src/common/regex.cpp

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        src/common/regex.cpp
   3 // Purpose:     regular expression matching
   4 // Author:      Karsten Ballüder and Vadim Zeitlin
   5 // Modified by:
   6 // Created:     13.07.01
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 2000 Karsten Ballüder <ballueder@gmx.net>
   9 //                  2001 Vadim Zeitlin <vadim@wxwindows.org>
  10 // Licence:     wxWindows licence
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 // ============================================================================
  14 // declarations
  15 // ============================================================================
  16
  17 // ----------------------------------------------------------------------------
  18 // headers
  19 // ----------------------------------------------------------------------------
  20
  21 #ifdef __GNUG__
  22     #pragma implementation "regex.h"
  23 #endif
  24
  25 // For compilers that support precompilation, includes "wx.h".
  26 #include "wx/wxprec.h"
  27
  28 #ifdef __BORLANDC__
  29     #pragma hdrstop
  30 #endif
  31
  32 #if wxUSE_REGEX
  33
  34 #ifndef WX_PRECOMP
  35     #include "wx/object.h"
  36     #include "wx/string.h"
  37     #include "wx/log.h"
  38     #include "wx/intl.h"
  39 #endif //WX_PRECOMP
  40
  41 // FreeBSD & Watcom require this, it probably doesn't hurt for others
  42 #if defined(__UNIX__) || defined(__WATCOMC__) || defined(__DIGITALMARS__)
  43     #include <sys/types.h>
  44 #endif
  45
  46 #include <regex.h>
  47
  48 #include "wx/regex.h"
  49
  50 // ----------------------------------------------------------------------------
  51 // private classes
  52 // ----------------------------------------------------------------------------
  53
  54 // the real implementation of wxRegEx
  55 class wxRegExImpl
  56 {
  57 public:
  58     // ctor and dtor
  59     wxRegExImpl();
  60     ~wxRegExImpl();
  61
  62     // return TRUE if Compile() had been called successfully
  63     bool IsValid() const { return m_isCompiled; }
  64
  65     // RE operations
  66     bool Compile(const wxString& expr, int flags = 0);
  67     bool Matches(const wxChar *str, int flags = 0) const;
  68     bool GetMatch(size_t *start, size_t *len, size_t index = 0) const;
  69     int Replace(wxString *pattern, const wxString& replacement,
  70                 size_t maxMatches = 0) const;
  71
  72 private:
  73     // return the string containing the error message for the given err code
  74     wxString GetErrorMsg(int errorcode) const;
  75
  76     // init the members
  77     void Init()
  78     {
  79         m_isCompiled = FALSE;
  80         m_Matches = NULL;
  81         m_nMatches = 0;
  82     }
  83
  84     // free the RE if compiled
  85     void Free()
  86     {
  87         if ( IsValid() )
  88         {
  89             regfree(&m_RegEx);
  90         }
  91
  92         delete [] m_Matches;
  93     }
  94
  95     // free the RE if any and reinit the members
  96     void Reinit()
  97     {
  98         Free();
  99         Init();
 100     }
 101
 102
 103     // compiled RE
 104     regex_t     m_RegEx;
 105
 106     // the subexpressions data
 107     regmatch_t *m_Matches;
 108     size_t      m_nMatches;
 109
 110     // TRUE if m_RegEx is valid
 111     bool        m_isCompiled;
 112 };
 113
 114 // ============================================================================
 115 // implementation
 116 // ============================================================================
 117
 118 // ----------------------------------------------------------------------------
 119 // wxRegExImpl
 120 // ----------------------------------------------------------------------------
 121
 122 wxRegExImpl::wxRegExImpl()
 123 {
 124     Init();
 125 }
 126
 127 wxRegExImpl::~wxRegExImpl()
 128 {
 129     Free();
 130 }
 131
 132 wxString wxRegExImpl::GetErrorMsg(int errorcode) const
 133 {
 134     wxString msg;
 135
 136     // first get the string length needed
 137     int len = regerror(errorcode, &m_RegEx, NULL, 0);
 138     if ( len > 0 )
 139     {
 140         len++;
 141
 142 #if wxUSE_UNICODE
 143         wxCharBuffer buf(len);
 144
 145         (void)regerror(errorcode, &m_RegEx, (char *)buf.data(), len);
 146
 147         msg = wxString(buf.data(), wxConvLibc);
 148 #else // !Unicode
 149         (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
 150
 151         msg.UngetWriteBuf();
 152 #endif // Unicode/!Unicode
 153     }
 154     else // regerror() returned 0
 155     {
 156         msg = _("unknown error");
 157     }
 158
 159     return msg;
 160 }
 161
 162 bool wxRegExImpl::Compile(const wxString& expr, int flags)
 163 {
 164     Reinit();
 165
 166     // translate our flags to regcomp() ones
 167     wxASSERT_MSG( !(flags &
 168                         ~(wxRE_BASIC | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)),
 169                   _T("unrecognized flags in wxRegEx::Compile") );
 170
 171     int flagsRE = 0;
 172     if ( !(flags & wxRE_BASIC) )
 173         flagsRE |= REG_EXTENDED;
 174     if ( flags & wxRE_ICASE )
 175         flagsRE |= REG_ICASE;
 176     if ( flags & wxRE_NOSUB )
 177         flagsRE |= REG_NOSUB;
 178     if ( flags & wxRE_NEWLINE )
 179         flagsRE |= REG_NEWLINE;
 180
 181     // compile it
 182     int errorcode = regcomp(&m_RegEx, expr.mb_str(), flagsRE);
 183     if ( errorcode )
 184     {
 185         wxLogError(_("Invalid regular expression '%s': %s"),
 186                    expr.c_str(), GetErrorMsg(errorcode).c_str());
 187
 188         m_isCompiled = FALSE;
 189     }
 190     else // ok
 191     {
 192         // don't allocate the matches array now, but do it later if necessary
 193         if ( flags & wxRE_NOSUB )
 194         {
 195             // we don't need it at all
 196             m_nMatches = 0;
 197         }
 198         else
 199         {
 200             // we will alloc the array later (only if really needed) but count
 201             // the number of sub-expressions in the regex right now
 202
 203             // there is always one for the whole expression
 204             m_nMatches = 1;
 205
 206             // and some more for bracketed subexperessions
 207             for ( const wxChar *cptr = expr.c_str(); *cptr; cptr++ )
 208             {
 209                 if ( *cptr == _T('\\') )
 210                 {
 211                     // in basic RE syntax groups are inside \(...\)
 212                     if ( *++cptr == _T('(') && (flags & wxRE_BASIC) )
 213                     {
 214                         m_nMatches++;
 215                     }
 216                 }
 217                 else if ( *cptr == _T('(') && !(flags & wxRE_BASIC) )
 218                 {
 219                     // we know that the previous character is not an unquoted
 220                     // backslash because it would have been eaten above, so we
 221                     // have a bar '(' and this indicates a group start for the
 222                     // extended syntax
 223                     m_nMatches++;
 224                 }
 225             }
 226         }
 227
 228         m_isCompiled = TRUE;
 229     }
 230
 231     return IsValid();
 232 }
 233
 234 bool wxRegExImpl::Matches(const wxChar *str, int flags) const
 235 {
 236     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 237
 238     // translate our flags to regexec() ones
 239     wxASSERT_MSG( !(flags & ~(wxRE_NOTBOL | wxRE_NOTEOL)),
 240                   _T("unrecognized flags in wxRegEx::Matches") );
 241
 242     int flagsRE = 0;
 243     if ( flags & wxRE_NOTBOL )
 244         flagsRE |= REG_NOTBOL;
 245     if ( flags & wxRE_NOTEOL )
 246         flagsRE |= REG_NOTEOL;
 247
 248     // allocate matches array if needed
 249     wxRegExImpl *self = wxConstCast(this, wxRegExImpl);
 250     if ( !m_Matches && m_nMatches )
 251     {
 252         self->m_Matches = new regmatch_t[m_nMatches];
 253     }
 254
 255     // do match it
 256     int rc = regexec(&self->m_RegEx, wxConvertWX2MB(str), m_nMatches, m_Matches, flagsRE);
 257
 258     switch ( rc )
 259     {
 260         case 0:
 261             // matched successfully
 262             return TRUE;
 263
 264         default:
 265             // an error occured
 266             wxLogError(_("Failed to match '%s' in regular expression: %s"),
 267                        str, GetErrorMsg(rc).c_str());
 268             // fall through
 269
 270         case REG_NOMATCH:
 271             // no match
 272             return FALSE;
 273     }
 274 }
 275
 276 bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
 277 {
 278     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 279     wxCHECK_MSG( m_Matches, FALSE, _T("can't use with wxRE_NOSUB") );
 280     wxCHECK_MSG( index < m_nMatches, FALSE, _T("invalid match index") );
 281
 282     const regmatch_t& match = m_Matches[index];
 283
 284     if ( start )
 285         *start = match.rm_so;
 286     if ( len )
 287         *len = match.rm_eo - match.rm_so;
 288
 289     return TRUE;
 290 }
 291
 292 int wxRegExImpl::Replace(wxString *text,
 293                          const wxString& replacement,
 294                          size_t maxMatches) const
 295 {
 296     wxCHECK_MSG( text, -1, _T("NULL text in wxRegEx::Replace") );
 297     wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
 298
 299     // the replacement text
 300     wxString textNew;
 301
 302     // attempt at optimization: don't iterate over the string if it doesn't
 303     // contain back references at all
 304     bool mayHaveBackrefs =
 305         replacement.find_first_of(_T("\\&")) != wxString::npos;
 306
 307     if ( !mayHaveBackrefs )
 308     {
 309         textNew = replacement;
 310     }
 311
 312     // the position where we start looking for the match
 313     //
 314     // NB: initial version had a nasty bug because it used a wxChar* instead of
 315     //     an index but the problem is that replace() in the loop invalidates
 316     //     all pointers into the string so we have to use indices instead
 317     size_t matchStart = 0;
 318
 319     // number of replacement made: we won't make more than maxMatches of them
 320     // (unless maxMatches is 0 which doesn't limit the number of replacements)
 321     size_t countRepl = 0;
 322
 323     // note that "^" shouldn't match after the first call to Matches() so we
 324     // use wxRE_NOTBOL to prevent it from happening
 325     while ( (!maxMatches || countRepl < maxMatches) &&
 326             Matches(text->c_str() + matchStart, countRepl ? wxRE_NOTBOL : 0) )
 327     {
 328         // the string possibly contains back references: we need to calculate
 329         // the replacement text anew after each match
 330         if ( mayHaveBackrefs )
 331         {
 332             mayHaveBackrefs = FALSE;
 333             textNew.clear();
 334             textNew.reserve(replacement.length());
 335
 336             for ( const wxChar *p = replacement.c_str(); *p; p++ )
 337             {
 338                 size_t index = (size_t)-1;
 339
 340                 if ( *p == _T('\\') )
 341                 {
 342                     if ( wxIsdigit(*++p) )
 343                     {
 344                         // back reference
 345                         wxChar *end;
 346                         index = (size_t)wxStrtoul(p, &end, 10);
 347                         p = end - 1; // -1 to compensate for p++ in the loop
 348                     }
 349                     //else: backslash used as escape character
 350                 }
 351                 else if ( *p == _T('&') )
 352                 {
 353                     // treat this as "\0" for compatbility with ed and such
 354                     index = 0;
 355                 }
 356
 357                 // do we have a back reference?
 358                 if ( index != (size_t)-1 )
 359                 {
 360                     // yes, get its text
 361                     size_t start, len;
 362                     if ( !GetMatch(&start, &len, index) )
 363                     {
 364                         wxFAIL_MSG( _T("invalid back reference") );
 365
 366                         // just eat it...
 367                     }
 368                     else
 369                     {
 370                         textNew += wxString(text->c_str() + matchStart + start,
 371                                             len);
 372
 373                         mayHaveBackrefs = TRUE;
 374                     }
 375                 }
 376                 else // ordinary character
 377                 {
 378                     textNew += *p;
 379                 }
 380             }
 381         }
 382
 383         size_t start, len;
 384         if ( !GetMatch(&start, &len) )
 385         {
 386             // we did have match as Matches() returned true above!
 387             wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") );
 388
 389             return -1;
 390         }
 391
 392         matchStart += start;
 393         text->replace(matchStart, len, textNew);
 394
 395         countRepl++;
 396
 397         matchStart += textNew.length();
 398     }
 399
 400     return countRepl;
 401 }
 402
 403 // ----------------------------------------------------------------------------
 404 // wxRegEx: all methods are mostly forwarded to wxRegExImpl
 405 // ----------------------------------------------------------------------------
 406
 407 void wxRegEx::Init()
 408 {
 409     m_impl = NULL;
 410 }
 411
 412
 413 wxRegEx::~wxRegEx()
 414 {
 415     delete m_impl;
 416 }
 417
 418 bool wxRegEx::Compile(const wxString& expr, int flags)
 419 {
 420     if ( !m_impl )
 421     {
 422         m_impl = new wxRegExImpl;
 423     }
 424
 425     if ( !m_impl->Compile(expr, flags) )
 426     {
 427         // error message already given in wxRegExImpl::Compile
 428         delete m_impl;
 429         m_impl = NULL;
 430
 431         return FALSE;
 432     }
 433
 434     return TRUE;
 435 }
 436
 437 bool wxRegEx::Matches(const wxChar *str, int flags) const
 438 {
 439     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 440
 441     return m_impl->Matches(str, flags);
 442 }
 443
 444 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const
 445 {
 446     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 447
 448     return m_impl->GetMatch(start, len, index);
 449 }
 450
 451 wxString wxRegEx::GetMatch(const wxString& text, size_t index) const
 452 {
 453     size_t start, len;
 454     if ( !GetMatch(&start, &len, index) )
 455         return wxEmptyString;
 456
 457     return text.Mid(start, len);
 458 }
 459
 460 int wxRegEx::Replace(wxString *pattern,
 461                      const wxString& replacement,
 462                      size_t maxMatches) const
 463 {
 464     wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
 465
 466     return m_impl->Replace(pattern, replacement, maxMatches);
 467 }
 468
 469 #endif // wxUSE_REGEX