unicode compilation fix

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

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index c03e61260143e2e163b053f225d3499c82143f7b..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
 
@@ -38,7 +38,14 @@
     #include "wx/intl.h"
 #endif //WX_PRECOMP
 
+// FreeBSD & Watcom require this, it probably doesn't hurt for others
+#if defined(__UNIX__) || defined(__WATCOMC__) || defined(__DIGITALMARS__)
+    #include <sys/types.h>
+#endif
+
+#ifndef __WXWINCE__
 #include <regex.h>
+#endif
 
 #include "wx/regex.h"
 
@@ -68,17 +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;
 
@@ -100,15 +123,12 @@ private:
 
 wxRegExImpl::wxRegExImpl()
 {
-    m_isCompiled = FALSE;
-    m_Matches = NULL;
+    Init();
 }
 
 wxRegExImpl::~wxRegExImpl()
 {
     Free();
-
-    delete [] m_Matches;
 }
 
 wxString wxRegExImpl::GetErrorMsg(int errorcode) const
@@ -121,9 +141,15 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode) const
     {
         len++;
 
-        (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
+#if wxUSE_UNICODE
+        wxCharBuffer buf(len);
 
-        msg.UngetWriteBuf();
+        (void)regerror(errorcode, &m_RegEx, (char *)buf.data(), len);
+
+        msg = wxString(buf.data(), wxConvLibc);
+#else // !Unicode
+        (void)regerror(errorcode, &m_RegEx, wxStringBuffer(msg, len), len);
+#endif // Unicode/!Unicode
     }
     else // regerror() returned 0
     {
@@ -135,7 +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 &
@@ -153,7 +179,7 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
         flagsRE |= REG_NEWLINE;
 
     // compile it
-    int errorcode = regcomp(&m_RegEx, expr, flagsRE);
+    int errorcode = regcomp(&m_RegEx, expr.mb_str(), flagsRE);
     if ( errorcode )
     {
         wxLogError(_("Invalid regular expression '%s': %s"),
@@ -171,8 +197,32 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
         }
         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;
@@ -203,7 +253,7 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
     }
 
     // do match it
-    int rc = regexec(&self->m_RegEx, str, m_nMatches, m_Matches, flagsRE);
+    int rc = regexec(&self->m_RegEx, wxConvertWX2MB(str), m_nMatches, m_Matches, flagsRE);
 
     switch ( rc )
     {
@@ -230,8 +280,6 @@ bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
     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;
@@ -313,17 +361,7 @@ int wxRegExImpl::Replace(wxString *text,
                     size_t start, len;
                     if ( !GetMatch(&start, &len, index) )
                     {
-                        // we can't do it because GetMatch() returns FALSE
-                        // even for a valid back reference index if it didn't
-                        // match for this expression (e.g. it when alternative
-                        // branches were used and the one contained the back
-                        // ref didn't match)
-                        //
-                        // it would be better to distinguish between this case
-                        // and really invalid index, but I don't know how to
-                        // do it
-
-                        //wxFAIL_MSG( _T("invalid back reference") );
+                        wxFAIL_MSG( _T("invalid back reference") );
 
                         // just eat it...
                     }