X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/7553e636a2d186784b2a6148c9557a410599956e..8c36c49496660f0779584c6a318c7416c519f84d:/src/common/regex.cpp

diff --git a/src/common/regex.cpp b/src/common/regex.cpp
index c646c915bb..58c50f399d 100644
--- a/src/common/regex.cpp
+++ b/src/common/regex.cpp
@@ -32,6 +32,7 @@
     #include "wx/string.h"
     #include "wx/log.h"
     #include "wx/intl.h"
+    #include "wx/crt.h"
 #endif //WX_PRECOMP
 
 // FreeBSD, Watcom and DMars require this, CW doesn't have nor need it.
@@ -56,18 +57,22 @@
 #ifdef __REG_NOFRONT
 #   define WXREGEX_USING_BUILTIN
 #   define WXREGEX_IF_NEED_LEN(x) ,x
-#   define WXREGEX_CHAR(x) x
+#   if wxUSE_UNICODE
+#       define WXREGEX_CHAR(x) (x).wc_str()
+#   else
+#       define WXREGEX_CHAR(x) (x).mb_str()
+#   endif
 #else
 #   ifdef HAVE_RE_SEARCH
 #       define WXREGEX_IF_NEED_LEN(x) ,x
 #       define WXREGEX_USING_RE_SEARCH
-#   else        
+#   else
 #       define WXREGEX_IF_NEED_LEN(x)
 #   endif
 #   if wxUSE_UNICODE
 #       define WXREGEX_CONVERT_TO_MB
 #   endif
-#   define WXREGEX_CHAR(x) wxConvertWX2MB(x)
+#   define WXREGEX_CHAR(x) (x).mb_str()
 #   define wx_regfree regfree
 #   define wx_regerror regerror
 #endif
@@ -84,11 +89,22 @@ class wxRegExMatches
 public:
     typedef regmatch_t *match_type;
 
-    wxRegExMatches(size_t n)        { m_matches = new regmatch_t[n]; } 
+    wxRegExMatches(size_t n)        { m_matches = new regmatch_t[n]; }
     ~wxRegExMatches()               { delete [] m_matches; }
 
-    size_t Start(size_t n) const    { return m_matches[n].rm_so; }
-    size_t End(size_t n) const      { return m_matches[n].rm_eo; }
+    // we just use casts here because the fields of regmatch_t struct may be 64
+    // bit but we're limited to size_t in our public API and are not going to
+    // change it because operating on strings longer than 4GB using it is
+    // absolutely impractical anyhow
+    size_t Start(size_t n) const
+    {
+        return wx_truncate_cast(size_t, m_matches[n].rm_so);
+    }
+
+    size_t End(size_t n) const
+    {
+        return wx_truncate_cast(size_t, m_matches[n].rm_eo);
+    }
 
     regmatch_t *get() const         { return m_matches; }
 
@@ -238,7 +254,7 @@ wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const
 
         (void)wx_regerror(errorcode, &m_RegEx, szcmbError, len);
 
-        szError = wxConvertMB2WX(szcmbError);
+        szError = wxConvLibc.cMB2WX(szcmbError);
         delete [] szcmbError;
     }
     else // regerror() returned 0
@@ -282,8 +298,11 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
     // compile it
 #ifdef WXREGEX_USING_BUILTIN
     bool conv = true;
-    int errorcode = wx_re_comp(&m_RegEx, expr, expr.length(), flagsRE);
+    // FIXME-UTF8: use wc_str() after removing ANSI build
+    int errorcode = wx_re_comp(&m_RegEx, expr.c_str(), expr.length(), flagsRE);
 #else
+    // FIXME-UTF8: this is potentially broken, we shouldn't even try it
+    //             and should always use builtin regex library (or PCRE?)
     const wxWX2MBbuf conv = expr.mbc_str();
     int errorcode = conv ? regcomp(&m_RegEx, conv, flagsRE) : REG_BADPAT;
 #endif
@@ -494,7 +513,11 @@ int wxRegExImpl::Replace(wxString *text,
     // note that "^" shouldn't match after the first call to Matches() so we
     // use wxRE_NOTBOL to prevent it from happening
     while ( (!maxMatches || countRepl < maxMatches) &&
+#ifndef WXREGEX_CONVERT_TO_MB
             Matches(textstr + matchStart,
+#else
+            Matches(textstr.data() + matchStart,
+#endif 
                     countRepl ? wxRE_NOTBOL : 0
                     WXREGEX_IF_NEED_LEN(textlen - matchStart)) )
     {
@@ -540,7 +563,12 @@ int wxRegExImpl::Replace(wxString *text,
                     }
                     else
                     {
+#ifndef WXREGEX_CONVERT_TO_MB
                         textNew += wxString(textstr + matchStart + start,
+#else
+                        textNew += wxString(textstr.data() + matchStart +
+					    start,
+#endif
                                             *wxConvCurrent, len);
 
                         mayHaveBackrefs = true;
@@ -570,7 +598,8 @@ int wxRegExImpl::Replace(wxString *text,
 #ifndef WXREGEX_CONVERT_TO_MB
         result.append(*text, matchStart, start);
 #else
-        result.append(wxString(textstr + matchStart, *wxConvCurrent, start));
+        result.append(wxString(textstr.data() + matchStart, *wxConvCurrent, 
+			       start));
 #endif
         matchStart += start;
         result.append(textNew);
@@ -583,7 +612,7 @@ int wxRegExImpl::Replace(wxString *text,
 #ifndef WXREGEX_CONVERT_TO_MB
     result.append(*text, matchStart, wxString::npos);
 #else
-    result.append(wxString(textstr + matchStart, *wxConvCurrent));
+    result.append(wxString(textstr.data() + matchStart, *wxConvCurrent));
 #endif
     *text = result;
 
@@ -623,21 +652,12 @@ bool wxRegEx::Compile(const wxString& expr, int flags)
     return true;
 }
 
-bool wxRegEx::Matches(const wxChar *str, int flags, size_t len) const
-{
-    wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
-    (void)len;
-
-    return m_impl->Matches(WXREGEX_CHAR(str), flags WXREGEX_IF_NEED_LEN(len));
-}
-
-bool wxRegEx::Matches(const wxChar *str, int flags) const
+bool wxRegEx::Matches(const wxString& str, int flags) const
 {
     wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
 
-    return m_impl->Matches(WXREGEX_CHAR(str),
-                           flags
-                           WXREGEX_IF_NEED_LEN(wxStrlen(str)));
+    return m_impl->Matches(WXREGEX_CHAR(str), flags
+                            WXREGEX_IF_NEED_LEN(str.length()));
 }
 
 bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const