--- parser3/src/main/pa_charset.C	2002/02/27 16:37:38	1.21
+++ parser3/src/main/pa_charset.C	2002/07/30 14:28:47	1.26
@@ -4,7 +4,7 @@
 	Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexander Petrosyan<paf@design.ru>(http://paf.design.ru)
 
-	$Id: pa_charset.C,v 1.21 2002/02/27 16:37:38 paf Exp $
+	$Id: pa_charset.C,v 1.26 2002/07/30 14:28:47 paf Exp $
 */
 
 #include "pa_charset.h"
@@ -153,7 +153,7 @@ void Charset::loadDefinition(const Strin
 			case 8:
 				// charset
 				if(tables.toTableSize>MAX_CHARSET_UNI_CODES)
-					throw Exception(0, 0,
+					throw Exception("parser.runtime",
 						&request_file_spec,
 						"charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES);
 
@@ -191,7 +191,8 @@ void Charset::sort_ToTable() {
 }
 
 static XMLByte xlatOneTo(const XMLCh toXlat,
-						 const Charset::Tables& tables) {
+						 const Charset::Tables& tables,
+						 XMLByte not_found) {
     unsigned int    lowOfs = 0;
     unsigned int    hiOfs = tables.toTableSize - 1;
     XMLByte         curByte = 0;
@@ -210,7 +211,7 @@ static XMLByte xlatOneTo(const XMLCh toX
 			return tables.toTable[midOfs].extCh;
 	} while(lowOfs+1<hiOfs);
 
-    return '?';
+    return not_found;
 }
 
 void Charset::transcode(Pool& pool,
@@ -362,6 +363,7 @@ static int transcodeToUTF8(
 
 	return srcPtr==srcEnd?(int)toFillLen:-1;
 }
+/// @todo digital entites only when xml/html output [at output in html/xml mode, in html part of a letter]
 static size_t transcodeFromUTF8(
 								const XMLByte *srcData, size_t& srcLen,
 								XMLByte* toFill, size_t& toFillLen,
@@ -404,7 +406,7 @@ static size_t transcodeFromUTF8(
                      break;
 
             default:
-                throw Exception(0, 0,
+                throw Exception(0,
 					0,
 					"transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes);
         }
@@ -413,10 +415,13 @@ static size_t transcodeFromUTF8(
         //  If it will fit into a single char, then put it in. Otherwise
         //  fail [*encode it as a surrogate pair. If its not valid, use the
         //  replacement char.*]
-        if(!(tmpVal & 0xFFFF0000))
-            *outPtr++= xlatOneTo(tmpVal, tables);
-		else
-			throw Exception(0, 0,
+        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);
 	}
@@ -434,7 +439,7 @@ static size_t transcodeFromUTF8(
 void Charset::transcodeToUTF8(Pool& pool,
 								 const void *source_body, size_t source_content_length,
 								 const void *& adest_body, size_t& dest_content_length) const {
-	dest_content_length=source_content_length*6/*so that surly enough*/;
+	dest_content_length=source_content_length*6/*so that surly enough, max utf8 seq len=6*/;
 	XMLByte *dest_body=(XMLByte*)pool.malloc(dest_content_length);
 
 	if(::transcodeToUTF8(
@@ -448,11 +453,10 @@ void Charset::transcodeToUTF8(Pool& pool
 	// return
 	adest_body=dest_body;
 }
-/// @test buf overflow
 void Charset::transcodeFromUTF8(Pool& pool,
 								   const void *source_body, size_t source_content_length,
 								   const void *& adest_body, size_t& dest_content_length) const {
-	dest_content_length=source_content_length/*surly enough*/;
+	dest_content_length=source_content_length*6/*so that surly enough, "&#255;" has max ratio */;
 	XMLByte *dest_body=(XMLByte*)pool.malloc(dest_content_length);
 
 	if(::transcodeFromUTF8(
@@ -485,7 +489,7 @@ void Charset::transcodeToCharset(Pool& p
 		for(XMLByte* outPtr=dest_body; srcPtr<srcEnd; srcPtr++) {
 			XMLCh curVal = tables.fromTable[*srcPtr];
 			if(curVal) 
-				*outPtr++=xlatOneTo(curVal, dest_charset.tables);
+				*outPtr++=xlatOneTo(curVal, dest_charset.tables, '?');
 			else {
 				// use the replacement character
 				*outPtr++= '?';
@@ -540,7 +544,7 @@ void Charset::initTranscoder(const Strin
 
 xmlCharEncodingHandler *Charset::transcoder(const String *source) {
 	if(!ftranscoder)
-		throw Exception(0, 0,
+		throw Exception("parser.runtime",
 			source,
 			"unsupported encoding");
 	return ftranscoder;
@@ -563,7 +567,7 @@ const char *Charset::transcode_cstr(xmlC
 	} else
 		memcpy(out, s, size=inlen);
 	if(size<0)
-		throw Exception(0, 0,
+		throw Exception(0,
 			0,
 			"transcode_cstr failed (%d)", size);
 
@@ -581,7 +585,7 @@ String& Charset::transcode(GdomeDOMStrin
 }
 
 /// @test less memory using -maybe- xmlParserInputBufferCreateMem
-GdomeDOMString_auto_ptr Charset::transcode_buf(const char *buf, size_t buf_size) { 
+xmlChar *Charset::transcode_buf2xchar(const char *buf, size_t buf_size) {
 	int outlen=buf_size*6/*max*/+1;
 	unsigned char *out=(unsigned char*)malloc(outlen*sizeof(unsigned char));
 
@@ -595,16 +599,19 @@ GdomeDOMString_auto_ptr Charset::transco
 		memcpy(out, buf, size=buf_size);
 	
 	if(size<0)
-		throw Exception(0, 0,
+		throw Exception(0,
 			0,
 			"transcode_buf failed (%d)", size);
 
 	out[size]=0;
-	return GdomeDOMString_auto_ptr((gchar*)out);
+	return (xmlChar *)out;
+}
+GdomeDOMString_auto_ptr Charset::transcode_buf2dom(const char *buf, size_t buf_size) { 
+	return GdomeDOMString_auto_ptr((gchar*)transcode_buf2xchar(buf, buf_size));
 }
 GdomeDOMString_auto_ptr Charset::transcode(const String& s) { 
 	const char *cstr=s.cstr(String::UL_UNSPECIFIED);
 
-	return transcode_buf(cstr, strlen(cstr)); 
+	return transcode_buf2dom(cstr, strlen(cstr)); 
 }
 #endif