(mdatabase__dir_list): Renamed from mdb_dir_list.

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

diff --git a/src/mtext.c b/src/mtext.c
index f117cf2..e397c36 100644 (file)
--- a/src/mtext.c
+++ b/src/mtext.c
@@ -96,14 +96,6 @@ static M17NObjectArray mtext_table;
 
 static MSymbol M_charbag;
 
-#ifdef WORDS_BIGENDIAN
-static enum MTextFormat default_utf_16 = MTEXT_FORMAT_UTF_16BE;
-static enum MTextFormat default_utf_32 = MTEXT_FORMAT_UTF_32BE;
-#else
-static enum MTextFormat default_utf_16 = MTEXT_FORMAT_UTF_16LE;
-static enum MTextFormat default_utf_32 = MTEXT_FORMAT_UTF_32LE;
-#endif
-
 /** Increment character position CHAR_POS and unit position UNIT_POS
     so that they point to the next character in M-text MT.  No range
     check for CHAR_POS and UNIT_POS.  */
@@ -121,7 +113,7 @@ static enum MTextFormat default_utf_32 = MTEXT_FORMAT_UTF_32LE;
       {                                                                \
        c = ((unsigned short *) ((mt)->data))[(unit_pos)];      \
                                                                \
-       if ((mt)->format != default_utf_16)                     \
+       if ((mt)->format != MTEXT_FORMAT_UTF_16)                \
          c = SWAP_16 (c);                                      \
        (unit_pos) += CHAR_UNITS_BY_HEAD_UTF16 (c);             \
       }                                                                \
@@ -149,7 +141,7 @@ static enum MTextFormat default_utf_32 = MTEXT_FORMAT_UTF_32LE;
       {                                                                        \
        int c = ((unsigned short *) ((mt)->data))[(unit_pos) - 1];      \
                                                                        \
-       if ((mt)->format != default_utf_16)                             \
+       if ((mt)->format != MTEXT_FORMAT_UTF_16)                        \
          c = SWAP_16 (c);                                              \
        (unit_pos) -= 2 - (c < 0xD800 || c >= 0xE000);                  \
       }                                                                        \
@@ -251,8 +243,8 @@ insert (MText *mt1, int pos, MText *mt2, int from, int to)
       /* Be sure to make mt1->format sufficient to contain all
         characters in mt2.  */
       if (mt1->format == MTEXT_FORMAT_UTF_8
-         || mt1->format == default_utf_32
-         || (mt1->format == default_utf_16
+         || mt1->format == MTEXT_FORMAT_UTF_32
+         || (mt1->format == MTEXT_FORMAT_UTF_16
              && mt2->format <= MTEXT_FORMAT_UTF_16BE
              && mt2->format != MTEXT_FORMAT_UTF_8))
        ;
@@ -260,8 +252,8 @@ insert (MText *mt1, int pos, MText *mt2, int from, int to)
        {
          if (mt2->format == MTEXT_FORMAT_UTF_8)
            mt1->format = MTEXT_FORMAT_UTF_8;
-         else if (mt2->format == default_utf_16
-                  || mt2->format == default_utf_32)
+         else if (mt2->format == MTEXT_FORMAT_UTF_16
+                  || mt2->format == MTEXT_FORMAT_UTF_32)
            mtext__adjust_format (mt1, mt2->format);
          else
            mtext__adjust_format (mt1, MTEXT_FORMAT_UTF_8);
@@ -313,7 +305,7 @@ insert (MText *mt1, int pos, MText *mt2, int from, int to)
          p += CHAR_STRING_UTF8 (c, p);
        }
     }
-  else if (mt1->format == default_utf_16)
+  else if (mt1->format == MTEXT_FORMAT_UTF_16)
     {
       unsigned short *p;
       int total_bytes, i, c;
@@ -335,7 +327,7 @@ insert (MText *mt1, int pos, MText *mt2, int from, int to)
          p += CHAR_STRING_UTF16 (c, p);
        }
     }
-  else                         /* default_utf_32 */
+  else                         /* MTEXT_FORMAT_UTF_32 */
     {
       unsigned int *p;
       int total_bytes, i;
@@ -457,22 +449,18 @@ count_utf_16_chars (const void *data, int nitems, int swap)
       if (prev_surrogate)
        {
          if (c < 0xDC00 || c >= 0xE000)
-           return -1;
-         prev_surrogate = 0;
+           /* Invalid surrogate */
+           nchars++;
        }
       else
        {
-         if (c < 0xD800)
-           ;
-         else if (c < 0xDC00)
+         if (c >= 0xD800 && c < 0xDC00)
            prev_surrogate = 1;
-         else if (c < 0xE000)
-           return -1;
          nchars++;
        }
     }
   if (prev_surrogate)
-    return -1;
+    nchars++;
   return nchars;
 }
 
@@ -492,7 +480,7 @@ find_char_forward (MText *mt, int from, int to, int c)
     {
       unsigned short *p = (unsigned short *) (mt->data) + from_byte;
 
-      if (mt->format == default_utf_16)
+      if (mt->format == MTEXT_FORMAT_UTF_16)
        while (from < to && STRING_CHAR_ADVANCE_UTF16 (p) != c) from++;
       else if (c < 0x10000)
        {
@@ -524,7 +512,7 @@ find_char_forward (MText *mt, int from, int to, int c)
       unsigned *p = (unsigned *) (mt->data) + from_byte;
       unsigned c1 = c;
 
-      if (mt->format != default_utf_32)
+      if (mt->format != MTEXT_FORMAT_UTF_32)
        c1 = SWAP_32 (c1);
       while (from < to && *p++ != c1) from++;
     }
@@ -554,7 +542,7 @@ find_char_backward (MText *mt, int from, int to, int c)
     {
       unsigned short *p = (unsigned short *) (mt->data) + to_byte;
 
-      if (mt->format == default_utf_16)
+      if (mt->format == MTEXT_FORMAT_UTF_16)
        {
          while (from < to)
            {
@@ -594,7 +582,7 @@ find_char_backward (MText *mt, int from, int to, int c)
       unsigned *p = (unsigned *) (mt->data) + to_byte;
       unsigned c1 = c;
 
-      if (mt->format != default_utf_32)
+      if (mt->format != MTEXT_FORMAT_UTF_32)
        c1 = SWAP_32 (c1);
       while (from < to && p[-1] != c1) to--, p--;
     }
@@ -887,7 +875,7 @@ mtext__from_data (const void *data, int nitems, enum MTextFormat format,
   else if (format <= MTEXT_FORMAT_UTF_16BE)
     {
       if ((nchars = count_utf_16_chars (data, nitems,
-                                       format != default_utf_16)) < 0)
+                                       format != MTEXT_FORMAT_UTF_16)) < 0)
        MERROR (MERROR_MTEXT, NULL);
       nbytes = USHORT_SIZE * nitems;
       unit_bytes = USHORT_SIZE;
@@ -956,7 +944,7 @@ mtext__adjust_format (MText *mt, enum MTextFormat format)
        }
 
       default:
-       if (format == default_utf_16)
+       if (format == MTEXT_FORMAT_UTF_16)
          {
            unsigned short *p0, *p1;
 
@@ -1022,7 +1010,8 @@ mtext__bol (MText *mt, int pos)
   else if (mt->format <= MTEXT_FORMAT_UTF_16BE)
     {
       unsigned short *p = ((unsigned short *) (mt->data)) + byte_pos;
-      unsigned short newline = mt->format == default_utf_16 ? 0x0A00 : 0x000A;
+      unsigned short newline = (mt->format == MTEXT_FORMAT_UTF_16
+                               ? 0x0A00 : 0x000A);
 
       if (p[-1] == newline)
        return pos;
@@ -1037,7 +1026,8 @@ mtext__bol (MText *mt, int pos)
   else
     {
       unsigned *p = ((unsigned *) (mt->data)) + byte_pos;
-      unsigned newline = mt->format == default_utf_32 ? 0x0A000000 : 0x0000000A;
+      unsigned newline = (mt->format == MTEXT_FORMAT_UTF_32
+                         ? 0x0A000000 : 0x0000000A);
 
       if (p[-1] == newline)
        return pos;
@@ -1080,7 +1070,8 @@ mtext__eol (MText *mt, int pos)
     {
       unsigned short *p = ((unsigned short *) (mt->data)) + byte_pos;
       unsigned short *endp;
-      unsigned short newline = mt->format == default_utf_16 ? 0x0A00 : 0x000A;
+      unsigned short newline = (mt->format == MTEXT_FORMAT_UTF_16
+                               ? 0x0A00 : 0x000A);
 
       if (*p == newline)
        return pos + 1;
@@ -1097,7 +1088,8 @@ mtext__eol (MText *mt, int pos)
     {
       unsigned *p = ((unsigned *) (mt->data)) + byte_pos;
       unsigned *endp;
-      unsigned newline = mt->format == default_utf_32 ? 0x0A000000 : 0x0000000A;
+      unsigned newline = (mt->format == MTEXT_FORMAT_UTF_32
+                         ? 0x0A000000 : 0x0000000A);
 
       if (*p == newline)
        return pos + 1;
@@ -1115,6 +1107,18 @@ mtext__eol (MText *mt, int pos)
 \f
 /* External API */
 
+#ifdef WORDS_BIGENDIAN
+const int MTEXT_FORMAT_UTF_16 = MTEXT_FORMAT_UTF_16BE;
+#else
+const int MTEXT_FORMAT_UTF_16 = MTEXT_FORMAT_UTF_16LE;
+#endif
+
+#ifdef WORDS_BIGENDIAN
+const int MTEXT_FORMAT_UTF_32 = MTEXT_FORMAT_UTF_32BE;
+#else
+const int MTEXT_FORMAT_UTF_32 = MTEXT_FORMAT_UTF_32LE;
+#endif
+
 /*** @addtogroup m17nMtext */
 /*** @{ */
 /*=*/
@@ -1174,7 +1178,7 @@ mtext ()
     The contents of $DATA must not be modified while the M-text is alive.
 
     The allocated M-text will not be freed unless the user explicitly
-    does so with the m17n_object_free () function.  Even in that case,
+    does so with the m17n_object_unref () function.  Even in that case,
     $DATA is not freed.
 
     @return
@@ -1202,7 +1206,7 @@ mtext ()
     ³ä¤êÅö¤Æ¤é¤ì¤¿ M-text ¤Îʸ»úÎó¤ÏÊѹ¹¤Ç¤­¤Ê¤¤¡£$DATA ¤ÎÆâÍÆ¤Ï 
     M-text ¤¬Í­¸ú¤Ê´Ö¤ÏÊѹ¹¤·¤Æ¤Ï¤Ê¤é¤Ê¤¤¡£
 
-    ³ä¤êÅö¤Æ¤é¤ì¤¿ M-text ¤Ï¡¢´Ø¿ô m17n_object_free () ¤Ë¤è¤Ã¤Æ¥æ¡¼¥¶
+    ³ä¤êÅö¤Æ¤é¤ì¤¿ M-text ¤Ï¡¢´Ø¿ô m17n_object_unref () ¤Ë¤è¤Ã¤Æ¥æ¡¼¥¶
     ¤¬ÌÀ¼¨Åª¤Ë¹Ô¤Ê¤ï¤Ê¤¤¸Â¤ê¡¢²òÊü¤µ¤ì¤Ê¤¤¡£¤½¤Î¾ì¹ç¤Ç¤â $DATA ¤Ï²òÊü
     ¤µ¤ì¤Ê¤¤¡£
 
@@ -1285,7 +1289,7 @@ mtext_ref_char (MText *mt, int pos)
        = (unsigned short *) (mt->data) + POS_CHAR_TO_BYTE (mt, pos);
       unsigned short p1[2];
 
-      if (mt->format != default_utf_16)
+      if (mt->format != MTEXT_FORMAT_UTF_16)
        {
          p1[0] = SWAP_16 (*p);
          if (p1[0] >= 0xD800 || p1[0] < 0xDC00)
@@ -1297,7 +1301,7 @@ mtext_ref_char (MText *mt, int pos)
   else
     {
       c = ((unsigned *) (mt->data))[pos];
-      if (mt->format != default_utf_32)
+      if (mt->format != MTEXT_FORMAT_UTF_32)
        c = SWAP_32 (c);
     }
   return c;
@@ -1355,11 +1359,11 @@ mtext_set_char (MText *mt, int pos, int c)
     {
       if (c >= 0x110000)
        mtext__adjust_format (mt, MTEXT_FORMAT_UTF_8);
-      else if (mt->format != default_utf_16)
-       mtext__adjust_format (mt, default_utf_16);
+      else if (mt->format != MTEXT_FORMAT_UTF_16)
+       mtext__adjust_format (mt, MTEXT_FORMAT_UTF_16);
     }
-  else if (mt->format != default_utf_32)
-    mtext__adjust_format (mt, default_utf_32);
+  else if (mt->format != MTEXT_FORMAT_UTF_32)
+    mtext__adjust_format (mt, MTEXT_FORMAT_UTF_32);
 
   unit_bytes = UNIT_BYTES (mt->format);
   pos_unit = POS_CHAR_TO_BYTE (mt, pos);
@@ -1397,7 +1401,7 @@ mtext_set_char (MText *mt, int pos, int c)
        break;
       }
     default:
-      if (mt->format == default_utf_16)
+      if (mt->format == MTEXT_FORMAT_UTF_16)
        {
          unsigned short *p = (unsigned short *) mt->data + pos_unit;
 
@@ -1458,13 +1462,13 @@ mtext_cat_char (MText *mt, int c)
     }
   else if (mt->format >= MTEXT_FORMAT_UTF_32LE)
     {
-      if (mt->format != default_utf_32)
-       mtext__adjust_format (mt, default_utf_32);
+      if (mt->format != MTEXT_FORMAT_UTF_32)
+       mtext__adjust_format (mt, MTEXT_FORMAT_UTF_32);
     }
   else if (mt->format >= MTEXT_FORMAT_UTF_16LE)
     {
-      if (mt->format != default_utf_16)
-       mtext__adjust_format (mt, default_utf_16);
+      if (mt->format != MTEXT_FORMAT_UTF_16)
+       mtext__adjust_format (mt, MTEXT_FORMAT_UTF_16);
     }
 
   nunits = CHAR_UNITS (c, mt->format);
@@ -1480,7 +1484,7 @@ mtext_cat_char (MText *mt, int c)
       p += CHAR_STRING_UTF8 (c, p);
       *p = 0;
     }
-  else if (mt->format == default_utf_16)
+  else if (mt->format == MTEXT_FORMAT_UTF_16)
     {
       unsigned short *p = (unsigned short *) mt->data + mt->nbytes;
       p += CHAR_STRING_UTF16 (c, p);
@@ -1992,13 +1996,13 @@ mtext_ins_char (MText *mt, int pos, int c, int n)
     }
   else if (mt->format >= MTEXT_FORMAT_UTF_32LE)
     {
-      if (mt->format != default_utf_32)
-       mtext__adjust_format (mt, default_utf_32);
+      if (mt->format != MTEXT_FORMAT_UTF_32)
+       mtext__adjust_format (mt, MTEXT_FORMAT_UTF_32);
     }
   else if (mt->format >= MTEXT_FORMAT_UTF_16LE)
     {
-      if (mt->format != default_utf_16)
-       mtext__adjust_format (mt, default_utf_16);
+      if (mt->format != MTEXT_FORMAT_UTF_16)
+       mtext__adjust_format (mt, MTEXT_FORMAT_UTF_16);
     }
 
   nunits = CHAR_UNITS (c, mt->format);
@@ -2023,7 +2027,7 @@ mtext_ins_char (MText *mt, int pos, int c, int n)
       for (i = 0; i < n; i++)
        p += CHAR_STRING_UTF8 (c, p);
     }
-  else if (mt->format == default_utf_16)
+  else if (mt->format == MTEXT_FORMAT_UTF_16)
     {
       unsigned short *p = (unsigned short *) mt->data + pos_unit;
 
@@ -2752,26 +2756,24 @@ mdebug_dump_mtext (MText *mt, int indent, int fullp)
   memset (prefix, 32, indent);
   prefix[indent] = 0;
 
+  fprintf (stderr,
+          "(mtext (size %d %d %d) (cache %d %d)",
+          mt->nchars, mt->nbytes, mt->allocated,
+          mt->cache_char_pos, mt->cache_byte_pos);
   if (! fullp)
     {
-      fprintf (stderr, "\"");
-      for (i = 0; i < mt->nbytes; i++)
+      fprintf (stderr, " \"");
+      for (i = 0; i < mt->nchars; i++)
        {
-         int c = mt->data[i];
+         int c = mtext_ref_char (mt, i);
          if (c >= ' ' && c < 127)
            fprintf (stderr, "%c", c);
          else
            fprintf (stderr, "\\x%02X", c);
        }
       fprintf (stderr, "\"");
-      return mt;
     }
-
-  fprintf (stderr,
-          "(mtext (size %d %d %d) (cache %d %d)",
-          mt->nchars, mt->nbytes, mt->allocated,
-          mt->cache_char_pos, mt->cache_byte_pos);
-  if (mt->nchars > 0)
+  else if (mt->nchars > 0)
     {
       fprintf (stderr, "\n%s (bytes \"", prefix);
       for (i = 0; i < mt->nbytes; i++)
@@ -2784,7 +2786,7 @@ mdebug_dump_mtext (MText *mt, int indent, int fullp)
          int len;
          int c = STRING_CHAR_AND_BYTES (p, len);
 
-         if (c >= ' ' && c < 127 && c != '\\' && c != '"')
+         if (c >= ' ' && c < 127 && c != '\\' && c != '\"')
            fputc (c, stderr);
          else
            fprintf (stderr, "\\x%X", c);