--- parser3/src/main/pa_charset.C	2003/11/06 08:22:49	1.39
+++ parser3/src/main/pa_charset.C	2005/08/09 08:14:51	1.52
@@ -1,11 +1,11 @@
 /** @file
 	Parser: Charset connection implementation.
 
-	Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
+	Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexander Petrosyan<paf@design.ru>(http://paf.design.ru)
 */
 
-static const char* IDENT_CHARSET_C="$Date: 2003/11/06 08:22:49 $";
+static const char * const IDENT_CHARSET_C="$Date: 2005/08/09 08:14:51 $";
 
 #include "pa_charset.h"
 #include "pa_charsets.h"
@@ -14,6 +14,8 @@ static const char* IDENT_CHARSET_C="$Dat
 #include "libxml/encoding.h"
 #endif
 
+//#define PA_PATCHED_LIBXML_BACKWARD
+
 // globals
 
 Charset::UTF8CaseTable::Rec UTF8CaseToUpperRecords[]={
@@ -126,7 +128,7 @@ void Charset::load_definition(Request_ch
 
 	// parse cells
 	char *row;
-	while(row=getrow(&data)) {
+	while((row=getrow(&data))) {
 		// remove empty&comment lines
 		if(!*row || *row=='#')
 			continue;
@@ -134,7 +136,7 @@ void Charset::load_definition(Request_ch
 		// char	white-space	digit	hex-digit	letter	word	lowercase	unicode1	unicode2	
 		unsigned int c=0;
 		char *cell;
-		for(int column=0; cell=lsplit(&row, '\t'); column++) {
+		for(int column=0; (cell=lsplit(&row, '\t')); column++) {
 			switch(column) {
 			case 0: c=to_wchar_code(cell); break;
 			// pcre_tables
@@ -391,7 +393,7 @@ static int transcodeFromUTF8(const XMLBy
 		default:
 			throw Exception(0,
 				0,
-				"transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes);
+				"transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); // never
 		}
 		tmpVal-=gUTFOffsets[trailingBytes];
 		
@@ -401,12 +403,14 @@ static int transcodeFromUTF8(const XMLBy
 		if(!(tmpVal & 0xFFFF0000)) {
 			if(XMLByte xlat=xlatOneTo(tmpVal, tables, 0))
 				*outPtr++=xlat;
-			else 
-				outPtr+=sprintf((char *)outPtr, "&#%d;", tmpVal); // &#decimal;
-		} else
-			throw Exception(0,
-			0,
-			"transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal);
+			else {
+				outPtr+=sprintf((char *)outPtr, "&#%u;", tmpVal); // &#decimal;
+			}
+		} else {
+			const XMLByte* recoverPtr=srcPtr-trailingBytes-1;
+			for(uint i=0; i<=trailingBytes; i++)
+				outPtr+=sprintf((char*)outPtr, "%%%02X", *recoverPtr++);
+		}
 	}
 	
 	// Update the bytes eaten
@@ -439,7 +443,7 @@ const String::C Charset::transcodeToUTF8
 		(XMLByte *)src.str, src_length,
 		dest_body, dest_length,
 		tables)<0)
-		throw(0, 0,
+		throw Exception(0,
 			0,
 			"Charset::transcodeToUTF8 buffer overflow");
 
@@ -519,13 +523,19 @@ static void change_case_UTF8(XMLCh src,
 						const Charset::UTF8CaseTable& table) {
 	store_UTF8(change_case_UTF8(src, table), outPtr);
 };
-void change_case_UTF8(const XMLByte* srcData, XMLByte* toFill, 
-							const Charset::UTF8CaseTable& table) {
+void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
+					  XMLByte* toFill, size_t toFillLen,
+					  const Charset::UTF8CaseTable& table) {
 	const XMLByte* srcPtr=srcData;
+	const XMLByte* srcEnd=srcData+srcLen;
 	XMLByte* outPtr=toFill;
+	XMLByte* outEnd=toFill+toFillLen;
+
+	//  We now loop until we either run out of input data, or room to store
+	while ((srcPtr < srcEnd) && (outPtr < outEnd)) {
+		// Get the next leading byte out
+		const XMLByte firstByte =* srcPtr;
 
-	// Get the next leading byte out
-	while (const XMLByte firstByte = *srcPtr) {
 		if(firstByte<= 127) {
 			change_case_UTF8(firstByte, outPtr, table);
 			srcPtr++;
@@ -583,7 +593,7 @@ const String::C Charset::transcodeFromUT
 		(XMLByte *)src.str, src_length,
 		dest_body, dest_length,
 		tables)<0)
-		throw(0, 0,
+		throw Exception(0, 
 			0,
 			"Charset::transcodeFromUTF8 buffer overflow");
 
@@ -618,6 +628,28 @@ const String::C Charset::transcodeToChar
 
 static const Charset::Tables* tables[MAX_CHARSETS];
 
+#ifdef PA_PATCHED_LIBXML_BACKWARD
+
+#define declareXml256ioFuncs(i) \
+	static int xml256CharEncodingInputFunc##i( \
+		unsigned char *out, int *outlen, \
+		const unsigned char *in, int *inlen, void*) { \
+		return transcodeToUTF8( \
+			in, *(size_t*)inlen, \
+			out, *(size_t*)outlen, \
+			*tables[i]); \
+	} \
+	static int xml256CharEncodingOutputFunc##i( \
+		unsigned char *out, int *outlen, \
+		const unsigned char *in, int *inlen, void*) { \
+		return transcodeFromUTF8( \
+			in, *(size_t*)inlen, \
+			out, *(size_t*)outlen, \
+			*tables[i]); \
+	}
+
+#else
+
 #define declareXml256ioFuncs(i) \
 	static int xml256CharEncodingInputFunc##i( \
 		unsigned char *out, int *outlen, \
@@ -636,6 +668,9 @@ static const Charset::Tables* tables[MAX
 			*tables[i]); \
 	}
 
+#endif
+
+
 declareXml256ioFuncs(0)	declareXml256ioFuncs(1)
 declareXml256ioFuncs(2)	declareXml256ioFuncs(3)
 declareXml256ioFuncs(4)	declareXml256ioFuncs(5)
@@ -665,7 +700,7 @@ void Charset::addEncoding(char *name_cst
 			"already allocated %d handlers, no space for new encoding '%s'",
 				MAX_CHARSETS, name_cstr);
 
-	xmlCharEncodingHandler* handler=new(PointerFreeGC) xmlCharEncodingHandler;
+	xmlCharEncodingHandler* handler=new(UseGC) xmlCharEncodingHandler;
 	{
 		handler->name=name_cstr;
 		handler->input=inputFuncs[handlers_count]; 
@@ -696,7 +731,7 @@ String::C Charset::transcode_cstr(xmlCha
 		return String::C("", 0);
 
 	int inlen=strlen((const char*)s);
-	int outlen=inlen; // max
+	int outlen=inlen*6/*strlen("&#255;")*/; // max
 #ifndef NDEBUG
 	int saved_outlen=outlen;
 #endif
@@ -706,7 +741,11 @@ String::C Charset::transcode_cstr(xmlCha
 	if(xmlCharEncodingOutputFunc output=transcoder(FNAME).output) {
 		error=output(
 			(unsigned char*)out, &outlen,
-			(const unsigned char*)s, &inlen);
+			(const unsigned char*)s, &inlen
+#ifdef PA_PATCHED_LIBXML_BACKWARD
+			,0
+#endif
+			);
 	} else {
 		memcpy(out, s, outlen=inlen);
 		error=0;
@@ -741,14 +780,18 @@ xmlChar* Charset::transcode_buf2xchar(co
 	int saved_outlen;
 #endif
 	if(xmlCharEncodingInputFunc input=transcoder(FNAME).input) {
-		outlen=buf_size*6/*max*/;
+		outlen=buf_size*6/*max UTF8 bytes per char*/;
 #ifndef NDEBUG
 		saved_outlen=outlen;
 #endif
 		out=(xmlChar*)xmlMalloc(outlen+1);
 		error=input(
 			out, &outlen,
-			(const unsigned char*)buf, (int*)&buf_size);
+			(const unsigned char*)buf, (int*)&buf_size
+#ifdef PA_PATCHED_LIBXML_BACKWARD
+			,0
+#endif
+			);
 	} else {
 		outlen=buf_size;
 #ifndef NDEBUG
@@ -818,7 +861,7 @@ struct Transcode_pair_info {
 	const Charset* dest_transcoder;
 };
 #endif
-static void transcode_pair(const String::Body akey, 
+static void transcode_pair(const String::Body /*akey*/, 
 			 String::Body& avalue, 
 			 Transcode_pair_info* info) {
 	avalue=Charset::transcode(avalue,