Fix: CP932 doesn't imply -x.

[nkf/nkf.git] / nkf.c

diff --git a/nkf.c b/nkf.c
index da1c766..aab1bcd 100644 (file)
--- a/nkf.c
+++ b/nkf.c
@@ -21,7 +21,7 @@
  * 3. This notice may not be removed or altered from any source distribution.
  */
 #define NKF_VERSION "2.0.9"
-#define NKF_RELEASE_DATE "2009-01-20"
+#define NKF_RELEASE_DATE "2009-06-23"
 #define COPY_RIGHT \
     "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa).\n" \
     "Copyright (C) 1996-2009, The nkf Project."
@@ -295,7 +295,7 @@ struct {
                               && (c != '(') && (c != ')') && (c != '.') && (c != 0x22)))
 
 #define is_ibmext_in_sjis(c2) (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END)
-#define nkf_byte_jisx0201_katakana_p(c) (SP <= c && c < (0xE0&0x7F))
+#define nkf_byte_jisx0201_katakana_p(c) (SP <= c && c <= 0x5F)
 
 #define         HOLD_SIZE       1024
 #if defined(INT_IS_SHORT)
@@ -468,6 +468,8 @@ struct input_code input_code_list[] = {
     {"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0},
 #ifdef UTF8_INPUT_ENABLE
     {"UTF-8",     0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0},
+    {"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}
 };
@@ -1204,6 +1206,7 @@ set_input_encoding(nkf_encoding *enc)
     case CP50220:
     case CP50221:
     case CP50222:
+       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1225,6 +1228,7 @@ set_input_encoding(nkf_encoding *enc)
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
+       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1246,6 +1250,7 @@ set_input_encoding(nkf_encoding *enc)
     case EUCJP_NKF:
        break;
     case CP51932:
+       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1325,6 +1330,7 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case CP50221:
+       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1332,6 +1338,11 @@ set_output_encoding(nkf_encoding *enc)
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
        break;
+    case ISO_2022_JP:
+#ifdef SHIFTJIS_CP932
+       if (cp932inv_f == TRUE) cp932inv_f = FALSE;
+#endif
+       break;
     case ISO_2022_JP_1:
        x0212_f = TRUE;
 #ifdef SHIFTJIS_CP932
@@ -1348,6 +1359,7 @@ set_output_encoding(nkf_encoding *enc)
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
+       x0201_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
@@ -1376,6 +1388,7 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case CP51932:
+       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1426,6 +1439,7 @@ set_output_encoding(nkf_encoding *enc)
        output_endian = ENDIAN_LITTLE;
        output_bom_f = TRUE;
        break;
+    case UTF_32:
     case UTF_32BE_BOM:
        output_bom_f = TRUE;
        break;
@@ -1652,7 +1666,7 @@ nkf_unicode_to_utf8(nkf_char val, nkf_char *p1, nkf_char *p2, nkf_char *p3, nkf_
        *p3 = 0x80 | ( val        & 0x3f);
        *p4 = 0;
     } else if (nkf_char_unicode_value_p(val)) {
-       *p1 = 0xe0 |  (val >> 16);
+       *p1 = 0xf0 |  (val >> 18);
        *p2 = 0x80 | ((val >> 12) & 0x3f);
        *p3 = 0x80 | ((val >>  6) & 0x3f);
        *p4 = 0x80 | ( val        & 0x3f);
@@ -3128,18 +3142,16 @@ h_conv(FILE *f, int c1, int c2)
            } else if ((c3 = (*i_getc)(f)) == EOF) {
                ret = EOF;
                break;
-           } else {
-               code_status(c3);
-               if (hold_index < hold_count){
-                   c4 = hold_buf[hold_index++];
-               } else if ((c4 = (*i_getc)(f)) == EOF) {
-                   c3 = ret = EOF;
-                   break;
-               } else {
-                   code_status(c4);
-                   (*iconv)(c1, c2, (c3<<8)|c4);
-               }
            }
+           code_status(c3);
+           if (hold_index < hold_count){
+               c4 = hold_buf[hold_index++];
+           } else if ((c4 = (*i_getc)(f)) == EOF) {
+               c3 = ret = EOF;
+               break;
+           }
+           code_status(c4);
+           (*iconv)(c1, c2, (c3<<8)|c4);
            break;
        case -1:
            /* 3 bytes EUC or UTF-8 */
@@ -5444,6 +5456,12 @@ kanji_convert(FILE *f)
            if (input_mode == JIS_X_0208 && DEL <= c1 && c1 < 0x92) {
                /* CP5022x */
                MORE;
+           }else if (input_codename && input_codename[0] == 'I' &&
+                   0xA1 <= c1 && c1 <= 0xDF) {
+               /* JIS X 0201 Katakana in 8bit JIS */
+               c2 = JIS_X_0201_1976_K;
+               c1 &= 0x7f;
+               SEND;
            } else if (c1 > DEL) {
                /* 8 bit code */
                if (!estab_f && !iso8859_f) {
@@ -5518,7 +5536,7 @@ kanji_convert(FILE *f)
                SKIP;
            } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {
                if ((c1 = (*i_getc)(f)) == EOF) {
-                   /*  (*oconv)(0, ESC); don't send bogus code */
+                   (*oconv)(0, ESC);
                    LAST;
                }
                else if (c1 == '&') {
@@ -5648,7 +5666,7 @@ kanji_convert(FILE *f)
            } else if (c1 == ESC && iconv == s_iconv) {
                /* ESC in Shift_JIS */
                if ((c1 = (*i_getc)(f)) == EOF) {
-                   /*  (*oconv)(0, ESC); don't send bogus code */
+                   (*oconv)(0, ESC);
                    LAST;
                } else if (c1 == '$') {
                    /* J-PHONE emoji */
@@ -6159,9 +6177,6 @@ options(unsigned char *cp)
                    output_endian = ENDIAN_LITTLE;
                } else if (cp[0] == 'B') {
                    cp++;
-               } else {
-                   output_encoding = nkf_enc_from_index(enc_idx);
-                   continue;
                }
                if (cp[0] == '0'){
                    cp++;
@@ -6229,10 +6244,10 @@ options(unsigned char *cp)
               bit:3   Convert HTML Entity
               bit:4   Convert JIS X 0208 Katakana to JIS X 0201 Katakana
             */
-           while ('0'<= *cp && *cp <='9') {
+           while ('0'<= *cp && *cp <='4') {
                alpha_f |= 1 << (*cp++ - '0');
            }
-           if (!alpha_f) alpha_f = 1;
+           alpha_f |= 1;
            continue;
        case 'x':   /* Convert X0201 kana to X0208 or X0201 Conversion */
            x0201_f = FALSE;    /* No X0201->X0208 conversion */