(Mdetail_text, minput_get_description, minput_get_commands)

[m17n/m17n-lib.git] / src / coding.c

diff --git a/src/coding.c b/src/coding.c
index 533970e..986958b 100644 (file)
--- a/src/coding.c
+++ b/src/coding.c
@@ -99,7 +99,7 @@ typedef struct
       reseting a converter.  */
   int (*resetter) (MConverter *converter);
 
-  int (*decoder) (unsigned char *str, int str_bytes, MText *mt,
+  int (*decoder) (const unsigned char *str, int str_bytes, MText *mt,
                  MConverter *converter);
 
   int (*encoder) (MText *mt, int from, int to,
@@ -390,8 +390,8 @@ encode_unsupporeted_char (int c, unsigned char *dst, unsigned char *dst_end,
 
 static int
 finish_decoding (MText *mt, MConverter *converter, int nchars,
-                unsigned char *source, unsigned char *src_end,
-                unsigned char *src,
+                const unsigned char *source, const unsigned char *src_end,
+                const unsigned char *src,
                 int error)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
@@ -420,7 +420,7 @@ finish_decoding (MText *mt, MConverter *converter, int nchars,
     {
       unsigned char *dst = mt->data + mt->nbytes;
       unsigned char *dst_end = mt->data + mt->allocated;
-      unsigned char *src_stop = src_end;
+      const unsigned char *src_stop = src_end;
       int c;
       int last_nchars = nchars;
 
@@ -511,15 +511,15 @@ reset_coding_charset (MConverter *converter)
 }
 
 static int
-decode_coding_charset (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_charset (const unsigned char *source, int src_bytes, MText *mt,
                       MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
   MCodingSystem *coding = internal->coding;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated;
   int nchars = 0;
@@ -712,15 +712,15 @@ encode_coding_charset (MText *mt, int from, int to,
 
 
 static int
-decode_coding_utf_8 (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_utf_8 (const unsigned char *source, int src_bytes, MText *mt,
                     MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
   MCodingSystem *coding = internal->coding;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated;
   int nchars = 0;
@@ -921,14 +921,14 @@ reset_coding_utf (MConverter *converter)
 }
 
 static int
-decode_coding_utf_16 (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_utf_16 (const unsigned char *source, int src_bytes, MText *mt,
                      MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated;
   int nchars = 0;
@@ -1025,14 +1025,14 @@ decode_coding_utf_16 (unsigned char *source, int src_bytes, MText *mt,
 
 
 static int
-decode_coding_utf_32 (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_utf_32 (const unsigned char *source, int src_bytes, MText *mt,
                      MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated;
   int nchars = 0;
@@ -1524,15 +1524,15 @@ find_ctext_non_standard_charset (char *charset_name)
 }
 
 static int
-decode_coding_iso_2022 (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_iso_2022 (const unsigned char *source, int src_bytes, MText *mt,
                       MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
   MCodingSystem *coding = internal->coding;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated;
   int nchars = 0;
@@ -2645,15 +2645,15 @@ reset_coding_sjis (MConverter *converter)
 }
 
 static int
-decode_coding_sjis (unsigned char *source, int src_bytes, MText *mt,
+decode_coding_sjis (const unsigned char *source, int src_bytes, MText *mt,
                    MConverter *converter)
 {
   MConverterStatus *internal = (MConverterStatus *) converter->internal_info;
   MCodingSystem *coding = internal->coding;
-  unsigned char *src = internal->carryover;
-  unsigned char *src_stop = src + internal->carryover_bytes;
-  unsigned char *src_end = source + src_bytes;
-  unsigned char *src_base;
+  const unsigned char *src = internal->carryover;
+  const unsigned char *src_stop = src + internal->carryover_bytes;
+  const unsigned char *src_end = source + src_bytes;
+  const unsigned char *src_base;
   unsigned char *dst = mt->data + mt->nbytes;
   unsigned char *dst_end = mt->data + mt->allocated - MAX_UTF8_CHAR_BYTES;
   int nchars = 0;
@@ -3774,9 +3774,9 @@ MSymbol Mcoding;
     @c MERROR_CODING  */
 
 MSymbol
-mconv_define_coding (char *name, MPlist *plist,
+mconv_define_coding (const char *name, MPlist *plist,
                     int (*resetter) (MConverter *),
-                    int (*decoder) (unsigned char *, int, MText *,
+                    int (*decoder) (const unsigned char *, int, MText *,
                                     MConverter *),
                     int (*encoder) (MText *, int, int,
                                     unsigned char *, int,
@@ -4049,11 +4049,11 @@ mconv_list_codings (MSymbol **symbols)
     @brief Create a code converter bound to a buffer.
 
     The mconv_buffer_converter () function creates a pointer to a code
-    converter for coding system $CODING.  The code converter is bound
+    converter for coding system $NAME.  The code converter is bound
     to buffer area of $N bytes pointed to by $BUF.  Subsequent
     decodings and encodings are done to/from this buffer area.
 
-    $CODING can be #Mnil.  In this case, a coding system associated
+    $NAME can be #Mnil.  In this case, a coding system associated
     with the current locale (LC_CTYPE) is used.
 
     @return
@@ -4064,12 +4064,12 @@ mconv_list_codings (MSymbol **symbols)
 /***ja
     @brief ¥Ð¥Ã¥Õ¥¡¤Ë·ë¤ÓÉÕ¤±¤é¤ì¤¿¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤òºî¤ë.
 
-    ´Ø¿ô mconv_buffer_converter () ¤Ï¡¢¥³¡¼¥É·Ï $CODING ÍѤΥ³¡¼¥É¥³¥ó
+    ´Ø¿ô mconv_buffer_converter () ¤Ï¡¢¥³¡¼¥É·Ï $NAME ÍѤΥ³¡¼¥É¥³¥ó
     ¥Ð¡¼¥¿¤òºî¤ë¡£¤³¤Î¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤Ï¡¢$BUF ¤Ç¼¨¤µ¤ì¤ëÂ礭¤µ $N ¥Ð
     ¥¤¥È¤Î¥Ð¥Ã¥Õ¥¡Îΰè¤Ë·ë¤ÓÉÕ¤±¤é¤ì¤ë¡£¤³¤ì°Ê¹ß¤Î¥Ç¥³¡¼¥É¤ª¤è¤Ó
     ¥¨¥ó¥³¡¼¥É¤Ï¡¢¤³¤Î¥Ð¥Ã¥Õ¥¡Îΰè¤ËÂФ·¤Æ¹Ô¤Ê¤ï¤ì¤ë¡£
 
-    $CODING ¤Ï #Mnil ¤Ç¤¢¤Ã¤Æ¤â¤è¤¤¡£¤³¤Î¾ì¹ç¤Ï¸½ºß¤Î¥í¥±¡¼¥ë 
+    $NAME ¤Ï #Mnil ¤Ç¤¢¤Ã¤Æ¤â¤è¤¤¡£¤³¤Î¾ì¹ç¤Ï¸½ºß¤Î¥í¥±¡¼¥ë 
     (LC_CTYPE) ¤Ë´ØÏ¢ÉÕ¤±¤é¤ì¤¿¥³¡¼¥É·Ï¤¬»È¤ï¤ì¤ë¡£
 
     @return
@@ -4126,12 +4126,12 @@ mconv_buffer_converter (MSymbol name, unsigned char *buf, int n)
 /***en
     @brief Create a code converter bound to a stream.
 
-    The mconv_stream_converter () function create a pointer to a code
-    converter for coding system $CODING.  The code converter is bound
+    The mconv_stream_converter () function creates a pointer to a code
+    converter for coding system $NAME.  The code converter is bound
     to stream $FP.  Subsequent decodings and encodings are done
     to/from this stream.
 
-    $CODING can be #Mnil.  In this case, a coding system associated
+    $NAME can be #Mnil.  In this case, a coding system associated
     with the current locale (LC_CTYPE) is used.
 
     @return If the operation was successful, mconv_stream_converter ()
@@ -4142,12 +4142,12 @@ mconv_buffer_converter (MSymbol name, unsigned char *buf, int n)
 /***ja
     @brief ¥¹¥È¥ê¡¼¥à¤Ë·ë¤ÓÉÕ¤±¤é¤ì¤¿¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤òºî¤ë.
 
-    ´Ø¿ô mconv_stream_converter () ¤Ï¡¢¥³¡¼¥É·Ï $CODING ÍѤΥ³¡¼¥É¥³¥ó
+    ´Ø¿ô mconv_stream_converter () ¤Ï¡¢¥³¡¼¥É·Ï $NAME ÍѤΥ³¡¼¥É¥³¥ó
     ¥Ð¡¼¥¿¤òºî¤ë¡£¤³¤Î¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤Ï¡¢¥¹¥È¥ê¡¼¥à $FP ¤Ë·ë¤ÓÉÕ¤±¤é
     ¤ì¤ë¡£¤³¤ì°Ê¹ß¤Î¥Ç¥³¡¼¥É¤ª¤è¤Ó¥¨¥ó¥³¡¼¥É¤Ï¡¢¤³¤Î¥¹¥È¥ê¡¼¥à¤ËÂФ·¤Æ
     ¹Ô¤Ê¤ï¤ì¤ë¡£
 
-    $CODING ¤Ï #Mnil ¤Ç¤¢¤Ã¤Æ¤â¤è¤¤¡£¤³¤Î¾ì¹ç¤Ï¸½ºß¤Î¥í¥±¡¼¥ë 
+    $NAME ¤Ï #Mnil ¤Ç¤¢¤Ã¤Æ¤â¤è¤¤¡£¤³¤Î¾ì¹ç¤Ï¸½ºß¤Î¥í¥±¡¼¥ë 
     (LC_CTYPE) ¤Ë´ØÏ¢ÉÕ¤±¤é¤ì¤¿¥³¡¼¥É·Ï¤¬»È¤ï¤ì¤ë¡£
 
     @return 
@@ -4367,7 +4367,8 @@ mconv_rebind_stream (MConverter *converter, FILE *fp)
 
     The mconv_decode () function decodes a byte sequence and appends
     the result at the end of M-text $MT.  The source byte sequence is
-    taken from currently bound the buffer area or the stream.
+    taken from either the buffer area or the stream that is currently
+    bound to $CONVERTER.
 
     @return
     If the operation was successful, mconv_decode () returns updated
@@ -4378,8 +4379,8 @@ mconv_rebind_stream (MConverter *converter, FILE *fp)
     @brief ¥Ð¥¤¥ÈÎó¤ò M-text ¤Ë¥Ç¥³¡¼¥É¤¹¤ë.
 
     ´Ø¿ô mconv_decode () ¤Ï¡¢¥Ð¥¤¥ÈÎó¤ò¥Ç¥³¡¼¥É¤·¤Æ¤½¤Î·ë²Ì¤ò M-text
-    $MT ¤ÎËöÈø¤ËÄɲ乤롣¥Ç¥³¡¼¥É¸µ¤Î¥Ð¥¤¥ÈÎó¤Ï¡¢¸½ºß·ë¤ÓÉÕ¤±¤é¤ì¤Æ¤¤¤ë
-    ¥Ð¥Ã¥Õ¥¡Îΰ褢¤ë¤¤¤Ï¥¹¥È¥ê¡¼¥à¤«¤é¼è¤é¤ì¤ë¡£
+    $MT ¤ÎËöÈø¤ËÄɲ乤롣¥Ç¥³¡¼¥É¸µ¤Î¥Ð¥¤¥ÈÎó¤Ï¡¢$CONVERTER ¤Ë¸½ºß·ë
+    ¤ÓÉÕ¤±¤é¤ì¤Æ¤¤¤ë¥Ð¥Ã¥Õ¥¡Îΰ褢¤ë¤¤¤Ï¥¹¥È¥ê¡¼¥à¤«¤é¼è¤é¤ì¤ë¡£
 
     @return
     ¤â¤·½èÍý¤¬À®¸ù¤¹¤ì¤Ð¡¢mconv_decode () ¤Ï¹¹¿·¤µ¤ì¤¿ $MT ¤òÊÖ¤¹¡£¤½
@@ -4404,6 +4405,9 @@ mconv_decode (MConverter *converter, MText *mt)
 
   M_CHECK_READONLY (mt, NULL);
 
+  if (mt->format != MTEXT_FORMAT_UTF_8)
+    mtext__adjust_format (mt, MTEXT_FORMAT_UTF_8);
+
   if (! mt->data)
     mtext__enlarge (mt, MAX_UTF8_CHAR_BYTES);
 
@@ -4503,7 +4507,7 @@ mconv_decode (MConverter *converter, MText *mt)
 /***en
     @brief Decode a buffer area based on a coding system.
 
-    The mconv_decode_buffer () function decodes $N bytes of buffer
+    The mconv_decode_buffer () function decodes $N bytes of the buffer
     area pointed to by $BUF based on the coding system $NAME.  A
     temporary code converter for decoding is automatically created
     and freed.
@@ -4656,7 +4660,7 @@ mconv_encode (MConverter *converter, MText *mt)
     error code to the external variable #merror_code.  */
 
 /***ja
-    @brief M-text ¤Î°ìÉô¤ò¤ò¥Ð¥¤¥ÈÎó¤Ë¥¨¥ó¥³¡¼¥É¤¹¤ë.
+    @brief M-text ¤Î°ìÉô¤ò¥Ð¥¤¥ÈÎó¤Ë¥¨¥ó¥³¡¼¥É¤¹¤ë.
 
     ´Ø¿ô mconv_encode_range () ¤Ï¡¢M-text $MT ¤Î $FROM ¡Ê´Þ¤à¡Ë¤«¤é 
     $TO ¡Ê´Þ¤Þ¤Ê¤¤¡Ë¤Þ¤Ç¤ÎÈϰϤΥƥ­¥¹¥È¤ò¥¨¥ó¥³¡¼¥É¤·¤Æ¡¢¥³¡¼¥É¥³¥ó¥Ð¡¼
@@ -4957,7 +4961,7 @@ mconv_ungetc (MConverter *converter, int c)
     an error code to the external variable #merror_code.  */
 
 /***ja
-    @brief ¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤ò·Ðͳ¤Ç1ʸ»ú½ñ¤¯.
+    @brief ¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿¤ò·Ðͳ¤·¤Æ1ʸ»ú½ñ¤¯.
 
     ´Ø¿ô mconv_putc () ¤Ï¡¢¥³¡¼¥É¥³¥ó¥Ð¡¼¥¿ $CONVERTER ¤Ë¸½ºß·ë¤ÓÉÕ¤±
     ¤é¤ì¤Æ¤¤¤ë¥Ð¥Ã¥Õ¥¡Îΰ褢¤ë¤¤¤Ï¥¹¥È¥ê¡¼¥à¤Ëʸ»ú $C ¤ò½ñ¤­½Ð¤¹¡£Ê¸»ú
@@ -5037,6 +5041,9 @@ mconv_gets (MConverter *converter, MText *mt)
   int c;
 
   M_CHECK_READONLY (mt, NULL);
+  if (mt->format != MTEXT_FORMAT_UTF_8)
+    mtext__adjust_format (mt, MTEXT_FORMAT_UTF_8);
+
   while (1)
     {
       c = mconv_getc (converter);