WIP IME

CMakeLists.txt

@@ -61,8 +61,15 @@ if(SJ4_BUILD_TOOLS OR (NOT CMAKE_BINARY_DIR MATCHES CMAKE_CURRENT_SOURCE_DIR AND
 	endif()
 
 	if(NOT SJ4_GLOBAL)
-		file(GLOB_RECURSE SJ4TEST_SRCS src/sj4test/**.c)
+		macro(sj4_test name)
-		add_executable(sj4test ${SJ4TEST_SRCS})
+			file(GLOB_RECURSE ${name}_SRCS src/${name}/**.c)
+			add_executable(${name} ${${name}_SRCS})
+		endmacro()
+
+		sj4_test(sj4test)
 		target_link_libraries(sj4test PRIVATE sj4lib)
+
+		sj4_test(sj4test2)
+		target_link_libraries(sj4test2 PRIVATE sj4ime)
 	endif()
 endif()

include/sj4charset/sj4charset.h

@@ -1,5 +1,5 @@
-#ifndef __SJ_CHARSET_H__
+#ifndef __SJ4CHARSET_H__
-#define __SJ_CHARSET_H__
+#define __SJ4CHARSET_H__
 
 #include <sj_types.h>
 

include/sj4ime/sj4ime.h

@@ -1,6 +1,8 @@
 #ifndef __SJ4IME_H__
 #define __SJ4IME_H__
 
+#include <sj4charset.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -8,6 +10,7 @@ extern "C" {
 typedef struct sj4ime Sj4Ime;
 
 Sj4Ime* sj4_ime(int charset, const char* dict);
+void	sj4_ime_key(Sj4Ime* ime, int key);
 
 #ifdef __cplusplus
 }

include/sj4lib/sj4lib.h

@@ -1,7 +1,7 @@
 #ifndef __SJ4LIB_H__
 #define __SJ4LIB_H__
 
-#include <sj_charset.h>
+#include <sj4charset.h>
 #include <wchar.h>
 
 /* increase this as needed... */

lib/sj4charset/conv.c

@@ -1,4 +1,4 @@
-#include "sj_charset.h"
+#include "sj4charset.h"
 
 #include "sj_euc2ucs.h"
 #include "sj_ucs2euc.h"

lib/sj4ime/KanaTable

@@ -215,7 +215,6 @@
 {"QU", "\244\257\244\245", ""},
 {"QE", "\244\257\244\247", ""},
 {"QO", "\244\257\244\251", ""},
-{"N", "\244\363", ""},
 {"NN", "\244\363", ""},
 {"KK", "\244\303", "K"},
 {"SS", "\244\303", "S"},

lib/sj4ime/ime.c

@@ -1,11 +1,31 @@
 #include <sj4ime.h>
 
 #include <sj4lib.h>
+#include <sj4charset.h>
 
+#include <stdio.h>
+#include <wchar.h>
+#include <string.h>
 #include <stdlib.h>
+#include <ctype.h>
 
 struct sj4ime {
 	Sj4Lib* lib;
+	int	charset;
+
+	Sj4Kouho     kouho;
+	unsigned int last;
+
+	u_char romabuf[SJ4BUFSZ];
+	u_char kanabuf[SJ4BUFSZ];
+};
+
+static struct sj4table {
+	const char* roma;
+	u_char*	    kana;
+	const char* roma2;
+} kanatbl[] = {
+#include "KanaTable"
 };
 
 Sj4Ime* sj4_ime(int charset, const char* dict) {
@@ -18,4 +38,110 @@ Sj4Ime* sj4_ime(int charset, const char* dict) {
 
 		return NULL;
 	}
+
+	ime->charset = charset;
+}
+
+static int w_len(wchar_t* wc) {
+	int i;
+
+	for(i = 0; wc[i] != 0; i++);
+
+	return i;
+}
+
+static int b_len(int charset, Sj4KouhoBuffer* buffer) {
+	if(charset == SJ4UTF16) {
+		return w_len(buffer->utf16);
+	} else {
+		return strlen(buffer->raw);
+	}
+}
+
+static void concat(int charset, u_char* out, u_char* in) {
+	int i;
+
+#ifdef UCS
+	if(charset == SJ4UTF16) {
+		wchar_t* w_out = (wchar_t*)out;
+
+		i = w_len(w_out);
+
+		w_out[i + sj4_charset_to(charset, (u_char*)(w_out + i), in, strlen(in))] = 0;
+	} else
+#else
+	{
+		i = strlen(out);
+
+		out[i + sj4_charset_to(charset, out + i, in, strlen(in))] = 0;
+	}
+#endif
+}
+
+static void comp(Sj4Ime* ime) {
+	ime->kanabuf[0] = 0;
+	ime->last	= 0;
+}
+
+#define ZEROBUF memset(&ime->kouho.buffer, 0, sizeof(ime->kouho.buffer))
+
+void sj4_ime_key(Sj4Ime* ime, int key) {
+	if(key == '\n') {
+		comp(ime);
+
+		ZEROBUF;
+	} else if(key == ' ') {
+	reconvert:
+		if((ime->last = b_len(ime->charset, &ime->kouho.buffer)) == 0) {
+			sj4_getkan(ime->lib, ime->kanabuf, strlen(ime->kanabuf), &ime->kouho);
+		} else {
+			if((ime->last = sj4_nextkan(ime->lib)) == 0) {
+				ZEROBUF;
+
+				goto reconvert;
+			}
+		}
+
+		printf("%s\n", ime->kouho.buffer.utf8);
+	} else if('a' <= tolower(key) && tolower(key) <= 'z') {
+		int i;
+
+		if(ime->last != 0) comp(ime);
+
+		key = toupper(key);
+
+		if(ime->romabuf[strlen(ime->romabuf) - 1] == 'N' && !(key == 'A' || key == 'I' || key == 'U' || key == 'E' || key == 'O' || key == 'N')) {
+			ime->romabuf[strlen(ime->romabuf) + 1] = 0;
+			ime->romabuf[strlen(ime->romabuf)]     = 'N';
+		}
+
+		ime->romabuf[strlen(ime->romabuf) + 1] = 0;
+		ime->romabuf[strlen(ime->romabuf)]     = key;
+
+	reduce:
+		for(i = 0; i < sizeof(kanatbl) / sizeof(kanatbl[0]); i++) {
+			int n = strlen(kanatbl[i].roma);
+
+			if(n > strlen(ime->romabuf)) continue;
+
+			if(memcmp(ime->romabuf, kanatbl[i].roma, n) == 0) {
+				int j;
+
+				n = strlen(kanatbl[i].roma) - strlen(kanatbl[i].roma2);
+
+				memmove(ime->romabuf - (n < 0 ? n : 0), ime->romabuf + (n > 0 ? n : 0), SJ4BUFSZ - n);
+				if(n > 0) {
+					for(j = SJ4BUFSZ - n; j < SJ4BUFSZ; j++) ime->romabuf[j] = 0;
+				} else {
+					for(j = 0; j < -n; j++) ime->romabuf[j] = 0;
+				}
+
+				memcpy(ime->romabuf, kanatbl[i].roma2, strlen(kanatbl[i].roma2));
+
+				concat(ime->charset, ime->kanabuf, kanatbl[i].kana);
+
+				goto reduce;
+			}
+		}
+	}
 }

lib/sj4lib/sj.c

@@ -2,7 +2,7 @@
 #include <sj_kanakan.h>
 #include <sj_string.h>
 
-#include <sj_charset.h>
+#include <sj4charset.h>
 
 #include <stdlib.h>
 

src/sj4mkdic/char.c

@@ -38,7 +38,7 @@
 
 #include "sj4mkdic.h"
 
-#include "sj_charset.h"
+#include "sj4charset.h"
 
 int cnvyomi(int code) {
 	u_short hh;

src/sj4mkdic/readline.c

@@ -154,8 +154,6 @@ skip_blank() {
 	return c;
 }
 
-#include "sj_charset.h"
-
 static int
 readhinsi() {
 	static int    c = 0;

src/sj4test2/main.c

@@ -0,0 +1,22 @@
+#include <sj4ime.h>
+#include <stdio.h>
+
+#ifdef _WIN32
+#define CHARSET SJ4SJIS
+#else
+#define CHARSET SJ4UTF8
+#endif
+
+int main() {
+	Sj4Ime* ime = sj4_ime(CHARSET, "sj4main.dic");
+	char	b;
+
+	if(ime == NULL) {
+		fprintf(stderr, "Failed to create context\n");
+		return 1;
+	}
+
+	while(fread(&b, 1, sizeof(b), stdin)) {
+		sj4_ime_key(ime, b);
+	}
+}