--- parser3/src/include/pa_charset.h	2003/07/24 11:31:21	1.23
+++ parser3/src/include/pa_charset.h	2012/06/09 12:13:50	1.49
@@ -1,14 +1,14 @@
 /** @file
 	Parser: Charset connection decl.
 
-	Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
+	Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com)
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 */
 
 #ifndef PA_CHARSET_H
 #define PA_CHARSET_H
 
-static const char* IDENT_CHARSET_H="$Date: 2003/07/24 11:31:21 $";
+#define IDENT_PA_CHARSET_H "$Id: pa_charset.h,v 1.49 2012/06/09 12:13:50 moko Exp $"
 
 
 #include "pa_exception.h"
@@ -16,11 +16,10 @@ static const char* IDENT_CHARSET_H="$Dat
 #include "pa_hash.h"
 #include "pa_array.h"
 
-// hiding into namespace so to avoid stupid conflict
-namespace PCRE {
+// pcre.h must be included BEFORE pcre_internal.h
 #include "pcre.h"
-#	include "internal.h"
-};
+// have to call config.h for using pcre_internal.h as well
+#include "pcre_internal.h"
 
 #ifdef XML
 #include "libxml/encoding.h"
@@ -32,25 +31,16 @@ namespace PCRE {
 
 #define MAX_CHARSET_UNI_CODES 500
 
-#	ifndef XMLCh 
-		typedef unsigned int XMLCh;
-#	endif
-#	ifndef XMLByte
-		typedef unsigned char XMLByte;
-#	endif
-
-// forwards
-
-class GdomeDOMString_auto_ptr;
+#ifndef XMLCh 
+	typedef unsigned int XMLCh;
+#endif
+#ifndef XMLByte
+	typedef unsigned char XMLByte;
+#endif
 
 // helpers
 
-struct Charset_TransRec {
-	XMLCh intCh;
-	XMLByte extCh;
-};
-
-typedef Hash<const StringBody, StringBody> HashStringString;
+typedef HashString<String::Body> HashStringString;
 
 /**	charset holds name & transcode tables 
 	registers libxml transcoders
@@ -58,22 +48,22 @@ typedef Hash<const StringBody, StringBod
 class Charset: public PA_Object {
 public:
 
-	Charset(Request_charsets* charsets, const StringBody ANAME, const String* afile_spec);
+	Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec);
 	
-	const StringBody NAME() const { return FNAME; }
+	const String::Body NAME() const { return FNAME; }
+	const char* NAME_CSTR() const { return FNAME_CSTR; }
 
 	bool isUTF8() const { return fisUTF8; }
 
 	static String::C transcode(const String::C src,
-		const Charset& source_transcoder, 
-		const Charset& dest_transcoder
-	);
+		const Charset& source_charset, 
+		const Charset& dest_charset);
 
 	static String& transcode(const String& src,
 		const Charset& source_transcoder, 
 		const Charset& dest_transcoder);
 
-	static StringBody transcode(const StringBody src,
+	static String::Body transcode(const String::Body src,
 		const Charset& source_transcoder, 
 		const Charset& dest_transcoder);
 
@@ -85,8 +75,24 @@ public:
 		const Charset& source_transcoder, 
 		const Charset& dest_transcoder);
 
+	static String::C escape(const String::C src,
+		const Charset& source_charset);
+	static String::Body escape(const String::Body src,
+		const Charset& source_charset);
+	static String& escape(const String& src,
+		const Charset& source_charset);
+
+	static String::C escape_JSON(const String::C src,
+		const Charset& source_charset);
+	static String::Body escape_JSON(const String::Body src,
+		const Charset& source_charset);
+	static String& escape_JSON(const String& src,
+		const Charset& source_charset);
+
+	void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found);
+
 #ifdef XML
-	xmlCharEncodingHandler& transcoder(const StringBody NAME);
+	xmlCharEncodingHandler& transcoder(const String::Body NAME);
 #endif
 
 public:
@@ -107,44 +113,64 @@ private:
 public:
 
 	struct Tables {
+		struct Rec {
+			XMLCh intCh;
+			XMLByte extCh;
+		};
+
 		XMLCh fromTable[0x100];
-		Charset_TransRec toTable[MAX_CHARSET_UNI_CODES];
+		Rec toTable[MAX_CHARSET_UNI_CODES];
 		uint toTableSize;
 	};
 
+	struct UTF8CaseTable {
+		struct Rec {
+			XMLCh from, to;
+		};
+
+		uint size;
+		Rec* records;
+	};
+
 private:
 
-	const StringBody FNAME;
+	const String::Body FNAME;
 	char* FNAME_CSTR;
 	bool fisUTF8;
 	Tables tables;
 
+	static size_t calc_escaped_length_UTF8(XMLByte* src, size_t src_length);
+	static size_t calc_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
+	static size_t calc_escaped_length(const String::C src, const Charset& source_charset);
+	static size_t escape_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
+	static size_t escape(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
+
+	static size_t calc_JSON_escaped_length_UTF8(XMLByte* src, size_t src_length);
+	static size_t calc_JSON_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
+	static size_t calc_JSON_escaped_length(const String::C src, const Charset& source_charset);
+	static size_t escape_JSON_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
+	static size_t escape_JSON(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
+
 #ifdef XML
 
 private:
 	void addEncoding(char* name_cstr);
-	void initTranscoder(const StringBody name, const char* name_cstr);
-	
+	void initTranscoder(const String::Body name, const char* name_cstr);
+
 public:
-	/// converts GdomeDOMString string to char* 
-	String::C transcode_cstr(GdomeDOMString* s);
-	/// converts GdomeDOMString string to parser String
-	const String& transcode(GdomeDOMString* s);
 	/// converts xmlChar* null-terminated string to char* 
-	String::C transcode_cstr(xmlChar* s);
+	String::C transcode_cstr(const xmlChar* s);
 	/// converts xmlChar* null-terminated string to parser String
-	const String& transcode(xmlChar* s);
+	const String& transcode(const xmlChar* s);
 
 	/** converts sized char*  to xmlChar*
 		@returns xmlChar*  WHICH CALLER SHOULD FREE
 	*/
 	xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size);
-	/// converts char*  to GdomeDOMString
-	GdomeDOMString_auto_ptr transcode_buf2dom(const char* buf, size_t buf_size);
-	/// converts parser String to GdomeDOMString
-	GdomeDOMString_auto_ptr transcode(const String& s);
-	/// converts parser StringBody to GdomeDOMString
-	GdomeDOMString_auto_ptr transcode(const StringBody s);
+	/// converts parser String to xmlChar*
+	xmlChar* transcode(const String& s);
+	/// converts parser String::Body to xmlChar*
+	xmlChar* transcode(const String::Body s);
 
 private:
 
@@ -154,52 +180,35 @@ private:
 
 };
 
-#ifdef XML
-/// Auto-object used to track GdomeDOMString usage
-class GdomeDOMString_auto_ptr {
-	GdomeDOMString* fstring;
-public:
-	/// frees astring afterwards!!!
-	explicit GdomeDOMString_auto_ptr(xmlChar* astring) : fstring(gdome_str_mkref_xml(astring)) {}
-	explicit GdomeDOMString_auto_ptr(GdomeDOMString* astring=0) : fstring(astring) {
-		// not ref-ing, owning
-	}
-	~GdomeDOMString_auto_ptr() {
-		if(fstring)
-			gdome_str_unref(fstring);
-	}
-/*	GdomeDOMString* get() {
-		return fstring;
-	}*/
-	GdomeDOMString* use() {
-		if(fstring)
-			gdome_str_ref(fstring);
-		return fstring;
-	}
-	GdomeDOMString* operator->() {
-		return fstring;
-	}
-/*	GdomeDOMString& operator*() {
-		return* fstring;
-	}*/
-
-	// copying
-	GdomeDOMString_auto_ptr(const GdomeDOMString_auto_ptr& src) : fstring(src.fstring) {
-		gdome_str_ref(fstring);
-	}
-	GdomeDOMString_auto_ptr& operator =(const GdomeDOMString_auto_ptr& src) {
-		if(this == &src)
-			return* this;
-
-		if(fstring)
-			gdome_str_unref(fstring);
-		fstring=src.fstring;
-		if(fstring)
-			gdome_str_ref(fstring);
 
-		return* this;
-	}
+// externs
+
+extern Charset::UTF8CaseTable UTF8CaseToUpper;
+extern Charset::UTF8CaseTable UTF8CaseToLower;
+void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
+					XMLByte* toFill, size_t toFillLen,
+					const Charset::UTF8CaseTable& table);
+size_t getUTF8BytePos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t charPos/*position in characters*/);
+size_t getUTF8CharPos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t bytePos/*position in bytes*/);
+size_t lengthUTF8(const XMLByte* srcBegin, const XMLByte* srcEnd);
+unsigned int lengthUTF8Char(const XMLByte c);
+
+
+class UTF8_string_iterator {
+	public:
+		UTF8_string_iterator(const String& astring): fsrcPtr((XMLByte*)astring.cstr()), fsrcEnd(fsrcPtr + astring.length()) {}
+		UTF8_string_iterator(XMLByte* asrcPtr, size_t length): fsrcPtr(asrcPtr), fsrcEnd(fsrcPtr + length) {}
+
+		bool has_next();
+		XMLCh next() { return fUTF8Char; }
+		XMLByte getFirstByte(){ return ffirstByte; }
+		size_t getCharSize(){ return fcharSize; }
+	private:
+		const XMLByte* fsrcPtr;
+		const XMLByte* fsrcEnd;
+		size_t fcharSize;
+		XMLByte ffirstByte;
+		XMLCh fUTF8Char;
 };
-#endif
 
 #endif