--- parser3/src/include/pa_charset.h 2001/12/26 08:46:12 1.2 +++ parser3/src/include/pa_charset.h 2009/04/10 11:31:34 1.41 @@ -1,62 +1,95 @@ /** @file Parser: Charset connection decl. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) - - $Id: pa_charset.h,v 1.2 2001/12/26 08:46:12 paf Exp $ + Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ #ifndef PA_CHARSET_H #define PA_CHARSET_H -#include "pa_pool.h" +static const char * const IDENT_CHARSET_H="$Date: 2009/04/10 11:31:34 $"; + + #include "pa_exception.h" #include "pa_common.h" +#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 "../lib/pcre/config.h" +#include "pcre_internal.h" + +#ifdef XML +#include "libxml/encoding.h" +#endif // defines -#ifdef XML -# include -# include -# define Charset_TransRec XMLTransService::TransRec -#else -# ifndef XMLCh -# define XMLCh unsigned int -# endif -# ifndef XMLByte -# define XMLByte unsigned char -# endif - - struct Charset_TransRec { - XMLCh intCh; - XMLByte extCh; - }; +#define MAX_CHARSETS 10 + +#define MAX_CHARSET_UNI_CODES 500 + +#ifndef XMLCh + typedef unsigned int XMLCh; +#endif +#ifndef XMLByte + typedef unsigned char XMLByte; #endif +// helpers + +typedef Hash HashStringString; + /** charset holds name & transcode tables - registers Xerces transcoders + registers libxml transcoders */ -class Charset : public Pooled { +class Charset: public PA_Object { public: - Charset(Pool& apool, const String& aname, const String *request_file_spec); - Charset::~Charset(); + Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec); - const String& 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 void transcode(Pool& pool, - const Charset& source_transcoder, const void *source_body, size_t source_content_length, - const Charset& dest_transcoder, const void *& dest_body, size_t& dest_content_length - ); + static String::C transcode(const String::C src, + const Charset& source_charset, + const Charset& dest_charset); + + static String& transcode(const String& src, + const Charset& source_transcoder, + const Charset& dest_transcoder); + + static String::Body transcode(const String::Body src, + const Charset& source_transcoder, + const Charset& dest_transcoder); + + static void transcode(ArrayString& src, + const Charset& source_transcoder, + const Charset& dest_transcoder); + + static void transcode(HashStringString& src, + 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); + + void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found); + +#ifdef XML + xmlCharEncodingHandler& transcoder(const String::Body NAME); +#endif public: @@ -64,51 +97,83 @@ public: private: - void loadDefinition(const String& request_file_spec); + void load_definition(Request_charsets& charsets, const String& afile_spec); void sort_ToTable(); - XMLByte xlatOneTo(const XMLCh toXlat) const; - void transcodeToUTF8(Pool& pool, - const void *source_body, size_t source_content_length, - const void *& dest_body, size_t& dest_content_length) const; - void transcodeFromUTF8(Pool& pool, - const void *source_body, size_t source_content_length, - const void *& dest_body, size_t& dest_content_length) const; - void transcodeToCharset(Pool& pool, - const Charset& dest_transcoder, - const void *source_body, size_t source_content_length, - const void *& dest_body, size_t& dest_content_length) const; + const String::C transcodeToUTF8(const String::C src) const; + const String::C transcodeFromUTF8(const String::C src) const; + + const String::C transcodeToCharset(const String::C src, + const Charset& dest_transcoder) const; + +public: + + struct Tables { + struct Rec { + XMLCh intCh; + XMLByte extCh; + }; + + XMLCh fromTable[0x100]; + Rec toTable[MAX_CHARSET_UNI_CODES]; + uint toTableSize; + }; + + struct UTF8CaseTable { + struct Rec { + XMLCh from, to; + }; + + uint size; + Rec* records; + }; private: - const String& fname; + const String::Body FNAME; + char* FNAME_CSTR; bool fisUTF8; - XMLCh fromTable[0x100]; - Charset_TransRec *toTable; - uint toTableSize; + Tables tables; #ifdef XML private: - void addEncoding(const char *name_cstr); - void initTranscoder(const String *source, const char *name_cstr); - + void addEncoding(char* name_cstr); + void initTranscoder(const String::Body name, const char* name_cstr); + public: - /// converts Xalan string to char * - const char *transcode_cstr(const XalanDOMString& s); - /// converts Xalan string to parser String - String& transcode(const XalanDOMString& s); - /// converts char * to Xalan string - std::auto_ptr transcode_buf(const char *buf, size_t buf_size); - /// converts parser String to Xalan string - std::auto_ptr transcode(const String& s); + /// converts xmlChar* null-terminated string to char* + String::C transcode_cstr(const xmlChar* s); + /// converts xmlChar* null-terminated string to parser String + 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 parser String to xmlChar* + xmlChar* transcode(const String& s); + /// converts parser String::Body to xmlChar* + xmlChar* transcode(const String::Body s); private: - XMLTranscoder *transcoder; + xmlCharEncodingHandler* ftranscoder; #endif }; + +// 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); + #endif