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 #ifndef __WXWINCE__
  47 #include <regex.h>
  48 #endif
  49
  50 #include "wx/regex.h"
  51
  52 // ----------------------------------------------------------------------------
  53 // private classes
  54 // ----------------------------------------------------------------------------
  55
  56 // the real implementation of wxRegEx
  57 class wxRegExImpl
  58 {
  59 public:
  60     // ctor and dtor
  61     wxRegExImpl();
  62     ~wxRegExImpl();
  63
  64     // return TRUE if Compile() had been called successfully
  65     bool IsValid() const { return m_isCompiled; }
  66
  67     // RE operations
  68     bool Compile(const wxString& expr, int flags = 0);
  69     bool Matches(const wxChar *str, int flags = 0) const;
  70     bool GetMatch(size_t *start, size_t *len, size_t index = 0) const;
  71     int Replace(wxString *pattern, const wxString& replacement,
  72                 size_t maxMatches = 0) const;
  73
  74 private:
  75     // return the string containing the error message for the given err code
  76     wxString GetErrorMsg(int errorcode) const;
  77
  78     // init the members
  79     void Init()
  80     {
  81         m_isCompiled = FALSE;
  82         m_Matches = NULL;
  83         m_nMatches = 0;
  84     }
  85
  86     // free the RE if compiled
  87     void Free()
  88     {
  89         if ( IsValid() )
  90         {
  91             regfree(&m_RegEx);
  92         }
  93
  94         delete [] m_Matches;
  95     }
  96
  97     // free the RE if any and reinit the members
  98     void Reinit()
  99     {
 100         Free();
 101         Init();
 102     }
 103
 104
 105     // compiled RE
 106     regex_t     m_RegEx;
 107
 108     // the subexpressions data
 109     regmatch_t *m_Matches;
 110     size_t      m_nMatches;
 111
 112     // TRUE if m_RegEx is valid
 113     bool        m_isCompiled;
 114 };
 115
 116 // ============================================================================
 117 // implementation
 118 // ============================================================================
 119
 120 // ----------------------------------------------------------------------------
 121 // wxRegExImpl
 122 // ----------------------------------------------------------------------------
 123
 124 wxRegExImpl::wxRegExImpl()
 125 {
 126     Init();
 127 }
 128
 129 wxRegExImpl::~wxRegExImpl()
 130 {
 131     Free();
 132 }
 133
 134 wxString wxRegExImpl::GetErrorMsg(int errorcode) const
 135 {
 136     wxString msg;
 137
 138     // first get the string length needed
 139     int len = regerror(errorcode, &m_RegEx, NULL, 0);
 140     if ( len > 0 )
 141     {
 142         len++;
 143
 144 #if wxUSE_UNICODE
 145         wxCharBuffer buf(len);
 146
 147         (void)regerror(errorcode, &m_RegEx, (char *)buf.data(), len);
 148
 149         msg = wxString(buf.data(), wxConvLibc);
 150 #else // !Unicode
 151         (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
 152
 153         msg.UngetWriteBuf();
 154 #endif // Unicode/!Unicode
 155     }
 156     else // regerror() returned 0
 157     {
 158         msg = _("unknown error");
 159     }
 160
 161     return msg;
 162 }
 163
 164 bool wxRegExImpl::Compile(const wxString& expr, int flags)
 165 {
 166     Reinit();
 167
 168     // translate our flags to regcomp() ones
 169     wxASSERT_MSG( !(flags &
 170                         ~(wxRE_BASIC | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)),
 171                   _T("unrecognized flags in wxRegEx::Compile") );
 172
 173     int flagsRE = 0;
 174     if ( !(flags & wxRE_BASIC) )
 175         flagsRE |= REG_EXTENDED;
 176     if ( flags & wxRE_ICASE )
 177         flagsRE |= REG_ICASE;
 178     if ( flags & wxRE_NOSUB )
 179         flagsRE |= REG_NOSUB;
 180     if ( flags & wxRE_NEWLINE )
 181         flagsRE |= REG_NEWLINE;
 182
 183     // compile it
 184     int errorcode = regcomp(&m_RegEx, expr.mb_str(), flagsRE);
 185     if ( errorcode )
 186     {
 187         wxLogError(_("Invalid regular expression '%s': %s"),
 188                    expr.c_str(), GetErrorMsg(errorcode).c_str());
 189
 190         m_isCompiled = FALSE;
 191     }
 192     else // ok
 193     {
 194         // don't allocate the matches array now, but do it later if necessary
 195         if ( flags & wxRE_NOSUB )
 196         {
 197             // we don't need it at all
 198             m_nMatches = 0;
 199         }
 200         else
 201         {
 202             // we will alloc the array later (only if really needed) but count
 203             // the number of sub-expressions in the regex right now
 204
 205             // there is always one for the whole expression
 206             m_nMatches = 1;
 207
 208             // and some more for bracketed subexperessions
 209             for ( const wxChar *cptr = expr.c_str(); *cptr; cptr++ )
 210             {
 211                 if ( *cptr == _T('\\') )
 212                 {
 213                     // in basic RE syntax groups are inside \(...\)
 214                     if ( *++cptr == _T('(') && (flags & wxRE_BASIC) )
 215                     {
 216                         m_nMatches++;
 217                     }
 218                 }
 219                 else if ( *cptr == _T('(') && !(flags & wxRE_BASIC) )
 220                 {
 221                     // we know that the previous character is not an unquoted
 222                     // backslash because it would have been eaten above, so we
 223                     // have a bar '(' and this indicates a group start for the
 224                     // extended syntax
 225                     m_nMatches++;
 226                 }
 227             }
 228         }
 229
 230         m_isCompiled = TRUE;
 231     }
 232
 233     return IsValid();
 234 }
 235
 236 bool wxRegExImpl::Matches(const wxChar *str, int flags) const
 237 {
 238     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 239
 240     // translate our flags to regexec() ones
 241     wxASSERT_MSG( !(flags & ~(wxRE_NOTBOL | wxRE_NOTEOL)),
 242                   _T("unrecognized flags in wxRegEx::Matches") );
 243
 244     int flagsRE = 0;
 245     if ( flags & wxRE_NOTBOL )
 246         flagsRE |= REG_NOTBOL;
 247     if ( flags & wxRE_NOTEOL )
 248         flagsRE |= REG_NOTEOL;
 249
 250     // allocate matches array if needed
 251     wxRegExImpl *self = wxConstCast(this, wxRegExImpl);
 252     if ( !m_Matches && m_nMatches )
 253     {
 254         self->m_Matches = new regmatch_t[m_nMatches];
 255     }
 256
 257     // do match it
 258     int rc = regexec(&self->m_RegEx, wxConvertWX2MB(str), m_nMatches, m_Matches, flagsRE);
 259
 260     switch ( rc )
 261     {
 262         case 0:
 263             // matched successfully
 264             return TRUE;
 265
 266         default:
 267             // an error occured
 268             wxLogError(_("Failed to match '%s' in regular expression: %s"),
 269                        str, GetErrorMsg(rc).c_str());
 270             // fall through
 271
 272         case REG_NOMATCH:
 273             // no match
 274             return FALSE;
 275     }
 276 }
 277
 278 bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
 279 {
 280     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 281     wxCHECK_MSG( m_Matches, FALSE, _T("can't use with wxRE_NOSUB") );
 282     wxCHECK_MSG( index < m_nMatches, FALSE, _T("invalid match index") );
 283
 284     const regmatch_t& match = m_Matches[index];
 285
 286     if ( start )
 287         *start = match.rm_so;
 288     if ( len )
 289         *len = match.rm_eo - match.rm_so;
 290
 291     return TRUE;
 292 }
 293
 294 int wxRegExImpl::Replace(wxString *text,
 295                          const wxString& replacement,
 296                          size_t maxMatches) const
 297 {
 298     wxCHECK_MSG( text, -1, _T("NULL text in wxRegEx::Replace") );
 299     wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
 300
 301     // the replacement text
 302     wxString textNew;
 303
 304     // attempt at optimization: don't iterate over the string if it doesn't
 305     // contain back references at all
 306     bool mayHaveBackrefs =
 307         replacement.find_first_of(_T("\\&")) != wxString::npos;
 308
 309     if ( !mayHaveBackrefs )
 310     {
 311         textNew = replacement;
 312     }
 313
 314     // the position where we start looking for the match
 315     //
 316     // NB: initial version had a nasty bug because it used a wxChar* instead of
 317     //     an index but the problem is that replace() in the loop invalidates
 318     //     all pointers into the string so we have to use indices instead
 319     size_t matchStart = 0;
 320
 321     // number of replacement made: we won't make more than maxMatches of them
 322     // (unless maxMatches is 0 which doesn't limit the number of replacements)
 323     size_t countRepl = 0;
 324
 325     // note that "^" shouldn't match after the first call to Matches() so we
 326     // use wxRE_NOTBOL to prevent it from happening
 327     while ( (!maxMatches || countRepl < maxMatches) &&
 328             Matches(text->c_str() + matchStart, countRepl ? wxRE_NOTBOL : 0) )
 329     {
 330         // the string possibly contains back references: we need to calculate
 331         // the replacement text anew after each match
 332         if ( mayHaveBackrefs )
 333         {
 334             mayHaveBackrefs = FALSE;
 335             textNew.clear();
 336             textNew.reserve(replacement.length());
 337
 338             for ( const wxChar *p = replacement.c_str(); *p; p++ )
 339             {
 340                 size_t index = (size_t)-1;
 341
 342                 if ( *p == _T('\\') )
 343                 {
 344                     if ( wxIsdigit(*++p) )
 345                     {
 346                         // back reference
 347                         wxChar *end;
 348                         index = (size_t)wxStrtoul(p, &end, 10);
 349                         p = end - 1; // -1 to compensate for p++ in the loop
 350                     }
 351                     //else: backslash used as escape character
 352                 }
 353                 else if ( *p == _T('&') )
 354                 {
 355                     // treat this as "\0" for compatbility with ed and such
 356                     index = 0;
 357                 }
 358
 359                 // do we have a back reference?
 360                 if ( index != (size_t)-1 )
 361                 {
 362                     // yes, get its text
 363                     size_t start, len;
 364                     if ( !GetMatch(&start, &len, index) )
 365                     {
 366                         wxFAIL_MSG( _T("invalid back reference") );
 367
 368                         // just eat it...
 369                     }
 370                     else
 371                     {
 372                         textNew += wxString(text->c_str() + matchStart + start,
 373                                             len);
 374
 375                         mayHaveBackrefs = TRUE;
 376                     }
 377                 }
 378                 else // ordinary character
 379                 {
 380                     textNew += *p;
 381                 }
 382             }
 383         }
 384
 385         size_t start, len;
 386         if ( !GetMatch(&start, &len) )
 387         {
 388             // we did have match as Matches() returned true above!
 389             wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") );
 390
 391             return -1;
 392         }
 393
 394         matchStart += start;
 395         text->replace(matchStart, len, textNew);
 396
 397         countRepl++;
 398
 399         matchStart += textNew.length();
 400     }
 401
 402     return countRepl;
 403 }
 404
 405 // ----------------------------------------------------------------------------
 406 // wxRegEx: all methods are mostly forwarded to wxRegExImpl
 407 // ----------------------------------------------------------------------------
 408
 409 void wxRegEx::Init()
 410 {
 411     m_impl = NULL;
 412 }
 413
 414
 415 wxRegEx::~wxRegEx()
 416 {
 417     delete m_impl;
 418 }
 419
 420 bool wxRegEx::Compile(const wxString& expr, int flags)
 421 {
 422     if ( !m_impl )
 423     {
 424         m_impl = new wxRegExImpl;
 425     }
 426
 427     if ( !m_impl->Compile(expr, flags) )
 428     {
 429         // error message already given in wxRegExImpl::Compile
 430         delete m_impl;
 431         m_impl = NULL;
 432
 433         return FALSE;
 434     }
 435
 436     return TRUE;
 437 }
 438
 439 bool wxRegEx::Matches(const wxChar *str, int flags) const
 440 {
 441     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 442
 443     return m_impl->Matches(str, flags);
 444 }
 445
 446 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const
 447 {
 448     wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
 449
 450     return m_impl->GetMatch(start, len, index);
 451 }
 452
 453 wxString wxRegEx::GetMatch(const wxString& text, size_t index) const
 454 {
 455     size_t start, len;
 456     if ( !GetMatch(&start, &len, index) )
 457         return wxEmptyString;
 458
 459     return text.Mid(start, len);
 460 }
 461
 462 int wxRegEx::Replace(wxString *pattern,
 463                      const wxString& replacement,
 464                      size_t maxMatches) const
 465 {
 466     wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
 467
 468     return m_impl->Replace(pattern, replacement, maxMatches);
 469 }
 470
 471 #endif // wxUSE_REGEX