initialize m_initialized to false to avoid spurious asserts

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

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index db11e97cec0cde16f08437dc22cda8cc67aa8198..2cd6afafce6618554716818ad94964f6ab542bad 100644 (file)
--- a/src/common/regex.cpp
+++ b/src/common/regex.cpp
@@ -18,7 +18,7 @@
 // headers
 // ----------------------------------------------------------------------------
 
 // headers
 // ----------------------------------------------------------------------------
 

-#ifdef __GNUG__
+#if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)

     #pragma implementation "regex.h"
 #endif
 
     #pragma implementation "regex.h"
 #endif
 


@@ -38,13 +38,15 @@
     #include "wx/intl.h"
 #endif //WX_PRECOMP
 
     #include "wx/intl.h"
 #endif //WX_PRECOMP
 

-// FreeBSD requires this, it probably doesn't hurt for others
-#ifdef __UNIX__
-    #include <sys/types.h>
+// 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
 
 #include <regex.h>
 #endif
 
 #include <regex.h>

-

 #include "wx/regex.h"
 
 // ----------------------------------------------------------------------------
 #include "wx/regex.h"
 
 // ----------------------------------------------------------------------------


@@ -71,7 +73,15 @@ public:
 
 private:
     // return the string containing the error message for the given err code
 
 private:
     // return the string containing the error message for the given err code

-    wxString GetErrorMsg(int errorcode) const;
+    wxString GetErrorMsg(int errorcode, bool badconv) const;
+
+    // init the members
+    void Init()
+    {
+        m_isCompiled = FALSE;
+        m_Matches = NULL;
+        m_nMatches = 0;
+    }

 
     // free the RE if compiled
     void Free()
 
     // free the RE if compiled
     void Free()


@@ -79,11 +89,19 @@ private:
         if ( IsValid() )
         {
             regfree(&m_RegEx);
         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;
 
     // compiled RE
     regex_t     m_RegEx;
 


@@ -105,51 +123,71 @@ private:
 
 wxRegExImpl::wxRegExImpl()
 {
 
 wxRegExImpl::wxRegExImpl()
 {

-    m_isCompiled = FALSE;
-    m_Matches = NULL;
+    Init();

 }
 
 wxRegExImpl::~wxRegExImpl()
 {
     Free();
 }
 
 wxRegExImpl::~wxRegExImpl()
 {
     Free();

-
-    delete [] m_Matches;

 }
 
 }
 

-wxString wxRegExImpl::GetErrorMsg(int errorcode) const
+wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const

 {
 {

-    wxString msg;
+#if wxUSE_UNICODE && !defined(__REG_NOFRONT)
+    // currently only needed when using system library in Unicode mode
+    if ( badconv )
+    {
+        return _("conversion to 8-bit encoding failed");
+    }
+#else
+    // 'use' badconv to avoid a compiler warning
+    (void)badconv;
+#endif
+
+    wxString szError;

 
     // first get the string length needed
     int len = regerror(errorcode, &m_RegEx, NULL, 0);
     if ( len > 0 )
     {
 
     // 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
     {
     }
     else // regerror() returned 0
     {

-        msg = _("unknown error");
+        szError = _("unknown error");

     }
 
     }
 

-    return msg;
+    return szError;

 }
 
 bool wxRegExImpl::Compile(const wxString& expr, int flags)
 {
 }
 
 bool wxRegExImpl::Compile(const wxString& expr, int flags)
 {

-    Free();
-
-    // translate our flags to regcomp() ones
-    wxASSERT_MSG( !(flags &
-                        ~(wxRE_BASIC | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)),
+    Reinit();
+
+#ifdef WX_NO_REGEX_ADVANCED
+#   define FLAVORS wxRE_BASIC
+#else
+#   define FLAVORS (wxRE_ADVANCED | wxRE_BASIC)
+    wxASSERT_MSG( (flags & FLAVORS) != FLAVORS,
+                  _T("incompatible flags in wxRegEx::Compile") );
+#endif
+    wxASSERT_MSG( !(flags & ~(FLAVORS | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)),

                   _T("unrecognized flags in wxRegEx::Compile") );
 
                   _T("unrecognized flags in wxRegEx::Compile") );
 

+    // translate our flags to regcomp() ones

     int flagsRE = 0;
     if ( !(flags & wxRE_BASIC) )
     int flagsRE = 0;
     if ( !(flags & wxRE_BASIC) )

-        flagsRE |= REG_EXTENDED;
+#ifndef WX_NO_REGEX_ADVANCED
+        if (flags & wxRE_ADVANCED)
+            flagsRE |= REG_ADVANCED;
+        else
+#endif
+            flagsRE |= REG_EXTENDED;

     if ( flags & wxRE_ICASE )
         flagsRE |= REG_ICASE;
     if ( flags & wxRE_NOSUB )
     if ( flags & wxRE_ICASE )
         flagsRE |= REG_ICASE;
     if ( flags & wxRE_NOSUB )


@@ -158,11 +196,18 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
         flagsRE |= REG_NEWLINE;
 
     // compile it
         flagsRE |= REG_NEWLINE;
 
     // compile it

-    int errorcode = regcomp(&m_RegEx, expr, flagsRE);
+#ifdef __REG_NOFRONT
+    bool conv = true;
+    int errorcode = re_comp(&m_RegEx, expr, expr.length(), flagsRE);
+#else
+    const wxWX2MBbuf conv = expr.mbc_str();
+    int errorcode = conv ? regcomp(&m_RegEx, conv, flagsRE) : REG_BADPAT;
+#endif
+

     if ( errorcode )
     {
         wxLogError(_("Invalid regular expression '%s': %s"),
     if ( errorcode )
     {
         wxLogError(_("Invalid regular expression '%s': %s"),

-                   expr.c_str(), GetErrorMsg(errorcode).c_str());
+                   expr.c_str(), GetErrorMsg(errorcode, !conv).c_str());

 
         m_isCompiled = FALSE;
     }
 
         m_isCompiled = FALSE;
     }


@@ -176,8 +221,32 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
         }
         else
         {
         }
         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;
         }
 
         m_isCompiled = TRUE;


@@ -208,7 +277,13 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
     }
 
     // do match it
     }
 
     // do match it

-    int rc = regexec(&self->m_RegEx, str, m_nMatches, m_Matches, flagsRE);
+#ifdef __REG_NOFRONT
+    bool conv = true;
+    int rc = re_exec(&self->m_RegEx, str, wxStrlen(str), NULL, m_nMatches, m_Matches, flagsRE);
+#else
+    const wxWX2MBbuf conv = wxConvertWX2MB(str);
+    int rc = conv ? regexec(&self->m_RegEx, conv, m_nMatches, m_Matches, flagsRE) : REG_BADPAT;
+#endif

 
     switch ( rc )
     {
 
     switch ( rc )
     {


@@ -219,7 +294,7 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
         default:
             // an error occured
             wxLogError(_("Failed to match '%s' in regular expression: %s"),
         default:
             // an error occured
             wxLogError(_("Failed to match '%s' in regular expression: %s"),

-                       str, GetErrorMsg(rc).c_str());
+                       str, GetErrorMsg(rc, !conv).c_str());

             // fall through
 
         case REG_NOMATCH:
             // fall through
 
         case REG_NOMATCH:


@@ -235,8 +310,6 @@ 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];
     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;
 
     if ( start )
         *start = match.rm_so;


@@ -318,17 +391,7 @@ int wxRegExImpl::Replace(wxString *text,
                     size_t start, len;
                     if ( !GetMatch(&start, &len, index) )
                     {
                     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...
                     }
 
                         // just eat it...
                     }