tests/regex/regextest.cpp

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        tests/regex/regex.cpp
   3 // Purpose:     Test the built-in regex lib and wxRegEx
   4 // Author:      Mike Wetherell
   5 // RCS-ID:      $Id$
   6 // Copyright:   (c) 2004 Mike Wetherell
   7 // Licence:     wxWidgets licence
   8 ///////////////////////////////////////////////////////////////////////////////
   9
  10 //
  11 // Notes:
  12 //
  13 // To run just one section, say wx_1, do this:
  14 //  test regex.wx_1
  15 //
  16 // To run all the regex tests:
  17 //  test regex
  18 //
  19 // Some tests must be skipped since they use features which we do not make
  20 // available through wxRegEx. To see the list of tests that have been skipped
  21 // turn on verbose logging, e.g.:
  22 //  test --verbose regex
  23 //
  24 // The tests here are for the builtin library, tests for wxRegEx in general
  25 // should go in wxregex.cpp
  26 //
  27 // The tests are generated from Henry Spencer's reg.test, additional test
  28 // can be added in wxreg.test. These test files are then turned into a C++
  29 // include file 'regex.inc' (included below) using a script 'regex.pl'.
  30 //
  31
  32 // For compilers that support precompilation, includes "wx/wx.h".
  33 #include "wx/wxprec.h"
  34
  35 #ifdef __BORLANDC__
  36     #pragma hdrstop
  37 #endif
  38
  39 // for all others, include the necessary headers
  40 #ifndef WX_PRECOMP
  41     #include "wx/wx.h"
  42 #endif
  43
  44
  45 // many of the tests are specific to the builtin regex lib, so only attempts
  46 // to do them when using the builtin regex lib.
  47 //
  48 #ifdef wxHAS_REGEX_ADVANCED
  49
  50 #include "wx/regex.h"
  51 #include "wx/cppunit.h"
  52 #include <string>
  53 #include <vector>
  54
  55 using CppUnit::Test;
  56 using CppUnit::TestCase;
  57 using CppUnit::TestSuite;
  58 using CppUnit::Exception;
  59
  60 using namespace std;
  61
  62 ///////////////////////////////////////////////////////////////////////////////
  63 // The test case - an instance represents a single test
  64
  65 class RegExTestCase : public TestCase
  66 {
  67 public:
  68     // constructor - create a single testcase
  69     RegExTestCase(
  70         const string& name,
  71         const char *mode,
  72         const char *id,
  73         const char *flags,
  74         const char *pattern,
  75         const char *data,
  76         const vector<const char *>& expected);
  77
  78 protected:
  79     // run this testcase
  80     void runTest();
  81
  82 private:
  83     // workers
  84     wxString Conv(const char *str);
  85     void parseFlags(const wxString& flags);
  86     void doTest(int flavor);
  87     static wxString quote(const wxString& arg);
  88     const wxChar *convError() const { return _T("<cannot convert>"); }
  89
  90     // assertions - adds some information about the test that failed
  91     void fail(const wxString& msg) const;
  92     void failIf(bool condition, const wxString& msg) const
  93         { if (condition) fail(msg); }
  94
  95     // mode, id, flags, pattern, test data, expected results...
  96     int m_mode;
  97     wxString m_id;
  98     wxString m_flags;
  99     wxString m_pattern;
 100     wxString m_data;
 101     wxArrayString m_expected;
 102
 103     // the flag decoded
 104     int m_compileFlags;
 105     int m_matchFlags;
 106     bool m_basic;
 107     bool m_extended;
 108     bool m_advanced;
 109 };
 110
 111 // constructor - throws Exception on failure
 112 //
 113 RegExTestCase::RegExTestCase(
 114     const string& name,
 115     const char *mode,
 116     const char *id,
 117     const char *flags,
 118     const char *pattern,
 119     const char *data,
 120     const vector<const char *>& expected)
 121   :
 122     TestCase(name),
 123     m_mode(mode[0]),
 124     m_id(Conv(id)),
 125     m_flags(Conv(flags)),
 126     m_pattern(Conv(pattern)),
 127     m_data(Conv(data)),
 128     m_compileFlags(0),
 129     m_matchFlags(0),
 130     m_basic(false),
 131     m_extended(false),
 132     m_advanced(false)
 133 {
 134     bool badconv = m_pattern == convError() || m_data == convError();
 135     vector<const char *>::const_iterator it;
 136
 137     for (it = expected.begin(); it != expected.end(); ++it) {
 138         m_expected.push_back(Conv(*it));
 139         badconv = badconv || *m_expected.rbegin() == convError();
 140     }
 141
 142     failIf(badconv, _T("cannot convert to default character encoding"));
 143
 144     // the flags need further parsing...
 145     parseFlags(m_flags);
 146
 147 #ifndef wxHAS_REGEX_ADVANCED
 148     failIf(!m_basic && !m_extended, _T("advanced regexs not available"));
 149 #endif
 150 }
 151
 152 int wxWcscmp(const wchar_t* s1, const wchar_t* s2)
 153 {
 154     size_t nLen1 = wxWcslen(s1);
 155     size_t nLen2 = wxWcslen(s2);
 156
 157     if (nLen1 != nLen2)
 158         return nLen1 - nLen2;
 159
 160     return wxMemcmp(s1, s2, nLen1);
 161 }
 162
 163 // convert a string from UTF8 to the internal encoding
 164 //
 165 wxString RegExTestCase::Conv(const char *str)
 166 {
 167     const wxWCharBuffer wstr = wxConvUTF8.cMB2WC(str);
 168     const wxWC2WXbuf buf = wxConvCurrent->cWC2WX(wstr);
 169
 170     if (!buf || wxWcscmp(wxConvCurrent->cWX2WC(buf), wstr) != 0)
 171         return convError();
 172     else
 173         return buf;
 174 }
 175
 176 // Parse flags
 177 //
 178 void RegExTestCase::parseFlags(const wxString& flags)
 179 {
 180     for (const wxChar *p = flags; *p; p++) {
 181         switch (*p) {
 182             // noop
 183             case '-': break;
 184
 185             // we don't fully support these flags, but they don't stop us
 186             // checking for success of failure of the match, so treat as noop
 187             case 'A': case 'B': case 'E': case 'H':
 188             case 'I': case 'L': case 'M': case 'N':
 189             case 'P': case 'Q': case 'R': case 'S':
 190             case 'T': case 'U': case '%':
 191                 break;
 192
 193             // match options
 194             case '^': m_matchFlags |= wxRE_NOTBOL; break;
 195             case '$': m_matchFlags |= wxRE_NOTEOL; break;
 196 #if wxUSE_UNICODE
 197             case '*': break;
 198 #endif
 199             // compile options
 200             case '&': m_advanced = m_basic = true; break;
 201             case 'b': m_basic = true; break;
 202             case 'e': m_extended = true; break;
 203             case 'i': m_compileFlags |= wxRE_ICASE; break;
 204             case 'o': m_compileFlags |= wxRE_NOSUB; break;
 205             case 'n': m_compileFlags |= wxRE_NEWLINE; break;
 206             case 't': if (strchr("ep", m_mode)) break; // else fall through...
 207
 208             // anything else we must skip the test
 209             default:
 210                 fail(wxString::Format(
 211                      _T("requires unsupported flag '%c'"), *p));
 212         }
 213     }
 214 }
 215
 216 // Try test for all flavours of expression specified
 217 //
 218 void RegExTestCase::runTest()
 219 {
 220     if (m_basic)
 221         doTest(wxRE_BASIC);
 222     if (m_extended)
 223         doTest(wxRE_EXTENDED);
 224 #ifdef wxHAS_REGEX_ADVANCED
 225     if (m_advanced || (!m_basic && !m_extended))
 226         doTest(wxRE_ADVANCED);
 227 #endif
 228 }
 229
 230 // Try the test for a single flavour of expression
 231 //
 232 void RegExTestCase::doTest(int flavor)
 233 {
 234     wxRegEx re(m_pattern, m_compileFlags | flavor);
 235
 236     // 'e' - test that the pattern fails to compile
 237     if (m_mode == 'e') {
 238         failIf(re.IsValid(), _T("compile succeeded (should fail)"));
 239         return;
 240     }
 241     failIf(!re.IsValid(), _T("compile failed"));
 242
 243     bool matches = re.Matches(m_data, m_matchFlags);
 244
 245     // 'f' or 'p' - test that the pattern does not match
 246     if (m_mode == 'f' || m_mode == 'p') {
 247         failIf(matches, _T("match succeeded (should fail)"));
 248         return;
 249     }
 250
 251     // otherwise 'm' or 'i' - test the pattern does match
 252     failIf(!matches, _T("match failed"));
 253
 254     if (m_compileFlags & wxRE_NOSUB)
 255         return;
 256
 257     // check wxRegEx has correctly counted the number of subexpressions
 258     failIf(m_expected.size() != re.GetMatchCount(),
 259            wxString::Format(_T("GetMatchCount() == %d, expected %d"),
 260                             re.GetMatchCount(), m_expected.size()));
 261
 262     wxString result;
 263     size_t start, len;
 264
 265     for (size_t i = 0; i < m_expected.size(); i++) {
 266         failIf(!re.GetMatch(&start, &len, i), wxString::Format(
 267                 _T("wxRegEx::GetMatch failed for match %d"), i));
 268
 269         // m - check the match returns the strings given
 270         if (m_mode == 'm')
 271             if (start < INT_MAX)
 272                 result = m_data.substr(start, len);
 273             else
 274                 result = _T("");
 275
 276         // i - check the match returns the offsets given
 277         else if (m_mode == 'i')
 278             if (start < INT_MAX)
 279                 result = wxString::Format(_T("%d %d"), start, start + len - 1);
 280             else
 281                 result = _T("-1 -1");
 282
 283         failIf(result != m_expected[i], wxString::Format(
 284                 _T("match(%d) == %s, expected == %s"), i,
 285                 quote(result).c_str(), quote(m_expected[i]).c_str()));
 286     }
 287 }
 288
 289 // assertion - adds some information about the test that failed
 290 //
 291 void RegExTestCase::fail(const wxString& msg) const
 292 {
 293     wxString str;
 294     wxArrayString::const_iterator it;
 295
 296     str << (wxChar)m_mode << _T(" ") << m_id << _T(" ") << m_flags << _T(" ")
 297         << quote(m_pattern) << _T(" ") << quote(m_data);
 298
 299     for (it = m_expected.begin(); it != m_expected.end(); ++it)
 300         str << _T(" ") << quote(*it);
 301
 302     if (str.length() > 77)
 303         str = str.substr(0, 74) + _T("...");
 304
 305     str << _T("\n ") << msg;
 306
 307     // no lossy convs so using utf8
 308     CPPUNIT_FAIL(string(str.mb_str(wxConvUTF8)));
 309 }
 310
 311 // quote a string so that it can be displayed (static)
 312 //
 313 wxString RegExTestCase::quote(const wxString& arg)
 314 {
 315     const wxChar *needEscape = _T("\a\b\t\n\v\f\r\"\\");
 316     const wxChar *escapes = _T("abtnvfr\"\\");
 317     wxString str;
 318
 319     for (size_t i = 0; i < arg.length(); i++) {
 320         wxUChar ch = arg[i];
 321         const wxChar *p = wxStrchr(needEscape, ch);
 322
 323         if (p)
 324             str += wxString::Format(_T("\\%c"), escapes[p - needEscape]);
 325         else if (wxIscntrl(ch))
 326             str += wxString::Format(_T("\\%03o"), ch);
 327         else
 328             str += ch;
 329     }
 330
 331     return str.length() == arg.length() && str.find(' ') == wxString::npos ?
 332         str : _T("\"") + str + _T("\"");
 333 }
 334
 335
 336 ///////////////////////////////////////////////////////////////////////////////
 337 // Test suite
 338
 339 class RegExTestSuite : public TestSuite
 340 {
 341 public:
 342     RegExTestSuite(string name) : TestSuite(name) { }
 343     void add(const char *mode, const char *id, const char *flags,
 344              const char *pattern, const char *data, const char *expected, ...);
 345 };
 346
 347 // Add a testcase to the suite
 348 //
 349 void RegExTestSuite::add(
 350     const char *mode,
 351     const char *id,
 352     const char *flags,
 353     const char *pattern,
 354     const char *data,
 355     const char *expected, ...)
 356 {
 357     string name = getName() + "." + id;
 358
 359     vector<const char *> expected_results;
 360     va_list ap;
 361
 362     for (va_start(ap, expected); expected; expected = va_arg(ap, const char *))
 363         expected_results.push_back(expected);
 364
 365     va_end(ap);
 366
 367     try {
 368         addTest(new RegExTestCase(
 369             name, mode, id, flags, pattern, data, expected_results));
 370     }
 371     catch (Exception& e) {
 372         wxLogInfo(wxString::Format(_T("skipping: %s\n %s\n"),
 373             wxString(name.c_str(), wxConvUTF8).c_str(),
 374             wxString(e.what(), wxConvUTF8).c_str()));
 375     }
 376 }
 377
 378
 379 // Include the generated tests
 380 //
 381 #include "regex.inc"
 382
 383
 384 #endif // wxHAS_REGEX_ADVANCED