don't write the strings to the stream one char at a time, it's *horribly* slow

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

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index c03e61260143e2e163b053f225d3499c82143f7b..7b480f24c80e18a686e0a84bd3812c4c96f25b48 100644 (file)
--- a/src/common/regex.cpp
+++ b/src/common/regex.cpp
@@ -38,6 +38,11 @@
     #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>
+#endif
+
 #include <regex.h>
 
 #include "wx/regex.h"
@@ -102,6 +107,7 @@ wxRegExImpl::wxRegExImpl()
 {
     m_isCompiled = FALSE;
     m_Matches = NULL;
+    m_nMatches = 0;
 }
 
 wxRegExImpl::~wxRegExImpl()
@@ -121,9 +127,17 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode) const
     {
         len++;
 
+#if wxUSE_UNICODE
+        wxCharBuffer buf(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
     }
     else // regerror() returned 0
     {
@@ -153,7 +167,7 @@ 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 +185,29 @@ 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
+            const wxChar *cptr = expr.c_str();
+            wxChar prev = _T('\0');
+            while ( *cptr != _T('\0') )
+            {
+                // 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('\\')) )
+                {
+                    m_nMatches++;
+                }
+
+                prev = *cptr;
+                cptr++;
+            }
         }
 
         m_isCompiled = TRUE;
@@ -203,7 +238,7 @@ 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 +265,6 @@ 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 +346,7 @@ 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...
                     }