3 files changed, 31 insertions, 16 deletions

diff --git a/include/mupdf/fitz/string.h b/include/mupdf/fitz/string.h
index 81429b67..8daa28fc 100644
--- a/include/mupdf/fitz/string.h
+++ b/include/mupdf/fitz/string.h
@@ -103,6 +103,16 @@ int fz_runetochar(char *str, int rune);
 int fz_runelen(int rune);
 
 /*
+	fz_utflen: Count how many runes the UTF-8 encoded string
+	consists of.
+
+	s: The UTF-8 encoded, NUL-terminated text string.
+
+	Returns the number of runes in the string.
+*/
+int fz_utflen(const char *s);
+
+/*
 	fz_strtod/fz_strtof: Locale-independent decimal to binary
 	conversion. On overflow return (-)INFINITY and set errno to ERANGE. On
 	underflow return 0 and set errno to ERANGE. Special inputs (case
diff --git a/source/fitz/string.c b/source/fitz/string.c
index d22efe8e..025b612c 100644
--- a/source/fitz/string.c
+++ b/source/fitz/string.c
@@ -352,6 +352,24 @@ fz_runelen(int c)
 	return fz_runetochar(str, c);
 }
 
+int
+fz_utflen(const char *s)
+{
+	int c, n, rune;
+	n = 0;
+	for(;;) {
+		c = *(const unsigned char*)s;
+		if(c < Runeself) {
+			if(c == 0)
+				return n;
+			s++;
+		} else
+			s += fz_chartorune(&rune, s);
+		n++;
+	}
+	return 0;
+}
+
 float fz_atof(const char *s)
 {
 	float result;
diff --git a/source/html/html-layout.c b/source/html/html-layout.c
index 5b84df6b..7be32592 100644
--- a/source/html/html-layout.c
+++ b/source/html/html-layout.c
@@ -1808,19 +1808,6 @@ typedef struct
 	uni_buf *buffer;
 } bidi_data;
 
-static size_t utf8len(const char *text)
-{
-	size_t len = 0;
-
-	while (*text)
-	{
-		int rune;
-		text += fz_chartorune(&rune, text);
-		len++;
-	}
-	return len;
-}
-
 static void newFragCb(const uint32_t *fragment,
 			size_t fragment_len,
 			int block_r2l,
@@ -1855,12 +1842,12 @@ static void newFragCb(const uint32_t *fragment,
 		else
 		{
 			/* Must be text */
-			len = utf8len(data->flow->content.text);
+			len = fz_utflen(data->flow->content.text);
 			if (len > fragment_len)
 			{
 				/* We need to split this flow box */
 				(void)split_flow(data->ctx, data->pool, data->flow, fragment_len);
-				len = utf8len(data->flow->content.text);
+				len = fz_utflen(data->flow->content.text);
 			}
 		}
 
@@ -1905,7 +1892,7 @@ detect_flow_directionality(fz_context *ctx, fz_pool *pool, uni_buf *buffer, fz_b
 			switch (end->type)
 			{
 			case FLOW_WORD:
-				len = utf8len(end->content.text);
+				len = fz_utflen(end->content.text);
 				text = end->content.text;
 				break;
 			case FLOW_SPACE: