unittests/Support/ConvertUTFTest.cpp

   1 //===- llvm/unittest/Support/ConvertUTFTest.cpp - ConvertUTF tests --------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9
  10 #include "llvm/Support/ConvertUTF.h"
  11 #include "gtest/gtest.h"
  12 #include <string>
  13
  14 using namespace llvm;
  15
  16 TEST(ConvertUTFTest, ConvertUTF16LittleEndianToUTF8String) {
  17   // Src is the look of disapproval.
  18   static const char Src[] = "\xff\xfe\xa0\x0c_\x00\xa0\x0c";
  19   ArrayRef<char> Ref(Src, sizeof(Src) - 1);
  20   std::string Result;
  21   bool Success = convertUTF16ToUTF8String(Ref, Result);
  22   EXPECT_TRUE(Success);
  23   std::string Expected("\xe0\xb2\xa0_\xe0\xb2\xa0");
  24   EXPECT_EQ(Expected, Result);
  25 }
  26
  27 TEST(ConvertUTFTest, ConvertUTF16BigEndianToUTF8String) {
  28   // Src is the look of disapproval.
  29   static const char Src[] = "\xfe\xff\x0c\xa0\x00_\x0c\xa0";
  30   ArrayRef<char> Ref(Src, sizeof(Src) - 1);
  31   std::string Result;
  32   bool Success = convertUTF16ToUTF8String(Ref, Result);
  33   EXPECT_TRUE(Success);
  34   std::string Expected("\xe0\xb2\xa0_\xe0\xb2\xa0");
  35   EXPECT_EQ(Expected, Result);
  36 }
  37
  38 TEST(ConvertUTFTest, OddLengthInput) {
  39   std::string Result;
  40   bool Success = convertUTF16ToUTF8String(ArrayRef<char>("xxxxx", 5), Result);
  41   EXPECT_FALSE(Success);
  42 }
  43
  44 TEST(ConvertUTFTest, Empty) {
  45   std::string Result;
  46   bool Success = convertUTF16ToUTF8String(ArrayRef<char>(), Result);
  47   EXPECT_TRUE(Success);
  48   EXPECT_TRUE(Result.empty());
  49 }
  50
  51 TEST(ConvertUTFTest, HasUTF16BOM) {
  52   bool HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>("\xff\xfe", 2));
  53   EXPECT_TRUE(HasBOM);
  54   HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>("\xfe\xff", 2));
  55   EXPECT_TRUE(HasBOM);
  56   HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>("\xfe\xff ", 3));
  57   EXPECT_TRUE(HasBOM); // Don't care about odd lengths.
  58   HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>("\xfe\xff\x00asdf", 6));
  59   EXPECT_TRUE(HasBOM);
  60
  61   HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>());
  62   EXPECT_FALSE(HasBOM);
  63   HasBOM = hasUTF16ByteOrderMark(ArrayRef<char>("\xfe", 1));
  64   EXPECT_FALSE(HasBOM);
  65 }