--- parser3/src/main/pa_charset.C	2002/01/25 09:32:07	1.18
+++ parser3/src/main/pa_charset.C	2003/03/21 09:43:48	1.34
@@ -1,13 +1,15 @@
 /** @file
 	Parser: Charset connection implementation.
 
-	Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com)
+	Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexander Petrosyan<paf@design.ru>(http://paf.design.ru)
-
-	$Id: pa_charset.C,v 1.18 2002/01/25 09:32:07 paf Exp $
 */
 
+static const char* IDENT_CHARSET_C="$Date: 2003/03/21 09:43:48 $";
+
 #include "pa_charset.h"
+#include "pa_array.h"
+#include "pa_hash.h"
 
 #ifdef XML
 #include "libxml/encoding.h"
@@ -153,7 +155,7 @@ 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 +193,8 @@ 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 +213,7 @@ static XMLByte xlatOneTo(const XMLCh toX
 			return tables.toTable[midOfs].extCh;
 	} while(lowOfs+1<hiOfs);
 
-    return '?';
+    return not_found;
 }
 
 void Charset::transcode(Pool& pool,
@@ -360,9 +363,18 @@ static int transcodeToUTF8(
     // Return the characters read
     toFillLen = outPtr - toFill;
 
-	return srcPtr==srcEnd?(int)toFillLen:-1;
+	//return srcPtr==srcEnd?(int)toFillLen:-1;
+/*
+xmlCharEncodingInputFunc
+Returns :
+the number of byte written, or -1 by lack of space, or -2 if the transcoding failed. The value of inlen after return is the
+number of octets consumed as the return value is positive, else unpredictiable. The value of outlen after return is the number
+of ocetes consumed.
+*/
+	return 0;
 }
-static size_t transcodeFromUTF8(
+/// @todo digital entites only when xml/html output [at output in html/xml mode, in html part of a letter]
+static int transcodeFromUTF8(
 								const XMLByte *srcData, size_t& srcLen,
 								XMLByte* toFill, size_t& toFillLen,
 								const Charset::Tables& tables) {
@@ -404,7 +416,7 @@ static size_t transcodeFromUTF8(
                      break;
 
             default:
-                throw Exception(0, 0,
+                throw Exception(0,
 					0,
 					"transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes);
         }
@@ -413,10 +425,13 @@ 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);
 	}
@@ -427,14 +442,22 @@ static size_t transcodeFromUTF8(
     // Return the characters read
     toFillLen = outPtr - toFill;
 
-	return srcPtr==srcEnd?(int)toFillLen:-1;
+	//return srcPtr==srcEnd?(int)toFillLen:-1;
+/*
+xmlCharEncodingOutputFunc
+Returns :
+the number of byte written, or -1 by lack of space, or -2 if the transcoding failed. The value of inlen after return is the
+number of octets consumed as the return value is positive, else unpredictiable. The value of outlen after return is the number
+of ocetes consumed.
+*/
+	return 0;
 }
 
 /// @todo not so memory-hungry with prescan
 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 +471,10 @@ 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 +507,7 @@ 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++= '?';
@@ -505,8 +527,8 @@ static int         xml256CharEncodingInp
 													int *inlen, 
 													void *info) {
 	return transcodeToUTF8(
-							in, *(unsigned int*)inlen,
-							out, *(unsigned int*)outlen,
+							in, *(size_t*)inlen,
+							out, *(size_t*)outlen,
 							*(const Charset::Tables *)info);
 }
 
@@ -517,8 +539,8 @@ static int         xml256CharEncodingOut
 													int *inlen, 
 													void *info) {
 	return transcodeFromUTF8(
-							in, *(unsigned int*)inlen,
-							out, *(unsigned int*)outlen,
+							in, *(size_t*)inlen,
+							out, *(size_t*)outlen,
 							*(const Charset::Tables *)info);
 }
 
@@ -540,7 +562,7 @@ 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;
@@ -554,57 +576,127 @@ const char *Charset::transcode_cstr(xmlC
 	int outlen=inlen+1; // max
 	char *out=(char *)malloc(outlen*sizeof(char));
 	
-	int size;
+	int error;
 	if(xmlCharEncodingOutputFunc output=transcoder(0)->output) {
-		size=output(
+		error=output(
 			(unsigned char*)out, &outlen,
 			(const unsigned char*)s, &inlen,
 			transcoder(0)->outputInfo);
-	} else
-		memcpy(out, s, size=inlen);
-	if(size<0)
-		throw Exception(0, 0,
+	} else {
+		memcpy(out, s, outlen=inlen);
+		error=0;
+	}
+	if(error<0)
+		throw Exception(0,
 			0,
-			"transcode_cstr failed (%d)", size);
+			"transcode_cstr failed (%d)", error);
 
-	out[size]=0;
+	out[outlen/*surely would be less then on input*/]=0;
 	return out;
 }
-String& Charset::transcode(xmlChar *s) { 
-	return *NEW String(pool(), transcode_cstr(s)); 
+String& Charset::transcode(xmlChar *s
+#ifndef NO_STRING_ORIGIN
+		, const String *origin
+#endif
+						   ) { 
+	String& result=*NEW String(pool());
+	result.APPEND_CLEAN(transcode_cstr(s), 0/*auto-size*/, origin->origin().file, origin->origin().line);
+	return result; 
 }
 const char *Charset::transcode_cstr(GdomeDOMString *s) { 
 	return s?transcode_cstr(BAD_CAST s->str):"";
 }
-String& Charset::transcode(GdomeDOMString *s) { 
-	return *NEW String(pool(), transcode_cstr(s)); 
+String& Charset::transcode(GdomeDOMString *s
+#ifndef NO_STRING_ORIGIN
+		, const String *origin
+#endif
+						   ) { 
+	String& result=*NEW String(pool());
+	result.APPEND_CLEAN(transcode_cstr(s), 0/*auto-size*/, origin->origin().file, origin->origin().line);
+	return result; 
 }
 
 /// @test less memory using -maybe- xmlParserInputBufferCreateMem
-GdomeDOMString_auto_ptr Charset::transcode_buf(const char *buf, size_t buf_size) { 
-	int outlen=buf_size*6/*max*/+1;
-	unsigned char *out=(unsigned char*)malloc(outlen*sizeof(unsigned char));
-
-	int size;
+xmlChar *Charset::transcode_buf2xchar(const char *buf, size_t buf_size) {
+	unsigned char *out;
+	int outlen;
+	int error;
 	if(xmlCharEncodingInputFunc input=transcoder(0)->input) {
-		size=input(
+		outlen=buf_size*6/*max*/;
+		out=(unsigned char*)malloc((outlen+1)*sizeof(unsigned char));
+		error=input(
 			out, &outlen,
 			(const unsigned char *)buf,  (int *)&buf_size,
 			transcoder(0)->inputInfo);
-	} else
-		memcpy(out, buf, size=buf_size);
+	} else {
+		outlen=buf_size;
+		out=(unsigned char*)malloc((outlen+1)*sizeof(unsigned char));
+		memcpy(out, buf, outlen);
+		error=0;
+	}
 	
-	if(size<0)
-		throw Exception(0, 0,
+	if(error<0)
+		throw Exception(0,
 			0,
-			"transcode_buf failed (%d)", size);
+			"transcode_buf failed (%d)", error);
 
-	out[size]=0;
-	return GdomeDOMString_auto_ptr((gchar*)out);
+	out[outlen/*surely would be less then on input*/]=0;
+	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
+
+String& Charset::transcode(Pool& pool, 
+	const Charset& source_transcoder, 
+	const Charset& dest_transcoder, 
+	const String& src) {
+
+	const char *src_ptr=src.cstr(String::UL_UNSPECIFIED);
+	size_t src_size=strlen(src_ptr);
+
+	const void *dest_ptr;
+	size_t dest_size;
+
+	Charset::transcode(pool, 
+		source_transcoder, (const void*)src_ptr, src_size,
+		dest_transcoder, dest_ptr, dest_size);
+
+	return *new(pool) String(pool, (const char*)dest_ptr, dest_size);
+}
+
+void Charset::transcode(Pool& pool, 
+	const Charset& source_transcoder, 
+	const Charset& dest_transcoder, 
+	Array& src) {
+	for(int i=0; i<src.size(); i++)
+		src.put(i, &transcode(pool, source_transcoder, dest_transcoder, *src.get_string(i)));
+}
+
+#ifndef DOXYGEN
+struct Transcode_pair_info {
+	Pool* pool;
+	const Charset* source_transcoder;
+	const Charset* dest_transcoder;
+};
+#endif
+static void transcode_pair(const Hash::Key& key, Hash::Val *& value, void *raw_info) {
+	Transcode_pair_info& info=*static_cast<Transcode_pair_info*>(raw_info);
+	value=&Charset::transcode(*info.pool, 
+		*info.source_transcoder, 
+		*info.dest_transcoder, 
+		*static_cast<String*>(value));
+}
+void Charset::transcode(Pool& pool, 
+	const Charset& source_transcoder, 
+	const Charset& dest_transcoder, 
+	Hash& src) {
+	Transcode_pair_info info={&pool, &source_transcoder, &dest_transcoder};
+	src.for_each(transcode_pair, &info);
+}