--- parser3/src/include/pa_charset.h 2003/09/24 14:32:05 1.23.4.1 +++ parser3/src/include/pa_charset.h 2009/11/06 05:01:24 1.44 @@ -1,14 +1,14 @@ /** @file Parser: Charset connection decl. - Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + 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 -static const char* IDENT_CHARSET_H="$Date: 2003/09/24 14:32:05 $"; +static const char * const IDENT_CHARSET_H="$Date: 2009/11/06 05:01:24 $"; #include "pa_exception.h" @@ -16,11 +16,11 @@ static const char* IDENT_CHARSET_H="$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 "../lib/pcre/config.h" +#include "pcre_internal.h" #ifdef XML #include "libxml/encoding.h" @@ -32,25 +32,16 @@ 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 HashStringString; +typedef HashString HashStringString; /** charset holds name & transcode tables registers libxml transcoders @@ -61,13 +52,13 @@ public: Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec); 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, @@ -85,6 +76,17 @@ 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); + + void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found); + #ifdef XML xmlCharEncodingHandler& transcoder(const String::Body NAME); #endif @@ -107,11 +109,25 @@ 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 String::Body FNAME; @@ -126,25 +142,19 @@ private: 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 String::Body to GdomeDOMString - GdomeDOMString_auto_ptr transcode(const String::Body 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 +164,35 @@ 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