From cd159b6775b16cc8c2449422a4ed0d0e9f66df57 Mon Sep 17 00:00:00 2001
From: Michael Drake <tlsa@netsurf-browser.org>
Date: Wed, 18 Jun 2014 13:12:08 +0100
Subject: Don't display certain invisible characters.

  Fixes display of U+200E code points all over Google
  search results.
---
 framebuffer/font_internal.c | 29 ++++++++++++++++++++++++++---
 framebuffer/font_internal.h |  5 +++++
 framebuffer/framebuffer.c   |  3 +++
 3 files changed, 34 insertions(+), 3 deletions(-)

diff --git a/framebuffer/font_internal.c b/framebuffer/font_internal.c
index 5aada8f11..e514b12d9 100644
--- a/framebuffer/font_internal.c
+++ b/framebuffer/font_internal.c
@@ -287,11 +287,24 @@ static nserror utf8_from_local(const char *string,
 	return NSERROR_OK;
 }
 
+
 static bool nsfont_width(const plot_font_style_t *fstyle,
                          const char *string, size_t length,
                          int *width)
 {
-        *width = FB_FONT_WIDTH * utf8_bounded_length(string, length);
+        size_t nxtchr = 0;
+
+	*width = 0;
+        while (nxtchr < length) {
+		uint32_t ucs4;
+		ucs4 = utf8_to_ucs4(string + nxtchr, length - nxtchr);
+		if (codepoint_displayable(ucs4)) {
+			*width += FB_FONT_WIDTH;
+		}
+
+		nxtchr = utf8_next(string, length, nxtchr);
+        }
+
 	return true;
 }
 
@@ -315,10 +328,15 @@ static bool nsfont_position_in_string(const plot_font_style_t *fstyle,
 	int x_pos = 0;
 
         while (nxtchr < length) {
+		uint32_t ucs4;
                 if (abs(x_pos - x) <= (FB_FONT_WIDTH / 2))
                         break;
 
-                x_pos += FB_FONT_WIDTH;
+		ucs4 = utf8_to_ucs4(string + nxtchr, length - nxtchr);
+		if (codepoint_displayable(ucs4)) {
+			x_pos += FB_FONT_WIDTH;
+		}
+
                 nxtchr = utf8_next(string, length, nxtchr);
         }
 
@@ -363,13 +381,18 @@ static bool nsfont_split(const plot_font_style_t *fstyle,
 
         *actual_x = 0;
         while (nxtchr < length) {
+		uint32_t ucs4;
 
                 if (string[nxtchr] == ' ') {
                         last_space_x = *actual_x;
                         last_space_idx = nxtchr;
                 }
 
-                *actual_x += FB_FONT_WIDTH;
+		ucs4 = utf8_to_ucs4(string + nxtchr, length - nxtchr);
+		if (codepoint_displayable(ucs4)) {
+			*actual_x += FB_FONT_WIDTH;
+		}
+
                 if (*actual_x > x && last_space_idx != 0) {
                         /* string has exceeded available width and we've
                          * found a space; return previous space */
diff --git a/framebuffer/font_internal.h b/framebuffer/font_internal.h
index 236183ace..ead179c99 100644
--- a/framebuffer/font_internal.h
+++ b/framebuffer/font_internal.h
@@ -19,6 +19,8 @@
 #ifndef NETSURF_FB_FONT_INTERNAL_H
 #define NETSURF_FB_FONT_INTERNAL_H
 
+#include <stdbool.h>
+
 struct fb_font_desc {
     const char *name;
     int width, height, pitch;
@@ -39,5 +41,8 @@ enum fb_font_style fb_get_font_style(const plot_font_style_t *fstyle);
 
 const uint8_t * fb_get_glyph(uint32_t ucs4, enum fb_font_style style);
 
+#define codepoint_displayable(u) \
+	(!(u >= 0x200b && u <= 0x200f))
+
 #endif /* NETSURF_FB_FONT_INTERNAL_H */
 
diff --git a/framebuffer/framebuffer.c b/framebuffer/framebuffer.c
index 0115d838c..4cd064c34 100644
--- a/framebuffer/framebuffer.c
+++ b/framebuffer/framebuffer.c
@@ -140,6 +140,9 @@ static bool framebuffer_plot_text(int x, int y, const char *text, size_t length,
         ucs4 = utf8_to_ucs4(text + nxtchr, length - nxtchr);
         nxtchr = utf8_next(text, length, nxtchr);
 
+	if (!codepoint_displayable(ucs4))
+		continue;
+
         loc.x0 = x;
         loc.y0 = y;
         loc.x1 = loc.x0 + FB_FONT_WIDTH;
-- 
cgit v1.2.3