bvt/bvt16.cc

   1 // This file is part of the ustl library, an STL implementation.
   2 //
   3 // Copyright (C) 2005 by Mike Sharov <msharov@users.sourceforge.net>
   4 // This file is free software, distributed under the MIT License.
   5 //
   6
   7 #include "stdtest.h"
   8
   9 static void Widen (const string& str, vector<wchar_t>& result)
  10 {
  11     result.clear();
  12     result.resize (str.length());
  13     copy (str.utf8_begin(), str.utf8_end(), result.begin());
  14 }
  15
  16 static void DumpWchars (const vector<wchar_t>& v)
  17 {
  18     foreach (vector<wchar_t>::const_iterator, i, v)
  19         cout.format (" %u", uint32_t(*i));
  20 }
  21
  22 void TestUTF8 (void)
  23 {
  24     cout << "Generating Unicode characters ";
  25     vector<wchar_t> srcChars;
  26     srcChars.resize (0xFFFF);
  27     iota (srcChars.begin(), srcChars.end(), 0);
  28     cout.format ("%zu - %zu\n", size_t(srcChars[0]), size_t(srcChars.back()));
  29
  30     cout << "Encoding to utf8.\n";
  31     string encoded;
  32     encoded.reserve (srcChars.size() * 4);
  33     copy (srcChars, utf8out (back_inserter(encoded)));
  34     static const char c_ProperEncoding[] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 };
  35     if (encoded.compare (encoded.begin(), encoded.begin() + VectorSize(c_ProperEncoding), VectorRange(c_ProperEncoding))) {
  36         cout << "Encoding failed: ";
  37         for (string::const_iterator i = encoded.begin(); i != encoded.begin() + VectorSize(c_ProperEncoding); ++ i)
  38             cout << uint32_t(*i);
  39         cout << endl;
  40     }
  41
  42     cout << "Decoding back.\n";
  43     vector<wchar_t> decChars;
  44     Widen (encoded, decChars);
  45
  46     cout.format ("Comparing.\nsrc = %zu chars, encoded = %zu chars, decoded = %zu\n", srcChars.size(), encoded.size(), decChars.size());
  47     size_t nDiffs = 0;
  48     for (uoff_t i = 0; i < min (srcChars.size(), decChars.size()); ++ i) {
  49         if (srcChars[i] != decChars[i]) {
  50             cout.format ("%u != %u\n", uint32_t(srcChars[i]), uint32_t(decChars[i]));
  51             ++ nDiffs;
  52         }
  53     }
  54     cout.format ("%zu differences between src and decoded.\n", nDiffs);
  55
  56     cout << "Testing wide character string::insert\n";
  57     string ws ("1234567890", 10);
  58
  59     ws.insert (ws.find('1'), wchar_t(1234));
  60     static const wchar_t c_WChars[2] = { 3456, 4567 };
  61     ws.insert (ws.find('3'), VectorRange(c_WChars), 2);
  62     ws.insert (ws.find('3'), wchar_t(2345));
  63     ws.insert (ws.size(), wchar_t(5678));
  64     cout.format ("Values[%zu]:", ws.length());
  65     for (string::utf8_iterator j = ws.utf8_begin(); j < ws.utf8_end(); ++ j)
  66         cout.format (" %u", uint32_t(*j));
  67     cout << endl;
  68
  69     cout << "Character offsets:";
  70     for (string::utf8_iterator k = ws.utf8_begin(); k < ws.utf8_end(); ++ k)
  71         cout.format (" %zu", distance (ws.begin(), k.base()));
  72     cout << endl;
  73
  74     cout.format ("Erasing character %zu: ", ws.length() - 1);
  75     ws.erase (ws.wiat(ws.length() - 1), ws.end());
  76     Widen (ws, decChars);
  77     DumpWchars (decChars);
  78     cout << endl;
  79
  80     cout << "Erasing 2 characters after '2': ";
  81     ws.erase (ws.find('2')+1, Utf8Bytes(VectorRange(c_WChars)));
  82     Widen (ws, decChars);
  83     DumpWchars (decChars);
  84     cout << endl;
  85 }
  86
  87 StdBvtMain (TestUTF8)
  88