commit 0f3267590980bccd39f2d954320aa97cf1901c0b
author Ryan Mitchell <rtmitchell@google.com> Mon Mar 11 11:00:25 2019 -0700
committer Ryan Mitchell <rtmitchell@google.com> Mon Mar 11 11:00:25 2019 -0700
tree 2c265050794cdf04c8d3c7aa0d7c0b259a47958e
parent 41469c3b522f28c4590398ad6e35193b2142dda7

Do not convert whitespace chars above max char to regular space

Using isspace and iswspace on characters above the maximum char value is
undefined. This change makes aapt2 use isspace like aapt and only trims
whitespace characters at or below the maximum char value like aapt2.

Bug: 121017795
Test: aapt2_tests
Change-Id: I015e4b77f0ff53e409e880fcf9ae104ba3444d1a

tools/aapt2/ResourceParser_test.cpp

diff --git a/tools/aapt2/ResourceParser_test.cpp b/tools/aapt2/ResourceParser_test.cpp
index 251ca0c..5336141 100644
--- a/tools/aapt2/ResourceParser_test.cpp
+++ b/tools/aapt2/ResourceParser_test.cpp
@@ -183,11 +183,11 @@
   EXPECT_THAT(str->untranslatable_sections, IsEmpty());
 
   // Preserve non-ASCII whitespace including extended ASCII characters
-  EXPECT_TRUE(TestParse(R"(<string name="foo3">&#160;Hello&#160;</string>)"));
+  EXPECT_TRUE(TestParse(R"(<string name="foo3">&#160;Hello&#x202F;World&#160;</string>)"));
 
   str = test::GetValue<String>(&table_, "string/foo3");
   ASSERT_THAT(str, NotNull());
-  EXPECT_THAT(*str->value, StrEq("\xC2\xA0Hello\xC2\xA0"));
+  EXPECT_THAT(*str->value, StrEq("\xC2\xA0Hello\xE2\x80\xAFWorld\xC2\xA0"));
   EXPECT_THAT(str->untranslatable_sections, IsEmpty());
 
   EXPECT_TRUE(TestParse(R"(<string name="foo4">2005年6月1日</string>)"));

tools/aapt2/ResourceUtils.cpp

diff --git a/tools/aapt2/ResourceUtils.cpp b/tools/aapt2/ResourceUtils.cpp
index f0e4d9e..a571aee 100644
--- a/tools/aapt2/ResourceUtils.cpp
+++ b/tools/aapt2/ResourceUtils.cpp
@@ -40,8 +40,6 @@
 namespace aapt {
 namespace ResourceUtils {
 
-constexpr int32_t kNonBreakingSpace = 0xa0;
-
 Maybe<ResourceName> ToResourceName(
     const android::ResTable::resource_name& name_in) {
   // TODO: Remove this when ResTable and AssetManager(1) are removed from AAPT2
@@ -854,7 +852,8 @@
   Utf8Iterator iter(text);
   while (iter.HasNext()) {
     char32_t codepoint = iter.Next();
-    if (!preserve_spaces_ && !quote_ && codepoint != kNonBreakingSpace && iswspace(codepoint)) {
+    if (!preserve_spaces_ && !quote_ && (codepoint <= std::numeric_limits<char>::max())
+                                         && isspace(static_cast<char>(codepoint))) {
       if (!last_codepoint_was_space_) {
         // Emit a space if it's the first.
         xml_string_.text += ' ';