patch [#29859] UTF-8 support for text editor.

This also fixes cursor movement in the beginning of line and adds do_versions block for converting text files with old extended ascii encoding into UTF-8.

6 files changed, 869 insertions, 554 deletions

diff --git a/source/blender/blenkernel/BKE_text.h b/source/blender/blenkernel/BKE_text.h
index a951347d946..ffea8e0c8e2 100644
--- a/source/blender/blenkernel/BKE_text.h
+++ b/source/blender/blenkernel/BKE_text.h
@@ -46,6 +46,7 @@ void			free_text		(struct Text *text);
 void 			txt_set_undostate	(int u);
 int 			txt_get_undostate	(void);
 struct Text*	add_empty_text	(const char *name);
+int             txt_extended_ascii_as_utf8(char **str);
 int	            reopen_text		(struct Text *text);
 struct Text*	add_text		(const char *file, const char *relpath); 
 struct Text*	copy_text		(struct Text *ta);
@@ -59,6 +60,8 @@ void	txt_order_cursors	(struct Text *text);
 int		txt_find_string		(struct Text *text, const char *findstr, int wrap, int match_case);
 int		txt_has_sel			(struct Text *text);
 int		txt_get_span		(struct TextLine *from, struct TextLine *to);
+int     txt_utf8_offset_to_index(char *str, int offset);
+int     txt_utf8_index_to_offset(char *str, int index);
 void	txt_move_up			(struct Text *text, short sel);
 void	txt_move_down		(struct Text *text, short sel);
 void	txt_move_left		(struct Text *text, short sel);
@@ -86,9 +89,9 @@ void	txt_do_redo			(struct Text *text);
 void	txt_split_curline	(struct Text *text);
 void	txt_backspace_char	(struct Text *text);
 void	txt_backspace_word	(struct Text *text);
-int		txt_add_char		(struct Text *text, char add);
-int		txt_add_raw_char	(struct Text *text, char add);
-int		txt_replace_char	(struct Text *text, char add);
+int		txt_add_char		(struct Text *text, unsigned int add);
+int		txt_add_raw_char	(struct Text *text, unsigned int add);
+int		txt_replace_char	(struct Text *text, unsigned int add);
 void	txt_export_to_object(struct Text *text);
 void	txt_export_to_objects(struct Text *text);
 void	txt_unindent		(struct Text *text);
@@ -127,34 +130,48 @@ int text_check_whitespace(char ch);
 #define UNDO_SLEFT		005
 #define UNDO_SRIGHT		006
 #define UNDO_SUP		007
-#define UNDO_SDOWN		021
+#define UNDO_SDOWN		010
 
 /* Complex movement (opcode is followed
  * by 4 character line ID + a 2 character
  * position ID and opcode (repeat)) */
-#define UNDO_CTO		022
-#define UNDO_STO		023
-
-/* Complex editing (opcode is followed
- * by 1 character ID and opcode (repeat)) */
-#define UNDO_INSERT		024
-#define UNDO_BS			025
-#define UNDO_DEL		026
+#define UNDO_CTO		011
+#define UNDO_STO		012
+
+/* Complex editing */
+/* 1 - opcode is followed by 1 byte for ascii character and opcode (repeat)) */
+/* 2 - opcode is followed by 2 bytes for utf-8 character and opcode (repeat)) */
+/* 3 - opcode is followed by 3 bytes for utf-8 character and opcode (repeat)) */
+/* 4 - opcode is followed by 4 bytes for unicode character and opcode (repeat)) */
+#define UNDO_INSERT_1   013
+#define UNDO_INSERT_2   014
+#define UNDO_INSERT_3   015
+#define UNDO_INSERT_4   016
+
+#define UNDO_BS_1       017
+#define UNDO_BS_2       020
+#define UNDO_BS_3       021
+#define UNDO_BS_4       022
+
+#define UNDO_DEL_1      023
+#define UNDO_DEL_2      024
+#define UNDO_DEL_3      025
+#define UNDO_DEL_4      026
 
 /* Text block (opcode is followed
  * by 4 character length ID + the text
  * block itself + the 4 character length
  * ID (repeat) and opcode (repeat)) */
-#define UNDO_DBLOCK		027 /* Delete block */
-#define UNDO_IBLOCK		030 /* Insert block */
+#define UNDO_DBLOCK	    027 /* Delete block */
+#define UNDO_IBLOCK	    030 /* Insert block */
 
 /* Misc */
-#define UNDO_SWAP		031	/* Swap cursors */
+#define UNDO_SWAP       031	/* Swap cursors */
 
-#define UNDO_INDENT		032
-#define UNDO_UNINDENT		033
-#define UNDO_COMMENT		034
-#define UNDO_UNCOMMENT		035
+#define UNDO_INDENT     032
+#define UNDO_UNINDENT   033
+#define UNDO_COMMENT    034
+#define UNDO_UNCOMMENT  035
 
 /* Marker flags */
 #define TMARK_TEMP		0x01	/* Remove on non-editing events, don't save */
diff --git a/source/blender/blenkernel/intern/text.c b/source/blender/blenkernel/intern/text.c
index 7e102bc9854..bc86a53c35e 100644
--- a/source/blender/blenkernel/intern/text.c
+++ b/source/blender/blenkernel/intern/text.c
@@ -33,6 +33,8 @@
 #include <string.h> /* strstr */
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <wchar.h>
+#include <wctype.h>
 
 #include "MEM_guardedalloc.h"
 
@@ -215,8 +217,48 @@ Text *add_empty_text(const char *name)
 	return ta;
 }
 
+/* this function replaces extended ascii characters */
+/* to a valid utf-8 sequences */
+int txt_extended_ascii_as_utf8(char **str)
+{
+	int bad_char, added= 0, i= 0;
+	int length = strlen(*str);
+
+	while ((*str)[i]) {
+		if((bad_char= BLI_utf8_invalid_byte(*str+i, length)) == -1)
+		    break;
+
+		added++;
+		i+= bad_char + 1;
+	}
+	
+	if (added != 0) {
+		char *newstr = MEM_mallocN(length+added+1, "text_line");
+		int mi = 0;
+		i= 0;
+		
+		while ((*str)[i]) {
+			if((bad_char= BLI_utf8_invalid_byte((*str)+i, length)) == -1) {
+				memcpy(newstr+mi, (*str)+i, length - i + 1);
+				break;
+			}
+			
+			memcpy(newstr+mi, (*str)+i, bad_char);
+
+			BLI_str_utf8_from_unicode((*str)[i+bad_char], newstr+mi+bad_char);
+			i+= bad_char+1;
+			mi+= bad_char+2;
+		}
+		newstr[length+added] = '\0';
+		MEM_freeN(*str);
+		*str = newstr;
+	}
+	
+	return added;
+}
+
 // this function removes any control characters from
-// a textline
+// a textline and fixes invalid utf-8 sequences
 
 static void cleanup_textline(TextLine * tl)
 {
@@ -229,6 +271,7 @@ static void cleanup_textline(TextLine * tl)
 			i--;
 		}
 	}
+	tl->len+= txt_extended_ascii_as_utf8(&tl->line);
 }
 
 int reopen_text(Text *text)
@@ -689,16 +732,10 @@ static void txt_make_dirty (Text *text)
 }
 
 /* 0:whitespace, 1:punct, 2:alphanumeric */
-static short txt_char_type (char ch)
-{
-	if (ch <= ' ') return 0; /* 32 */
-	if (ch <= '/') return 1; /* 47 */
-	if (ch <= '9') return 2; /* 57 */
-	if (ch <= '@') return 1; /* 64 */
-	if (ch <= 'Z') return 2; /* 90 */
-	if (ch == '_') return 2; /* 95, dont delimit '_' */
-	if (ch <= '`') return 1; /* 96 */
-	if (ch <= 'z') return 2; /* 122 */
+static short txt_char_type(unsigned int ch)
+{
+	if (iswspace(ch)) return 0;
+	if (iswalpha(ch) || iswdigit(ch)) return 2;
 	return 1;
 }
 
@@ -731,9 +768,42 @@ static void txt_curs_first (Text *text, TextLine **linep, int *charp)
 	}
 }
 
-/****************************/
+/*****************************/
 /* Cursor movement functions */
-/****************************/
+/*****************************/
+
+int txt_utf8_offset_to_index(char *str, int offset)
+{
+	int index= 0, pos= 0;
+	while (pos != offset) {
+		pos += BLI_str_utf8_size(str + pos);
+		index++;
+	}
+	return index;
+}
+
+int txt_utf8_index_to_offset(char *str, int index)
+{
+	int offset= 0, pos= 0;
+	while (pos != index) {
+		offset += BLI_str_utf8_size(str + offset);
+		pos++;
+	}
+	return offset;
+}
+
+/* returns the real number of characters in string */
+/* not the same as BLI_strlen_utf8, which returns length for wide characters */
+static int txt_utf8_len(const char *src)
+{
+	int len;
+
+	for (len=0; *src; len++) {
+		src += BLI_str_utf8_size(src);
+	}
+
+	return len;
+}
 
 void txt_move_up(Text *text, short sel)
 {
@@ -747,13 +817,13 @@ void txt_move_up(Text *text, short sel)
 	old= *charp;
 
 	if((*linep)->prev) {
+		int index = txt_utf8_offset_to_index((*linep)->line, *charp);
 		*linep= (*linep)->prev;
-		if (*charp > (*linep)->len) {
-			*charp= (*linep)->len;
-			if(!undoing) txt_undo_add_toop(text, sel?UNDO_STO:UNDO_CTO, txt_get_span(text->lines.first, (*linep)->next), old, txt_get_span(text->lines.first, *linep), (unsigned short) *charp);
-		} else {
-			if(!undoing) txt_undo_add_op(text, sel?UNDO_SUP:UNDO_CUP);
-		}
+		if (index > txt_utf8_len((*linep)->line)) *charp= (*linep)->len;
+		else *charp= txt_utf8_index_to_offset((*linep)->line, index);
+		
+		if(!undoing)
+			txt_undo_add_op(text, sel?UNDO_SUP:UNDO_CUP);
 	} else {
 		txt_move_bol(text, sel);
 	}
@@ -773,12 +843,13 @@ void txt_move_down(Text *text, short sel)
 	old= *charp;
 
 	if((*linep)->next) {
+		int index = txt_utf8_offset_to_index((*linep)->line, *charp);
 		*linep= (*linep)->next;
-		if (*charp > (*linep)->len) {
-			*charp= (*linep)->len;
-			if(!undoing) txt_undo_add_toop(text, sel?UNDO_STO:UNDO_CTO, txt_get_span(text->lines.first, (*linep)->prev), old, txt_get_span(text->lines.first, *linep), (unsigned short)*charp);
-		} else
-			if(!undoing) txt_undo_add_op(text, sel?UNDO_SDOWN:UNDO_CDOWN);	
+		if (index > txt_utf8_len((*linep)->line)) *charp= (*linep)->len;
+		else *charp= txt_utf8_index_to_offset((*linep)->line, index);
+		
+		if(!undoing)
+			txt_undo_add_op(text, sel?UNDO_SDOWN:UNDO_CDOWN);
 	} else {
 		txt_move_eol(text, sel);
 	}
@@ -790,7 +861,7 @@ void txt_move_left(Text *text, short sel)
 {
 	TextLine **linep;
 	int *charp, oundoing= undoing;
-	int tabsize = 1, i=0;
+	int tabsize= 0, i= 0;
 	
 	if (!text) return;
 	if(sel) txt_curs_sel(text, &linep, &charp);
@@ -799,32 +870,36 @@ void txt_move_left(Text *text, short sel)
 
 	undoing= 1;
 
-	// do nice left only if there are only spaces
-	// TXT_TABSIZE hardcoded in DNA_text_types.h
-	if (text->flags & TXT_TABSTOSPACES) {
-		tabsize = TXT_TABSIZE;
-
-		if (*charp < tabsize)
-			tabsize = *charp;
-		else {
-			for (i=0;i<(*charp);i++)
+	if (*charp== 0) {
+		if ((*linep)->prev) {
+			txt_move_up(text, sel);
+			*charp= (*linep)->len;
+		}
+	}
+	else {
+		// do nice left only if there are only spaces
+		// TXT_TABSIZE hardcoded in DNA_text_types.h
+		if (text->flags & TXT_TABSTOSPACES) {
+			tabsize= (*charp < TXT_TABSIZE) ? *charp : TXT_TABSIZE;
+			
+			for (i=0; i<(*charp); i++)
 				if ((*linep)->line[i] != ' ') {
-					tabsize = 1;
+					tabsize= 0;
 					break;
 				}
+			
 			// if in the middle of the space-tab
-			if ((*charp) % tabsize != 0)
-					tabsize = ((*charp) % tabsize);
+			if (tabsize && (*charp) % TXT_TABSIZE != 0)
+				tabsize= ((*charp) % TXT_TABSIZE);
 		}
-	}
-
-	if (*charp== 0) {
-		if ((*linep)->prev) {
-			txt_move_up(text, sel);
-			*charp= (*linep)->len;
+		
+		if (tabsize)
+			(*charp)-= tabsize;
+		else {
+			const char *prev= BLI_str_prev_char_utf8((*linep)->line + *charp);
+			*charp= prev - (*linep)->line;
 		}
 	}
-	else (*charp)-= tabsize;
 
 	undoing= oundoing;
 	if(!undoing) txt_undo_add_op(text, sel?UNDO_SLEFT:UNDO_CLEFT);
@@ -835,8 +910,7 @@ void txt_move_left(Text *text, short sel)
 void txt_move_right(Text *text, short sel) 
 {
 	TextLine **linep;
-	int *charp, oundoing= undoing;
-	int tabsize=1, i=0;
+	int *charp, oundoing= undoing, do_tab= 0, i;
 	
 	if (!text) return;
 	if(sel) txt_curs_sel(text, &linep, &charp);
@@ -845,32 +919,33 @@ void txt_move_right(Text *text, short sel)
 
 	undoing= 1;
 
-	// do nice right only if there are only spaces
-	// spaces hardcoded in DNA_text_types.h
-	if (text->flags & TXT_TABSTOSPACES) {
-		tabsize = TXT_TABSIZE;
-
-		if ((*charp) + tabsize > (*linep)->len)
-			tabsize = 1;
-		else {
-			for (i=0;i<(*charp) + tabsize - ((*charp) % tabsize);i++)
-				if ((*linep)->line[i] != ' ') {
-					tabsize = 1;
-					break;
-				}
-			// if in the middle of the space-tab
-			tabsize -= (*charp) % tabsize;
-		}
-	}
-
 	if (*charp== (*linep)->len) {
 		if ((*linep)->next) {
 			txt_move_down(text, sel);
 			*charp= 0;
 		}
-	} else {
-		(*charp)+=tabsize;
+	} 
+	else {
+		// do nice right only if there are only spaces
+		// spaces hardcoded in DNA_text_types.h
+		if (text->flags & TXT_TABSTOSPACES && (*linep)->line[*charp]== ' ') {
+			do_tab= 1;
+			for (i=0; i<*charp; i++)
+				if ((*linep)->line[i]!= ' ') {
+					do_tab= 0;
+					break;
+				}
+		}
+		
+		if (do_tab) {
+			int tabsize= (*charp) % TXT_TABSIZE + 1;
+			for (i=*charp+1; (*linep)->line[i]==' ' && tabsize<TXT_TABSIZE; i++)
+				tabsize++;
+			(*charp)= i;
+		}
+		else (*charp)+= BLI_str_utf8_size((*linep)->line + *charp);
 	}
+	
 	undoing= oundoing;
 	if(!undoing) txt_undo_add_op(text, sel?UNDO_SRIGHT:UNDO_CRIGHT);
 
@@ -896,9 +971,12 @@ void txt_jump_left(Text *text, short sel)
 	count= 0;
 	for (i=0; i<3; i++) {
 		if (count < 2) {
-			while (*charp>0 && txt_char_type((*linep)->line[*charp-1])==i) {
-				txt_move_left(text, sel);
-				count++;
+			while (*charp>0) {
+				char *sym= BLI_str_prev_char_utf8((*linep)->line + *charp);
+				if (txt_char_type(BLI_str_utf8_as_unicode(sym))==i) {
+					txt_move_left(text, sel);
+					count++;
+				} else break;
 			}
 		}
 	}
@@ -927,9 +1005,12 @@ void txt_jump_right(Text *text, short sel)
 	count= 0;
 	for (i=0; i<3; i++) {
 		if (count < 2) {
-			while (*charp<(*linep)->len && txt_char_type((*linep)->line[*charp])==i) {
-				txt_move_right(text, sel);
-				count++;
+			while (*charp<(*linep)->len) {
+				char *sym= (*linep)->line + *charp;
+				if (txt_char_type(BLI_str_utf8_as_unicode(sym))==i) {
+					txt_move_right(text, sel);
+					count++;
+				} else break;
 			}
 		}
 	}
@@ -1014,6 +1095,7 @@ void txt_move_toline (Text *text, unsigned int line, short sel)
 	txt_move_to(text, line, 0, sel);
 }
 
+/* Moves to a certain byte in a line, not a certain utf8-character! */
 void txt_move_to (Text *text, unsigned int line, unsigned int ch, short sel)
 {
 	TextLine **linep, *oldl;
@@ -1396,42 +1478,45 @@ static void txt_shift_markers(Text *text, int lineno, int count)
 
 void txt_insert_buf(Text *text, const char *in_buffer)
 {
-	int i=0, l=0, j, u, len, lineno= -1, count= 0;
+	int l=0, u, len, lineno= -1, count= 0;
+	size_t i=0, j;
 	TextLine *add;
+	char *buffer;
 
 	if (!text) return;
 	if (!in_buffer) return;
 
 	txt_delete_sel(text);
 	
-	if(!undoing) txt_undo_add_block (text, UNDO_IBLOCK, in_buffer); 	
+	len= strlen(in_buffer);
+	buffer= BLI_strdupn(in_buffer, len);
+	len+= txt_extended_ascii_as_utf8(&buffer);
+	
+	if(!undoing) txt_undo_add_block(text, UNDO_IBLOCK, buffer);
 
 	u= undoing;
 	undoing= 1;
 
 	/* Read the first line (or as close as possible */
-	while (in_buffer[i] && in_buffer[i]!='\n') {
-		txt_add_raw_char(text, in_buffer[i]);
-		i++;
-	}
+	while (buffer[i] && buffer[i]!='\n')
+		txt_add_raw_char(text, BLI_str_utf8_as_unicode_step(buffer, &i));
 	
-	if (in_buffer[i]=='\n') txt_split_curline(text);
-	else { undoing = u; return; }
+	if (buffer[i]=='\n') txt_split_curline(text);
+	else { undoing = u; MEM_freeN(buffer); return; }
 	i++;
 
 	/* Read as many full lines as we can */
-	len= strlen(in_buffer);
 	lineno= txt_get_span(text->lines.first, text->curl);
 
 	while (i<len) {
 		l=0;
 
-		while (in_buffer[i] && in_buffer[i]!='\n') {
+		while (buffer[i] && buffer[i]!='\n') {
 			i++; l++;
 		}
 	
-		if(in_buffer[i]=='\n') {
-			add= txt_new_linen(in_buffer +(i-l), l);
+		if(buffer[i]=='\n') {
+			add= txt_new_linen(buffer +(i-l), l);
 			BLI_insertlinkbefore(&text->lines, text->curl, add);
 			i++;
 			count++;
@@ -1441,21 +1526,19 @@ void txt_insert_buf(Text *text, const char *in_buffer)
 				count= 0;
 			}
 
-			for (j= i-l; j<i && j<(int)strlen(in_buffer); j++) {
-				txt_add_raw_char(text, in_buffer[j]);
-			}
+			for (j= i-l; j<i && j<len; )
+				txt_add_raw_char(text, BLI_str_utf8_as_unicode_step(buffer, &j));
 			break;
 		}
 	}
+	
+	MEM_freeN(buffer);
 
 	if(count) {
 		txt_shift_markers(text, lineno, count);
-		count= 0;
 	}
 
 	undoing= u;
-
-	(void)count;
 }
 
 /******************/
@@ -1525,12 +1608,30 @@ void txt_print_undo(Text *text)
 			ops= "Selection ";
 		} else if (op==UNDO_CTO) {
 			ops= "Cursor ";
-		} else if (op==UNDO_INSERT) {
-			ops= "Insert";
-		} else if (op==UNDO_BS) {
-			ops= "Backspace";
-		} else if (op==UNDO_DEL) {
-			ops= "Delete";
+		} else if (op==UNDO_INSERT_1) {
+			ops= "Insert ascii ";
+		} else if (op==UNDO_INSERT_2) {
+			ops= "Insert 2 bytes ";
+		} else if (op==UNDO_INSERT_3) {
+			ops= "Insert 3 bytes ";
+		} else if (op==UNDO_INSERT_4) {
+			ops= "Insert unicode ";
+		} else if (op==UNDO_BS_1) {
+			ops= "Backspace for ascii ";
+		} else if (op==UNDO_BS_2) {
+			ops= "Backspace for 2 bytes ";
+		} else if (op==UNDO_BS_3) {
+			ops= "Backspace for 3 bytes ";
+		} else if (op==UNDO_BS_4) {
+			ops= "Backspace for unicode ";
+		} else if (op==UNDO_DEL_1) {
+			ops= "Delete ascii ";
+		} else if (op==UNDO_DEL_2) {
+			ops= "Delete 2 bytes ";
+		} else if (op==UNDO_DEL_3) {
+			ops= "Delete 3 bytes ";
+		} else if (op==UNDO_DEL_4) {
+			ops= "Delete unicode ";
 		} else if (op==UNDO_SWAP) {
 			ops= "Cursor swap";
 		} else if (op==UNDO_DBLOCK) {
@@ -1550,10 +1651,35 @@ void txt_print_undo(Text *text)
 		}
 		
 		printf ("Op (%o) at %d = %s", op, i, ops);
-		if (op==UNDO_INSERT || op==UNDO_BS || op==UNDO_DEL) {
-			i++;
-			printf (" - Char is %c", text->undo_buf[i]);  
+		if (op >= UNDO_INSERT_1 && op <= UNDO_DEL_4) {
 			i++;
+			printf (" - Char is ");
+			switch (op) {
+				case UNDO_INSERT_1: case UNDO_BS_1: case UNDO_DEL_1:
+					printf ("%c", text->undo_buf[i]);  
+					i++;
+					break;
+				case UNDO_INSERT_2: case UNDO_BS_2: case UNDO_DEL_2:
+					printf ("%c%c", text->undo_buf[i], text->undo_buf[i+1]);  
+					i+=2;
+					break;
+				case UNDO_INSERT_3: case UNDO_BS_3: case UNDO_DEL_3:
+					printf ("%c%c%c", text->undo_buf[i], text->undo_buf[i+1], text->undo_buf[i+2]);  
+					i+=3;
+					break;
+				case UNDO_INSERT_4: case UNDO_BS_4: case UNDO_DEL_4: {
+					unsigned int uc;
+					char c[BLI_UTF8_MAX+1];
+					size_t c_len;
+					uc= text->undo_buf[i]; i++;
+					uc= uc+(text->undo_buf[i]<<8); i++;
+					uc= uc+(text->undo_buf[i]<<16); i++;
+					uc= uc+(text->undo_buf[i]<<24); i++;
+					c_len= BLI_str_utf8_from_unicode(uc, c);
+					c[c_len]= '\0';
+					printf ("%s", c);
+				}
+			}
 		} else if (op==UNDO_STO || op==UNDO_CTO) {
 			i++;
 
@@ -1635,40 +1761,43 @@ static void txt_undo_add_op(Text *text, int op)
 	text->undo_buf[text->undo_pos+1]= 0;
 }
 
+static void txt_undo_store_uint16(char *undo_buf, int *undo_pos, unsigned short value) 
+{
+	undo_buf[*undo_pos]= (value)&0xff;
+	(*undo_pos)++;
+	undo_buf[*undo_pos]= (value>>8)&0xff;
+	(*undo_pos)++;
+}
+
+static void txt_undo_store_uint32(char *undo_buf, int *undo_pos, unsigned int value) 
+{
+	undo_buf[*undo_pos]= (value)&0xff;
+	(*undo_pos)++;
+	undo_buf[*undo_pos]= (value>>8)&0xff;
+	(*undo_pos)++;
+	undo_buf[*undo_pos]= (value>>16)&0xff;
+	(*undo_pos)++;
+	undo_buf[*undo_pos]= (value>>24)&0xff;
+	(*undo_pos)++;
+}
+
 static void txt_undo_add_block(Text *text, int op, const char *buf)
 {
-	int length;
-	
-	length= strlen(buf);
+	unsigned int length= strlen(buf);
 	
 	if(!max_undo_test(text, length+11))
 		return;
 
 	text->undo_pos++;
 	text->undo_buf[text->undo_pos]= op;
-	
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>8)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>16)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>24)&0xff;
-
 	text->undo_pos++;
+	
+	txt_undo_store_uint32(text->undo_buf, &text->undo_pos, length);
+	
 	strncpy(text->undo_buf+text->undo_pos, buf, length);
 	text->undo_pos+=length;
 
-	text->undo_buf[text->undo_pos]= (length)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>8)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>16)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (length>>24)&0xff;
-
-	text->undo_pos++;
+	txt_undo_store_uint32(text->undo_buf, &text->undo_pos, length);
 	text->undo_buf[text->undo_pos]= op;
 	
 	text->undo_buf[text->undo_pos+1]= 0;
@@ -1685,51 +1814,139 @@ void txt_undo_add_toop(Text *text, int op, unsigned int froml, unsigned short fr
 	text->undo_buf[text->undo_pos]= op;
 
 	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (fromc)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (fromc>>8)&0xff;
+	
+	txt_undo_store_uint16(text->undo_buf, &text->undo_pos, fromc);
+	txt_undo_store_uint32(text->undo_buf, &text->undo_pos, froml);
+	txt_undo_store_uint16(text->undo_buf, &text->undo_pos, toc);
+	txt_undo_store_uint32(text->undo_buf, &text->undo_pos, tol);
+		
+	text->undo_buf[text->undo_pos]= op;
 
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (froml)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (froml>>8)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (froml>>16)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (froml>>24)&0xff;
+	text->undo_buf[text->undo_pos+1]= 0;
+}
 
+static void txt_undo_add_charop(Text *text, int op_start, unsigned int c)
+{
+	char utf8[BLI_UTF8_MAX];
+	size_t i, utf8_size = BLI_str_utf8_from_unicode(c, utf8);
+	
+	if(!max_undo_test(text, 3 + utf8_size))
+		return;
+	
 	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (toc)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (toc>>8)&0xff;
+	
+	if (utf8_size < 4) {
+		text->undo_buf[text->undo_pos]= op_start + utf8_size - 1;
+		text->undo_pos++;
+		
+		for (i = 0; i < utf8_size; i++) {
+			text->undo_buf[text->undo_pos]= utf8[i];
+			text->undo_pos++;
+		}
+		
+		text->undo_buf[text->undo_pos]= op_start + utf8_size - 1;
+	} else {
+		text->undo_buf[text->undo_pos]= op_start + 3;
+		text->undo_pos++;
+		txt_undo_store_uint32(text->undo_buf, &text->undo_pos, c);
+		text->undo_buf[text->undo_pos]= op_start + 3;
+	}
+	
+	text->undo_buf[text->undo_pos+1]= 0;
+}
 
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (tol)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (tol>>8)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (tol>>16)&0xff;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= (tol>>24)&0xff;
+static unsigned short txt_undo_read_uint16(const char *undo_buf, int *undo_pos)
+{
+	unsigned short val;
+	val= undo_buf[*undo_pos]; (*undo_pos)--;
+	val= (val<<8)+undo_buf[*undo_pos]; (*undo_pos)--;
+	return val;
+}
 
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= op;
+static unsigned int txt_undo_read_uint32(const char *undo_buf, int *undo_pos)
+{
+	unsigned int val;
+	val= undo_buf[*undo_pos]; (*undo_pos)--;
+	val= (val<<8)+undo_buf[*undo_pos]; (*undo_pos)--;
+	val= (val<<8)+undo_buf[*undo_pos]; (*undo_pos)--;
+	val= (val<<8)+undo_buf[*undo_pos]; (*undo_pos)--;
+	return val;
+}
 
-	text->undo_buf[text->undo_pos+1]= 0;
+static unsigned int txt_undo_read_unicode(const char *undo_buf, int *undo_pos, short bytes)
+{
+	unsigned int unicode;
+	char utf8[BLI_UTF8_MAX+1];
+	
+	switch (bytes) {
+		case 1: /* ascii */
+			unicode = undo_buf[*undo_pos]; (*undo_pos)--; 
+			break;
+		case 2: /* 2-byte symbol */
+			utf8[2] = '\0';
+			utf8[1] = undo_buf[*undo_pos]; (*undo_pos)--;
+			utf8[0] = undo_buf[*undo_pos]; (*undo_pos)--;
+			unicode= BLI_str_utf8_as_unicode(utf8);
+			break;
+		case 3: /* 3-byte symbol */
+			utf8[3] = '\0';
+			utf8[2] = undo_buf[*undo_pos]; (*undo_pos)--;
+			utf8[1] = undo_buf[*undo_pos]; (*undo_pos)--;
+			utf8[0] = undo_buf[*undo_pos]; (*undo_pos)--;
+			unicode= BLI_str_utf8_as_unicode(utf8);
+			break;
+		case 4: /* 32-bit unicode symbol */
+			unicode= txt_undo_read_uint32(undo_buf, undo_pos);
+	}
+	
+	return unicode;
 }
 
-static void txt_undo_add_charop(Text *text, int op, char c)
+static unsigned short txt_redo_read_uint16(const char *undo_buf, int *undo_pos)
 {
-	if(!max_undo_test(text, 4))
-		return;
+	unsigned short val;
+	val = undo_buf[*undo_pos]; (*undo_pos)++;
+	val = val+(undo_buf[*undo_pos]<<8); (*undo_pos)++;
+	return val;
+}
 
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= op;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= c;
-	text->undo_pos++;
-	text->undo_buf[text->undo_pos]= op;
-	text->undo_buf[text->undo_pos+1]= 0;
+static unsigned int txt_redo_read_uint32(const char *undo_buf, int *undo_pos)
+{
+	unsigned int val;
+	val= undo_buf[*undo_pos]; (*undo_pos)++;
+	val= val+(undo_buf[*undo_pos]<<8); (*undo_pos)++;
+	val= val+(undo_buf[*undo_pos]<<16); (*undo_pos)++;
+	val= val+(undo_buf[*undo_pos]<<24); (*undo_pos)++;
+	return val;
+}
+
+static unsigned int txt_redo_read_unicode(const char *undo_buf, int *undo_pos, short bytes)
+{
+	unsigned int unicode;
+	char utf8[BLI_UTF8_MAX+1];
+	
+	switch (bytes) {
+		case 1: /* ascii */
+			unicode = undo_buf[*undo_pos]; (*undo_pos)++; 
+			break;
+		case 2: /* 2-byte symbol */
+			utf8[0] = undo_buf[*undo_pos]; (*undo_pos)++;
+			utf8[1] = undo_buf[*undo_pos]; (*undo_pos)++;
+			utf8[2] = '\0';
+			unicode= BLI_str_utf8_as_unicode(utf8);
+			break;
+		case 3: /* 3-byte symbol */
+			utf8[0] = undo_buf[*undo_pos]; (*undo_pos)++;
+			utf8[1] = undo_buf[*undo_pos]; (*undo_pos)++;
+			utf8[2] = undo_buf[*undo_pos]; (*undo_pos)++;
+			utf8[3] = '\0';
+			unicode= BLI_str_utf8_as_unicode(utf8);
+			break;
+		case 4: /* 32-bit unicode symbol */
+			unicode= txt_undo_read_uint32(undo_buf, undo_pos);
+	}
+	
+	return unicode;
 }
 
 void txt_do_undo(Text *text)
@@ -1792,13 +2009,8 @@ void txt_do_undo(Text *text)
 			text->undo_pos--;
 			text->undo_pos--;
 		
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-
-			charp= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			charp= (charp<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			linep= txt_undo_read_uint32(text->undo_buf, &text->undo_pos);
+			charp= txt_undo_read_uint16(text->undo_buf, &text->undo_pos);
 			
 			if (op==UNDO_CTO) {
 				txt_move_toline(text, linep, 0);
@@ -1812,23 +2024,23 @@ void txt_do_undo(Text *text)
 			text->undo_pos--;
 			break;
 			
-		case UNDO_INSERT:
+		case UNDO_INSERT_1: case UNDO_INSERT_2: case UNDO_INSERT_3: case UNDO_INSERT_4:
 			txt_backspace_char(text);
-			text->undo_pos--;
+			text->undo_pos-= op - UNDO_INSERT_1 + 1;
 			text->undo_pos--;
 			break;
 
-		case UNDO_BS:
-			txt_add_char(text, text->undo_buf[text->undo_pos]);
-			text->undo_pos--;
+		case UNDO_BS_1: case UNDO_BS_2: case UNDO_BS_3: case UNDO_BS_4:
+			charp = op - UNDO_BS_1 + 1;
+			txt_add_char(text, txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp));
 			text->undo_pos--;
-			break;
-
-		case UNDO_DEL:
-			txt_add_char(text, text->undo_buf[text->undo_pos]);
+			break;		
+			
+		case UNDO_DEL_1: case UNDO_DEL_2: case UNDO_DEL_3: case UNDO_DEL_4: 
+			charp = op - UNDO_DEL_1 + 1;
+			txt_add_char(text, txt_undo_read_unicode(text->undo_buf, &text->undo_pos, charp));
 			txt_move_left(text, 0);
 			text->undo_pos--;
-			text->undo_pos--;
 			break;
 
 		case UNDO_SWAP:
@@ -1836,10 +2048,7 @@ void txt_do_undo(Text *text)
 			break;
 
 		case UNDO_DBLOCK:
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			linep= txt_undo_read_uint32(text->undo_buf, &text->undo_pos);
 
 			buf= MEM_mallocN(linep+1, "dblock buffer");
 			for (i=0; i < linep; i++){
@@ -1863,25 +2072,31 @@ void txt_do_undo(Text *text)
 			}
 			text->curc= holdc;
 
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			text->undo_pos--;
+			text->undo_pos--;
+			text->undo_pos--; 
+			text->undo_pos--;
 
 			text->undo_pos--;
 			
 			break;
 
 		case UNDO_IBLOCK:
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep= (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-
+			linep= txt_undo_read_uint32(text->undo_buf, &text->undo_pos);
 			txt_delete_sel(text);
+
+			/* txt_backspace_char removes utf8-characters, not bytes */
+			buf= MEM_mallocN(linep+1, "iblock buffer");
+			for (i=0; i < linep; i++){
+				buf[(linep-1)-i]= text->undo_buf[text->undo_pos]; 
+				text->undo_pos--;
+			}
+			buf[i]= 0;
+			linep= txt_utf8_len(buf);
+			MEM_freeN(buf);
+
 			while (linep>0) {
 				txt_backspace_char(text);
-				text->undo_pos--;
 				linep--;
 			}
 
@@ -1897,30 +2112,23 @@ void txt_do_undo(Text *text)
 		case UNDO_UNINDENT:
 		case UNDO_COMMENT:
 		case UNDO_UNCOMMENT:
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			linep= txt_undo_read_uint32(text->undo_buf, &text->undo_pos);
 			//linep is now the end line of the selection
 			
-			charp = text->undo_buf[text->undo_pos]; text->undo_pos--;
-			charp = (charp<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			charp = txt_undo_read_uint16(text->undo_buf, &text->undo_pos);
 			//charp is the last char selected or text->line->len
-			//set the selcetion for this now
+			
+			//set the selection for this now
 			text->selc = charp;
 			text->sell = text->lines.first;
 			for (i= 0; i < linep; i++) {
 				text->sell = text->sell->next;
 			}
 
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
-			linep = (linep<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			linep= txt_undo_read_uint32(text->undo_buf, &text->undo_pos);
 			//first line to be selected
 			
-			charp = text->undo_buf[text->undo_pos]; text->undo_pos--;
-			charp = (charp<<8)+text->undo_buf[text->undo_pos]; text->undo_pos--;
+			charp = txt_undo_read_uint16(text->undo_buf, &text->undo_pos);
 			//first postion to be selected
 			text->curc = charp;
 			text->curl = text->lines.first;
@@ -2014,22 +2222,22 @@ void txt_do_redo(Text *text)
 			txt_move_down(text, 1);
 			break;
 		
-		case UNDO_INSERT:
-			text->undo_pos++;
-			txt_add_char(text, text->undo_buf[text->undo_pos]);
+		case UNDO_INSERT_1: case UNDO_INSERT_2: case UNDO_INSERT_3: case UNDO_INSERT_4:
 			text->undo_pos++;
+			charp = op - UNDO_INSERT_1 + 1;
+			txt_add_char(text, txt_redo_read_unicode(text->undo_buf, &text->undo_pos, charp));
 			break;
 
-		case UNDO_BS:
+		case UNDO_BS_1: case UNDO_BS_2: case UNDO_BS_3: case UNDO_BS_4:
 			text->undo_pos++;
 			txt_backspace_char(text);
-			text->undo_pos++;
+			text->undo_pos+= op - UNDO_BS_1 + 1;
 			break;
 
-		case UNDO_DEL:
+		case UNDO_DEL_1: case UNDO_DEL_2: case UNDO_DEL_3: case UNDO_DEL_4:
 			text->undo_pos++;
 			txt_delete_char(text);
-			text->undo_pos++;
+			text->undo_pos+= op - UNDO_DEL_1 + 1;
 			break;
 
 		case UNDO_SWAP:
@@ -2049,15 +2257,8 @@ void txt_do_redo(Text *text)
 
 			text->undo_pos++;
 
-			charp= text->undo_buf[text->undo_pos];
-			text->undo_pos++;
-			charp= charp+(text->undo_buf[text->undo_pos]<<8);
-
-			text->undo_pos++;
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
+			charp= txt_redo_read_uint16(text->undo_buf, &text->undo_pos);
+			linep= txt_redo_read_uint32(text->undo_buf, &text->undo_pos);
 			
 			if (op==UNDO_CTO) {
 				txt_move_toline(text, linep, 0);
@@ -2072,12 +2273,9 @@ void txt_do_redo(Text *text)
 
 		case UNDO_DBLOCK:
 			text->undo_pos++;
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
-
+			linep= txt_redo_read_uint32(text->undo_buf, &text->undo_pos);
 			txt_delete_sel(text);
+			
 			text->undo_pos+=linep;
 
 			text->undo_pos++;
@@ -2089,10 +2287,7 @@ void txt_do_redo(Text *text)
 
 		case UNDO_IBLOCK:
 			text->undo_pos++;
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
+			linep= txt_redo_read_uint32(text->undo_buf, &text->undo_pos);
 
 			buf= MEM_mallocN(linep+1, "iblock buffer");
 			memcpy (buf, &text->undo_buf[text->undo_pos], linep);
@@ -2102,26 +2297,21 @@ void txt_do_redo(Text *text)
 			txt_insert_buf(text, buf);			
 			MEM_freeN(buf);
 
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep= linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
-			(void)linep;
-
+			text->undo_pos++;
+			text->undo_pos++;
+			text->undo_pos++; 
+			text->undo_pos++;
 			break;
+			
 		case UNDO_INDENT:
 		case UNDO_UNINDENT:
 		case UNDO_COMMENT:
 		case UNDO_UNCOMMENT:
 			text->undo_pos++;
-			charp = text->undo_buf[text->undo_pos]; text->undo_pos++;
-			charp = charp+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
+			charp = txt_redo_read_uint16(text->undo_buf, &text->undo_pos);
 			//charp is the first char selected or 0
 			
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
+			linep= txt_redo_read_uint32(text->undo_buf, &text->undo_pos);
 			//linep is now the first line of the selection			
 			//set the selcetion for this now
 			text->curc = charp;
@@ -2130,13 +2320,10 @@ void txt_do_redo(Text *text)
 				text->curl = text->curl->next;
 			}
 			
-			charp = text->undo_buf[text->undo_pos]; text->undo_pos++;
-			charp = charp+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
+			charp = txt_redo_read_uint16(text->undo_buf, &text->undo_pos);
 			//last postion to be selected
-			linep= text->undo_buf[text->undo_pos]; text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<8); text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<16); text->undo_pos++;
-			linep = linep+(text->undo_buf[text->undo_pos]<<24); text->undo_pos++;
+			
+			linep= txt_redo_read_uint32(text->undo_buf, &text->undo_pos);
 			//Last line to be selected
 			
 			text->selc = charp;
@@ -2203,8 +2390,7 @@ void txt_split_curline (Text *text)
 	left[text->curc]=0;
 	
 	right= MEM_mallocN(text->curl->len - text->curc+1, "textline_string");
-	if (text->curl->len - text->curc) memcpy(right, text->curl->line+text->curc, text->curl->len-text->curc);
-	right[text->curl->len - text->curc]=0;
+	memcpy(right, text->curl->line+text->curc, text->curl->len-text->curc+1);
 
 	MEM_freeN(text->curl->line);
 	if (text->curl->format) MEM_freeN(text->curl->format);
@@ -2228,7 +2414,7 @@ void txt_split_curline (Text *text)
 	txt_clean_text(text);
 	
 	txt_pop_sel(text);
-	if(!undoing) txt_undo_add_charop(text, UNDO_INSERT, '\n');
+	if(!undoing) txt_undo_add_charop(text, UNDO_INSERT_1, '\n');
 }
 
 static void txt_delete_line (Text *text, TextLine *line) 
@@ -2296,9 +2482,9 @@ static void txt_combine_lines (Text *text, TextLine *linea, TextLine *lineb)
 	txt_clean_text(text);
 }
 
-void txt_delete_char (Text *text) 
+void txt_delete_char(Text *text) 
 {
-	char c='\n';
+	unsigned int c='\n';
 	
 	if (!text) return;
 	if (!text->curl) return;
@@ -2314,12 +2500,14 @@ void txt_delete_char (Text *text)
 			txt_pop_sel(text);
 		}
 	} else { /* Just deleting a char */
-		int i= text->curc;
+		size_t c_len = 0;
+		TextMarker *mrk;
+		c= BLI_str_utf8_as_unicode_and_size(text->curl->line + text->curc, &c_len);
 
-		TextMarker *mrk= txt_find_marker_region(text, text->curl, i-1, text->curl->len, 0, 0);
+		mrk= txt_find_marker_region(text, text->curl, text->curc - c_len, text->curl->len, 0, 0);
 		if (mrk) {
 			int lineno= mrk->lineno;
-			if (mrk->end==i) {
+			if (mrk->end==text->curc) {
 				if ((mrk->flags & TMARK_TEMP) && !(mrk->flags & TMARK_EDITALL)) {
 					txt_clear_markers(text, mrk->group, TMARK_TEMP);
 				} else {
@@ -2328,18 +2516,15 @@ void txt_delete_char (Text *text)
 				return;
 			}
 			do {
-				if (mrk->start>i) mrk->start--;
-				mrk->end--;
+				if (mrk->start>text->curc) mrk->start-= c_len;
+				mrk->end-= c_len;
 				mrk= mrk->next;
 			} while (mrk && mrk->lineno==lineno);
 		}
 		
-		c= text->curl->line[i];
-		while(i< text->curl->len) {
-			text->curl->line[i]= text->curl->line[i+1];
-			i++;
-		}
-		text->curl->len--;
+		memmove(text->curl->line+text->curc, text->curl->line+text->curc+c_len, text->curl->len-text->curc-c_len+1);
+
+		text->curl->len-= c_len;
 
 		txt_pop_sel(text);
 	}
@@ -2347,7 +2532,7 @@ void txt_delete_char (Text *text)
 	txt_make_dirty(text);
 	txt_clean_text(text);
 	
-	if(!undoing) txt_undo_add_charop(text, UNDO_DEL, c);
+	if(!undoing) txt_undo_add_charop(text, UNDO_DEL_1, c);
 }
 
 void txt_delete_word (Text *text) 
@@ -2358,7 +2543,7 @@ void txt_delete_word (Text *text)
 
 void txt_backspace_char (Text *text) 
 {
-	char c='\n';
+	unsigned int c='\n';
 	
 	if (!text) return;
 	if (!text->curl) return;
@@ -2378,12 +2563,15 @@ void txt_backspace_char (Text *text)
 		txt_pop_sel(text);
 	}
 	else { /* Just backspacing a char */
-		int i= text->curc-1;
+		size_t c_len = 0;
+		TextMarker *mrk;
+		char *prev = BLI_str_prev_char_utf8(text->curl->line + text->curc);
+		c= BLI_str_utf8_as_unicode_and_size(prev, &c_len);
 
-		TextMarker *mrk= txt_find_marker_region(text, text->curl, i, text->curl->len, 0, 0);
+		mrk= txt_find_marker_region(text, text->curl, text->curc - c_len, text->curl->len, 0, 0);
 		if (mrk) {
 			int lineno= mrk->lineno;
-			if (mrk->start==i+1) {
+			if (mrk->start==text->curc) {
 				if ((mrk->flags & TMARK_TEMP) && !(mrk->flags & TMARK_EDITALL)) {
 					txt_clear_markers(text, mrk->group, TMARK_TEMP);
 				} else {
@@ -2392,19 +2580,16 @@ void txt_backspace_char (Text *text)
 				return;
 			}
 			do {
-				if (mrk->start>i) mrk->start--;
-				mrk->end--;
+				if (mrk->start>text->curc - c_len) mrk->start-= c_len;
+				mrk->end-= c_len;
 				mrk= mrk->next;
 			} while (mrk && mrk->lineno==lineno);
 		}
 		
-		c= text->curl->line[i];
-		while(i< text->curl->len) {
-			text->curl->line[i]= text->curl->line[i+1];
-			i++;
-		}
-		text->curl->len--;
-		text->curc--;
+		memcpy(text->curl->line + text->curc - c_len, text->curl->line + text->curc, text->curl->len-text->curc+1);
+
+		text->curl->len-= c_len;
+		text->curc-= c_len;
 
 		txt_pop_sel(text);
 	}
@@ -2412,7 +2597,7 @@ void txt_backspace_char (Text *text)
 	txt_make_dirty(text);
 	txt_clean_text(text);
 	
-	if(!undoing) txt_undo_add_charop(text, UNDO_BS, c);
+	if(!undoing) txt_undo_add_charop(text, UNDO_BS_1, c);
 }
 
 void txt_backspace_word (Text *text) 
@@ -2436,11 +2621,12 @@ static void txt_convert_tab_to_spaces (Text *text)
 	txt_insert_buf(text, sb);
 }
 
-static int txt_add_char_intern (Text *text, char add, int replace_tabs)
+static int txt_add_char_intern (Text *text, unsigned int add, int replace_tabs)
 {
-	int len, lineno;
-	char *tmp;
+	int lineno;
+	char *tmp, ch[BLI_UTF8_MAX];
 	TextMarker *mrk;
+	size_t add_len;
 	
 	if (!text) return 0;
 	if (!text->curl) return 0;
@@ -2458,43 +2644,42 @@ static int txt_add_char_intern (Text *text, char add, int replace_tabs)
 
 	txt_delete_sel(text);
 	
+	add_len = BLI_str_utf8_from_unicode(add, ch);
 	mrk= txt_find_marker_region(text, text->curl, text->curc-1, text->curl->len, 0, 0);
 	if (mrk) {
 		lineno= mrk->lineno;
 		do {
-			if (mrk->start>text->curc) mrk->start++;
-			mrk->end++;
+			if (mrk->start>text->curc) mrk->start+= add_len;
+			mrk->end+= add_len;
 			mrk= mrk->next;
 		} while (mrk && mrk->lineno==lineno);
 	}
 	
-	tmp= MEM_mallocN(text->curl->len+2, "textline_string");
-	
-	if(text->curc) memcpy(tmp, text->curl->line, text->curc);
-	tmp[text->curc]= add;
+	tmp= MEM_mallocN(text->curl->len+add_len+1, "textline_string");
 	
-	len= text->curl->len - text->curc;
-	if(len>0) memcpy(tmp+text->curc+1, text->curl->line+text->curc, len);
-	tmp[text->curl->len+1]=0;
+	memcpy(tmp, text->curl->line, text->curc);
+	memcpy(tmp+text->curc, ch, add_len);
+	memcpy(tmp+text->curc+add_len, text->curl->line+text->curc, text->curl->len-text->curc+1);
+
 	make_new_line(text->curl, tmp);
 		
-	text->curc++;
+	text->curc+= add_len;
 
 	txt_pop_sel(text);
 	
 	txt_make_dirty(text);
 	txt_clean_text(text);
 
-	if(!undoing) txt_undo_add_charop(text, UNDO_INSERT, add);
+	if(!undoing) txt_undo_add_charop(text, UNDO_INSERT_1, add);
 	return 1;
 }
 
-int txt_add_char (Text *text, char add)
+int txt_add_char (Text *text, unsigned int add)
 {
 	return txt_add_char_intern(text, add, text->flags & TXT_TABSTOSPACES);
 }
 
-int txt_add_raw_char (Text *text, char add)
+int txt_add_raw_char (Text *text, unsigned int add)
 {
 	return txt_add_char_intern(text, add, 0);
 }
@@ -2505,34 +2690,48 @@ void txt_delete_selected(Text *text)
 	txt_make_dirty(text);
 }
 
-int txt_replace_char (Text *text, char add)
+int txt_replace_char (Text *text, unsigned int add)
 {
-	char del;
+	unsigned int del;
+	size_t del_size = 0, add_size;
+	char ch[BLI_UTF8_MAX];
 	
 	if (!text) return 0;
 	if (!text->curl) return 0;
 
 	/* If text is selected or we're at the end of the line just use txt_add_char */
 	if (text->curc==text->curl->len || txt_has_sel(text) || add=='\n') {
-		TextMarker *mrk;
 		int i= txt_add_char(text, add);
-		mrk= txt_find_marker(text, text->curl, text->curc, 0, 0);
-		if (mrk && mrk->end==text->curc) mrk->end--;
+		TextMarker *mrk= txt_find_marker(text, text->curl, text->curc, 0, 0);
+		if (mrk) BLI_freelinkN(&text->markers, mrk);
 		return i;
 	}
 	
-	del= text->curl->line[text->curc];
-	text->curl->line[text->curc]= (unsigned char) add;
-	text->curc++;
-	txt_pop_sel(text);
+	del= BLI_str_utf8_as_unicode_and_size(text->curl->line + text->curc, &del_size);
+	add_size= BLI_str_utf8_from_unicode(add, ch);
 	
+	if (add_size > del_size) {
+		char *tmp= MEM_mallocN(text->curl->len+add_size-del_size+1, "textline_string");
+		memcpy(tmp, text->curl->line, text->curc);
+		memcpy(tmp+text->curc+add_size, text->curl->line+text->curc+del_size, text->curl->len-text->curc-del_size+1);
+		MEM_freeN(text->curl->line);
+		text->curl->line = tmp;
+	} else if (add_size < del_size) {
+		char *tmp= text->curl->line;
+		memmove(tmp+text->curc+add_size, tmp+text->curc+del_size, text->curl->len-text->curc-del_size+1);
+	}
+	
+	memcpy(text->curl->line + text->curc, ch, add_size);
+	text->curc+= add_size;
+	
+	txt_pop_sel(text);
 	txt_make_dirty(text);
 	txt_clean_text(text);
 
 	/* Should probably create a new op for this */
 	if(!undoing) {
-		txt_undo_add_charop(text, UNDO_DEL, del);
-		txt_undo_add_charop(text, UNDO_INSERT, add);
+		txt_undo_add_charop(text, UNDO_DEL_1, del);
+		txt_undo_add_charop(text, UNDO_INSERT_1, add);
 	}
 	return 1;
 }
diff --git a/source/blender/blenloader/intern/readfile.c b/source/blender/blenloader/intern/readfile.c
index 0c05655b8d7..1aa71732697 100644
--- a/source/blender/blenloader/intern/readfile.c
+++ b/source/blender/blenloader/intern/readfile.c
@@ -135,6 +135,7 @@
 #include "BKE_scene.h"
 #include "BKE_screen.h"
 #include "BKE_sequencer.h"
+#include "BKE_text.h" // for txt_extended_ascii_as_utf8
 #include "BKE_texture.h" // for open_plugin_tex
 #include "BKE_tracking.h"
 #include "BKE_utildefines.h" // SWITCH_INT DATA ENDB DNA1 O_BINARY GLOB USER TEST REND
@@ -12976,6 +12977,23 @@ static void do_versions(FileData *fd, Library *lib, Main *main)
 	
 	/* put compatibility code here until next subversion bump */
 	{
+		{
+			/* convert extended ascii to utf-8 for text editor */
+			Text *text;
+			for (text= main->text.first; text; text= text->id.next)
+				if(!(text->flags & TXT_ISEXT)) {
+					TextLine *tl;
+					
+					for (tl= text->lines.first; tl; tl= tl->next) {
+						int added= txt_extended_ascii_as_utf8(&tl->line);
+						tl->len+= added;
+						
+						/* reset cursor position if line was changed */
+						if (added && tl == text->curl)
+							text->curc = 0;
+					}
+				}
+		}
 	}
 
 	/* WATCH IT!!!: pointers from libdata have not been converted yet here! */
diff --git a/source/blender/editors/space_text/text_draw.c b/source/blender/editors/space_text/text_draw.c
index 25b9c2f2864..8a78f236fea 100644
--- a/source/blender/editors/space_text/text_draw.c
+++ b/source/blender/editors/space_text/text_draw.c
@@ -72,7 +72,7 @@ static void text_font_end(SpaceText *UNUSED(st))
 {
 }
 
-static int text_font_draw(SpaceText *UNUSED(st), int x, int y, char *str)
+static int text_font_draw(SpaceText *UNUSED(st), int x, int y, const char *str)
 {
 	BLF_position(mono, x, y, 0);
 	BLF_draw(mono, str, BLF_DRAW_STR_DUMMY_MAX);
@@ -92,19 +92,28 @@ static int text_font_draw_character(SpaceText *st, int x, int y, char c)
 	return st->cwidth;
 }
 
-int text_font_width(SpaceText *UNUSED(st), const char *str)
+static int text_font_draw_character_utf8(SpaceText *st, int x, int y, const char *c)
 {
-	return BLF_width(mono, str);
+	char str[BLI_UTF8_MAX+1];
+	size_t len = BLI_str_utf8_size(c);
+	memcpy(str, c, len);
+	str[len]= '\0';
+
+	BLF_position(mono, x, y, 0);
+	BLF_draw(mono, str, len);
+
+	return st->cwidth;
 }
 
 /****************** flatten string **********************/
 
-static void flatten_string_append(FlattenString *fs, char c, int accum) 
+static void flatten_string_append(FlattenString *fs, const char *c, int accum, int len) 
 {
-	if(fs->pos>=fs->len && fs->pos>=sizeof(fs->fixedbuf)-1) {
+	int i;
+	
+	if(fs->pos+len > fs->len) {
 		char *nbuf; int *naccum;
-		if(fs->len) fs->len*= 2;
-		else fs->len= sizeof(fs->fixedbuf) * 2;
+		fs->len*= 2;
 
 		nbuf= MEM_callocN(sizeof(*fs->buf)*fs->len, "fs->buf");
 		naccum= MEM_callocN(sizeof(*fs->accum)*fs->len, "fs->accum");
@@ -121,35 +130,45 @@ static void flatten_string_append(FlattenString *fs, char c, int accum)
 		fs->accum= naccum;
 	}
 	
-	fs->buf[fs->pos]= c;	
-	fs->accum[fs->pos]= accum;
-	
-	fs->pos++;
+	for (i = 0; i < len; i++)
+	{
+		fs->buf[fs->pos+i]= c[i];
+		fs->accum[fs->pos+i]= accum;
+	}
+
+	fs->pos+= len;
 }
 
 int flatten_string(SpaceText *st, FlattenString *fs, const char *in)
 {
-	int r = 0, i = 0;
+	int r, i, total = 0;
 
 	memset(fs, 0, sizeof(FlattenString));
 	fs->buf= fs->fixedbuf;
 	fs->accum= fs->fixedaccum;
-	
-	for(r=0, i=0; *in; r++, in++) {
-		if(*in=='\t') {
-			if(fs->pos && *(in-1)=='\t')
-				i= st->tabnumber;
-			else if(st->tabnumber > 0)
-				i= st->tabnumber - (fs->pos%st->tabnumber);
+	fs->len = sizeof(fs->fixedbuf);
 
+	for(r = 0, i = 0; *in; r++) {
+		if(*in=='\t') {
+			i= st->tabnumber - (total%st->tabnumber);
+			total+= i;
+			
 			while(i--)
-				flatten_string_append(fs, ' ', r);
+				flatten_string_append(fs, " ", r, 1);
+			
+			in++;
+		}
+		else {
+			size_t len= BLI_str_utf8_size(in);
+			flatten_string_append(fs, in, r, len);
+			in += len;
+			total++;
 		}
-		else
-			flatten_string_append(fs, *in, r);
 	}
+	
+	flatten_string_append(fs, "\0", r, 1);
 
-	return fs->pos;
+	return total;
 }
 
 void flatten_string_free(FlattenString *fs)
@@ -304,9 +323,8 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 	}
 	else orig = 0xFF;
 
-	flatten_string(st, &fs, line->line);
+	len = flatten_string(st, &fs, line->line);
 	str = fs.buf;
-	len = strlen(str);
 	if(!text_check_format_len(line, len)) {
 		flatten_string_free(&fs);
 		return;
@@ -318,7 +336,7 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 		if(*str == '\\') {
 			*fmt = prev; fmt++; str++;
 			if(*str == '\0') break;
-			*fmt = prev; fmt++; str++;
+			*fmt = prev; fmt++; str += BLI_str_utf8_size(str);
 			continue;
 		}
 		/* Handle continuations */
@@ -339,14 +357,16 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 			}
 
 			*fmt = 'l';
+			str += BLI_str_utf8_size(str) - 1;
 		}
 		/* Not in a string... */
 		else {
 			/* Deal with comments first */
-			if(prev == '#' || *str == '#')
+			if(prev == '#' || *str == '#') {
 				*fmt = '#';
-			/* Strings */
-			else if(*str == '"' || *str == '\'') {
+				str += BLI_str_utf8_size(str) - 1;
+			} else if(*str == '"' || *str == '\'') {
+				/* Strings */
 				find = *str;
 				cont = (*str== '"') ? TXT_DBLQUOTSTR : TXT_SNGQUOTSTR;
 				if(*(str+1) == find && *(str+2) == find) {
@@ -371,14 +391,18 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 					}
 					*fmt = 'n';
 				}
-				else
+				else {
+					str += BLI_str_utf8_size(str) - 1;
 					*fmt = 'q';
+				}
 			/* Punctuation */
 			else if(text_check_delim(*str))
 				*fmt = '!';
 			/* Identifiers and other text (no previous ws. or delims. so text continues) */
-			else if(prev == 'q')
+			else if(prev == 'q') {
+				str += BLI_str_utf8_size(str) - 1;
 				*fmt = 'q';
+			}
 			/* Not ws, a digit, punct, or continuing text. Must be new, check for special words */
 			else {
 				/* Special vars(v) or built-in keywords(b) */
@@ -395,8 +419,10 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 					}
 					*fmt = prev;
 				}
-				else
+				else {
+					str += BLI_str_utf8_size(str) - 1;
 					*fmt = 'q';
+				}
 			}
 		}
 		prev = *fmt;
@@ -408,14 +434,11 @@ static void txt_format_line(SpaceText *st, TextLine *line, int do_next)
 	*fmt = '\0'; fmt++;
 	*fmt = cont;
 
-	/* Debugging */
-	//print_format(st, line);
-
 	/* If continuation has changed and we're allowed, process the next line */
 	if(cont!=orig && do_next && line->next) {
 		txt_format_line(st, line->next, do_next);
 	}
-
+	
 	flatten_string_free(&fs);
 }
 
@@ -539,13 +562,14 @@ void wrap_offset(SpaceText *st, ARegion *ar, TextLine *linein, int cursin, int *
 	}
 
 	max= wrap_width(st, ar);
+	cursin = txt_utf8_offset_to_index(linein->line, cursin);
 
 	while(linep) {
 		start= 0;
 		end= max;
 		chop= 1;
 		*offc= 0;
-		for(i=0, j=0; linep->line[j]!='\0'; j++) {
+		for(i=0, j=0; linep->line[j]; j+=BLI_str_utf8_size(linep->line+j)) {
 			int chars;
 
 			/* Mimic replacement of tabs */
@@ -591,6 +615,7 @@ void wrap_offset(SpaceText *st, ARegion *ar, TextLine *linein, int cursin, int *
 	}
 }
 
+/* cursin - mem, offc - view */
 void wrap_offset_in_line(SpaceText *st, ARegion *ar, TextLine *linein, int cursin, int *offl, int *offc)
 {
 	int i, j, start, end, chars, max, chop;
@@ -607,8 +632,9 @@ void wrap_offset_in_line(SpaceText *st, ARegion *ar, TextLine *linein, int cursi
 	end= max;
 	chop= 1;
 	*offc= 0;
+	cursin = txt_utf8_offset_to_index(linein->line, cursin);
 
-	for(i=0, j=0; linein->line[j]!='\0'; j++) {
+	for(i=0, j=0; linein->line[j]; j += BLI_str_utf8_size(linein->line + j)) {
 
 		/* Mimic replacement of tabs */
 		ch= linein->line[j];
@@ -653,7 +679,7 @@ int text_get_char_pos(SpaceText *st, const char *line, int cur)
 {
 	int a=0, i;
 	
-	for(i=0; i<cur && line[i]; i++) {
+	for(i=0; i<cur && line[i]; i += BLI_str_utf8_size(line + i)) {
 		if(line[i]=='\t')
 			a += st->tabnumber-a%st->tabnumber;
 		else
@@ -662,54 +688,65 @@ int text_get_char_pos(SpaceText *st, const char *line, int cur)
 	return a;
 }
 
-static int text_draw_wrapped(SpaceText *st, char *str, int x, int y, int w, char *format, int skip)
+static const char *txt_utf8_get_nth(const char *str, int n)
+{
+	int pos= 0;
+	while (str[pos] && n--) {
+		pos+= BLI_str_utf8_size(str + pos);
+	}
+	return str + pos;
+}
+
+static int text_draw_wrapped(SpaceText *st, const char *str, int x, int y, int w, const char *format, int skip)
 {
 	FlattenString fs;
-	int basex, i, a, len, start, end, max, lines;
+	int basex, i, a, start, end, max, lines; /* view */
+	int mi, ma, mstart, mend;                /* mem */
 	
-	len= flatten_string(st, &fs, str);
+	flatten_string(st, &fs, str);
 	str= fs.buf;
 	max= w/st->cwidth;
 	if(max<8) max= 8;
 	basex= x;
-
 	lines= 1;
-	start= 0;
-	end= max;
-	for(i=0; i<len; i++) {
+	
+	start= 0; mstart= 0;
+	end= max; mend= txt_utf8_get_nth(str, max) - str;
+	
+	for(i=0, mi=0; str[mi]; i++, mi+=BLI_str_utf8_size(str+mi)) {
 		if(i-start >= max) {
 			/* skip hidden part of line */
 			if(skip) {
 				skip--;
-				start= end;
-				end += max;
+				start= end; mstart= mend;
+				end += max; mend= txt_utf8_get_nth(str+mend, max) - str;
 				continue;
 			}
 
 			/* Draw the visible portion of text on the overshot line */
-			for(a=start; a<end; a++) {
+			for(a=start, ma=mstart; a<end; a++, ma+=BLI_str_utf8_size(str+ma)) {
 				if(st->showsyntax && format) format_draw_color(format[a]);
-				x += text_font_draw_character(st, x, y, str[a]);
+				x += text_font_draw_character_utf8(st, x, y, str + ma);
 			}
 			y -= st->lheight;
 			x= basex;
 			lines++;
-			start= end;
-			end += max;
+			start= end; mstart= mend;
+			end += max; mend= txt_utf8_get_nth(str+mend, max) - str;
 
 			if(y<=0) break;
 		}
-		else if(str[i]==' ' || str[i]=='-') {
-			end = i+1;
+		else if(str[mi]==' ' || str[mi]=='-') {
+			end = i+1; mend = mi+1;
 		}
 	}
 
 	/* Draw the remaining text */
-	for(a=start; a<len && y > 0; a++) {
+	for(a=start, ma=mstart; str[ma] && y > 0; a++, ma+=BLI_str_utf8_size(str+ma)) {
 		if(st->showsyntax && format)
 			format_draw_color(format[a]);
 
-		x += text_font_draw_character(st, x, y, str[a]);
+		x += text_font_draw_character_utf8(st, x, y, str+ma);
 	}
 
 	flatten_string_free(&fs);
@@ -717,45 +754,36 @@ static int text_draw_wrapped(SpaceText *st, char *str, int x, int y, int w, char
 	return lines;
 }
 
-static int text_draw(SpaceText *st, char *str, int cshift, int maxwidth, int draw, int x, int y, char *format)
+static int text_draw(SpaceText *st, char *str, int cshift, int maxwidth, int draw, int x, int y, const char *format)
 {
 	FlattenString fs;
-	int r=0, w= 0, amount;
-	int *acc;
-	char *in;
+	int *acc, r=0;
+	const char *in;
 
-	w= flatten_string(st, &fs, str);
+	int w= flatten_string(st, &fs, str);
 	if(w < cshift) {
 		flatten_string_free(&fs);
 		return 0; /* String is shorter than shift */
 	}
 	
-	in= fs.buf+cshift;
+	in= txt_utf8_get_nth(fs.buf, cshift);
 	acc= fs.accum+cshift;
 	w= w-cshift;
 
 	if(draw) {
+		int amount = maxwidth ? MIN2(w, maxwidth) : w;
+		
 		if(st->showsyntax && format) {
-			int a;
+			int a, str_shift= 0;
 			format = format+cshift;
-		
-			amount = strlen(in);
-			if(maxwidth)
-				amount= MIN2(amount, maxwidth);
-			
+
 			for(a = 0; a < amount; a++) {
 				format_draw_color(format[a]);
-				x += text_font_draw_character(st, x, y, in[a]);
+				x += text_font_draw_character_utf8(st, x, y, in + str_shift);
+				str_shift += BLI_str_utf8_size(in + str_shift);
 			}
 		}
-		else {
-			amount = strlen(in);
-			if(maxwidth)
-				amount= MIN2(amount, maxwidth);
-
-			in[amount]= 0;
-			text_font_draw(st, x, y, in);
-		}
+		else text_font_draw(st, x, y, in);
 	}
 	else {
 		while(w-- && *acc++ < maxwidth)
@@ -976,8 +1004,8 @@ int text_get_visible_lines(SpaceText *st, ARegion *ar, const char *str)
 	max= wrap_width(st, ar);
 	lines= 1;
 	start= 0;
-	end= max;
-	for(i= 0, j= 0; str[j] != '\0'; j++) {
+	end= max; 
+	for(i= 0, j= 0; str[j]; j+=BLI_str_utf8_size(str+j)) {
 		/* Mimic replacement of tabs */
 		ch= str[j];
 		if(ch=='\t') {
@@ -1421,7 +1449,7 @@ static void draw_suggestion_list(SpaceText *st, ARegion *ar)
 
 		BLI_strncpy(str, item->name, SUGG_LIST_WIDTH);
 
-		w = text_font_width(st, str);
+		w = BLF_width(mono, str);
 		
 		if(item == sel) {
 			UI_ThemeColor(TH_SHADE2);
@@ -1496,8 +1524,6 @@ static void draw_cursor(SpaceText *st, ARegion *ar)
 				glRecti(x-4, y, ar->winx, y-st->lheight),  y-=st->lheight;
 
 			glRecti(x-4, y, x+toc*st->cwidth, y-st->lheight);  y-=st->lheight;
-
-			(void)y;
 		}
 	}
 	else {
@@ -1569,8 +1595,9 @@ static void draw_brackets(SpaceText *st, ARegion *ar)
 {
 	TextLine *startl, *endl, *linep;
 	Text *text = st->text;
-	int b, c, startc, endc, find, stack;
-	int viewc, viewl, offl, offc, x, y;
+	int b, fc, find, stack, viewc, viewl, offl, offc, x, y;
+	int startc, endc, c;
+	
 	char ch;
 
 	// showsyntax must be on or else the format string will be null
@@ -1584,21 +1611,23 @@ static void draw_brackets(SpaceText *st, ARegion *ar)
 	
 	linep= startl;
 	c= startc;
+	fc= txt_utf8_offset_to_index(linep->line, startc);
 	endl= NULL;
 	endc= -1;
 	find= -b;
 	stack= 0;
 	
 	/* Dont highlight backets if syntax HL is off or bracket in string or comment. */
-	if(!linep->format || linep->format[c] == 'l' || linep->format[c] == '#')
+	if(!linep->format || linep->format[fc] == 'l' || linep->format[fc] == '#')
 		return;
 
 	if(b>0) {
 		/* opening bracket, search forward for close */
-		c++;
+		fc++;
+		c+= BLI_str_utf8_size(linep->line+c);
 		while(linep) {
 			while(c<linep->len) {
-				if(linep->format && linep->format[c] != 'l' && linep->format[c] != '#') {
+				if(linep->format && linep->format[fc] != 'l' && linep->format[fc] != '#') {
 					b= text_check_bracket(linep->line[c]);
 					if(b==find) {
 						if(stack==0) {
@@ -1612,19 +1641,22 @@ static void draw_brackets(SpaceText *st, ARegion *ar)
 						stack++;
 					}
 				}
-				c++;
+				fc++;
+				c+= BLI_str_utf8_size(linep->line+c);
 			}
 			if(endl) break;
 			linep= linep->next;
 			c= 0;
+			fc= 0;
 		}
 	}
 	else {
 		/* closing bracket, search backward for open */
-		c--;
+		fc--;
+		if (c>0) c -= linep->line+c-BLI_str_prev_char_utf8(linep->line+c);
 		while(linep) {
-			while(c>=0) {
-				if(linep->format && linep->format[c] != 'l' && linep->format[c] != '#') {
+			while(fc>=0) {
+				if(linep->format && linep->format[fc] != 'l' && linep->format[fc] != '#') {
 					b= text_check_bracket(linep->line[c]);
 					if(b==find) {
 						if(stack==0) {
@@ -1638,11 +1670,17 @@ static void draw_brackets(SpaceText *st, ARegion *ar)
 						stack++;
 					}
 				}
-				c--;
+				fc--;
+				if (c>0) c -= linep->line+c-BLI_str_prev_char_utf8(linep->line+c);
 			}
 			if(endl) break;
 			linep= linep->prev;
-			if(linep) c= linep->len-1;
+			if(linep) {
+				if (linep->format) fc= strlen(linep->format)-1;
+				else fc= -1;
+				if (linep->len) c= BLI_str_prev_char_utf8(linep->line+linep->len)-linep->line;
+				else fc= -1;
+			}
 		}
 	}
 
diff --git a/source/blender/editors/space_text/text_intern.h b/source/blender/editors/space_text/text_intern.h
index 62cd4fedf0e..043fb97547b 100644
--- a/source/blender/editors/space_text/text_intern.h
+++ b/source/blender/editors/space_text/text_intern.h
@@ -47,9 +47,6 @@ struct wmWindowManager;
 /* text_draw.c */
 void draw_text_main(struct SpaceText *st, struct ARegion *ar);
 
-int text_font_width_character(struct SpaceText *st);
-int text_font_width(struct SpaceText *st, const char *str);
-
 void text_update_line_edited(struct TextLine *line);
 void text_update_edited(struct Text *text);
 void text_update_character_width(struct SpaceText *st);
diff --git a/source/blender/editors/space_text/text_ops.c b/source/blender/editors/space_text/text_ops.c
index 19f0c9bba61..eab06474546 100644
--- a/source/blender/editors/space_text/text_ops.c
+++ b/source/blender/editors/space_text/text_ops.c
@@ -1420,7 +1420,7 @@ static int text_get_cursor_rel(SpaceText* st, ARegion *ar, TextLine *linein, int
 	end= max;
 	chop= loop= 1;
 
-	for(i=0, j=0; loop; j++) {
+	for(i=0, j=0; loop; j+=BLI_str_utf8_size(linein->line+j)) {
 		int chars;
 		/* Mimic replacement of tabs */
 		ch= linein->line[j];
@@ -1595,7 +1595,7 @@ static void txt_wrap_move_bol(SpaceText *st, ARegion *ar, short sel)
 	chop= loop= 1;
 	*charp= 0;
 
-	for(i=0, j=0; loop; j++) {
+	for(i=0, j=0; loop; j+=BLI_str_utf8_size((*linep)->line+j)) {
 		int chars;
 		/* Mimic replacement of tabs */
 		ch= (*linep)->line[j];
@@ -1610,7 +1610,7 @@ static void txt_wrap_move_bol(SpaceText *st, ARegion *ar, short sel)
 				*charp= endj;
 
 				if(j>=oldc) {
-					if(ch=='\0') *charp= start;
+					if(ch=='\0') *charp= txt_utf8_index_to_offset((*linep)->line, start);
 					loop= 0;
 					break;
 				}
@@ -1623,7 +1623,7 @@ static void txt_wrap_move_bol(SpaceText *st, ARegion *ar, short sel)
 			}
 			else if(ch==' ' || ch=='-' || ch=='\0') {
 				if(j>=oldc) {
-					*charp= start;
+					*charp= txt_utf8_index_to_offset((*linep)->line, start);
 					loop= 0;
 					break;
 				}
@@ -1663,7 +1663,7 @@ static void txt_wrap_move_eol(SpaceText *st, ARegion *ar, short sel)
 	chop= loop= 1;
 	*charp= 0;
 
-	for(i=0, j=0; loop; j++) {
+	for(i=0, j=0; loop; j+=BLI_str_utf8_size((*linep)->line+j)) {
 		int chars;
 		/* Mimic replacement of tabs */
 		ch= (*linep)->line[j];
@@ -1675,7 +1675,7 @@ static void txt_wrap_move_eol(SpaceText *st, ARegion *ar, short sel)
 
 		while(chars--) {
 			if(i-start>=max) {
-				if(chop) endj= j-1;
+				if(chop) endj= BLI_str_prev_char_utf8((*linep)->line+j)-(*linep)->line;
 
 				if(endj>=oldc) {
 					if(ch=='\0') *charp= (*linep)->len;
@@ -2364,148 +2364,183 @@ typedef struct SetSelection {
 	short old[2];
 } SetSelection;
 
-static void text_cursor_set_to_pos(SpaceText *st, ARegion *ar, int x, int y, int sel)
+static int flatten_len(SpaceText *st, const char *str)
 {
-	FlattenString fs;
-	Text *text= st->text;
-	TextLine **linep;
-	int *charp;
-	int w;
-
-	text_update_character_width(st);
+	int i, total = 0;
 
-	if(sel) { linep= &text->sell; charp= &text->selc; } 
-	else { linep= &text->curl; charp= &text->curc; }
+	for(i = 0; str[i]; i += BLI_str_utf8_size(str+i)) {
+		if(str[i]=='\t') {
+			total += st->tabnumber - total%st->tabnumber;
+		}
+		else total++;
+	}
 	
-	y= (ar->winy - 2 - y)/st->lheight;
-
-	if(st->showlinenrs)
-		x-= TXT_OFFSET+TEXTXLOC;
-	else
-		x-= TXT_OFFSET;
+	return total;
+}
 
-	if(x<0) x= 0;
-	x = (x/st->cwidth) + st->left;
+static int flatten_index_to_offset(SpaceText *st, const char *str, int index)
+{
+	int i, j;
+	for (i= 0, j= 0; i < index; j += BLI_str_utf8_size(str+j))
+		if(str[j]=='\t')
+			i += st->tabnumber - i%st->tabnumber;
+		else
+			i++;
 	
-	if(st->wordwrap) {
-		int i, j, endj, curs, max, chop, start, end, loop, found;
-		char ch;
-
-		/* Point to first visible line */
-		*linep= text->lines.first;
-		i= st->top;
-		while(i>0 && *linep) {
-			int lines= text_get_visible_lines(st, ar, (*linep)->line);
+	return j;
+}
 
-			if (i-lines<0) {
-				y+= i;
-				break;
-			} else {
-				*linep= (*linep)->next;
-				i-= lines;
-			}
+static TextLine *get_first_visible_line(SpaceText *st, ARegion *ar, int *y)
+{
+	TextLine *linep = st->text->lines.first;
+	int i;
+	for (i = st->top; i > 0 && linep; ) {
+		int lines = text_get_visible_lines(st, ar, linep->line);
+		
+		if (i-lines < 0) {
+			*y += i;
+			break;
+		} else {
+			linep = linep->next;
+			i -= lines;
 		}
+	}
+	return linep;
+}
 
-		max= wrap_width(st, ar);
-
-		loop= 1;
-		found= 0;
-		while(loop && *linep) {
-			start= 0;
-			end= max;
-			chop= 1;
-			curs= 0;
-			endj= 0;
-			for(i=0, j=0; loop; j++) {
-				int chars;
-
-				/* Mimic replacement of tabs */
-				ch= (*linep)->line[j];
-				if(ch=='\t') {
-					chars= st->tabnumber-i%st->tabnumber;
-					ch= ' ';
-				}
-				else
-					chars= 1;
-
-				while(chars--) {
-					/* Gone too far, go back to last wrap point */
-					if(y<0) {
-						*charp= endj;
-						loop= 0;
-						break;
+static void text_cursor_set_to_pos_wrapped(SpaceText *st, ARegion *ar, int x, int y, int sel)
+{
+	Text *text = st->text;
+	int max = wrap_width(st, ar); /* view */
+	int charp;                    /* mem */
+	int loop = 1, found = 0;      /* flags */
+	char ch;
+	
+	/* Point to first visible line */
+	TextLine *linep = get_first_visible_line(st, ar, &y);
+	
+	while(loop && linep) {
+		int i = 0, start = 0, end = max; /* view */
+		int j = 0, curs = 0, endj = 0;   /* mem */
+		int chop = 1;                    /* flags */
+		
+		for (; loop; j += BLI_str_utf8_size(linep->line+j)) {
+			int chars;
+			
+			/* Mimic replacement of tabs */
+			ch = linep->line[j];
+			if(ch == '\t') {
+				chars = st->tabnumber - i%st->tabnumber;
+				ch = ' ';
+			}
+			else chars = 1;
+			
+			while (chars--) {
+				/* Gone too far, go back to last wrap point */
+				if (y < 0) {
+					charp = endj;
+					loop = 0;
+					break;
 					/* Exactly at the cursor */
-					}
-					else if(y==0 && i-start==x) {
-						/* current position could be wrapped to next line */
-						/* this should be checked when end of current line would be reached */
-						*charp= curs= j;
-						found= 1;
+				}
+				else if (y == 0 && i-start == x) {
+					/* current position could be wrapped to next line */
+					/* this should be checked when end of current line would be reached */
+					charp = curs= j;
+					found = 1;
 					/* Prepare curs for next wrap */
+				}
+				else if(i - end == x) {
+					curs = j;
+				}
+				if (i - start >= max) {
+					if (found) {
+						/* exact cursor position was found, check if it's */
+						/* still on needed line (hasn't been wrapped) */
+						if (charp > endj && !chop && ch!='\0') charp = endj;
+						loop = 0;
+						break;
 					}
-					else if(i-end==x) {
-						curs= j;
+					
+					if(chop) endj = j;
+					start = end;
+					end += max;
+					
+					if(j < linep->len)
+						y--;
+					
+					chop = 1;
+					if (y == 0 && i-start >= x) {
+						charp = curs;
+						loop = 0;
+						break;
 					}
-					if(i-start>=max) {
-						if(found) {
-							/* exact cursor position was found, check if it's */
-							/* still on needed line (hasn't been wrapped) */
-							if(*charp>endj && !chop && ch!='\0') (*charp)= endj;
-							loop= 0;
-							break;
-						}
-
-						if(chop) endj= j;
-						start= end;
-						end += max;
-
-						if(j<(*linep)->len)
-							y--;
-
-						chop= 1;
-						if(y==0 && i-start>=x) {
-							*charp= curs;
-							loop= 0;
-							break;
-						}
+				}
+				else if (ch == ' ' || ch == '-' || ch == '\0') {
+					if (found) {
+						loop = 0;
+						break;
 					}
-					else if(ch==' ' || ch=='-' || ch=='\0') {
-						if(found) {
-							loop= 0;
-							break;
-						}
-
-						if(y==0 && i-start>=x) {
-							*charp= curs;
-							loop= 0;
-							break;
-						}
-						end = i+1;
-						endj = j;
-						chop= 0;
+					
+					if(y == 0 && i-start >= x) {
+						charp = curs;
+						loop = 0;
+						break;
 					}
-					i++;
+					end = i + 1;
+					endj = j;
+					chop = 0;
 				}
-				if(ch=='\0') break;
+				i++;
 			}
-			if(!loop || found) break;
+			
+			if(ch == '\0') break;
+		}
+		
+		if(!loop || found) break;
+		
+		if(!linep->next) {
+			charp = linep->len;
+			break;
+		}
+		
+		/* On correct line but didn't meet cursor, must be at end */
+		if (y == 0) {
+			charp = linep->len;
+			break;
+		}
+		linep = linep->next;
+		
+		y--;
+	}
+	
+	if(sel) { text->sell = linep; text->selc = charp; } 
+	else { text->curl = linep; text->curc = charp; }
+}
 
-			if(!(*linep)->next) {
-				*charp= (*linep)->len;
-				break;
-			}
+static void text_cursor_set_to_pos(SpaceText *st, ARegion *ar, int x, int y, int sel)
+{
+	Text *text= st->text;
+	text_update_character_width(st);
+	y= (ar->winy - 2 - y)/st->lheight;
 
-			/* On correct line but didn't meet cursor, must be at end */
-			if(y==0) {
-				*charp= (*linep)->len;
-				break;
-			}
-			*linep= (*linep)->next;
-			y--;
-		}
+	if(st->showlinenrs) x-= TXT_OFFSET+TEXTXLOC;
+	else x-= TXT_OFFSET;
 
+	if(x<0) x= 0;
+	x = (x/st->cwidth) + st->left;
+	
+	if(st->wordwrap) {
+		text_cursor_set_to_pos_wrapped(st, ar, x, y, sel);
 	}
 	else {
+		TextLine **linep;
+		int *charp;
+		int w;
+		
+		if(sel) { linep= &text->sell; charp= &text->selc; } 
+		else { linep= &text->curl; charp= &text->curc; }
+		
 		y-= txt_get_span(text->lines.first, *linep) - st->top;
 		
 		if(y>0) {
@@ -2516,10 +2551,9 @@ static void text_cursor_set_to_pos(SpaceText *st, ARegion *ar, int x, int y, int
 		}
 
 		
-		w= flatten_string(st, &fs, (*linep)->line);
-		if(x<w) *charp= fs.accum[x];
+		w= flatten_len(st, (*linep)->line);
+		if(x<w) *charp= flatten_index_to_offset(st, (*linep)->line, x);
 		else *charp= (*linep)->len;
-		flatten_string_free(&fs);
 	}
 	if(!sel) txt_pop_sel(text);
 }
@@ -2762,19 +2796,23 @@ static int text_insert_exec(bContext *C, wmOperator *op)
 	SpaceText *st= CTX_wm_space_text(C);
 	Text *text= CTX_data_edit_text(C);
 	char *str;
-	int done = 0, i;
+	int done = 0;
+	size_t i = 0;
+	unsigned int code;
 
 	text_drawcache_tag_update(st, 0);
 
 	str= RNA_string_get_alloc(op->ptr, "text", NULL, 0);
 
 	if(st && st->overwrite) {
-		for(i=0; str[i]; i++) {
-			done |= txt_replace_char(text, str[i]);
+		while (str[i]) {
+			code = BLI_str_utf8_as_unicode_step(str, &i);
+			done |= txt_replace_char(text, code);
 		}
 	} else {
-		for(i=0; str[i]; i++) {
-			done |= txt_add_char(text, str[i]);
+		while (str[i]) {
+			code = BLI_str_utf8_as_unicode_step(str, &i);
+			done |= txt_add_char(text, code);
 		}
 	}
 
@@ -2802,9 +2840,17 @@ static int text_insert_invoke(bContext *C, wmOperator *op, wmEvent *event)
 			return OPERATOR_PASS_THROUGH;
 		}
 		else {
-			char str[2];
-			str[0]= event->ascii;
-			str[1]= '\0';
+			char str[BLI_UTF8_MAX+1];
+			size_t len;
+			
+			if (event->utf8_buf[0]) {
+				len = BLI_str_utf8_size(event->utf8_buf);
+				memcpy(str, event->utf8_buf, len);
+			} else {
+				/* in theory, ghost can set value to extended ascii here */
+				len = BLI_str_utf8_from_unicode(event->ascii, str);
+			}
+			str[len]= '\0';
 			RNA_string_set(op->ptr, "text", str);
 		}
 	}