Annotation of parser3/src/include/pa_charset.h, revision 1.34
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.31 paf 4: Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com)
1.14 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1 paf 6: */
7:
8: #ifndef PA_CHARSET_H
9: #define PA_CHARSET_H
1.17 paf 10:
1.34 ! misha 11: static const char * const IDENT_CHARSET_H="$Date: 2008-05-22 16:17:50 $";
1.23 paf 12:
1.1 paf 13:
14: #include "pa_exception.h"
15: #include "pa_common.h"
1.23 paf 16: #include "pa_hash.h"
17: #include "pa_array.h"
1.1 paf 18:
19: // hiding into namespace so to avoid stupid conflict
20: namespace PCRE {
21: #include "pcre.h"
22: # include "internal.h"
23: };
24:
1.3 paf 25: #ifdef XML
26: #include "libxml/encoding.h"
27: #endif
28:
1.1 paf 29: // defines
30:
1.23 paf 31: #define MAX_CHARSETS 10
32:
33: #define MAX_CHARSET_UNI_CODES 500
34:
1.3 paf 35: # ifndef XMLCh
1.10 paf 36: typedef unsigned int XMLCh;
1.3 paf 37: # endif
38: # ifndef XMLByte
1.11 paf 39: typedef unsigned char XMLByte;
1.3 paf 40: # endif
41:
1.23 paf 42: // helpers
43:
1.24 paf 44: typedef Hash<const String::Body, String::Body> HashStringString;
1.23 paf 45:
1.1 paf 46: /** charset holds name & transcode tables
1.23 paf 47: registers libxml transcoders
1.1 paf 48: */
1.23 paf 49: class Charset: public PA_Object {
1.1 paf 50: public:
51:
1.24 paf 52: Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec);
1.1 paf 53:
1.24 paf 54: const String::Body NAME() const { return FNAME; }
1.30 paf 55: const char* NAME_CSTR() const { return FNAME_CSTR; }
1.1 paf 56:
57: bool isUTF8() const { return fisUTF8; }
58:
1.23 paf 59: static String::C transcode(const String::C src,
60: const Charset& source_transcoder,
61: const Charset& dest_transcoder
1.1 paf 62: );
1.21 paf 63:
1.23 paf 64: static String& transcode(const String& src,
1.21 paf 65: const Charset& source_transcoder,
1.23 paf 66: const Charset& dest_transcoder);
1.21 paf 67:
1.24 paf 68: static String::Body transcode(const String::Body src,
1.21 paf 69: const Charset& source_transcoder,
1.23 paf 70: const Charset& dest_transcoder);
1.21 paf 71:
1.23 paf 72: static void transcode(ArrayString& src,
1.21 paf 73: const Charset& source_transcoder,
1.23 paf 74: const Charset& dest_transcoder);
75:
76: static void transcode(HashStringString& src,
77: const Charset& source_transcoder,
78: const Charset& dest_transcoder);
1.1 paf 79:
1.34 ! misha 80: void Charset::store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found);
1.33 misha 81:
1.9 paf 82: #ifdef XML
1.24 paf 83: xmlCharEncodingHandler& transcoder(const String::Body NAME);
1.9 paf 84: #endif
85:
1.1 paf 86: public:
87:
88: unsigned char pcre_tables[tables_length];
89:
90: private:
91:
1.23 paf 92: void load_definition(Request_charsets& charsets, const String& afile_spec);
1.1 paf 93: void sort_ToTable();
94:
1.23 paf 95: const String::C transcodeToUTF8(const String::C src) const;
96: const String::C transcodeFromUTF8(const String::C src) const;
97:
98: const String::C transcodeToCharset(const String::C src,
99: const Charset& dest_transcoder) const;
1.1 paf 100:
1.4 paf 101: public:
102:
103: struct Tables {
1.25 paf 104: struct Rec {
105: XMLCh intCh;
106: XMLByte extCh;
107: };
108:
1.4 paf 109: XMLCh fromTable[0x100];
1.25 paf 110: Rec toTable[MAX_CHARSET_UNI_CODES];
1.4 paf 111: uint toTableSize;
112: };
113:
1.25 paf 114: struct UTF8CaseTable {
115: struct Rec {
116: XMLCh from, to;
117: };
118:
119: uint size;
120: Rec* records;
121: };
122:
1.1 paf 123: private:
124:
1.24 paf 125: const String::Body FNAME;
1.23 paf 126: char* FNAME_CSTR;
1.1 paf 127: bool fisUTF8;
1.4 paf 128: Tables tables;
1.1 paf 129:
130: #ifdef XML
131:
132: private:
1.23 paf 133: void addEncoding(char* name_cstr);
1.24 paf 134: void initTranscoder(const String::Body name, const char* name_cstr);
1.9 paf 135:
1.1 paf 136: public:
1.23 paf 137: /// converts xmlChar* null-terminated string to char*
1.32 paf 138: String::C transcode_cstr(const xmlChar* s);
1.8 paf 139: /// converts xmlChar* null-terminated string to parser String
1.32 paf 140: const String& transcode(const xmlChar* s);
1.23 paf 141:
142: /** converts sized char* to xmlChar*
143: @returns xmlChar* WHICH CALLER SHOULD FREE
1.16 paf 144: */
1.23 paf 145: xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size);
1.32 paf 146: /// converts parser String to xmlChar*
147: xmlChar* transcode(const String& s);
148: /// converts parser String::Body to xmlChar*
149: xmlChar* transcode(const String::Body s);
1.1 paf 150:
151: private:
152:
1.23 paf 153: xmlCharEncodingHandler* ftranscoder;
1.1 paf 154:
155: #endif
156:
1.5 paf 157: };
1.25 paf 158:
159:
160: // externs
161:
162: extern Charset::UTF8CaseTable UTF8CaseToUpper;
163: extern Charset::UTF8CaseTable UTF8CaseToLower;
1.28 paf 164: void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
165: XMLByte* toFill, size_t toFillLen,
1.25 paf 166: const Charset::UTF8CaseTable& table);
167:
1.1 paf 168: #endif
E-mail: