[pango] Bug 668154 - GTK+ Crashes on Non BMP utf-8 charpoints

[Behdad Esfahbod <behdad src gnome org>]

commit 6de9b3b340e30856daefc418dc59545afbe62bf0
Author: Behdad Esfahbod <behdad behdad org>
Date:   Mon Feb 25 23:48:16 2013 -0500

    Bug 668154 - GTK+ Crashes on Non BMP utf-8 charpoints

 modules/basic/basic-win32.c |   45 +++++++++++++++++-------------------------
 1 files changed, 18 insertions(+), 27 deletions(-)
---
diff --git a/modules/basic/basic-win32.c b/modules/basic/basic-win32.c
index 2483022..f22f984 100644
--- a/modules/basic/basic-win32.c
+++ b/modules/basic/basic-win32.c
@@ -321,21 +321,6 @@ dump_glyphs_and_log_clusters (gboolean rtl,
 
 #endif /* BASIC_WIN32_DEBUGGING */
 
-static int
-unichar_index (wchar_t *wtext,
-              int      ix)
-{
-  int i, index;
-
-  index = 0;
-  for (i = 0; i < ix; i++)
-    /* Ignore the low surrogate */
-    if (!(wtext[i] >= 0xDC00 && wtext[i] < 0xE000))
-      index++;
-
-  return index;
-}
-
 static void
 set_up_pango_log_clusters (wchar_t *wtext,
                           gboolean rtl,
@@ -354,13 +339,13 @@ set_up_pango_log_clusters (wchar_t *wtext,
        * log_clusters array forwards.
        */
       int glyph0 = 0;
-      first_char_in_cluster = unichar_index (wtext, itemlen - 1);
+      first_char_in_cluster = itemlen - 1;
       for (j = itemlen - 1; j >= 0; j--)
        {
          if (j < itemlen - 1 && usp_log_clusters[j+1] != usp_log_clusters[j])
            {
              /* Cluster starts */
-             first_char_in_cluster = unichar_index (wtext, j);
+             first_char_in_cluster = j;
            }
          if (j == 0)
            {
@@ -371,7 +356,7 @@ set_up_pango_log_clusters (wchar_t *wtext,
          else if (usp_log_clusters[j-1] == usp_log_clusters[j])
            {
              /* Cluster continues */
-             first_char_in_cluster = unichar_index (wtext, j-1);
+             first_char_in_cluster = j-1;
            }
          else
            {
@@ -393,7 +378,7 @@ set_up_pango_log_clusters (wchar_t *wtext,
          if (j > 0 && usp_log_clusters[j-1] != usp_log_clusters[j])
            {
              /* Cluster starts */
-             first_char_in_cluster = unichar_index (wtext, j);
+             first_char_in_cluster = j;
            }
          if (j == itemlen - 1)
            {
@@ -418,12 +403,12 @@ set_up_pango_log_clusters (wchar_t *wtext,
 static void
 convert_log_clusters_to_byte_offsets (const char       *text,
                                      gint              length,
-                                     PangoGlyphString *glyphs)
+                                     PangoGlyphString *glyphs,
+                                     gint              utf16_len)
 {
   const char *p;
   int charix, glyphix;
-  int n_chars = g_utf8_strlen (text, length);
-  int *byte_offset = g_new (int, n_chars);
+  int *byte_offset = g_new (int, utf16_len);
 
   p = text;
   charix = 0;
@@ -431,14 +416,20 @@ convert_log_clusters_to_byte_offsets (const char       *text,
     {
       byte_offset[charix] = p - text;
       charix++;
+      if (g_utf8_get_char (p) > 0xFFFF)
+      {
+       byte_offset[charix] = p - text;
+       charix++;
+      }
       p = g_utf8_next_char (p);
     }
+  g_assert (charix <= utf16_len);
 
-  /* Convert char indexes in the log_clusters array to byte offsets.
+  /* Convert utf16 indexes in the log_clusters array to byte offsets.
    */
   for (glyphix = 0; glyphix < glyphs->num_glyphs; glyphix++)
     {
-      g_assert (glyphs->log_clusters[glyphix] < n_chars);
+      g_assert (glyphs->log_clusters[glyphix] < utf16_len);
       glyphs->log_clusters[glyphix] = byte_offset[glyphs->log_clusters[glyphix]];
     }
 
@@ -525,7 +516,7 @@ itemize_shape_and_place (PangoFont           *font,
        * count as two!
        */
       itemlen = items[item+1].iCharPos - items[item].iCharPos;
-      char_offset = unichar_index (wtext, items[item].iCharPos);
+      char_offset = items[item].iCharPos;
 
 #ifdef BASIC_WIN32_DEBUGGING
       if (pango_win32_debug)
@@ -673,7 +664,7 @@ uniscribe_shape (PangoFont           *font,
 
   if (retval)
     {
-      convert_log_clusters_to_byte_offsets (text, length, glyphs);
+      convert_log_clusters_to_byte_offsets (text, length, glyphs, wlen);
 #ifdef BASIC_WIN32_DEBUGGING
       if (pango_win32_debug)
        {
@@ -754,7 +745,7 @@ basic_engine_shape (PangoEngineShape        *engine,
       wc = g_utf8_get_char (p);
 
       if (analysis->level % 2)
-       if (pango_get_mirror_char (wc, &mirrored_ch))
+       if (g_unichar_get_mirror_char (wc, &mirrored_ch))
          wc = mirrored_ch;
 
       if (wc == 0xa0)  /* non-break-space */