < 0xf8) {
+			if(!utf8_later(in[1]) || !utf8_later(in[2]) || !utf8_later(in[3])) return 0;
+			if((in[0] & 0x07) == 0 && (in[1] & 0x30) == 0) return 0;
+
+			n = CAST_I32(in[0] & 0x07) << 18;
+			n |= CAST_I32(in[1] & 0x3f) << 12;
+			n |= CAST_I32(in[2] & 0x3f) << 6;
+			n |= CAST_I32(in[3] & 0x3f);
+
+			in += 4;
+		}
+
+		n = ucs_to_euc(n);
+
+		if(n <= 0xff) {
+			*out++ = n;
+		} else if(n <= 0xffff) {
+			*out++ = (n >> 8) & 0xff;
+			*out++ = (n >> 0) & 0xff;
+		} else {
+			*out++ = (n >> 16) & 0xff;
+			*out++ = (n >> 8) & 0xff;
+			*out++ = (n >> 0) & 0xff;
+		}
+	}
+
+	return out - o_out;
+}
+
+static int from_utf16(u_char* out, const u_char* in, int len) {
+}
+
+static int to_sjis(u_char* out, const u_char* in, int len) {
+}
+
+static int to_utf8(u_char* out, const u_char* in, int len) {
+	const u_char* o_in  = in;
+	u_char*	      o_out = out;
+
+	while((in - o_in) < len) {
+		u_char c = *in;
+		u_int  n = 0;
+
+		if(c <= 0x7f) {
+			n = c;
+
+			in += 1;
+		} else if(c != 0x8f) {
+			n = CAST_I32(in[0]) << 8;
+			n |= CAST_I32(in[1]);
+
+			in += 2;
+		} else {
+			n = CAST_I32(in[0]) << 16;
+			n |= CAST_I32(in[1]) << 8;
+			n |= CAST_I32(in[2]);
+
+			in += 3;
+		}
+	}
+
+	return out - o_out;
+}
+
+static int to_utf16(u_char* out, const u_char* in, int len) {
+}
+
+struct sj4lib {
+	Sj4Context* ctx;
+
+	int charset;
+
+	u_char inbuf[SJ4BUFSZ];
+	u_char outbuf[SJ4BUFSZ];
+};
+
+Sj4Lib* sj4_open(int charset, const char* dic) {
+	Sj4Lib* ctx;
+
+	if((ctx = calloc(1, sizeof(*ctx))) == NULL) return NULL;
+
+	if((ctx->ctx = alloccontext(dic)) == NULL) {
+		free(ctx);
+
+		return NULL;
+	}
+
+	ctx->charset = charset;
+
+	return ctx;
+}
+
+#define ICONV(to, from, len, mode) \
+	if(ctx->charset == SJ4EUCJP) { \
+		memcpy(to, from, len); \
+	} else if(ctx->charset == SJ4SJIS) { \
+		len = mode##_sjis(to, from, len); \
+	} else if(ctx->charset == SJ4UTF8) { \
+		len = mode##_utf8(to, from, len); \
+	} else if(ctx->charset == SJ4UTF16) { \
+		len = mode##_utf16(to, from, len); \
+	}
+
+int sj4_getkan(Sj4Lib* ctx, const void* input, int len, Sj4Kouho* kouho) {
+	memset(kouho, 0, sizeof(*kouho));
+
+	ICONV(ctx->inbuf, input, len, from);
+
+	len = cl2knj(ctx->ctx, ctx->inbuf, len, ctx->outbuf);
+
+	ICONV(kouho->buffer.raw, ctx->outbuf, len, to);
+
+	return len;
+}
+
+void sj4_close(Sj4Lib* ctx) {
+	free(ctx);
+}

src/sj4mkdic/string.c

@@ -37,7 +37,7 @@
 
 #include "sj4mkdic.h"
 
-#include "ucstable.h"
+#include "sj_ucs2euc.h"
 
 int bubun_str(u_char* p1, int l1, u_char* p2, int l2) {
 	u_char* p;
@@ -204,7 +204,7 @@ void utf8_print(char* out, int c) {
 
 int unicode_to_eucjp(int utf) {
 	if(0 <= utf && utf <= 0xffff) {
-		const T_BITMAP_INDEX* b = &utf16_to_euc_jp_table[(utf >> 8) & 0xff];
+		const T_U2E_BITMAP_INDEX* b = &utf16_to_euc_jp_table[(utf >> 8) & 0xff];
 
 		if(b->byType == 2) {
 			if(b->dwBitmapIndex != 0) {

src/sj4test/main.c

@@ -1,11 +1,12 @@
-#include <sj_kanakan.h>
+#include <sj4lib.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
 
 int main(int argc, char** argv) {
-	char	    out[64 * 1024];
+	char	c[8 * 1024];
-	int	    i = 0;
+	Sj4Lib* ctx = sj4_open(SJ4UTF8, "sj4main.dic");
-	char	    c[64 * 1024];
 	char	b;
-	Sj4Context* ctx = alloccontext("sj4main.dic");
 
 	if(ctx == NULL) {
 		fprintf(stderr, "Failed to create context\n");
@@ -15,7 +16,8 @@ int main(int argc, char** argv) {
 	c[0] = 0;
 
 	while(fread(&b, 1, 1, stdin) == 1) {
-		int n;
+		int	 n, i;
+		Sj4Kouho kouho;
 
 		if(b == '\n') {
 		} else if(b != '\r') {
@@ -28,22 +30,15 @@ int main(int argc, char** argv) {
 		}
 
 		i = 0;
-		while((strlen(c) - i) > 0 && (i += (n = cl2knj(ctx, (u_char*)c + i, strlen(c) - i, (u_char*)out)))) {
+		while((strlen(c) - i) > 0 && (n = sj4_getkan(ctx, c + i, strlen(c + i), &kouho))) {
-			STDYOUT* s   = (STDYOUT*)out;
+			printf("> %s\n", kouho.buffer);
-			char*	 str = malloc(n + 1);
 
-			memcpy(str, c + i - n, n);
+			i += n;
-			str[i] = 0;
-
-			printf("> %s -> %s\n", str, out + sizeof(STDYOUT));
-
-			free(str);
 		}
-
 		c[0] = 0;
 	}
 
-	free_context(ctx);
+	sj4_close(ctx);
 
 	return 0;
 }

tools/format.sh

@@ -2,4 +2,4 @@
 while [ ! -d .git ]; do
 	cd ..
 done
-clang-format --verbose -i `find src lib include "(" -name "*.c" -or -name "*.h" ")" -and -not -name ucstable.h -and -not -name maindic"*".c`
+clang-format --verbose -i `find src lib include "(" -name "*.c" -or -name "*.h" ")" -and -not -name sj_euc2ucs.h -and -not -name sj_ucs2euc.h  -and -not -name maindic"*".c`