applied patch 890642: wxRE_ADVANCED flag and docs

git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@25868 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775

build/bakefiles/regex.bkl

@@ -10,7 +10,7 @@
         <set var="wxUSE_REGEX">builtin</set>
     </if>
     <set var="LIB_REGEX">
-        <if cond="wxUSE_REGEX=='builtin'">wxregex$(WXDEBUGFLAG)$(HOST_SUFFIX)</if>
+        <if cond="wxUSE_REGEX=='builtin'">wxregex$(WXUNICODEFLAG)$(WXDEBUGFLAG)$(HOST_SUFFIX)</if>
     </set>
     <set var="INC_REGEX">
         <if cond="wxUSE_REGEX=='builtin'">$(TOP_SRCDIR)src/regex</if>
@@ -18,11 +18,12 @@
 
     <lib id="wxregex" template="msvc_setup_h,3rdparty_lib"
          cond="wxUSE_REGEX=='builtin' and BUILDING_LIB=='1'">
+        <libname>$(id)$(WXUNICODEFLAG)$(WXDEBUGFLAG)$(HOST_SUFFIX)</libname>
         <include cond="FORMAT!='autoconf'">$(TOP_SRCDIR)include</include>
         <include cond="FORMAT!='autoconf'">$(SETUPHDIR)</include>
         <define>$(UNICODE_DEFINE)</define>
         <dirname>$(LIBDIRNAME)</dirname>
-        <cflags-borland>-w-8004 -w-8012 -w-8057 -w-8066</cflags-borland> 
+        <cflags-borland>-w-8008 -w-8012 -w-8057 -w-8064 -w-8066 -w-8070</cflags-borland>
         <sources>
             src/regex/regcomp.c
             src/regex/regexec.c

include/wx/features.h

@@ -56,5 +56,13 @@
     #undef wxHAS_CRASH_REPORT
 #endif
 
+/*  wxRE_ADVANCED is not always available, depending on regex library used
+ *  (it's unavailable only if compiling via configure against system library) */
+#ifndef WX_NO_REGEX_ADVANCED
+    #define wxHAS_REGEX_ADVANCED
+#else
+    #undef wxHAS_REGEX_ADVANCED
+#endif
+
 #endif /*  _WX_FEATURES_H_ */
 

include/wx/regex.h

@@ -29,8 +29,13 @@ class WXDLLIMPEXP_BASE wxString;
 // flags for regex compilation: these can be used with Compile()
 enum
 {
-    // use extended regex syntax (default)
+    // use extended regex syntax
     wxRE_EXTENDED = 0,
+    
+    // use advanced RE syntax (built-in regex only)
+#ifdef wxHAS_REGEX_ADVANCED
+    wxRE_ADVANCED = 1,
+#endif
 
     // use basic RE syntax
     wxRE_BASIC    = 2,

src/common/regex.cpp

@@ -73,7 +73,7 @@ public:
 
 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()
@@ -131,8 +131,19 @@ wxRegExImpl::~wxRegExImpl()
     Free();
 }
 
-wxString wxRegExImpl::GetErrorMsg(int errorcode) const
+wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const
 {
+#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
@@ -158,14 +169,25 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
 {
     Reinit();
 
-    // translate our flags to regcomp() ones
+#ifdef WX_NO_REGEX_ADVANCED
-    wxASSERT_MSG( !(flags &
+#   define FLAVORS wxRE_BASIC
-                        ~(wxRE_BASIC | wxRE_ICASE | wxRE_NOSUB | wxRE_NEWLINE)),
+#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") );
 
+    // translate our flags to regcomp() ones
     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 )
@@ -173,14 +195,19 @@ bool wxRegExImpl::Compile(const wxString& expr, int flags)
     if ( flags & wxRE_NEWLINE )
         flagsRE |= REG_NEWLINE;
 
+    // compile it
+#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 )
-    int errorcode = regcomp(&m_RegEx, expr, flagsRE);
-
-     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;
     }
@@ -250,7 +277,13 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
     }
 
     // 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 )
     {
@@ -261,7 +294,7 @@ bool wxRegExImpl::Matches(const wxChar *str, int flags) const
         default:
             // an error occured
             wxLogError(_("Failed to match '%s' in regular expression: %s"),
-                       str, GetErrorMsg(rc).c_str());
+                       str, GetErrorMsg(rc, !conv).c_str());
             // fall through
 
         case REG_NOMATCH:

src/regex/splice.sh

@@ -0,0 +1,69 @@
+#!/bin/sh
+#############################################################################
+# Name:        splice
+# Purpose:     Splice a marked section of regcustom.h into regex.h
+# Author:      Mike Wetherell
+# RCS-ID:      $Id$
+# Copyright:   (c) 2004 Mike Wetherell
+# Licence:     wxWindows licence
+#############################################################################
+
+#
+# Works by greping for the marks then passing their line numbers to sed. Which
+# is slighly the long way round, but allows some error checking.
+#
+
+SRC=regcustom.h
+DST=regex.h
+MARK1='^/\* --- begin --- \*/$'
+MARK2='^/\* --- end --- \*/$'
+PROG=`basename $0`
+TMP=$DST.tmp
+
+# findline(pattern, file)
+# Prints the line number of the 1st line matching the pattern in the file
+#
+findline() {
+    if ! LINE=`grep -n -- "$1" "$2"`; then
+        echo "$PROG: marker '$1' not found in '$2'" >&2
+        return 1
+    fi
+    echo $LINE | sed -n '1s/[^0-9].*//p'    # take just the line number
+}
+
+# findmarkers([out] line1, [out] line2, pattern1, pattern2, file)
+# Returns (via the variables named in the 1st two parameters) the line
+# numbers of the lines matching the patterns in file. Checks pattern1 comes
+# before pattern2.
+#
+findmarkers() {
+    if ! LINE1=`findline "$3" "$5"` || ! LINE2=`findline "$4" "$5"`; then
+        return 1
+    fi
+    if [ $LINE1 -ge $LINE2 ]; then
+        echo "$PROG: marker '$3' not before '$4' in '$5'" >&2
+        return 1
+    fi
+    eval $1=$LINE1
+    eval $2=$LINE2
+}
+
+# find markers
+#
+if findmarkers SRCLINE1 SRCLINE2 "$MARK1" "$MARK2" $SRC &&
+   findmarkers DSTLINE1 DSTLINE2 "$MARK1" "$MARK2" $DST
+then
+    # do splice
+    #
+    if (sed $DSTLINE1,\$d $DST &&
+        sed -n $SRCLINE1,${SRCLINE2}p $SRC &&
+        sed 1,${DSTLINE2}d $DST) > $TMP
+    then
+        mv $TMP $DST
+        exit 0
+    else
+        rm $TMP
+    fi
+fi
+
+exit 1