Annotation of parser3/src/include/pa_charset.h, revision 1.44
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.41 misha 4: Copyright (c) 2001-2009 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:
1.44 ! misha 11: static const char * const IDENT_CHARSET_H="$Date: 2009-10-03 02:18:04 $";
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:
1.40 misha 19: // pcre.h must be included BEFORE pcre_internal.h
1.1 paf 20: #include "pcre.h"
1.40 misha 21: // have to call config.h for using pcre_internal.h as well
22: #include "../lib/pcre/config.h"
23: #include "pcre_internal.h"
1.1 paf 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:
1.41 misha 35: #ifndef XMLCh
36: typedef unsigned int XMLCh;
37: #endif
38: #ifndef XMLByte
39: typedef unsigned char XMLByte;
40: #endif
1.3 paf 41:
1.23 paf 42: // helpers
43:
1.42 misha 44: typedef HashString<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,
1.36 misha 60: const Charset& source_charset,
1.38 misha 61: const Charset& dest_charset);
1.21 paf 62:
1.23 paf 63: static String& transcode(const String& src,
1.21 paf 64: const Charset& source_transcoder,
1.23 paf 65: const Charset& dest_transcoder);
1.21 paf 66:
1.24 paf 67: static String::Body transcode(const String::Body src,
1.21 paf 68: const Charset& source_transcoder,
1.23 paf 69: const Charset& dest_transcoder);
1.21 paf 70:
1.23 paf 71: static void transcode(ArrayString& src,
1.21 paf 72: const Charset& source_transcoder,
1.23 paf 73: const Charset& dest_transcoder);
74:
75: static void transcode(HashStringString& src,
76: const Charset& source_transcoder,
77: const Charset& dest_transcoder);
1.1 paf 78:
1.36 misha 79: static String::C escape(const String::C src,
1.37 misha 80: const Charset& source_charset);
1.36 misha 81:
1.38 misha 82: static String::Body escape(const String::Body src,
83: const Charset& source_charset);
84:
85: static String& escape(const String& src,
86: const Charset& source_charset);
87:
1.35 misha 88: void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found);
1.33 misha 89:
1.9 paf 90: #ifdef XML
1.24 paf 91: xmlCharEncodingHandler& transcoder(const String::Body NAME);
1.9 paf 92: #endif
93:
1.1 paf 94: public:
95:
96: unsigned char pcre_tables[tables_length];
97:
98: private:
99:
1.23 paf 100: void load_definition(Request_charsets& charsets, const String& afile_spec);
1.1 paf 101: void sort_ToTable();
102:
1.23 paf 103: const String::C transcodeToUTF8(const String::C src) const;
104: const String::C transcodeFromUTF8(const String::C src) const;
105:
106: const String::C transcodeToCharset(const String::C src,
107: const Charset& dest_transcoder) const;
1.1 paf 108:
1.4 paf 109: public:
110:
111: struct Tables {
1.25 paf 112: struct Rec {
113: XMLCh intCh;
114: XMLByte extCh;
115: };
116:
1.4 paf 117: XMLCh fromTable[0x100];
1.25 paf 118: Rec toTable[MAX_CHARSET_UNI_CODES];
1.4 paf 119: uint toTableSize;
120: };
121:
1.25 paf 122: struct UTF8CaseTable {
123: struct Rec {
124: XMLCh from, to;
125: };
126:
127: uint size;
128: Rec* records;
129: };
130:
1.1 paf 131: private:
132:
1.24 paf 133: const String::Body FNAME;
1.23 paf 134: char* FNAME_CSTR;
1.1 paf 135: bool fisUTF8;
1.4 paf 136: Tables tables;
1.1 paf 137:
138: #ifdef XML
139:
140: private:
1.23 paf 141: void addEncoding(char* name_cstr);
1.24 paf 142: void initTranscoder(const String::Body name, const char* name_cstr);
1.9 paf 143:
1.1 paf 144: public:
1.23 paf 145: /// converts xmlChar* null-terminated string to char*
1.32 paf 146: String::C transcode_cstr(const xmlChar* s);
1.8 paf 147: /// converts xmlChar* null-terminated string to parser String
1.32 paf 148: const String& transcode(const xmlChar* s);
1.23 paf 149:
150: /** converts sized char* to xmlChar*
151: @returns xmlChar* WHICH CALLER SHOULD FREE
1.16 paf 152: */
1.23 paf 153: xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size);
1.32 paf 154: /// converts parser String to xmlChar*
155: xmlChar* transcode(const String& s);
156: /// converts parser String::Body to xmlChar*
157: xmlChar* transcode(const String::Body s);
1.1 paf 158:
159: private:
160:
1.23 paf 161: xmlCharEncodingHandler* ftranscoder;
1.1 paf 162:
163: #endif
164:
1.5 paf 165: };
1.25 paf 166:
167:
168: // externs
169:
170: extern Charset::UTF8CaseTable UTF8CaseToUpper;
171: extern Charset::UTF8CaseTable UTF8CaseToLower;
1.28 paf 172: void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
1.43 misha 173: XMLByte* toFill, size_t toFillLen,
174: const Charset::UTF8CaseTable& table);
1.37 misha 175: size_t getUTF8BytePos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t charPos/*position in characters*/);
176: size_t getUTF8CharPos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t bytePos/*position in bytes*/);
177: size_t lengthUTF8(const XMLByte* srcBegin, const XMLByte* srcEnd);
1.44 ! misha 178: unsigned int lengthUTF8Char(const XMLByte c);
1.25 paf 179:
1.43 misha 180:
181: class UTF8_string_iterator {
182: public:
183: UTF8_string_iterator(const String& astring): fsrcPtr((XMLByte*)astring.cstr()), fsrcEnd(fsrcPtr + astring.length()) {}
184: UTF8_string_iterator(XMLByte* asrcPtr, size_t length): fsrcPtr(asrcPtr), fsrcEnd(fsrcPtr + length) {}
185:
186: bool has_next();
187: XMLCh next() { return fUTF8Char; }
188: XMLByte getFirstByte(){ return ffirstByte; }
189: size_t getCharSize(){ return fcharSize; }
190: private:
191: const XMLByte* fsrcPtr;
192: const XMLByte* fsrcEnd;
193: size_t fcharSize;
194: XMLByte ffirstByte;
195: XMLCh fUTF8Char;
196: };
197:
1.1 paf 198: #endif
E-mail: