--- parser3/src/include/pa_charset.h 2002/01/14 17:48:56 1.8 +++ parser3/src/include/pa_charset.h 2004/02/11 15:33:13 1.29 @@ -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 1.8 2002/01/14 17:48:56 paf Exp $ + Copyright (c) 2001-2004 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: 2004/02/11 15:33:13 $"; + + #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 { @@ -22,142 +24,184 @@ namespace PCRE { #ifdef XML #include "libxml/encoding.h" +#include "gdome.h" #endif // defines -#ifdef XML -/*# include -# include -# define Charset_TransRec XMLTransService::TransRec -*/ +#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; - }; +// forwards -#else -# ifndef XMLCh -# define XMLCh unsigned int -# endif -# ifndef XMLByte -# define XMLByte unsigned char -# endif +class GdomeDOMString_auto_ptr; - struct Charset_TransRec { - XMLCh intCh; - XMLByte extCh; - }; -#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; } 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 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); + +#ifdef XML + xmlCharEncodingHandler& transcoder(const String::Body 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(); - 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]; - Charset_TransRec *toTable; + 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; Tables tables; #ifdef XML private: - void addEncoding(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 GdomeDOMString string to char * - const char *transcode_cstr(GdomeDOMString *s); + /// converts GdomeDOMString string to char* + String::C transcode_cstr(GdomeDOMString* s); /// converts GdomeDOMString string to parser String - String& transcode(GdomeDOMString *s); - /// converts xmlChar* null-terminated string to char * - const char *transcode_cstr(xmlChar *s); + 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 - String& transcode(xmlChar *s); - /// converts char * to GdomeDOMString - GdomeDOMString_auto_ptr transcode_buf(const char *buf, size_t buf_size); + 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 String::Body to GdomeDOMString + GdomeDOMString_auto_ptr transcode(const String::Body s); private: - xmlCharEncodingHandler *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); + + +#ifdef XML /// Auto-object used to track GdomeDOMString usage class GdomeDOMString_auto_ptr { - GdomeDOMString *fstring; + GdomeDOMString* fstring; public: - explicit GdomeDOMString_auto_ptr(gchar *astring) : fstring(gdome_str_mkref(astring)) {} - explicit GdomeDOMString_auto_ptr(GdomeDOMString *astring) : fstring(astring) { + /// 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() { - gdome_str_unref(fstring); + if(fstring) + gdome_str_unref(fstring); } - GdomeDOMString* get() { +/* GdomeDOMString* get() { + return fstring; + }*/ + GdomeDOMString* use() { + if(fstring) + gdome_str_ref(fstring); return fstring; } GdomeDOMString* operator->() { return fstring; } - GdomeDOMString& operator*() { - return *fstring; - } +/* GdomeDOMString& operator*() { + return* fstring; + }*/ // copying GdomeDOMString_auto_ptr(const GdomeDOMString_auto_ptr& src) : fstring(src.fstring) { @@ -165,14 +209,17 @@ public: } GdomeDOMString_auto_ptr& operator =(const GdomeDOMString_auto_ptr& src) { if(this == &src) - return *this; + return* this; - gdome_str_unref(fstring); + if(fstring) + gdome_str_unref(fstring); fstring=src.fstring; - gdome_str_ref(fstring); + if(fstring) + gdome_str_ref(fstring); - return *this; + return* this; } }; +#endif #endif