// headers
// ----------------------------------------------------------------------------
-#ifdef __GNUG__
+#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
#pragma implementation "regex.h"
#endif
#include "wx/intl.h"
#endif //WX_PRECOMP
+// FreeBSD, Watcom and DMars require this, CW doesn't have nor need it.
+// Others also don't seem to need it. If you have an error related to
+// (not) including <sys/types.h> please report details to
+// wx-dev@lists.wxwindows.org
+#if defined(__UNIX__) || defined(__WATCOMC__) || defined(__DIGITALMARS__)
+# include <sys/types.h>
+#endif
+
+#ifndef __WXWINCE__
#include <regex.h>
+#endif
#include "wx/regex.h"
bool IsValid() const { return m_isCompiled; }
// RE operations
- bool Compile(const wxString& expr, int flags);
- bool Matches(const wxString& str, int flags) const;
- bool GetMatch(size_t *start, size_t *len, size_t index) const;
- int Replace(wxString *pattern, const wxString& replacement) const;
+ bool Compile(const wxString& expr, int flags = 0);
+ bool Matches(const wxChar *str, int flags = 0) const;
+ bool GetMatch(size_t *start, size_t *len, size_t index = 0) const;
+ int Replace(wxString *pattern, const wxString& replacement,
+ size_t maxMatches = 0) const;
private:
// return the string containing the error message for the given err code
wxString GetErrorMsg(int errorcode) const;
+ // init the members
+ void Init()
+ {
+ m_isCompiled = FALSE;
+ m_Matches = NULL;
+ m_nMatches = 0;
+ }
+
// free the RE if compiled
void Free()
{
if ( IsValid() )
{
regfree(&m_RegEx);
-
- m_isCompiled = FALSE;
}
+
+ delete [] m_Matches;
}
+ // free the RE if any and reinit the members
+ void Reinit()
+ {
+ Free();
+ Init();
+ }
+
+
// compiled RE
regex_t m_RegEx;
wxRegExImpl::wxRegExImpl()
{
- m_isCompiled = FALSE;
- m_Matches = NULL;
+ Init();
}
wxRegExImpl::~wxRegExImpl()
{
Free();
-
- delete [] m_Matches;
}
wxString wxRegExImpl::GetErrorMsg(int errorcode) const
{
- wxString msg;
+ wxString szError;
// first get the string length needed
int len = regerror(errorcode, &m_RegEx, NULL, 0);
if ( len > 0 )
{
- len++;
+ char* szcmbError = new char[++len];
- (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
+ (void)regerror(errorcode, &m_RegEx, szcmbError, len);
- msg.UngetWriteBuf();
+ szError = wxConvertMB2WX(szcmbError);
+ delete [] szcmbError;
}
else // regerror() returned 0
{
- msg = _("unknown error");
+ szError = _("unknown error");
}
- return msg;
+ return szError;
}
bool wxRegExImpl::Compile(const wxString& expr, int flags)
{
- Free();
+ Reinit();
// translate our flags to regcomp() ones
wxASSERT_MSG( !(flags &
if ( flags & wxRE_NEWLINE )
flagsRE |= REG_NEWLINE;
- // compile it
+
+
int errorcode = regcomp(&m_RegEx, expr, flagsRE);
- if ( errorcode )
+
+ if ( errorcode )
{
wxLogError(_("Invalid regular expression '%s': %s"),
expr.c_str(), GetErrorMsg(errorcode).c_str());
}
else
{
- // will alloc later
- m_nMatches = WX_REGEX_MAXMATCHES;
+ // we will alloc the array later (only if really needed) but count
+ // the number of sub-expressions in the regex right now
+
+ // there is always one for the whole expression
+ m_nMatches = 1;
+
+ // and some more for bracketed subexperessions
+ for ( const wxChar *cptr = expr.c_str(); *cptr; cptr++ )
+ {
+ if ( *cptr == _T('\\') )
+ {
+ // in basic RE syntax groups are inside \(...\)
+ if ( *++cptr == _T('(') && (flags & wxRE_BASIC) )
+ {
+ m_nMatches++;
+ }
+ }
+ else if ( *cptr == _T('(') && !(flags & wxRE_BASIC) )
+ {
+ // we know that the previous character is not an unquoted
+ // backslash because it would have been eaten above, so we
+ // have a bar '(' and this indicates a group start for the
+ // extended syntax
+ m_nMatches++;
+ }
+ }
}
m_isCompiled = TRUE;
return IsValid();
}
-bool wxRegExImpl::Matches(const wxString& str, int flags) const
+bool wxRegExImpl::Matches(const wxChar *str, int flags) const
{
wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
default:
// an error occured
wxLogError(_("Failed to match '%s' in regular expression: %s"),
- str.c_str(), GetErrorMsg(rc).c_str());
+ str, GetErrorMsg(rc).c_str());
// fall through
case REG_NOMATCH:
wxCHECK_MSG( index < m_nMatches, FALSE, _T("invalid match index") );
const regmatch_t& match = m_Matches[index];
- if ( match.rm_so == -1 )
- return FALSE;
if ( start )
*start = match.rm_so;
return TRUE;
}
-int wxRegExImpl::Replace(wxString *pattern, const wxString& replacement) const
+int wxRegExImpl::Replace(wxString *text,
+ const wxString& replacement,
+ size_t maxMatches) const
{
- wxCHECK_MSG( pattern, -1, _T("NULL pattern in wxRegEx::Replace") );
+ wxCHECK_MSG( text, -1, _T("NULL text in wxRegEx::Replace") );
+ wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
- wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
+ // the replacement text
+ wxString textNew;
- int replaced = 0;
- size_t lastpos = 0;
- wxString newstring;
+ // attempt at optimization: don't iterate over the string if it doesn't
+ // contain back references at all
+ bool mayHaveBackrefs =
+ replacement.find_first_of(_T("\\&")) != wxString::npos;
- for ( size_t idx = 0;
- m_Matches[idx].rm_so != -1 && idx < m_nMatches;
- idx++ )
+ if ( !mayHaveBackrefs )
{
- // copy non-matching bits:
- newstring << pattern->Mid(lastpos, m_Matches[idx].rm_so - lastpos);
- // copy replacement:
- newstring << replacement;
- // remember how far we got:
- lastpos = m_Matches[idx].rm_eo;
- replaced ++;
+ textNew = replacement;
+ }
+
+ // the position where we start looking for the match
+ //
+ // NB: initial version had a nasty bug because it used a wxChar* instead of
+ // an index but the problem is that replace() in the loop invalidates
+ // all pointers into the string so we have to use indices instead
+ size_t matchStart = 0;
+
+ // number of replacement made: we won't make more than maxMatches of them
+ // (unless maxMatches is 0 which doesn't limit the number of replacements)
+ size_t countRepl = 0;
+
+ // note that "^" shouldn't match after the first call to Matches() so we
+ // use wxRE_NOTBOL to prevent it from happening
+ while ( (!maxMatches || countRepl < maxMatches) &&
+ Matches(text->c_str() + matchStart, countRepl ? wxRE_NOTBOL : 0) )
+ {
+ // the string possibly contains back references: we need to calculate
+ // the replacement text anew after each match
+ if ( mayHaveBackrefs )
+ {
+ mayHaveBackrefs = FALSE;
+ textNew.clear();
+ textNew.reserve(replacement.length());
+
+ for ( const wxChar *p = replacement.c_str(); *p; p++ )
+ {
+ size_t index = (size_t)-1;
+
+ if ( *p == _T('\\') )
+ {
+ if ( wxIsdigit(*++p) )
+ {
+ // back reference
+ wxChar *end;
+ index = (size_t)wxStrtoul(p, &end, 10);
+ p = end - 1; // -1 to compensate for p++ in the loop
+ }
+ //else: backslash used as escape character
+ }
+ else if ( *p == _T('&') )
+ {
+ // treat this as "\0" for compatbility with ed and such
+ index = 0;
+ }
+
+ // do we have a back reference?
+ if ( index != (size_t)-1 )
+ {
+ // yes, get its text
+ size_t start, len;
+ if ( !GetMatch(&start, &len, index) )
+ {
+ wxFAIL_MSG( _T("invalid back reference") );
+
+ // just eat it...
+ }
+ else
+ {
+ textNew += wxString(text->c_str() + matchStart + start,
+ len);
+
+ mayHaveBackrefs = TRUE;
+ }
+ }
+ else // ordinary character
+ {
+ textNew += *p;
+ }
+ }
+ }
+
+ size_t start, len;
+ if ( !GetMatch(&start, &len) )
+ {
+ // we did have match as Matches() returned true above!
+ wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") );
+
+ return -1;
+ }
+
+ matchStart += start;
+ text->replace(matchStart, len, textNew);
+
+ countRepl++;
+
+ matchStart += textNew.length();
}
- if(replaced > 0)
- *pattern = newstring;
- return replaced;
+
+ return countRepl;
}
// ----------------------------------------------------------------------------
return TRUE;
}
-bool wxRegEx::Matches(const wxString& str, int flags) const
+bool wxRegEx::Matches(const wxChar *str, int flags) const
{
wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const
{
- wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
+ wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
return m_impl->GetMatch(start, len, index);
}
return text.Mid(start, len);
}
-int wxRegEx::Replace(wxString *pattern, const wxString& replacement) const
+int wxRegEx::Replace(wxString *pattern,
+ const wxString& replacement,
+ size_t maxMatches) const
{
wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
- return m_impl->Replace(pattern, replacement);
+ return m_impl->Replace(pattern, replacement, maxMatches);
}
#endif // wxUSE_REGEX