Pass through non-character codepoints in UTF-8,16,32 and Wide conversion functions.
They're structurally valid code points unlike malformed byte/surrogate sequences. I believe it's better to leave them
alone in conversion functions.
This CL was triggered by file_util_unittest failure on Linux/Mac with my upcoming change
to file_util::ReplaceIllegalCharacters (a part of http://codereview.chromium.org/126223 ).
In addition, the upper bound for the output length in CodepageToWide was tightened.
TEST=pass string_util and file_util unittests
BUG=NONE
Review URL: http://codereview.chromium.org/147038
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@19132 0039d316-1c4b-4281-b951-d872f2087c98
CrOS-Libchrome-Original-Commit: 8df44a01ec210a3e0c04191fb34b392727017a2c
diff --git a/base/file_util_unittest.cc b/base/file_util_unittest.cc
index e0884f7..738bac8 100644
--- a/base/file_util_unittest.cc
+++ b/base/file_util_unittest.cc
@@ -839,11 +839,9 @@
#if defined(OS_WIN)
{L"bad*file\\name.jpg", L"bad-file-name.jpg"},
{L"\t bad*file\\name/.jpg ", L"bad-file-name-.jpg"},
- {L"bad\uFFFFfile\U0010FFFEname.jpg ", L"bad-file-name.jpg"},
#elif defined(OS_POSIX)
{L"bad*file?name.jpg", L"bad-file-name.jpg"},
{L"\t bad*file?name/.jpg ", L"bad-file-name-.jpg"},
- {L"bad\uFFFFfile-name.jpg ", L"bad-file-name.jpg"},
#endif
{L"this_file_name is okay!.mp3", L"this_file_name is okay!.mp3"},
{L"\u4E00\uAC00.mp3", L"\u4E00\uAC00.mp3"},
@@ -851,6 +849,9 @@
{L"\U00010330\U00010331.mp3", L"\U00010330\U00010331.mp3"},
// Unassigned codepoints are ok.
{L"\u0378\U00040001.mp3", L"\u0378\U00040001.mp3"},
+ // Non-characters are not allowed.
+ {L"bad\uFFFFfile\U0010FFFEname.jpg ", L"bad-file-name.jpg"},
+ {L"bad\uFDD0file\uFDEFname.jpg ", L"bad-file-name.jpg"},
};
TEST_F(FileUtilTest, ReplaceIllegalCharactersTest) {