--- parser3/src/include/pa_charset.h 2003/04/10 16:11:02 1.20.2.18.2.10 +++ parser3/src/include/pa_charset.h 2005/08/09 08:14:49 1.31 @@ -1,18 +1,20 @@ /** @file Parser: Charset connection decl. - Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ #ifndef PA_CHARSET_H #define PA_CHARSET_H -static const char* IDENT_CHARSET_H="$Date: 2003/04/10 16:11:02 $"; +static const char * const IDENT_CHARSET_H="$Date: 2005/08/09 08:14:49 $"; #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 { @@ -22,6 +24,7 @@ namespace PCRE { #ifdef XML #include "libxml/encoding.h" +#include "gdome.h" #endif // defines @@ -43,10 +46,7 @@ class GdomeDOMString_auto_ptr; // helpers -struct Charset_TransRec { - XMLCh intCh; - XMLByte extCh; -}; +typedef Hash HashStringString; /** charset holds name & transcode tables registers libxml transcoders @@ -54,9 +54,10 @@ struct Charset_TransRec { 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; } @@ -65,8 +66,24 @@ public: 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 StringBody NAME); + xmlCharEncodingHandler& transcoder(const String::Body NAME); #endif public: @@ -87,14 +104,28 @@ 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; @@ -103,7 +134,7 @@ private: 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* @@ -119,22 +150,12 @@ public: @returns xmlChar* WHICH CALLER SHOULD FREE */ xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size); - /** converts sized char* to gchar* - @returns gchar* WHICH CALLER SHOULD FREE - */ - gchar* transcode_buf2gchar(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: - - typedef void* (*transcode_buf_malloc_func)(size_t size); - void* transcode_buf2mchar(transcode_buf_malloc_func malloc_func, - const char* buf, size_t buf_size); + /// converts parser String::Body to GdomeDOMString + GdomeDOMString_auto_ptr transcode(const String::Body s); private: @@ -144,13 +165,23 @@ private: }; + +// 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; public: /// frees astring afterwards!!! - explicit GdomeDOMString_auto_ptr(gchar* astring) : fstring(gdome_str_mkref_own(astring)) {} + 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 } @@ -169,9 +200,9 @@ public: GdomeDOMString* operator->() { return fstring; } - GdomeDOMString& operator*() { +/* GdomeDOMString& operator*() { return* fstring; - } + }*/ // copying GdomeDOMString_auto_ptr(const GdomeDOMString_auto_ptr& src) : fstring(src.fstring) {