@@ -1,6 +1,7 @@

 *.o

 re

 tests

+tests_utf8

 *.log

 re.valgrind

 sdl.supp

@@ -132,8 +132,13 @@ undostack_t *redata_getstack(redata_t *redata, undo_t *undo);

 char *redata_generic_genname(char *filename,char *prefix, char *postfix, char *buf, int bufsize);

+/* utf8 convenience functions */

+

+/* calculate the number of utf8-charaters in buffer */

 int redata_generic_utf8len(char *ptr, int size);

+/* return a pointer to the "n"th ("col"th) utf8-character in buffer */

 char *redata_generic_utf8col(char *ptr, int size, int col);

+/* returns the len in bytes of the character starting at ptr (zero on error)*/

 int redata_generic_utf8charlen(char *ptr, int maxsize);

@@ -195,61 +195,4 @@ reui_printf(reui_t *ui, int x, int y, char *rgba, char *format, ...)

         return(0);

 }

-int

-reui_utf8len(reui_t *ui, char *ptr, int size)

-{

-        int len,i;

-        /* calculate the number of utf8-charaters in buffer */

-        if(ui==NULL || size<0 || (ptr==NULL && size!=0))

-                return(-1);

-        /* for now we only count the number of code points */

-        /* in UTF8: 0x00-0x7f single byte chars

-         *          0xc0-0xff leading bytes

-         *          0x80-0xbf continuation bytes (ignore these for len)*/

-/*#warning TODO: XXX support combining code points (at least U+0300 - U+036F ( https://en.wikipedia.org/wiki/Combining_character ) */

-        for(len=0,i=0;i<size;i++)

-                len+=((ptr[i]&0xc0)!=0x80)?1:0;

-        return(len);

-/*#warning TODO: XXX Also consider tabs*/

-}

-

-char *

-reui_utf8col(reui_t *ui, char *ptr, int size, int col)

-{

-        int len,i;

-        /* return a pointer to the "n"th ("col"th) utf8-character in buffer */

-        if(ui==NULL || size<0 || (ptr==NULL && size!=0))

-                return(NULL); /* sanity check failed */

-        /* see reui_utf8len() for explanation of algorithm */

-/*#warning TODO: support combining code points (at least U+0300 - U+036F ( https://en.wikipedia.org/wiki/Combining_character ) */

-        if(col>=size)

-                return(NULL);/* col greater than maximum possible char. count */

-        /* skip "col" amount of single byte chars and leading bytes */

-        for(len=0,i=0;len<col && i<size;i++)

-                len+=((ptr[i]&0xc0)!=0x80)?1:0;

-        /* if we landed in a continuation byte, advance until next single byte chars or leading byte */

-        while(i<size && (ptr[i]&0xc0)==0x80)

-                i++;

-        if(i>=size)

-                return(NULL); /* col is beyond end of string */

-        return(ptr+i);

-/*#warning TODO: XXX Also consider tabs*/

-}

-

-int

-reui_utf8charlen(reui_t *ui, char *ptr, int maxsize)

-{

-        int i;

-        /* returns the len in bytes of the character starting at ptr (zero on error)*/

-        if(ui==NULL || ptr==NULL || maxsize<1)

-                return(0); /* sanity check failed */

-/*#warning TODO: support combining code points (at least U+0300 - U+036F ( https://en.wikipedia.org/wiki/Combining_character ) */

-        if(((unsigned char *)ptr)[0]<0x80)

-                return(1); /* single byte char */

-        if((ptr[0]&0xc0)==0x80)

-                return(0); /* error: this is continuation, not leading byte */

-        for(i=1;i<maxsize && (ptr[i]&0xc0)==0x80;i++)

-                ;

-        return(i);

-}

@@ -45,12 +45,3 @@ int reui_printf(reui_t *ui, int x, int y, char *rgba, char *format, ...) __attri

 int reui_printf(reui_t *ui, int x, int y, char *rgba, char *format, ...);

 #endif

-/* calculate the number of utf8-charaters in buffer */

-int reui_utf8len(reui_t *ui, char *ptr, int size);

-

-/* return a pointer to the "n"th ("col"th) utf8-character in buffer */

-char *reui_utf8col(reui_t *ui, char *ptr, int size, int col);

-

-/* returns the len in bytes of the character starting at ptr (zero on error)*/

-int reui_utf8charlen(reui_t *ui, char *ptr, int maxsize);

-