Annotation of parser3/src/include/pa_charset.h, revision 1.30
1.1 paf 1: /** @file
2: Parser: Charset connection decl.
3:
1.29 paf 4: Copyright (c) 2001-2004 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:
1.30 ! paf 11: static const char * const IDENT_CHARSET_H="$Date: 2004/02/11 15:33:13 $";
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"
1.26 paf 27: #include "gdome.h"
1.3 paf 28: #endif
29:
1.1 paf 30: // defines
31:
1.23 paf 32: #define MAX_CHARSETS 10
33:
34: #define MAX_CHARSET_UNI_CODES 500
35:
1.3 paf 36: # ifndef XMLCh
1.10 paf 37: typedef unsigned int XMLCh;
1.3 paf 38: # endif
39: # ifndef XMLByte
1.11 paf 40: typedef unsigned char XMLByte;
1.3 paf 41: # endif
42:
1.23 paf 43: // forwards
44:
45: class GdomeDOMString_auto_ptr;
46:
47: // helpers
48:
1.24 paf 49: typedef Hash<const String::Body, String::Body> HashStringString;
1.23 paf 50:
1.1 paf 51: /** charset holds name & transcode tables
1.23 paf 52: registers libxml transcoders
1.1 paf 53: */
1.23 paf 54: class Charset: public PA_Object {
1.1 paf 55: public:
56:
1.24 paf 57: Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec);
1.1 paf 58:
1.24 paf 59: const String::Body NAME() const { return FNAME; }
1.30 ! paf 60: const char* NAME_CSTR() const { return FNAME_CSTR; }
1.1 paf 61:
62: bool isUTF8() const { return fisUTF8; }
63:
1.23 paf 64: static String::C transcode(const String::C src,
65: const Charset& source_transcoder,
66: const Charset& dest_transcoder
1.1 paf 67: );
1.21 paf 68:
1.23 paf 69: static String& transcode(const String& src,
1.21 paf 70: const Charset& source_transcoder,
1.23 paf 71: const Charset& dest_transcoder);
1.21 paf 72:
1.24 paf 73: static String::Body transcode(const String::Body src,
1.21 paf 74: const Charset& source_transcoder,
1.23 paf 75: const Charset& dest_transcoder);
1.21 paf 76:
1.23 paf 77: static void transcode(ArrayString& src,
1.21 paf 78: const Charset& source_transcoder,
1.23 paf 79: const Charset& dest_transcoder);
80:
81: static void transcode(HashStringString& src,
82: const Charset& source_transcoder,
83: const Charset& dest_transcoder);
1.1 paf 84:
1.9 paf 85: #ifdef XML
1.24 paf 86: xmlCharEncodingHandler& transcoder(const String::Body NAME);
1.9 paf 87: #endif
88:
1.1 paf 89: public:
90:
91: unsigned char pcre_tables[tables_length];
92:
93: private:
94:
1.23 paf 95: void load_definition(Request_charsets& charsets, const String& afile_spec);
1.1 paf 96: void sort_ToTable();
97:
1.23 paf 98: const String::C transcodeToUTF8(const String::C src) const;
99: const String::C transcodeFromUTF8(const String::C src) const;
100:
101: const String::C transcodeToCharset(const String::C src,
102: const Charset& dest_transcoder) const;
1.1 paf 103:
1.4 paf 104: public:
105:
106: struct Tables {
1.25 paf 107: struct Rec {
108: XMLCh intCh;
109: XMLByte extCh;
110: };
111:
1.4 paf 112: XMLCh fromTable[0x100];
1.25 paf 113: Rec toTable[MAX_CHARSET_UNI_CODES];
1.4 paf 114: uint toTableSize;
115: };
116:
1.25 paf 117: struct UTF8CaseTable {
118: struct Rec {
119: XMLCh from, to;
120: };
121:
122: uint size;
123: Rec* records;
124: };
125:
1.1 paf 126: private:
127:
1.24 paf 128: const String::Body FNAME;
1.23 paf 129: char* FNAME_CSTR;
1.1 paf 130: bool fisUTF8;
1.4 paf 131: Tables tables;
1.1 paf 132:
133: #ifdef XML
134:
135: private:
1.23 paf 136: void addEncoding(char* name_cstr);
1.24 paf 137: void initTranscoder(const String::Body name, const char* name_cstr);
1.9 paf 138:
1.1 paf 139: public:
1.23 paf 140: /// converts GdomeDOMString string to char*
141: String::C transcode_cstr(GdomeDOMString* s);
1.3 paf 142: /// converts GdomeDOMString string to parser String
1.23 paf 143: const String& transcode(GdomeDOMString* s);
144: /// converts xmlChar* null-terminated string to char*
145: String::C transcode_cstr(xmlChar* s);
1.8 paf 146: /// converts xmlChar* null-terminated string to parser String
1.23 paf 147: const String& transcode(xmlChar* s);
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);
153: /// converts char* to GdomeDOMString
154: GdomeDOMString_auto_ptr transcode_buf2dom(const char* buf, size_t buf_size);
1.3 paf 155: /// converts parser String to GdomeDOMString
1.5 paf 156: GdomeDOMString_auto_ptr transcode(const String& s);
1.24 paf 157: /// converts parser String::Body to GdomeDOMString
158: GdomeDOMString_auto_ptr transcode(const String::Body s);
1.1 paf 159:
160: private:
161:
1.23 paf 162: xmlCharEncodingHandler* ftranscoder;
1.1 paf 163:
164: #endif
165:
1.5 paf 166: };
1.25 paf 167:
168:
169: // externs
170:
171: extern Charset::UTF8CaseTable UTF8CaseToUpper;
172: extern Charset::UTF8CaseTable UTF8CaseToLower;
1.28 paf 173: void change_case_UTF8(const XMLByte* srcData, size_t srcLen,
174: XMLByte* toFill, size_t toFillLen,
1.25 paf 175: const Charset::UTF8CaseTable& table);
176:
1.5 paf 177:
1.12 paf 178: #ifdef XML
1.5 paf 179: /// Auto-object used to track GdomeDOMString usage
180: class GdomeDOMString_auto_ptr {
1.23 paf 181: GdomeDOMString* fstring;
1.5 paf 182: public:
1.23 paf 183: /// frees astring afterwards!!!
184: explicit GdomeDOMString_auto_ptr(xmlChar* astring) : fstring(gdome_str_mkref_xml(astring)) {}
185: explicit GdomeDOMString_auto_ptr(GdomeDOMString* astring=0) : fstring(astring) {
1.7 paf 186: // not ref-ing, owning
1.5 paf 187: }
188: ~GdomeDOMString_auto_ptr() {
1.23 paf 189: if(fstring)
190: gdome_str_unref(fstring);
1.5 paf 191: }
1.22 paf 192: /* GdomeDOMString* get() {
193: return fstring;
194: }*/
195: GdomeDOMString* use() {
196: if(fstring)
197: gdome_str_ref(fstring);
1.5 paf 198: return fstring;
199: }
200: GdomeDOMString* operator->() {
201: return fstring;
202: }
1.22 paf 203: /* GdomeDOMString& operator*() {
1.23 paf 204: return* fstring;
1.22 paf 205: }*/
1.5 paf 206:
207: // copying
208: GdomeDOMString_auto_ptr(const GdomeDOMString_auto_ptr& src) : fstring(src.fstring) {
209: gdome_str_ref(fstring);
210: }
211: GdomeDOMString_auto_ptr& operator =(const GdomeDOMString_auto_ptr& src) {
212: if(this == &src)
1.23 paf 213: return* this;
1.5 paf 214:
1.22 paf 215: if(fstring)
216: gdome_str_unref(fstring);
1.5 paf 217: fstring=src.fstring;
1.22 paf 218: if(fstring)
219: gdome_str_ref(fstring);
1.5 paf 220:
1.23 paf 221: return* this;
1.5 paf 222: }
1.1 paf 223: };
1.12 paf 224: #endif
1.1 paf 225:
226: #endif
E-mail: