(print_string): Don't ignore garbage bytes following a

[gnu-emacs] / src / casefiddle.c

diff --git a/src/casefiddle.c b/src/casefiddle.c
index 89d528bcabbcea9ea0377057f409a382838e0cec..b6d8b21f606674cf31ca72bc2d9f0d856ccbf7fc 100644 (file)
--- a/src/casefiddle.c
+++ b/src/casefiddle.c
@@ -1,5 +1,5 @@
 /* GNU Emacs case conversion functions.
-   Copyright (C) 1985, 1994 Free Software Foundation, Inc.
+   Copyright (C) 1985, 1994, 1997 Free Software Foundation, Inc.
 
 This file is part of GNU Emacs.
 
@@ -47,21 +47,29 @@ casify_object (flag, obj)
     {
       if (INTEGERP (obj))
        {
-         c = DOWNCASE (obj);
-         if (!inword && c == XFASTINT (obj))
-           c = UPCASE1 (obj);
-         XSETFASTINT (obj, c);
+         c = DOWNCASE (XFASTINT (obj));
+         if (inword)
+           XSETFASTINT (obj, c);
+         else if (c == XFASTINT (obj))
+           {
+             c = UPCASE1 (XFASTINT (obj));
+             XSETFASTINT (obj, c);
+           }
          return obj;
        }
+
       if (STRINGP (obj))
        {
-         int multibyte = !NILP (current_buffer->enable_multibyte_characters);
+         int multibyte = STRING_MULTIBYTE (obj);
 
          obj = Fcopy_sequence (obj);
-         len = XSTRING (obj)->size;
-         for (i = 0; i < len; i++)
+         len = STRING_BYTES (XSTRING (obj));
+
+         /* Scan all single-byte characters from start of string.  */
+         for (i = 0; i < len;)
            {
              c = XSTRING (obj)->data[i];
+
              if (multibyte && c >= 0x80)
                /* A multibyte character can't be handled in this
                    simple loop.  */
@@ -71,23 +79,34 @@ casify_object (flag, obj)
              else if (!UPPERCASEP (c)
                       && (!inword || flag != CASE_CAPITALIZE_UP))
                c = UPCASE1 (c);
+             /* If this char won't fit in a single-byte string.
+                fall out to the multibyte case.  */
+             if (multibyte ? ! ASCII_BYTE_P (c)
+                 : ! SINGLE_BYTE_CHAR_P (c))
+               break;
+
              XSTRING (obj)->data[i] = c;
              if ((int) flag >= (int) CASE_CAPITALIZE)
                inword = SYNTAX (c) == Sword;
+             i++;
            }
+
+         /* If we didn't do the whole string as single-byte,
+            scan the rest in a more complex way.  */
          if (i < len)
            {
              /* The work is not yet finished because of a multibyte
                 character just encountered.  */
-             int fromlen, tolen, j = i;
+             int fromlen, tolen, j = i, j_byte = i;
              char *buf
                = (char *) alloca ((len - i) * MAX_LENGTH_OF_MULTI_BYTE_FORM
                                   + i);
-             char *str, workbuf[4];
+             unsigned char *str, workbuf[4];
 
              /* Copy data already handled.  */
              bcopy (XSTRING (obj)->data, buf, i);
 
+             /* From now on, I counts bytes.  */
              while (i < len)
                {
                  c = STRING_CHAR_AND_LENGTH (XSTRING (obj)->data + i,
@@ -98,13 +117,15 @@ casify_object (flag, obj)
                           && (!inword || flag != CASE_CAPITALIZE_UP))
                    c = UPCASE1 (c);
                  tolen = CHAR_STRING (c, workbuf, str);
-                 bcopy (str, buf + j, tolen);
+                 bcopy (str, buf + j_byte, tolen);
                  i += fromlen;
-                 j += tolen;
+                 j++;
+                 j_byte += tolen;
                  if ((int) flag >= (int) CASE_CAPITALIZE)
                    inword = SYNTAX (c) == Sword;
                }
-             obj = make_string (buf, j);
+             obj = make_specified_string (buf, j, j_byte,
+                                          STRING_MULTIBYTE (obj));
            }
          return obj;
        }
@@ -161,6 +182,7 @@ The argument object is not altered--the value is a copy.")
 /* flag is CASE_UP, CASE_DOWN or CASE_CAPITALIZE or CASE_CAPITALIZE_UP.
    b and e specify range of buffer to operate on. */
 
+void
 casify_region (flag, b, e)
      enum case_action flag;
      Lisp_Object b, e;
@@ -170,6 +192,7 @@ casify_region (flag, b, e)
   register int inword = flag == CASE_DOWN;
   register int multibyte = !NILP (current_buffer->enable_multibyte_characters);
   int start, end;
+  int start_byte, end_byte;
   Lisp_Object ch, downch, val;
 
   if (EQ (b, e))
@@ -185,8 +208,10 @@ casify_region (flag, b, e)
   end = XFASTINT (e);
   modify_region (current_buffer, start, end);
   record_change (start, end - start);
+  start_byte = CHAR_TO_BYTE (start);
+  end_byte = CHAR_TO_BYTE (end);
 
-  for (i = start; i < end; i++)
+  for (i = start_byte; i < end_byte; i++)
     {
       c = FETCH_BYTE (i);
       if (multibyte && c >= 0x80)
@@ -201,13 +226,15 @@ casify_region (flag, b, e)
       if ((int) flag >= (int) CASE_CAPITALIZE)
        inword = SYNTAX (c) == Sword;
     }
-  if (i < end)
+  if (i < end_byte)
     {
       /* The work is not yet finished because of a multibyte character
         just encountered.  */
-      int opoint = PT, c2;
+      int opoint = PT;
+      int opoint_byte = PT_BYTE;
+      int c2;
 
-      while (i < end)
+      while (i < end_byte)
        {
          if ((c = FETCH_BYTE (i)) >= 0x80)
            c = FETCH_MULTIBYTE_CHAR (i);
@@ -220,7 +247,7 @@ casify_region (flag, b, e)
          if (c != c2)
            {
              int fromlen, tolen, j;
-             char workbuf[4], *str;
+             unsigned char workbuf[4], *str;
 
              /* Handle the most likely case */
              if (c < 0400 && c2 < 0400)
@@ -241,7 +268,7 @@ casify_region (flag, b, e)
                  else if (tolen > fromlen)
                    {
                      TEMP_SET_PT (i + fromlen);
-                     insert_1 (str + fromlen, tolen - fromlen, 1, 0);
+                     insert_1 (str + fromlen, tolen - fromlen, 1, 0, 0);
                    }
 #endif
                }
@@ -250,7 +277,7 @@ casify_region (flag, b, e)
            inword = SYNTAX (c2) == Sword;
          INC_POS (i);
        }
-      TEMP_SET_PT (opoint);
+      TEMP_SET_PT_BOTH (opoint, opoint_byte);
     }
 
   signal_after_change (start, end - start, end - start);
@@ -378,6 +405,7 @@ With negative argument, capitalize previous words but do not move.")
   return Qnil;
 }
 \f
+void
 syms_of_casefiddle ()
 {
   Qidentity = intern ("identity");
@@ -395,6 +423,7 @@ syms_of_casefiddle ()
   defsubr (&Scapitalize_word);
 }
 
+void
 keys_of_casefiddle ()
 {
   initial_define_key (control_x_map, Ctl('U'), "upcase-region");