Annotation of parser3/src/include/pa_charset.h, revision 1.59
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.59 ! moko 4: Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com)
1.57 moko 5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
1.1 paf 6: */
7:
8: #ifndef PA_CHARSET_H
9: #define PA_CHARSET_H
1.17 paf 10:
1.59 ! moko 11: #define IDENT_PA_CHARSET_H "$Id: pa_charset.h,v 1.58 2023/12/12 18:29:28 moko Exp $"
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:
1.58 moko 19: #ifdef HAVE_PCRE2
20: #include <pcre2.h>
21: #else
22: #include <pcre.h>
23: #endif
1.51 moko 24: // we are using some pcre_internal.h stuff as well
25: #include "../lib/pcre/pa_pcre_internal.h"
1.1 paf 26:
1.3 paf 27: #ifdef XML
28: #include "libxml/encoding.h"
29: #endif
30:
1.1 paf 31: // defines
32:
1.23 paf 33: #define MAX_CHARSETS 10
34:
35: #define MAX_CHARSET_UNI_CODES 500
36:
1.41 misha 37: #ifndef XMLCh
38: typedef unsigned int XMLCh;
39: #endif
40: #ifndef XMLByte
41: typedef unsigned char XMLByte;
42: #endif
1.3 paf 43:
1.23 paf 44: // helpers
45:
1.42 misha 46: typedef HashString<String::Body> HashStringString;
1.23 paf 47:
1.1 paf 48: /** charset holds name & transcode tables
1.23 paf 49: registers libxml transcoders
1.1 paf 50: */
1.23 paf 51: class Charset: public PA_Object {
1.1 paf 52: public:
53:
1.24 paf 54: Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec);
1.1 paf 55:
1.24 paf 56: const String::Body NAME() const { return FNAME; }
1.30 paf 57: const char* NAME_CSTR() const { return FNAME_CSTR; }
1.1 paf 58:
59: bool isUTF8() const { return fisUTF8; }
60:
1.53 moko 61: static String::C transcode(const String::C src, const Charset& source_charset, const Charset& dest_charset);
62: static String::Body transcode(const String::Body src, const Charset& source_transcoder, const Charset& dest_transcoder);
63: static String& transcode(const String& src, const Charset& source_transcoder, const Charset& dest_transcoder);
64: static void transcode(ArrayString& src, const Charset& source_transcoder, const Charset& dest_transcoder);
65: static void transcode(HashStringString& src, const Charset& source_transcoder, const Charset& dest_transcoder);
66:
67: static String::C escape(const String::C src, const Charset& source_charset);
68: static String::Body escape(const String::Body src, const Charset& source_charset);
69: static String& escape(const String& src, const Charset& source_charset);
70:
71: static String::C escape_JSON(const String::C src, const Charset& source_charset);
72: static String::Body escape_JSON(const String::Body src, const Charset& source_charset);
73: static String& escape_JSON(const String& src, const Charset& source_charset);
1.38 misha 74:
1.35 misha 75: void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found);
1.33 misha 76:
1.9 paf 77: #ifdef XML
1.24 paf 78: xmlCharEncodingHandler& transcoder(const String::Body NAME);
1.9 paf 79: #endif
80:
1.1 paf 81: public:
82:
83: unsigned char pcre_tables[tables_length];
84:
85: private:
86:
1.23 paf 87: void load_definition(Request_charsets& charsets, const String& afile_spec);
1.1 paf 88: void sort_ToTable();
89:
1.23 paf 90: const String::C transcodeToUTF8(const String::C src) const;
91: const String::C transcodeFromUTF8(const String::C src) const;
92:
93: const String::C transcodeToCharset(const String::C src,
94: const Charset& dest_transcoder) const;
1.1 paf 95:
1.4 paf 96: public:
97:
98: struct Tables {
1.25 paf 99: struct Rec {
100: XMLCh intCh;
101: XMLByte extCh;
102: };
103:
1.4 paf 104: XMLCh fromTable[0x100];
1.25 paf 105: Rec toTable[MAX_CHARSET_UNI_CODES];
1.4 paf 106: uint toTableSize;
107: };
108:
1.25 paf 109: struct UTF8CaseTable {
110: struct Rec {
111: XMLCh from, to;
112: };
113:
114: uint size;
115: Rec* records;
116: };
117:
1.1 paf 118: private:
119:
1.24 paf 120: const String::Body FNAME;
1.23 paf 121: char* FNAME_CSTR;
1.1 paf 122: bool fisUTF8;
1.4 paf 123: Tables tables;
1.1 paf 124:
1.45 misha 125: static size_t calc_escaped_length_UTF8(XMLByte* src, size_t src_length);
1.46 moko 126: static size_t calc_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
1.45 misha 127: static size_t calc_escaped_length(const String::C src, const Charset& source_charset);
128: static size_t escape_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
129: static size_t escape(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
130:
131: static size_t calc_JSON_escaped_length_UTF8(XMLByte* src, size_t src_length);
1.46 moko 132: static size_t calc_JSON_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
1.45 misha 133: static size_t calc_JSON_escaped_length(const String::C src, const Charset& source_charset);
134: static size_t escape_JSON_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
135: static size_t escape_JSON(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
136:
1.47 misha 137: #ifdef XML
138:
139: private:
140: void addEncoding(char* name_cstr);
141: void initTranscoder(const String::Body name, const char* name_cstr);
142:
1.1 paf 143: public:
1.23 paf 144: /// converts xmlChar* null-terminated string to char*
1.32 paf 145: String::C transcode_cstr(const xmlChar* s);
1.8 paf 146: /// converts xmlChar* null-terminated string to parser String
1.32 paf 147: const String& transcode(const xmlChar* s);
1.23 paf 148:
149: /** converts sized char* to xmlChar*
150: @returns xmlChar* WHICH CALLER SHOULD FREE
1.16 paf 151: */
1.23 paf 152: xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size);
1.32 paf 153: /// converts parser String to xmlChar*
154: xmlChar* transcode(const String& s);
155: /// converts parser String::Body to xmlChar*
156: xmlChar* transcode(const String::Body s);
1.1 paf 157:
158: private:
159:
1.23 paf 160: xmlCharEncodingHandler* ftranscoder;
1.1 paf 161:
162: #endif
163:
1.5 paf 164: };
1.25 paf 165:
166:
167: // externs
168:
169: extern Charset::UTF8CaseTable UTF8CaseToUpper;
170: extern Charset::UTF8CaseTable UTF8CaseToLower;
1.28 paf 171: void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
1.43 misha 172: XMLByte* toFill, size_t toFillLen,
173: const Charset::UTF8CaseTable& table);
1.37 misha 174: size_t getUTF8BytePos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t charPos/*position in characters*/);
175: size_t getUTF8CharPos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t bytePos/*position in bytes*/);
176: size_t lengthUTF8(const XMLByte* srcBegin, const XMLByte* srcEnd);
1.44 misha 177: unsigned int lengthUTF8Char(const XMLByte c);
1.25 paf 178:
1.52 moko 179: const char *fixUTF8(const char *src);
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: