--- parser3/src/include/pa_charset.h 2001/12/26 08:46:12 1.2 +++ parser3/src/include/pa_charset.h 2003/07/24 11:31:21 1.23 @@ -1,18 +1,20 @@ /** @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) 2001-2003 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* IDENT_CHARSET_H="$Date: 2003/07/24 11:31:21 $"; + + #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 { @@ -20,95 +22,184 @@ namespace PCRE { # include "internal.h" }; +#ifdef XML +#include "libxml/encoding.h" +#endif + // defines -#ifdef XML -# include -# include -# define Charset_TransRec XMLTransService::TransRec -#else +#define MAX_CHARSETS 10 + +#define MAX_CHARSET_UNI_CODES 500 + # ifndef XMLCh -# define XMLCh unsigned int + typedef unsigned int XMLCh; # endif # ifndef XMLByte -# define XMLByte unsigned char + typedef unsigned char XMLByte; # endif - struct Charset_TransRec { - XMLCh intCh; - XMLByte extCh; - }; -#endif +// forwards + +class GdomeDOMString_auto_ptr; + +// helpers + +struct Charset_TransRec { + XMLCh intCh; + XMLByte extCh; +}; + +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 StringBody ANAME, const String* afile_spec); - const String& name() const { return fname; } + const StringBody NAME() const { return FNAME; } 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_transcoder, + const Charset& dest_transcoder ); + static String& transcode(const String& src, + const Charset& source_transcoder, + const Charset& dest_transcoder); + + static StringBody transcode(const StringBody 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); + +#ifdef XML + xmlCharEncodingHandler& transcoder(const StringBody NAME); +#endif + public: unsigned char pcre_tables[tables_length]; 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 { + XMLCh fromTable[0x100]; + Charset_TransRec toTable[MAX_CHARSET_UNI_CODES]; + uint toTableSize; + }; private: - const String& fname; + const StringBody 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 StringBody 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 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); + /// converts xmlChar* null-terminated string to parser String + const String& transcode(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); private: - XMLTranscoder *transcoder; + xmlCharEncodingHandler* ftranscoder; #endif }; +#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; + } +}; +#endif + #endif