unicode compilation fix

[wxWidgets.git] / src / common / regex.cpp

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index cbaa57991dc9acfad1fdb91d09a2936e13bf5b53..9c7e55c04057f48b793d4af8ceca93c0fe780226 100644 (file)
--- a/src/common/regex.cpp
+++ b/src/common/regex.cpp
@@ -18,7 +18,7 @@
 // headers
 // ----------------------------------------------------------------------------
 
-#ifdef __GNUG__
+#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
     #pragma implementation "regex.h"
 #endif
 
@@ -39,11 +39,13 @@
 #endif //WX_PRECOMP
 
 // FreeBSD & Watcom require this, it probably doesn't hurt for others
-#if defined(__UNIX__) || defined(__WATCOMC__)
+#if defined(__UNIX__) || defined(__WATCOMC__) || defined(__DIGITALMARS__)
     #include <sys/types.h>
 #endif
 
+#ifndef __WXWINCE__
 #include <regex.h>
+#endif
 
 #include "wx/regex.h"
 
@@ -73,17 +75,33 @@ 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;
 
@@ -105,16 +123,12 @@ private:
 
 wxRegExImpl::wxRegExImpl()
 {
-    m_isCompiled = FALSE;
-    m_Matches = NULL;
-    m_nMatches = 0;
+    Init();
 }
 
 wxRegExImpl::~wxRegExImpl()
 {
     Free();
-
-    delete [] m_Matches;
 }
 
 wxString wxRegExImpl::GetErrorMsg(int errorcode) const
@@ -134,9 +148,7 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode) const
 
         msg = wxString(buf.data(), wxConvLibc);
 #else // !Unicode
-        (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
-
-        msg.UngetWriteBuf();
+        (void)regerror(errorcode, &m_RegEx, wxStringBuffer(msg, len), len);
 #endif // Unicode/!Unicode
     }
     else // regerror() returned 0
@@ -149,7 +161,7 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode) const
 
 bool wxRegExImpl::Compile(const wxString& expr, int flags)
 {
-    Free();
+    Reinit();
 
     // translate our flags to regcomp() ones
     wxASSERT_MSG( !(flags &
@@ -192,21 +204,24 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
             m_nMatches = 1;
 
             // and some more for bracketed subexperessions
-            const wxChar *cptr = expr.c_str();
-            wxChar prev = _T('\0');
-            while ( *cptr != _T('\0') )
+            for ( const wxChar *cptr = expr.c_str(); *cptr; cptr++ )
             {
-                // is this a subexpr start, i.e. "(" for extended regex or
-                // "\(" for a basic one?
-                if ( *cptr == _T('(') &&
-                     (flags & wxRE_BASIC ? prev == _T('\\')
-                                         : prev != _T('\\')) )
+                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++;
                 }
-
-                prev = *cptr;
-                cptr++;
             }
         }