[Autofill] Credit Card Autofill Last Used Date Experiment

base/i18n/time_formatting_unittest.cc

Index: base/i18n/time_formatting_unittest.cc
diff --git a/base/i18n/time_formatting_unittest.cc b/base/i18n/time_formatting_unittest.cc
index 06cd77284a61b388e0f12f13778dda8da3f4862b..74d0e4fa3897b1b8a395b9056809b555b25e5f1f 100644
--- a/base/i18n/time_formatting_unittest.cc
+++ b/base/i18n/time_formatting_unittest.cc
@@ -224,6 +224,33 @@ TEST(TimeFormattingTest, MAYBE_TimeFormatDateGB) {
             TimeFormatFriendlyDate(time));
 }
 
+TEST(TimeFormattingTest, TimeFormatWithPattern) {
+  test::ScopedRestoreICUDefaultLocale restore_locale;
+
+  Time time;
+  EXPECT_TRUE(Time::FromLocalExploded(kTestDateTimeExploded, &time));
+
+  i18n::SetICUDefaultLocale("en_US");

[jungshik at Google, 2017/01/26 19:29:14]

Thank for adding tests. 

I'm sorry that I forgot to tell you about "base/test/scoped_locale.h". 

To avoid any potential misbehavior, we'd better limit the locale change to a
given scope. 

That means, different locale cases need to be in separate functions or en-US,
en-GB and ja-JP have to be enclosed by braces ( { en-US }, {en-GB test} , {ja-JP
test} ). 

[jiahuiguo, 2017/01/27 09:31:05]

Thanks for the comments, separate to different unittests. Do we still need to
disable the test on Android?  

+  EXPECT_EQ(ASCIIToUTF16("Apr 30, 2011"), TimeFormatWithPattern(time, "yMMMd"));
+  EXPECT_EQ(ASCIIToUTF16("April 30, 3:42:07 PM"),
+            TimeFormatWithPattern(time, "MMMMdjmmss"));
+
+  i18n::SetICUDefaultLocale("en_GB");
+  EXPECT_EQ(ASCIIToUTF16("30 Apr 2011"), TimeFormatWithPattern(time, "yMMMd"));
+  EXPECT_EQ(ASCIIToUTF16("30 April, 15:42:07"),
+            TimeFormatWithPattern(time, "MMMMdjmmss"));
+
+  i18n::SetICUDefaultLocale("ja_JP");
+  EXPECT_EQ(WideToUTF16(L"2011\x5e74"
+                        L"4\x6708"
+                        L"30\x65e5"),

[jungshik at Google, 2017/01/26 19:29:14]

Now that every compiler we use support UTF-16 string literals and MSVC (Chromium
requires) finally supports the source code in UTF-8 without BOM, you can just
use u"2011年4月30日".  (I know that a lot of code uses hex-escapes and
WideToUTF16). 

[jiahuiguo, 2017/01/27 09:31:05]

Actually if we use u"2011年4月30日" as the first param, it returns a const char16_t
string, while a string16 is expected here, we still need to have a convertion.
Can you provide an example or suggest how to deal with this?  In this new patch
I use EXPECT_EQ(WideToUTF16(L"2011年4月30日"), TimeFormatWithPattern(time,
"yMMMd")) directly, will this be acceptable since we need an extra conversion
anyway?

+            TimeFormatWithPattern(time, "yMMMd"));
+  EXPECT_EQ(WideToUTF16(L"4\x6708"
+                        L"30\x65e5") +
+                ASCIIToUTF16(" 15:42:07"),
+            TimeFormatWithPattern(time, "MMMMdjmmss"));
+}
+
 TEST(TimeFormattingTest, TimeDurationFormat) {
   test::ScopedRestoreICUDefaultLocale restore_locale;
   TimeDelta delta = TimeDelta::FromMinutes(15 * 60 + 42);