sf3-003_0001-Correct-the-length-calculation.patch

From d45ffefae10a9a0fba279fb9ab249a70bd52e060 Mon Sep 17 00:00:00 2001
From: "Joshua J. Drake" <android-open-source@qoop.org>
Date: Sat, 15 Aug 2015 07:37:55 -0500
Subject: [PATCH] Correct the length calculation

In some cases the utf16_to_utf8_length incorrectly increments the src pointer.
This results in the length of the utf8 string being incorrect and can lead to
buffer problems in calling code.

Change-Id: Id1170658aa5b1d56acfd3d882e788632ca42b7eb
---
 libutils/Unicode.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libutils/Unicode.cpp b/libutils/Unicode.cpp
index 6f4b721..626456f 100644
--- a/libutils/Unicode.cpp
+++ b/libutils/Unicode.cpp
@@ -408,10 +408,10 @@ ssize_t utf16_to_utf8_length(const char16_t *src, size_t src_len)
     const char16_t* const end = src + src_len;
     while (src < end) {
         if ((*src & 0xFC00) == 0xD800 && (src + 1) < end
-                && (*++src & 0xFC00) == 0xDC00) {
+                && (*(src + 1) & 0xFC00) == 0xDC00) {
             // surrogate pairs are always 4 bytes.
             ret += 4;
-            src++;
+            src += 2;
         } else {
             ret += utf32_codepoint_utf8_length((char32_t) *src++);
         }
-- 
1.9.1