Annotation of parser3/src/include/pa_charset.h, revision 1.20.2.9
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.20 paf 4: Copyright (c) 2001, 2003 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:
1.20.2.9! paf 11: static const char* IDENT_CHARSET_H="$Date: 2003/01/30 14:59:43 $";
1.1 paf 12:
13: #include "pa_pool.h"
14: #include "pa_exception.h"
15: #include "pa_common.h"
16:
17: // hiding into namespace so to avoid stupid conflict
18: namespace PCRE {
19: #include "pcre.h"
20: # include "internal.h"
21: };
22:
1.3 paf 23: #ifdef XML
24: #include "libxml/encoding.h"
25: #endif
26:
1.1 paf 27: // defines
28:
1.3 paf 29: # ifndef XMLCh
1.10 paf 30: typedef unsigned int XMLCh;
1.3 paf 31: # endif
32: # ifndef XMLByte
1.11 paf 33: typedef unsigned char XMLByte;
1.3 paf 34: # endif
35:
1.10 paf 36: struct Charset_TransRec {
37: XMLCh intCh;
38: XMLByte extCh;
39: };
1.1 paf 40:
41: /** charset holds name & transcode tables
1.20.2.3 paf 42: registers libxml transcoders
1.1 paf 43: */
1.20.2.3 paf 44: class Charset: public PA_Object {
1.20.2.5 paf 45: Pool pool_for_load;
1.1 paf 46: public:
47:
1.20.2.9! paf 48: Charset(StringPtr aname, StringPtr afile_spec);
1.20.2.6 paf 49: override ~Charset();
1.1 paf 50:
1.20.2.9! paf 51: StringPtr name() const { return fname; }
1.1 paf 52:
53: bool isUTF8() const { return fisUTF8; }
54:
1.20.2.2 paf 55: static void transcode(Pool& pool,
1.1 paf 56: const Charset& source_transcoder, const void *source_body, size_t source_content_length,
57: const Charset& dest_transcoder, const void *& dest_body, size_t& dest_content_length
58: );
59:
1.9 paf 60: #ifdef XML
61: xmlCharEncodingHandler *transcoder(const String *source);
62: #endif
63:
1.1 paf 64: public:
65:
66: unsigned char pcre_tables[tables_length];
67:
68: private:
69:
1.20.2.9! paf 70: void load_definition(StringPtr afile_spec);
1.1 paf 71: void sort_ToTable();
72:
73: void transcodeToUTF8(Pool& pool,
74: const void *source_body, size_t source_content_length,
75: const void *& dest_body, size_t& dest_content_length) const;
76: void transcodeFromUTF8(Pool& pool,
77: const void *source_body, size_t source_content_length,
78: const void *& dest_body, size_t& dest_content_length) const;
1.15 paf 79:
1.1 paf 80: void transcodeToCharset(Pool& pool,
81: const Charset& dest_transcoder,
82: const void *source_body, size_t source_content_length,
83: const void *& dest_body, size_t& dest_content_length) const;
84:
1.4 paf 85: public:
86:
87: struct Tables {
88: XMLCh fromTable[0x100];
89: Charset_TransRec *toTable;
90: uint toTableSize;
91: };
92:
1.1 paf 93: private:
94:
1.20.2.9! paf 95: StringPtr fname;
1.20.2.2 paf 96: CharPtr fname_cstr;
1.1 paf 97: bool fisUTF8;
1.4 paf 98: Tables tables;
1.1 paf 99:
100: #ifdef XML
101:
102: private:
1.4 paf 103: void addEncoding(char *name_cstr);
1.1 paf 104: void initTranscoder(const String *source, const char *name_cstr);
1.9 paf 105:
1.1 paf 106: public:
1.3 paf 107: /// converts GdomeDOMString string to char *
108: const char *transcode_cstr(GdomeDOMString *s);
109: /// converts GdomeDOMString string to parser String
1.19 paf 110: String& transcode(GdomeDOMString *s
111: #ifndef NO_STRING_ORIGIN
112: , const String *origin
113: #endif
114: );
1.8 paf 115: /// converts xmlChar* null-terminated string to char *
116: const char *transcode_cstr(xmlChar *s);
117: /// converts xmlChar* null-terminated string to parser String
1.19 paf 118: String& transcode(xmlChar *s
119: #ifndef NO_STRING_ORIGIN
120: , const String *origin
121: #endif
122: );
1.16 paf 123: /** converts sized char * to xmlChar*
124: @returns xmlChar * which caller should free
125: */
126: xmlChar *transcode_buf2xchar(const char *buf, size_t buf_size);
1.3 paf 127: /// converts char * to GdomeDOMString
1.16 paf 128: GdomeDOMString_auto_ptr transcode_buf2dom(const char *buf, size_t buf_size);
1.3 paf 129: /// converts parser String to GdomeDOMString
1.5 paf 130: GdomeDOMString_auto_ptr transcode(const String& s);
1.1 paf 131:
132: private:
133:
1.9 paf 134: xmlCharEncodingHandler *ftranscoder;
1.1 paf 135:
136: #endif
137:
1.5 paf 138: };
1.20.2.8 paf 139: DECLARE_OBJECT_PTR(Charset);
1.20.2.4 paf 140:
141: extern CharsetPtr UTF8_charset;
1.5 paf 142:
1.12 paf 143: #ifdef XML
1.5 paf 144: /// Auto-object used to track GdomeDOMString usage
145: class GdomeDOMString_auto_ptr {
146: GdomeDOMString *fstring;
147: public:
1.6 paf 148: explicit GdomeDOMString_auto_ptr(gchar *astring) : fstring(gdome_str_mkref(astring)) {}
1.5 paf 149: explicit GdomeDOMString_auto_ptr(GdomeDOMString *astring) : fstring(astring) {
1.7 paf 150: // not ref-ing, owning
1.5 paf 151: }
152: ~GdomeDOMString_auto_ptr() {
153: gdome_str_unref(fstring);
154: }
155: GdomeDOMString* get() {
156: return fstring;
157: }
158: GdomeDOMString* operator->() {
159: return fstring;
160: }
161: GdomeDOMString& operator*() {
162: return *fstring;
163: }
164:
165: // copying
166: GdomeDOMString_auto_ptr(const GdomeDOMString_auto_ptr& src) : fstring(src.fstring) {
167: gdome_str_ref(fstring);
168: }
169: GdomeDOMString_auto_ptr& operator =(const GdomeDOMString_auto_ptr& src) {
170: if(this == &src)
171: return *this;
172:
173: gdome_str_unref(fstring);
174: fstring=src.fstring;
175: gdome_str_ref(fstring);
176:
177: return *this;
178: }
1.1 paf 179: };
1.12 paf 180: #endif
1.1 paf 181:
182: #endif
E-mail: