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