|
|
| version 1.23, 2003/07/24 11:31:21 | version 1.40, 2008/08/15 15:28:52 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: Charset connection decl. | 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 <paf@design.ru> (http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| */ | */ |
| #ifndef PA_CHARSET_H | #ifndef PA_CHARSET_H |
| #define PA_CHARSET_H | #define PA_CHARSET_H |
| static const char* IDENT_CHARSET_H="$Date$"; | static const char * const IDENT_CHARSET_H="$Date$"; |
| #include "pa_exception.h" | #include "pa_exception.h" |
| Line 16 static const char* IDENT_CHARSET_H="$Dat | Line 16 static const char* IDENT_CHARSET_H="$Dat |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_array.h" | #include "pa_array.h" |
| // hiding into namespace so to avoid stupid conflict | // pcre.h must be included BEFORE pcre_internal.h |
| namespace PCRE { | |
| #include "pcre.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 | #ifdef XML |
| #include "libxml/encoding.h" | #include "libxml/encoding.h" |
| Line 39 namespace PCRE { | Line 39 namespace PCRE { |
| typedef unsigned char XMLByte; | typedef unsigned char XMLByte; |
| # endif | # endif |
| // forwards | |
| class GdomeDOMString_auto_ptr; | |
| // helpers | // helpers |
| struct Charset_TransRec { | typedef Hash<const String::Body, String::Body> HashStringString; |
| XMLCh intCh; | |
| XMLByte extCh; | |
| }; | |
| typedef Hash<const StringBody, StringBody> HashStringString; | |
| /** charset holds name & transcode tables | /** charset holds name & transcode tables |
| registers libxml transcoders | registers libxml transcoders |
| Line 58 typedef Hash<const StringBody, StringBod | Line 49 typedef Hash<const StringBody, StringBod |
| class Charset: public PA_Object { | class Charset: public PA_Object { |
| public: | 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; } | bool isUTF8() const { return fisUTF8; } |
| static String::C transcode(const String::C src, | static String::C transcode(const String::C src, |
| const Charset& source_transcoder, | const Charset& source_charset, |
| const Charset& dest_transcoder | const Charset& dest_charset); |
| ); | |
| static String& transcode(const String& src, | static String& transcode(const String& src, |
| const Charset& source_transcoder, | const Charset& source_transcoder, |
| const Charset& dest_transcoder); | const Charset& dest_transcoder); |
| static StringBody transcode(const StringBody src, | static String::Body transcode(const String::Body src, |
| const Charset& source_transcoder, | const Charset& source_transcoder, |
| const Charset& dest_transcoder); | const Charset& dest_transcoder); |
| Line 85 public: | Line 76 public: |
| const Charset& source_transcoder, | const Charset& source_transcoder, |
| const Charset& dest_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 | #ifdef XML |
| xmlCharEncodingHandler& transcoder(const StringBody NAME); | xmlCharEncodingHandler& transcoder(const String::Body NAME); |
| #endif | #endif |
| public: | public: |
| Line 107 private: | Line 109 private: |
| public: | public: |
| struct Tables { | struct Tables { |
| struct Rec { | |
| XMLCh intCh; | |
| XMLByte extCh; | |
| }; | |
| XMLCh fromTable[0x100]; | XMLCh fromTable[0x100]; |
| Charset_TransRec toTable[MAX_CHARSET_UNI_CODES]; | Rec toTable[MAX_CHARSET_UNI_CODES]; |
| uint toTableSize; | uint toTableSize; |
| }; | }; |
| struct UTF8CaseTable { | |
| struct Rec { | |
| XMLCh from, to; | |
| }; | |
| uint size; | |
| Rec* records; | |
| }; | |
| private: | private: |
| const StringBody FNAME; | const String::Body FNAME; |
| char* FNAME_CSTR; | char* FNAME_CSTR; |
| bool fisUTF8; | bool fisUTF8; |
| Tables tables; | Tables tables; |
| Line 123 private: | Line 139 private: |
| private: | private: |
| void addEncoding(char* name_cstr); | 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: | 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* | /// 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 | /// converts xmlChar* null-terminated string to parser String |
| const String& transcode(xmlChar* s); | const String& transcode(const xmlChar* s); |
| /** converts sized char* to xmlChar* | /** converts sized char* to xmlChar* |
| @returns xmlChar* WHICH CALLER SHOULD FREE | @returns xmlChar* WHICH CALLER SHOULD FREE |
| */ | */ |
| xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size); | xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size); |
| /// converts char* to GdomeDOMString | /// converts parser String to xmlChar* |
| GdomeDOMString_auto_ptr transcode_buf2dom(const char* buf, size_t buf_size); | xmlChar* transcode(const String& s); |
| /// converts parser String to GdomeDOMString | /// converts parser String::Body to xmlChar* |
| GdomeDOMString_auto_ptr transcode(const String& s); | xmlChar* transcode(const String::Body s); |
| /// converts parser StringBody to GdomeDOMString | |
| GdomeDOMString_auto_ptr transcode(const StringBody s); | |
| private: | private: |
| Line 154 private: | Line 164 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; |
| #endif | 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 | #endif |