* nkf.c (numchar_getc): increase buffer size.

[nkf/nkf.git] / nkf.c

diff --git a/nkf.c b/nkf.c
index 22b5bb4..045d877 100644 (file)
--- a/nkf.c
+++ b/nkf.c
@@ -31,9 +31,9 @@
  * \e$B8=:_!"\e(Bnkf \e$B$O\e(B SorceForge \e$B$K$F%a%s%F%J%s%9$,B3$1$i$l$F$$$^$9!#\e(B
  * http://sourceforge.jp/projects/nkf/
  ***********************************************************************/
-#define NKF_IDENT "$Id: nkf.c,v 1.177 2008/02/08 12:08:50 naruse Exp $"
+#define NKF_IDENT "$Id: nkf.c,v 1.185 2008/10/28 13:42:25 naruse Exp $"
 #define NKF_VERSION "2.0.8"
-#define NKF_RELEASE_DATE "2008-02-08"
+#define NKF_RELEASE_DATE "2008-10-28"
 #define COPY_RIGHT \
     "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW\n" \
     "Copyright (C) 2002-2008 Kono, Furukawa, Naruse, mastodon"
@@ -41,6 +41,10 @@
 #include "config.h"
 #include "nkf.h"
 #include "utf8tbl.h"
+#ifdef __WIN32__
+#include <windows.h>
+#include <locale.h>
+#endif
 
 /* state of output_mode and input_mode
 
@@ -318,7 +322,7 @@ extern POINT _BufferSize;
 #endif
 
 struct input_code{
-    char *name;
+    const char *name;
     nkf_char stat;
     nkf_char score;
     nkf_char index;
@@ -328,11 +332,10 @@ struct input_code{
     int _file_stat;
 };
 
-static char *input_codename = NULL; /* NULL: unestablished, "": BINARY */
+static const char *input_codename = NULL; /* NULL: unestablished, "": BINARY */
 static nkf_encoding *input_encoding = NULL;
 static nkf_encoding *output_encoding = NULL;
 
-static int kanji_convert(FILE *f);
 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
 /* UCS Mapping
  * 0: Shift_JIS, eucJP-ascii
@@ -439,7 +442,7 @@ static nkf_char (*iconv_for_check)(nkf_char c2,nkf_char c1,nkf_char c0) = 0;
 #endif
 
 static int guess_f = 0; /* 0: OFF, 1: ON, 2: VERBOSE */
-static  void    set_input_codename(char *codename);
+static  void    set_input_codename(const char *codename);
 
 #ifdef EXEC_IO
 static int exec_f = 0;
@@ -653,14 +656,15 @@ static int             end_check;
 nkf_char std_gc_buf[STD_GC_BUFSIZE];
 nkf_char std_gc_ndx;
 
-static void
-nkf_str_upcase(const char *src, char *dest, size_t length)
+static int
+nkf_str_caseeql(const char *src, const char *target)
 {
-    int i = 0;
-    for (; i < length && src[i]; i++) {
-       dest[i] = nkf_toupper(src[i]);
+    int i;
+    for (i = 0; src[i] && target[i]; i++) {
+       if (nkf_toupper(src[i]) != nkf_toupper(target[i])) return FALSE;
     }
-    dest[i] = 0;
+    if (src[i] || target[i]) return FALSE;
+    else return TRUE;
 }
 
 static nkf_encoding*
@@ -675,14 +679,14 @@ nkf_enc_from_index(int idx)
 static int
 nkf_enc_find_index(const char *name)
 {
-    int i, index = -1;
-    if (*name == 'X' && *(name+1) == '-') name += 2;
+    int i;
+    if (name[0] == 'X' && *(name+1) == '-') name += 2;
     for (i = 0; encoding_name_to_id_table[i].id >= 0; i++) {
-       if (strcmp(name, encoding_name_to_id_table[i].name) == 0) {
+       if (nkf_str_caseeql(encoding_name_to_id_table[i].name, name)) {
            return encoding_name_to_id_table[i].id;
        }
     }
-    return index;
+    return -1;
 }
 
 static nkf_encoding*
@@ -718,7 +722,16 @@ nkf_locale_charmap()
 #ifdef HAVE_LANGINFO_H
     return nl_langinfo(CODESET);
 #elif defined(__WIN32__)
-    return sprintf("CP%d", GetACP());
+    char buf[16];
+    char *str;
+    int len = sprintf(buf, "CP%d", GetACP());
+    if (len > 0) {
+      str = malloc(len + 1);
+      strcpy(str, buf);
+      str[len] = '\0';
+      return str;
+    }
+    else return NULL;
 #else
     return NULL;
 #endif
@@ -914,8 +927,8 @@ get_backup_filename(const char *suffix, const char *filename)
        }
        backup_filename[j] = '\0';
     }else{
-       j = strlen(suffix) + filename_length;
-       backup_filename = malloc( + 1);
+       j = filename_length + strlen(suffix);
+       backup_filename = malloc(j + 1);
        strcpy(backup_filename, filename);
        strcat(backup_filename, suffix);
        backup_filename[j] = '\0';
@@ -1472,6 +1485,7 @@ s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
     nkf_char val;
 #endif
     static const char shift_jisx0213_s1a3_table[5][2] ={ { 1, 8}, { 3, 4}, { 5,12}, {13,14}, {15, 0} };
+    if (0xFC < c1) return 1;
 #ifdef SHIFTJIS_CP932
     if (!cp932inv_f && is_ibmext_in_sjis(c2)){
        val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
@@ -1482,10 +1496,10 @@ s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
     }
     if (cp932inv_f
        && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
-       nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
-       if (c){
-           c2 = c >> 8;
-           c1 = c & 0xff;
+       val = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
+       if (val){
+           c2 = val >> 8;
+           c1 = val & 0xff;
        }
     }
 #endif /* SHIFTJIS_CP932 */
@@ -2501,6 +2515,7 @@ w_oconv16(nkf_char c2, nkf_char c1)
        c1 = val & 0xff;
        if (!val) return;
     }
+
     if (output_endian == ENDIAN_LITTLE){
        (*o_putc)(c1);
        (*o_putc)(c2);
@@ -3499,7 +3514,7 @@ z_conv(nkf_char c2, nkf_char c1)
 
     if (alpha_f&8 && c2 == 0) {
        /* HTML Entity */
-       char *entity = 0;
+       const char *entity = 0;
        switch (c1){
        case '>': entity = "&gt;"; break;
        case '<': entity = "&lt;"; break;
@@ -3982,7 +3997,7 @@ debug(const char *str)
 #endif
 
 static void
-set_input_codename(char *codename)
+set_input_codename(const char *codename)
 {
     if (!input_codename) {
        input_codename = codename;
@@ -3991,7 +4006,7 @@ set_input_codename(char *codename)
     }
 }
 
-static char*
+static const char*
 get_guessed_code(void)
 {
     if (input_codename && !*input_codename) {
@@ -4098,7 +4113,7 @@ numchar_getc(FILE *f)
     nkf_char (*g)(FILE *) = i_ngetc;
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nungetc;
     int i = 0, j;
-    nkf_char buf[8];
+    nkf_char buf[12];
     long c = -1;
 
     buf[i] = (*g)(f);
@@ -5251,7 +5266,7 @@ kanji_convert(FILE *f)
 {
     nkf_char c1=0, c2=0, c3=0, c4=0;
     int shift_mode = 0; /* 0, 1, 2, 3 */
-    char g2 = 0;
+    int g2 = 0;
     int is_8bit = FALSE;
 
     if (input_encoding && !nkf_enc_asciicompat(input_encoding)) {
@@ -5693,7 +5708,6 @@ options(unsigned char *cp)
     nkf_char i, j;
     unsigned char *p;
     unsigned char *cp_back = NULL;
-    char codeset[32];
     nkf_encoding *enc;
 
     if (option_mode==1)
@@ -5733,15 +5747,13 @@ options(unsigned char *cp)
                cp = (unsigned char *)long_option[i].alias;
            }else{
                if (strcmp(long_option[i].name, "ic=") == 0){
-                   nkf_str_upcase((char *)p, codeset, 32);
-                   enc = nkf_enc_find(codeset);
+                   enc = nkf_enc_find((char *)p);
                    if (!enc) continue;
                    input_encoding = enc;
                    continue;
                }
                if (strcmp(long_option[i].name, "oc=") == 0){
-                   nkf_str_upcase((char *)p, codeset, 32);
-                   enc = nkf_enc_find(codeset);
+                   enc = nkf_enc_find((char *)p);
                    if (enc <= 0) continue;
                    output_encoding = enc;
                    continue;
@@ -6088,9 +6100,9 @@ options(unsigned char *cp)
                    cp++;
                    input_endian = ENDIAN_BIG;
                }
-               enc_idx = enc_idx == UTF_16
-                   ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
-                   : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
+               enc_idx = (enc_idx == UTF_16
+                   ? (input_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
+                   : (input_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE));
                input_encoding = nkf_enc_from_index(enc_idx);
            }
            continue;
@@ -6490,6 +6502,7 @@ main(int argc, char **argv)
                            fprintf(stderr, "Can't rename %s to %s\n",
                                    origfname, backup_filename);
                        }
+                       free(backup_filename);
                    }else{
 #ifdef MSDOS
                        if (unlink(origfname)){