Unicode fixes

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

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index 7b480f24c80e18a686e0a84bd3812c4c96f25b48..896d2ad248d0ddb74fff668efab5078c17abc4a3 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 & Watcom require this, it probably doesn't hurt for others
-#if defined(__UNIX__) || defined(__WATCOMC__)
-    #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"
 
 // ----------------------------------------------------------------------------


@@ -59,19 +61,28 @@ public:
     wxRegExImpl();
     ~wxRegExImpl();
 
     wxRegExImpl();
     ~wxRegExImpl();
 

-    // return TRUE if Compile() had been called successfully
+    // return true if Compile() had been called successfully

     bool IsValid() const { return m_isCompiled; }
 
     // RE operations
     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;
     bool IsValid() const { return m_isCompiled; }
 
     // RE operations
     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;

+    size_t GetMatchCount() 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
     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;
+    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 +90,19 @@ 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;
 


@@ -91,7 +110,7 @@ private:
     regmatch_t *m_Matches;
     size_t      m_nMatches;
 
     regmatch_t *m_Matches;
     size_t      m_nMatches;
 

-    // TRUE if m_RegEx is valid
+    // true if m_RegEx is valid

     bool        m_isCompiled;
 };
 
     bool        m_isCompiled;
 };
 


@@ -105,60 +124,71 @@ private:
 
 wxRegExImpl::wxRegExImpl()
 {
 
 wxRegExImpl::wxRegExImpl()
 {

-    m_isCompiled = FALSE;
-    m_Matches = NULL;
-    m_nMatches = 0;
+    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];

 
 

-#if wxUSE_UNICODE
-        wxCharBuffer buf(len);
+        (void)regerror(errorcode, &m_RegEx, szcmbError, len);

 
 

-        (void)regerror(errorcode, &m_RegEx, (char *)buf.data(), len);
-
-        msg = buf.data();
-#else // !Unicode
-        (void)regerror(errorcode, &m_RegEx, msg.GetWriteBuf(len), len);
-
-        msg.UngetWriteBuf();
-#endif // Unicode/!Unicode
+        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 )


@@ -167,13 +197,20 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
         flagsRE |= REG_NEWLINE;
 
     // compile it
         flagsRE |= REG_NEWLINE;
 
     // compile it

-    int errorcode = regcomp(&m_RegEx, expr.mb_str(), flagsRE);
+#ifdef __REG_NOFRONT
+    bool conv = true;
+    int errorcode = wx_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;

     }
     else // ok
     {
     }
     else // ok
     {


@@ -192,25 +229,31 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
             m_nMatches = 1;
 
             // and some more for bracketed subexperessions
             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('\\') )

                 {
                 {

-                    m_nMatches++;
+                    // 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 bare '(' and this indicates a group start for the
+                    // extended syntax. '(?' is used for extensions by perl-
+                    // like REs (e.g. advanced), and is not valid for POSIX
+                    // extended, so ignore them always.
+                    if ( cptr[1] != _T('?') )
+                        m_nMatches++;

                 }
                 }

-
-                prev = *cptr;
-                cptr++;

             }
         }
 
             }
         }
 

-        m_isCompiled = TRUE;
+        m_isCompiled = true;

     }
 
     return IsValid();
     }
 
     return IsValid();


@@ -218,7 +261,7 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
 
 bool wxRegExImpl::Matches(const wxChar *str, int flags) const
 {
 
 bool wxRegExImpl::Matches(const wxChar *str, int flags) const
 {

-    wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
+    wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );

 
     // translate our flags to regexec() ones
     wxASSERT_MSG( !(flags & ~(wxRE_NOTBOL | wxRE_NOTEOL)),
 
     // translate our flags to regexec() ones
     wxASSERT_MSG( !(flags & ~(wxRE_NOTBOL | wxRE_NOTEOL)),


@@ -238,31 +281,38 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
     }
 
     // do match it
     }
 
     // do match it

-    int rc = regexec(&self->m_RegEx, wxConvertWX2MB(str), m_nMatches, m_Matches, flagsRE);
+#ifdef __REG_NOFRONT
+    bool conv = true;
+    int rc = wx_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 )
     {
         case 0:
             // matched successfully
 
     switch ( rc )
     {
         case 0:
             // matched successfully

-            return TRUE;
+            return true;

 
         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:
             // no match
             // fall through
 
         case REG_NOMATCH:
             // no match

-            return FALSE;
+            return false;

     }
 }
 
 bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
 {
     }
 }
 
 bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
 {

-    wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
-    wxCHECK_MSG( m_Matches, FALSE, _T("can't use with wxRE_NOSUB") );
-    wxCHECK_MSG( index < m_nMatches, FALSE, _T("invalid match index") );
+    wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
+    wxCHECK_MSG( m_nMatches, false, _T("can't use with wxRE_NOSUB") );
+    wxCHECK_MSG( m_Matches, false, _T("must call Matches() first") );
+    wxCHECK_MSG( index < m_nMatches, false, _T("invalid match index") );

 
     const regmatch_t& match = m_Matches[index];
 
 
     const regmatch_t& match = m_Matches[index];
 


@@ -271,15 +321,23 @@ bool wxRegExImpl::GetMatch(size_t *start, size_t *len, size_t index) const
     if ( len )
         *len = match.rm_eo - match.rm_so;
 
     if ( len )
         *len = match.rm_eo - match.rm_so;
 

-    return TRUE;
+    return true;
+}
+
+size_t wxRegExImpl::GetMatchCount() const
+{
+    wxCHECK_MSG( IsValid(), 0, _T("must successfully Compile() first") );
+    wxCHECK_MSG( m_nMatches, 0, _T("can't use with wxRE_NOSUB") );
+
+    return m_nMatches;

 }
 
 int wxRegExImpl::Replace(wxString *text,
                          const wxString& replacement,
                          size_t maxMatches) const
 {
 }
 
 int wxRegExImpl::Replace(wxString *text,
                          const wxString& replacement,
                          size_t maxMatches) const
 {

-    wxCHECK_MSG( text, -1, _T("NULL text in wxRegEx::Replace") );
-    wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
+    wxCHECK_MSG( text, wxNOT_FOUND, _T("NULL text in wxRegEx::Replace") );
+    wxCHECK_MSG( IsValid(), wxNOT_FOUND, _T("must successfully Compile() first") );

 
     // the replacement text
     wxString textNew;
 
     // the replacement text
     wxString textNew;


@@ -314,7 +372,7 @@ int wxRegExImpl::Replace(wxString *text,
         // the replacement text anew after each match
         if ( mayHaveBackrefs )
         {
         // the replacement text anew after each match
         if ( mayHaveBackrefs )
         {

-            mayHaveBackrefs = FALSE;
+            mayHaveBackrefs = false;

             textNew.clear();
             textNew.reserve(replacement.length());
 
             textNew.clear();
             textNew.reserve(replacement.length());
 


@@ -355,7 +413,7 @@ int wxRegExImpl::Replace(wxString *text,
                         textNew += wxString(text->c_str() + matchStart + start,
                                             len);
 
                         textNew += wxString(text->c_str() + matchStart + start,
                                             len);
 

-                        mayHaveBackrefs = TRUE;
+                        mayHaveBackrefs = true;

                     }
                 }
                 else // ordinary character
                     }
                 }
                 else // ordinary character


@@ -371,7 +429,7 @@ int wxRegExImpl::Replace(wxString *text,
             // we did have match as Matches() returned true above!
             wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") );
 
             // we did have match as Matches() returned true above!
             wxFAIL_MSG( _T("internal logic error in wxRegEx::Replace") );
 

-            return -1;
+            return wxNOT_FOUND;

         }
 
         matchStart += start;
         }
 
         matchStart += start;


@@ -413,22 +471,22 @@ bool wxRegEx::Compile(const wxString& expr, int flags)
         delete m_impl;
         m_impl = NULL;
 
         delete m_impl;
         m_impl = NULL;
 

-        return FALSE;
+        return false;

     }
 
     }
 

-    return TRUE;
+    return true;

 }
 
 bool wxRegEx::Matches(const wxChar *str, int flags) const
 {
 }
 
 bool wxRegEx::Matches(const wxChar *str, int flags) const
 {

-    wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
+    wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );

 
     return m_impl->Matches(str, flags);
 }
 
 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const
 {
 
     return m_impl->Matches(str, flags);
 }
 
 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const
 {

-    wxCHECK_MSG( IsValid(), FALSE, _T("must successfully Compile() first") );
+    wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );

 
     return m_impl->GetMatch(start, len, index);
 }
 
     return m_impl->GetMatch(start, len, index);
 }


@@ -442,11 +500,18 @@ wxString wxRegEx::GetMatch(const wxString& text, size_t index) const
     return text.Mid(start, len);
 }
 
     return text.Mid(start, len);
 }
 

+size_t wxRegEx::GetMatchCount() const
+{
+    wxCHECK_MSG( IsValid(), 0, _T("must successfully Compile() first") );
+
+    return m_impl->GetMatchCount();
+}
+

 int wxRegEx::Replace(wxString *pattern,
                      const wxString& replacement,
                      size_t maxMatches) const
 {
 int wxRegEx::Replace(wxString *pattern,
                      const wxString& replacement,
                      size_t maxMatches) const
 {

-    wxCHECK_MSG( IsValid(), -1, _T("must successfully Compile() first") );
+    wxCHECK_MSG( IsValid(), wxNOT_FOUND, _T("must successfully Compile() first") );

 
     return m_impl->Replace(pattern, replacement, maxMatches);
 }
 
     return m_impl->Replace(pattern, replacement, maxMatches);
 }