Annotation of parser3/src/include/pa_charset.h, revision 1.56
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.56 ! moko 4: Copyright (c) 2001-2020 Art. Lebedev Studio (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:
1.56 ! moko 11: #define IDENT_PA_CHARSET_H "$Id: pa_charset.h,v 1.55 2017/02/07 22:00:31 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:
19: #include "pcre.h"
1.51 moko 20: // we are using some pcre_internal.h stuff as well
21: #include "../lib/pcre/pa_pcre_internal.h"
1.1 paf 22:
1.3 paf 23: #ifdef XML
24: #include "libxml/encoding.h"
25: #endif
26:
1.1 paf 27: // defines
28:
1.23 paf 29: #define MAX_CHARSETS 10
30:
31: #define MAX_CHARSET_UNI_CODES 500
32:
1.41 misha 33: #ifndef XMLCh
34: typedef unsigned int XMLCh;
35: #endif
36: #ifndef XMLByte
37: typedef unsigned char XMLByte;
38: #endif
1.3 paf 39:
1.23 paf 40: // helpers
41:
1.42 misha 42: typedef HashString<String::Body> HashStringString;
1.23 paf 43:
1.1 paf 44: /** charset holds name & transcode tables
1.23 paf 45: registers libxml transcoders
1.1 paf 46: */
1.23 paf 47: class Charset: public PA_Object {
1.1 paf 48: public:
49:
1.24 paf 50: Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec);
1.1 paf 51:
1.24 paf 52: const String::Body NAME() const { return FNAME; }
1.30 paf 53: const char* NAME_CSTR() const { return FNAME_CSTR; }
1.1 paf 54:
55: bool isUTF8() const { return fisUTF8; }
56:
1.53 moko 57: static String::C transcode(const String::C src, const Charset& source_charset, const Charset& dest_charset);
58: static String::Body transcode(const String::Body src, const Charset& source_transcoder, const Charset& dest_transcoder);
59: static String& transcode(const String& src, const Charset& source_transcoder, const Charset& dest_transcoder);
60: static void transcode(ArrayString& src, const Charset& source_transcoder, const Charset& dest_transcoder);
61: static void transcode(HashStringString& src, const Charset& source_transcoder, const Charset& dest_transcoder);
62:
63: static String::C escape(const String::C src, const Charset& source_charset);
64: static String::Body escape(const String::Body src, const Charset& source_charset);
65: static String& escape(const String& src, const Charset& source_charset);
66:
67: static String::C escape_JSON(const String::C src, const Charset& source_charset);
68: static String::Body escape_JSON(const String::Body src, const Charset& source_charset);
69: static String& escape_JSON(const String& src, const Charset& source_charset);
1.38 misha 70:
1.35 misha 71: void store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found);
1.33 misha 72:
1.9 paf 73: #ifdef XML
1.24 paf 74: xmlCharEncodingHandler& transcoder(const String::Body NAME);
1.9 paf 75: #endif
76:
1.1 paf 77: public:
78:
79: unsigned char pcre_tables[tables_length];
80:
81: private:
82:
1.23 paf 83: void load_definition(Request_charsets& charsets, const String& afile_spec);
1.1 paf 84: void sort_ToTable();
85:
1.23 paf 86: const String::C transcodeToUTF8(const String::C src) const;
87: const String::C transcodeFromUTF8(const String::C src) const;
88:
89: const String::C transcodeToCharset(const String::C src,
90: const Charset& dest_transcoder) const;
1.1 paf 91:
1.4 paf 92: public:
93:
94: struct Tables {
1.25 paf 95: struct Rec {
96: XMLCh intCh;
97: XMLByte extCh;
98: };
99:
1.4 paf 100: XMLCh fromTable[0x100];
1.25 paf 101: Rec toTable[MAX_CHARSET_UNI_CODES];
1.4 paf 102: uint toTableSize;
103: };
104:
1.25 paf 105: struct UTF8CaseTable {
106: struct Rec {
107: XMLCh from, to;
108: };
109:
110: uint size;
111: Rec* records;
112: };
113:
1.1 paf 114: private:
115:
1.24 paf 116: const String::Body FNAME;
1.23 paf 117: char* FNAME_CSTR;
1.1 paf 118: bool fisUTF8;
1.4 paf 119: Tables tables;
1.1 paf 120:
1.45 misha 121: static size_t calc_escaped_length_UTF8(XMLByte* src, size_t src_length);
1.46 moko 122: static size_t calc_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
1.45 misha 123: static size_t calc_escaped_length(const String::C src, const Charset& source_charset);
124: static size_t escape_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
125: static size_t escape(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
126:
127: static size_t calc_JSON_escaped_length_UTF8(XMLByte* src, size_t src_length);
1.46 moko 128: static size_t calc_JSON_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables);
1.45 misha 129: static size_t calc_JSON_escaped_length(const String::C src, const Charset& source_charset);
130: static size_t escape_JSON_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest);
131: static size_t escape_JSON(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables);
132:
1.47 misha 133: #ifdef XML
134:
135: private:
136: void addEncoding(char* name_cstr);
137: void initTranscoder(const String::Body name, const char* name_cstr);
138:
1.1 paf 139: public:
1.23 paf 140: /// converts xmlChar* null-terminated string to char*
1.32 paf 141: String::C transcode_cstr(const xmlChar* s);
1.8 paf 142: /// converts xmlChar* null-terminated string to parser String
1.32 paf 143: const String& transcode(const xmlChar* s);
1.23 paf 144:
145: /** converts sized char* to xmlChar*
146: @returns xmlChar* WHICH CALLER SHOULD FREE
1.16 paf 147: */
1.23 paf 148: xmlChar* transcode_buf2xchar(const char* buf, size_t buf_size);
1.32 paf 149: /// converts parser String to xmlChar*
150: xmlChar* transcode(const String& s);
151: /// converts parser String::Body to xmlChar*
152: xmlChar* transcode(const String::Body s);
1.1 paf 153:
154: private:
155:
1.23 paf 156: xmlCharEncodingHandler* ftranscoder;
1.1 paf 157:
158: #endif
159:
1.5 paf 160: };
1.25 paf 161:
162:
163: // externs
164:
165: extern Charset::UTF8CaseTable UTF8CaseToUpper;
166: extern Charset::UTF8CaseTable UTF8CaseToLower;
1.28 paf 167: void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
1.43 misha 168: XMLByte* toFill, size_t toFillLen,
169: const Charset::UTF8CaseTable& table);
1.37 misha 170: size_t getUTF8BytePos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t charPos/*position in characters*/);
171: size_t getUTF8CharPos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t bytePos/*position in bytes*/);
172: size_t lengthUTF8(const XMLByte* srcBegin, const XMLByte* srcEnd);
1.44 misha 173: unsigned int lengthUTF8Char(const XMLByte c);
1.25 paf 174:
1.52 moko 175: const char *fixUTF8(const char *src);
1.43 misha 176:
177: class UTF8_string_iterator {
178: public:
179: UTF8_string_iterator(const String& astring): fsrcPtr((XMLByte*)astring.cstr()), fsrcEnd(fsrcPtr + astring.length()) {}
180: UTF8_string_iterator(XMLByte* asrcPtr, size_t length): fsrcPtr(asrcPtr), fsrcEnd(fsrcPtr + length) {}
181:
182: bool has_next();
183: XMLCh next() { return fUTF8Char; }
184: XMLByte getFirstByte(){ return ffirstByte; }
185: size_t getCharSize(){ return fcharSize; }
186: private:
187: const XMLByte* fsrcPtr;
188: const XMLByte* fsrcEnd;
189: size_t fcharSize;
190: XMLByte ffirstByte;
191: XMLCh fUTF8Char;
192: };
193:
1.1 paf 194: #endif
E-mail: