tests/regex/regex.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 #include "wx/regex.h"
  45 #include "wx/cppunit.h"
  46 #include <stdexcept>
  47
  48 using namespace std;
  49 using namespace CppUnit;
  50
  51 // many of the tests are specific to the builtin regex lib, so only attempts
  52 // to do them when using the builtin regex lib.
  53 //
  54 #ifdef wxHAS_REGEX_ADVANCED
  55
  56
  57 ///////////////////////////////////////////////////////////////////////////////
  58 // The test case - an instance represents a single test
  59
  60 class RegExTestCase : public TestCase
  61 {
  62 public:
  63     // constructor - create a single testcase
  64     RegExTestCase(
  65         const string& name,
  66         const char *mode,
  67         const char *id,
  68         const char *flags,
  69         const char *pattern,
  70         const char *data,
  71         const vector<const char *>& expected);
  72
  73 protected:
  74     // run this testcase
  75     void runTest();
  76
  77 private:
  78     // workers
  79     wxString Conv(const char *str);
  80     void parseFlags(const wxString& flags);
  81     void doTest(int flavor);
  82     static wxString quote(const wxString& arg);
  83     const wxChar *convError() const { return _T("<cannot convert>"); }
  84
  85     // assertions - adds some information about the test that failed
  86     void fail(const wxString& msg) const;
  87     void failIf(bool condition, const wxString& msg) const
  88         { if (condition) fail(msg); }
  89
  90     // mode, id, flags, pattern, test data, expected results...
  91     int m_mode;
  92     wxString m_id;
  93     wxString m_flags;
  94     wxString m_pattern;
  95     wxString m_data;
  96     wxArrayString m_expected;
  97
  98     // the flag decoded
  99     int m_compileFlags;
 100     int m_matchFlags;
 101     bool m_basic;
 102     bool m_extended;
 103     bool m_advanced;
 104 };
 105
 106 // constructor - throws Exception on failure
 107 //
 108 RegExTestCase::RegExTestCase(
 109     const string& name,
 110     const char *mode,
 111     const char *id,
 112     const char *flags,
 113     const char *pattern,
 114     const char *data,
 115     const vector<const char *>& expected)
 116   :
 117     TestCase(name),
 118     m_mode(mode[0]),
 119     m_id(Conv(id)),
 120     m_flags(Conv(flags)),
 121     m_pattern(Conv(pattern)),
 122     m_data(Conv(data)),
 123     m_compileFlags(0),
 124     m_matchFlags(0),
 125     m_basic(false),
 126     m_extended(false),
 127     m_advanced(false)
 128 {
 129     bool badconv = m_pattern == convError() || m_data == convError();
 130     vector<const char *>::const_iterator it;
 131
 132     for (it = expected.begin(); it != expected.end(); ++it) {
 133         m_expected.push_back(Conv(*it));
 134         badconv = badconv || *m_expected.rbegin() == convError();
 135     }
 136
 137     failIf(badconv, _T("cannot convert to default character encoding"));
 138
 139     // the flags need further parsing...
 140     parseFlags(m_flags);
 141
 142 #ifndef wxHAS_REGEX_ADVANCED
 143     failIf(!m_basic && !m_extended, _T("advanced regexs not available"));
 144 #endif
 145 }
 146
 147 // convert a string from UTF8 to the internal encoding
 148 //
 149 wxString RegExTestCase::Conv(const char *str)
 150 {
 151     const wxWCharBuffer wstr = wxConvUTF8.cMB2WC(str);
 152     const wxWC2WXbuf buf = wxConvCurrent->cWC2WX(wstr);
 153
 154     if (!buf || wcscmp(wxConvCurrent->cWX2WC(buf), wstr) != 0)
 155         return convError();
 156     else
 157         return buf;
 158 }
 159
 160 // Parse flags
 161 //
 162 void RegExTestCase::parseFlags(const wxString& flags)
 163 {
 164     for (const wxChar *p = flags; *p; p++) {
 165         switch (*p) {
 166             // noop
 167             case '-': break;
 168
 169             // we don't fully support these flags, but they don't stop us
 170             // checking for success of failure of the match, so treat as noop
 171             case 'A': case 'B': case 'E': case 'H':
 172             case 'I': case 'L': case 'M': case 'N':
 173             case 'P': case 'Q': case 'R': case 'S':
 174             case 'T': case 'U': case '%':
 175                 break;
 176
 177             // match options
 178             case '^': m_matchFlags |= wxRE_NOTBOL; break;
 179             case '$': m_matchFlags |= wxRE_NOTEOL; break;
 180 #if wxUSE_UNICODE
 181             case '*': break;
 182 #endif
 183             // compile options
 184             case '&': m_advanced = m_basic = true; break;
 185             case 'b': m_basic = true; break;
 186             case 'e': m_extended = true; break;
 187             case 'i': m_compileFlags |= wxRE_ICASE; break;
 188             case 'o': m_compileFlags |= wxRE_NOSUB; break;
 189             case 'n': m_compileFlags |= wxRE_NEWLINE; break;
 190             case 't': if (strchr("ep", m_mode)) break; // else fall through...
 191
 192             // anything else we must skip the test
 193             default:
 194                 fail(wxString::Format(
 195                      _T("requires unsupported flag '%c'"), *p));
 196         }
 197     }
 198 }
 199
 200 // Try test for all flavours of expression specified
 201 //
 202 void RegExTestCase::runTest()
 203 {
 204     if (m_basic)
 205         doTest(wxRE_BASIC);
 206     if (m_extended)
 207         doTest(wxRE_EXTENDED);
 208 #ifdef wxHAS_REGEX_ADVANCED
 209     if (m_advanced || (!m_basic && !m_extended))
 210         doTest(wxRE_ADVANCED);
 211 #endif
 212 }
 213
 214 // Try the test for a single flavour of expression
 215 //
 216 void RegExTestCase::doTest(int flavor)
 217 {
 218     wxRegEx re(m_pattern, m_compileFlags | flavor);
 219
 220     // 'e' - test that the pattern fails to compile
 221     if (m_mode == 'e') {
 222         failIf(re.IsValid(), _T("compile succeeded (should fail)"));
 223         return;
 224     }
 225     failIf(!re.IsValid(), _T("compile failed"));
 226
 227     bool matches = re.Matches(m_data, m_matchFlags);
 228
 229     // 'f' or 'p' - test that the pattern does not match
 230     if (m_mode == 'f' || m_mode == 'p') {
 231         failIf(matches, _T("match succeeded (should fail)"));
 232         return;
 233     }
 234
 235     // otherwise 'm' or 'i' - test the pattern does match
 236     failIf(!matches, _T("match failed"));
 237
 238     if (m_compileFlags & wxRE_NOSUB)
 239         return;
 240
 241     // check wxRegEx has correctly counted the number of subexpressions
 242     failIf(m_expected.size() != re.GetMatchCount(),
 243            wxString::Format(_T("GetMatchCount() == %d, expected %d"),
 244                             re.GetMatchCount(), m_expected.size()));
 245
 246     wxString result;
 247     size_t start, len;
 248
 249     for (size_t i = 0; i < m_expected.size(); i++) {
 250         failIf(!re.GetMatch(&start, &len, i), wxString::Format(
 251                 _T("wxRegEx::GetMatch failed for match %d"), i));
 252
 253         // m - check the match returns the strings given
 254         if (m_mode == 'm')
 255             if (start < INT_MAX)
 256                 result = m_data.substr(start, len);
 257             else
 258                 result = _T("");
 259
 260         // i - check the match returns the offsets given
 261         else if (m_mode == 'i')
 262             if (start < INT_MAX)
 263                 result = wxString::Format(_T("%d %d"), start, start + len - 1);
 264             else
 265                 result = _T("-1 -1");
 266
 267         failIf(result != m_expected[i], wxString::Format(
 268                 _T("match(%d) == %s, expected == %s"), i,
 269                 quote(result).c_str(), quote(m_expected[i]).c_str()));
 270     }
 271 }
 272
 273 // assertion - adds some information about the test that failed
 274 //
 275 void RegExTestCase::fail(const wxString& msg) const
 276 {
 277     wxString str;
 278     wxArrayString::const_iterator it;
 279
 280     str << (wxChar)m_mode << _T(" ") << m_id << _T(" ") << m_flags << _T(" ")
 281         << quote(m_pattern) << _T(" ") << quote(m_data);
 282
 283     for (it = m_expected.begin(); it != m_expected.end(); ++it)
 284         str << _T(" ") << quote(*it);
 285
 286     if (str.length() > 77)
 287         str = str.substr(0, 74) + _T("...");
 288
 289     str << _T("\n ") << msg;
 290
 291     // no lossy convs so using utf8
 292     CPPUNIT_FAIL(string(str.mb_str(wxConvUTF8)));
 293 }
 294
 295 // quote a string so that it can be displayed (static)
 296 //
 297 wxString RegExTestCase::quote(const wxString& arg)
 298 {
 299     const wxChar *needEscape = _T("\a\b\t\n\v\f\r\"\\");
 300     const wxChar *escapes = _T("abtnvfr\"\\");
 301     wxString str;
 302
 303     for (size_t i = 0; i < arg.length(); i++) {
 304         wxUChar ch = arg[i];
 305         const wxChar *p = wxStrchr(needEscape, ch);
 306
 307         if (p)
 308             str += wxString::Format(_T("\\%c"), escapes[p - needEscape]);
 309         else if (wxIscntrl(ch))
 310             str += wxString::Format(_T("\\%03o"), ch);
 311         else
 312             str += ch;
 313     }
 314
 315     return str.length() == arg.length() && str.find(' ') == wxString::npos ?
 316         str : _T("\"") + str + _T("\"");
 317 }
 318
 319
 320 ///////////////////////////////////////////////////////////////////////////////
 321 // Test suite
 322 //
 323 // In a non-unicode build the regex is affected by the current locale, so
 324 // this derived TestSuite is used. It sets the locale in it's run() method
 325 // for the duration of the regex tests.
 326
 327 class RegExTestSuite : public TestSuite
 328 {
 329 public:
 330     RegExTestSuite(string name);
 331     void run(TestResult *result);
 332     void add(const char *mode, const char *id, const char *flags,
 333              const char *pattern, const char *data, const char *expected, ...);
 334 };
 335
 336 // constructor, sets the locale so that it is set when the tests are added
 337 //
 338 RegExTestSuite::RegExTestSuite(string name) : TestSuite(name)
 339 {
 340     setlocale(LC_ALL, "");
 341 }
 342
 343 // run the test suite, sets the locale again since it may have been changed
 344 // by another test since this suite was crated
 345 //
 346 void RegExTestSuite::run(TestResult *result)
 347 {
 348     setlocale(LC_ALL, "");
 349     TestSuite::run(result);
 350 }
 351
 352 // Add a testcase to the suite
 353 //
 354 void RegExTestSuite::add(
 355     const char *mode,
 356     const char *id,
 357     const char *flags,
 358     const char *pattern,
 359     const char *data,
 360     const char *expected, ...)
 361 {
 362     string name = getName() + "." + id;
 363
 364     vector<const char *> expected_results;
 365     va_list ap;
 366
 367     for (va_start(ap, expected); expected; expected = va_arg(ap, const char *))
 368         expected_results.push_back(expected);
 369
 370     va_end(ap);
 371
 372     try {
 373         addTest(new RegExTestCase(
 374             name, mode, id, flags, pattern, data, expected_results));
 375     }
 376     catch (Exception& e) {
 377         wxLogInfo(wxString::Format(_T("skipping: %s\n %s\n"),
 378             wxString(name.c_str(), wxConvUTF8).c_str(),
 379             wxString(e.what(), wxConvUTF8).c_str()));
 380     }
 381 }
 382
 383
 384 // Include the generated tests
 385 //
 386 #include "regex.inc"
 387
 388
 389 #endif // wxHAS_REGEX_ADVANCED