tests/mbconv/convautotest.cpp

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        tests/mbconv/convauto.cpp
   3 // Purpose:     wxConvAuto unit test
   4 // Author:      Vadim Zeitlin
   5 // Created:     2006-04-04
   6 // Copyright:   (c) 2006 Vadim Zeitlin
   7 ///////////////////////////////////////////////////////////////////////////////
   8
   9 // ----------------------------------------------------------------------------
  10 // headers
  11 // ----------------------------------------------------------------------------
  12
  13 #include "testprec.h"
  14
  15 #ifdef __BORLANDC__
  16     #pragma hdrstop
  17 #endif
  18
  19 #if wxUSE_UNICODE
  20
  21 #include "wx/convauto.h"
  22
  23 #include "wx/mstream.h"
  24 #include "wx/txtstrm.h"
  25
  26 // ----------------------------------------------------------------------------
  27 // test class
  28 // ----------------------------------------------------------------------------
  29
  30 class ConvAutoTestCase : public CppUnit::TestCase
  31 {
  32 public:
  33     ConvAutoTestCase() { }
  34
  35 private:
  36     CPPUNIT_TEST_SUITE( ConvAutoTestCase );
  37         CPPUNIT_TEST( Empty );
  38         CPPUNIT_TEST( Short );
  39         CPPUNIT_TEST( None );
  40         CPPUNIT_TEST( UTF32LE );
  41         CPPUNIT_TEST( UTF32BE );
  42         CPPUNIT_TEST( UTF16LE );
  43         CPPUNIT_TEST( UTF16BE );
  44         CPPUNIT_TEST( UTF8 );
  45         CPPUNIT_TEST( StreamUTF8NoBOM );
  46         CPPUNIT_TEST( StreamUTF8 );
  47         CPPUNIT_TEST( StreamUTF16LE );
  48         CPPUNIT_TEST( StreamUTF16BE );
  49         CPPUNIT_TEST( StreamUTF32LE );
  50         CPPUNIT_TEST( StreamUTF32BE );
  51     CPPUNIT_TEST_SUITE_END();
  52
  53     // real test function: check that converting the src multibyte string to
  54     // wide char using wxConvAuto yields wch as the first result
  55     //
  56     // the length of the string may need to be passed explicitly if it has
  57     // embedded NULs, otherwise it's not necessary
  58     void TestFirstChar(const char *src, wchar_t wch, size_t len = wxNO_LEN);
  59
  60     void Empty();
  61     void Short();
  62     void None();
  63     void UTF32LE();
  64     void UTF32BE();
  65     void UTF16LE();
  66     void UTF16BE();
  67     void UTF8();
  68
  69     // test whether two lines of text are converted properly from a stream
  70     void TestTextStream(const char *src,
  71                         size_t srclength,
  72                         const wxString& line1,
  73                         const wxString& line2);
  74
  75     void StreamUTF8NoBOM();
  76     void StreamUTF8();
  77     void StreamUTF16LE();
  78     void StreamUTF16BE();
  79     void StreamUTF32LE();
  80     void StreamUTF32BE();
  81 };
  82
  83 // register in the unnamed registry so that these tests are run by default
  84 CPPUNIT_TEST_SUITE_REGISTRATION(ConvAutoTestCase);
  85
  86 // also include in its own registry so that these tests can be run alone
  87 CPPUNIT_TEST_SUITE_NAMED_REGISTRATION(ConvAutoTestCase, "ConvAutoTestCase");
  88
  89 // ----------------------------------------------------------------------------
  90 // tests
  91 // ----------------------------------------------------------------------------
  92
  93 void ConvAutoTestCase::TestFirstChar(const char *src, wchar_t wch, size_t len)
  94 {
  95     wxWCharBuffer wbuf = wxConvAuto().cMB2WC(src, len, NULL);
  96     CPPUNIT_ASSERT( wbuf );
  97     CPPUNIT_ASSERT_EQUAL( wch, *wbuf );
  98 }
  99
 100 void ConvAutoTestCase::Empty()
 101 {
 102     CPPUNIT_ASSERT( !wxConvAuto().cMB2WC("") );
 103 }
 104
 105 void ConvAutoTestCase::Short()
 106 {
 107     TestFirstChar("1", wxT('1'));
 108 }
 109
 110 void ConvAutoTestCase::None()
 111 {
 112     TestFirstChar("Hello world", wxT('H'));
 113 }
 114
 115 void ConvAutoTestCase::UTF32LE()
 116 {
 117     TestFirstChar("\xff\xfe\0\0A\0\0\0", wxT('A'), 8);
 118 }
 119
 120 void ConvAutoTestCase::UTF32BE()
 121 {
 122     TestFirstChar("\0\0\xfe\xff\0\0\0B", wxT('B'), 8);
 123 }
 124
 125 void ConvAutoTestCase::UTF16LE()
 126 {
 127     TestFirstChar("\xff\xfeZ\0", wxT('Z'), 4);
 128 }
 129
 130 void ConvAutoTestCase::UTF16BE()
 131 {
 132     TestFirstChar("\xfe\xff\0Y", wxT('Y'), 4);
 133 }
 134
 135 void ConvAutoTestCase::UTF8()
 136 {
 137 #ifdef wxHAVE_U_ESCAPE
 138     TestFirstChar("\xef\xbb\xbf\xd0\x9f", L'\u041f');
 139 #endif
 140 }
 141
 142 void ConvAutoTestCase::TestTextStream(const char *src,
 143                                       size_t srclength,
 144                                       const wxString& line1,
 145                                       const wxString& line2)
 146 {
 147     wxMemoryInputStream instream(src, srclength);
 148     wxTextInputStream text(instream);
 149
 150     CPPUNIT_ASSERT_EQUAL( line1, text.ReadLine() );
 151     CPPUNIT_ASSERT_EQUAL( line2, text.ReadLine() );
 152 }
 153
 154 // the first line of the teststring used in the following functions is an
 155 // 'a' followed by a Japanese hiragana A (u+3042).
 156 // The second line is a single Greek beta (u+03B2). There is no blank line
 157 // at the end.
 158
 159 namespace
 160 {
 161
 162 const wxString line1 = wxString::FromUTF8("a\xe3\x81\x82");
 163 const wxString line2 = wxString::FromUTF8("\xce\xb2");
 164
 165 } // anonymous namespace
 166
 167 void ConvAutoTestCase::StreamUTF8NoBOM()
 168 {
 169     // currently this test doesn't work because without the BOM wxConvAuto
 170     // decides that the string is in Latin-1 after finding the first (but not
 171     // the two subsequent ones which are part of the same UTF-8 sequence!)
 172     // 8-bit character
 173     //
 174     // FIXME: we need to fix this at wxTextInputStream level, see #11570
 175 #if 0
 176     TestTextStream("\x61\xE3\x81\x82\x0A\xCE\xB2",
 177                    7, line1, line2);
 178 #endif
 179 }
 180
 181 void ConvAutoTestCase::StreamUTF8()
 182 {
 183     TestTextStream("\xEF\xBB\xBF\x61\xE3\x81\x82\x0A\xCE\xB2",
 184                    10, line1, line2);
 185 }
 186
 187 void ConvAutoTestCase::StreamUTF16LE()
 188 {
 189     TestTextStream("\xFF\xFE\x61\x00\x42\x30\x0A\x00\xB2\x03",
 190                    10, line1, line2);
 191 }
 192
 193 void ConvAutoTestCase::StreamUTF16BE()
 194 {
 195     TestTextStream("\xFE\xFF\x00\x61\x30\x42\x00\x0A\x03\xB2",
 196                    10, line1, line2);
 197 }
 198
 199 void ConvAutoTestCase::StreamUTF32LE()
 200 {
 201     TestTextStream("\xFF\xFE\0\0\x61\x00\0\0\x42\x30\0\0\x0A"
 202                    "\x00\0\0\xB2\x03\0\0",
 203                    20, line1, line2);
 204 }
 205
 206 void ConvAutoTestCase::StreamUTF32BE()
 207 {
 208     TestTextStream("\0\0\xFE\xFF\0\0\x00\x61\0\0\x30\x42\0\0\x00\x0A"
 209                    "\0\0\x03\xB2",
 210                    20, line1, line2);
 211 }
 212
 213 #endif // wxUSE_UNICODE