Show "(BOM)" if the byte sequence has BOM.

[nkf/nkf.git] / nkf.c

diff --git a/nkf.c b/nkf.c
index 465c7c8..ee855d8 100644 (file)
--- a/nkf.c
+++ b/nkf.c
@@ -20,11 +20,11 @@
  *
  * 3. This notice may not be removed or altered from any source distribution.
  */
-#define NKF_VERSION "2.1.1"
-#define NKF_RELEASE_DATE "2010-04-14"
+#define NKF_VERSION "2.1.3"
+#define NKF_RELEASE_DATE "2012-03-17"
 #define COPY_RIGHT \
     "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa).\n" \
-    "Copyright (C) 1996-2010, The nkf Project."
+    "Copyright (C) 1996-2012, The nkf Project."
 
 #include "config.h"
 #include "nkf.h"
@@ -356,6 +356,7 @@ static  int     no_cp932ext_f = FALSE;
 /* ignore ZERO WIDTH NO-BREAK SPACE */
 static  int     no_best_fit_chars_f = FALSE;
 static  int     input_endian = ENDIAN_BIG;
+static  int     input_bom_f = FALSE;
 static  nkf_char     unicode_subchar = '?'; /* the regular substitution character */
 static  void    (*encode_fallback)(nkf_char c) = NULL;
 static  void    w_status(struct input_code *, nkf_char);
@@ -477,7 +478,7 @@ struct input_code input_code_list[] = {
     {"UTF-16",     0, 0, 0, {0, 0, 0}, NULL, w_iconv16, 0},
     {"UTF-32",     0, 0, 0, {0, 0, 0}, NULL, w_iconv32, 0},
 #endif
-    {0}
+    {NULL,        0, 0, 0, {0, 0, 0}, NULL, NULL, 0}
 };
 
 static int              mimeout_mode = 0; /* 0, -1, 'Q', 'B', 1, 2 */
@@ -821,7 +822,7 @@ nkf_buf_new(int length)
     buf->capa = length;
     buf->len = 0;
     return buf;
-} 
+}
 
 #if 0
 static void
@@ -1209,10 +1210,10 @@ set_input_encoding(nkf_encoding *enc)
     case ISO_8859_1:
        iso8859_f = TRUE;
        break;
-    case CP50220:
     case CP50221:
     case CP50222:
        if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
+    case CP50220:
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1265,6 +1266,7 @@ set_input_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_MS:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
@@ -1273,6 +1275,7 @@ set_input_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_ASCII:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
@@ -1327,7 +1330,6 @@ set_output_encoding(nkf_encoding *enc)
 {
     switch (nkf_enc_to_index(enc)) {
     case CP50220:
-       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1403,12 +1405,14 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_MS:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_MS;
 #endif
        break;
     case EUCJP_ASCII:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_ASCII;
@@ -2199,8 +2203,8 @@ unicode_iconv(nkf_char wc)
     return 0;
 }
 
-#define NKF_ICONV_NEED_ONE_MORE_BYTE -1
-#define NKF_ICONV_NEED_TWO_MORE_BYTES -2
+#define NKF_ICONV_NEED_ONE_MORE_BYTE (size_t)-1
+#define NKF_ICONV_NEED_TWO_MORE_BYTES (size_t)-2
 #define UTF16_TO_UTF32(lead, trail) (((lead) << 10) + (trail) - NKF_INT32_C(0x35FDC00))
 static size_t
 nkf_iconv_utf_16(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
@@ -3199,6 +3203,7 @@ check_bom(FILE *f)
                        set_iconv(TRUE, w_iconv32);
                    }
                    if (iconv == w_iconv32) {
+                       input_bom_f = TRUE;
                        input_endian = ENDIAN_BIG;
                        return;
                    }
@@ -3229,6 +3234,7 @@ check_bom(FILE *f)
                    set_iconv(TRUE, w_iconv);
                }
                if (iconv == w_iconv) {
+                   input_bom_f = TRUE;
                    return;
                }
                (*i_ungetc)(0xBF,f);
@@ -3257,6 +3263,7 @@ check_bom(FILE *f)
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_BIG;
+               input_bom_f = TRUE;
                return;
            }
            (*i_ungetc)(0xFF,f);
@@ -3272,6 +3279,7 @@ check_bom(FILE *f)
                    }
                    if (iconv == w_iconv32) {
                        input_endian = ENDIAN_LITTLE;
+                       input_bom_f = TRUE;
                        return;
                    }
                    (*i_ungetc)(0x00,f);
@@ -3283,6 +3291,7 @@ check_bom(FILE *f)
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_LITTLE;
+               input_bom_f = TRUE;
                return;
            }
            (*i_ungetc)(0xFE,f);
@@ -3875,7 +3884,7 @@ static const unsigned char *mime_pattern[] = {
 
 /* \e$B3:Ev$9$k%3!<%I$NM%@hEY$r>e$2$k$?$a$NL\0u\e(B */
 nkf_char (*mime_priority_func[])(nkf_char c2, nkf_char c1, nkf_char c0) = {
-    e_iconv, s_iconv, 0, 0, 0, 0,
+    e_iconv, s_iconv, 0, 0, 0, 0, 0,
 #if defined(UTF8_INPUT_ENABLE)
     w_iconv, w_iconv,
 #endif
@@ -4187,8 +4196,13 @@ print_guessed_code(char *filename)
        if (guess_f == 1) {
            printf("%s\n", input_codename);
        } else {
-           printf("%s%s\n",
+           printf("%s%s%s%s\n",
                   input_codename,
+                  iconv != w_iconv16 && iconv != w_iconv32 ? "" :
+                  input_endian == ENDIAN_LITTLE ? " LE" :
+                  input_endian == ENDIAN_BIG ? " BE" :
+                  "[BUG]",
+                  input_bom_f ? " (BOM)" : "",
                   input_eol == CR   ? " (CR)" :
                   input_eol == LF   ? " (LF)" :
                   input_eol == CRLF ? " (CRLF)" :
@@ -4256,7 +4270,7 @@ numchar_getc(FILE *f)
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nungetc;
     int i = 0, j;
     nkf_char buf[12];
-    long c = -1;
+    nkf_char c = -1;
 
     buf[i] = (*g)(f);
     if (buf[i] == '&'){
@@ -4632,7 +4646,7 @@ static const char basis_64[] =
 
 #define MIMEOUT_BUF_LENGTH 74
 static struct {
-    char buf[MIMEOUT_BUF_LENGTH+1];
+    unsigned char buf[MIMEOUT_BUF_LENGTH+1];
     int count;
 } mimeout_state;
 
@@ -4944,7 +4958,7 @@ mime_putc(nkf_char c)
                    i = 0;
 
                    for (; i < mimeout_state.count - len; ++i) {
-                       if (!strncmp(mimeout_state.buf+i, str, len)) {
+                       if (!strncmp((char *)(mimeout_state.buf+i), str, len)) {
                            i += len - 2;
                            break;
                        }
@@ -5415,7 +5429,7 @@ noconvert(FILE *f)
 #define NEXT continue        /* no output, get next */
 #define SKIP c2=0;continue        /* no output, get next */
 #define MORE c2=c1;continue  /* need one more byte */
-#define SEND ;               /* output c1 and c2, get next */
+#define SEND (void)0         /* output c1 and c2, get next */
 #define LAST break           /* end of loop, go closing  */
 #define set_input_mode(mode) do { \
     input_mode = mode; \
@@ -5460,7 +5474,7 @@ kanji_convert(FILE *f)
     else if (iconv == w_iconv16) {
        while ((c1 = (*i_getc)(f)) != EOF &&
               (c2 = (*i_getc)(f)) != EOF) {
-           if (nkf_iconv_utf_16(c1, c2, 0, 0) == -2 &&
+           if (nkf_iconv_utf_16(c1, c2, 0, 0) == NKF_ICONV_NEED_TWO_MORE_BYTES &&
                (c3 = (*i_getc)(f)) != EOF &&
                (c4 = (*i_getc)(f)) != EOF) {
                nkf_iconv_utf_16(c1, c2, c3, c4);
@@ -5871,7 +5885,7 @@ finished:
 
 /*
  * int options(unsigned char *cp)
- * 
+ *
  * return values:
  *    0: success
  *   -1: ArgumentError
@@ -5900,7 +5914,7 @@ options(unsigned char *cp)
                option_mode = 1;
                return 0;
            }
-           for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
+           for (i=0;i<(int)(sizeof(long_option)/sizeof(long_option[0]));i++) {
                p = (unsigned char *)long_option[i].name;
                for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
                if (*p == cp[j] || cp[j] == SP){