Annotation of parser3/src/main/pa_charset_connection.C, revision 1.1
1.1 ! parser 1: /** @file
! 2: Parser: Charset connection implementation.
! 3:
! 4: Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
! 5: Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
! 6:
! 7: $Id: pa_sql_driver_manager.C,v 1.38 2001/09/26 10:32:26 parser Exp $
! 8: */
! 9:
! 10: #include "pa_charset_connection.h"
! 11: //#include "pa_exception.h"
! 12: //#include "pa_common.h"
! 13: //#include "pa_threads.h"
! 14:
! 15: #ifdef XML
! 16: # include <util/XercesDefs.hpp>
! 17: # include <util/TransENameMap.hpp>
! 18: # include <util/XML256TableTranscoder.hpp>
! 19: # include <util/PlatformUtils.hpp>
! 20: # include <PlatformSupport/XalanTranscodingServices.hpp>
! 21: #endif
! 22:
! 23:
! 24: // globals
! 25:
! 26:
! 27: // consts
! 28:
! 29: #define MAX_CHARSET_UNI_CODES 500
! 30:
! 31: //
! 32:
! 33: inline void prepare_case_tables(unsigned char *tables) {
! 34: unsigned char *lcc_table=tables+lcc_offset;
! 35: unsigned char *fcc_table=tables+fcc_offset;
! 36: for(int i=0; i<0x100; i++)
! 37: lcc_table[i]=fcc_table[i]=i;
! 38: }
! 39: inline void cstr2ctypes(unsigned char *tables, const unsigned char *cstr,
! 40: unsigned char bit) {
! 41: unsigned char *ctypes_table=tables+ctypes_offset;
! 42: ctypes_table[0]=bit;
! 43: for(; *cstr; cstr++) {
! 44: unsigned char c=*cstr;
! 45: ctypes_table[c]|=bit;
! 46: }
! 47: }
! 48: inline unsigned int to_wchar_code(const char *cstr) {
! 49: if(!cstr || !*cstr)
! 50: return 0;
! 51: if(cstr[1]==0)
! 52: return (unsigned int)(unsigned char)cstr[0];
! 53:
! 54: char *error_pos;
! 55: return (unsigned int)strtol(cstr, &error_pos, 0);
! 56: }
! 57: inline bool to_bool(const char *cstr) {
! 58: return cstr && *cstr!=0;
! 59: }
! 60: static void element2ctypes(unsigned char c, bool belongs,
! 61: unsigned char *tables, unsigned char bit, int group_offset=-1) {
! 62: if(!belongs)
! 63: return;
! 64:
! 65: unsigned char *ctypes_table=tables+ctypes_offset;
! 66:
! 67: ctypes_table[c]|=bit;
! 68: if(group_offset>=0)
! 69: tables[cbits_offset+group_offset+c/8] |= 1 << (c%8);
! 70: }
! 71: static void element2case(unsigned char from, unsigned char to,
! 72: unsigned char *tables) {
! 73: if(!to)
! 74: return;
! 75:
! 76: unsigned char *lcc_table=tables+lcc_offset;
! 77: unsigned char *fcc_table=tables+fcc_offset;
! 78: lcc_table[from]=to;
! 79: fcc_table[from]=to; fcc_table[to]=from;
! 80: }
! 81:
! 82: #ifdef XML
! 83:
! 84: static int sort_cmp_Trans_rec_intCh(const void *a, const void *b) {
! 85: const XMLCh ca=static_cast<const XMLTransService::TransRec *>(a)->intCh;
! 86: const XMLCh cb=static_cast<const XMLTransService::TransRec *>(b)->intCh;
! 87: // move zeros to end of table
! 88: if(ca==0)
! 89: return +1;
! 90: if(cb==0)
! 91: return -1;
! 92:
! 93: //
! 94: return ca-cb;
! 95: }
! 96:
! 97: template <class TType> class ENameMapFor2 : public ENameMap
! 98: {
! 99: public :
! 100: // -----------------------------------------------------------------------
! 101: // Constructors and Destructor
! 102: // -----------------------------------------------------------------------
! 103: ENameMapFor2(
! 104: const XMLCh* const encodingName
! 105: , const XMLCh* const fromTable
! 106: , const XMLTransService::TransRec* const toTable
! 107: , const unsigned int toTableSize
! 108: ) : ENameMap(encodingName),
! 109: ffromTable(fromTable),
! 110: ftoTable(toTable),
! 111: ftoTableSize(toTableSize) {}
! 112: ~ENameMapFor2() {}
! 113:
! 114: // -----------------------------------------------------------------------
! 115: // Implementation of virtual factory method
! 116: // -----------------------------------------------------------------------
! 117: virtual XMLTranscoder* makeNew(const unsigned int blockSize) const {
! 118: return new TType(
! 119: getKey(),
! 120: blockSize,
! 121: ffromTable,
! 122: ftoTable, ftoTableSize);
! 123: }
! 124: private:
! 125: const XMLCh* const ffromTable;
! 126: const XMLTransService::TransRec* const ftoTable;
! 127: const unsigned int ftoTableSize;
! 128:
! 129: private :
! 130: // -----------------------------------------------------------------------
! 131: // Unimplemented constructors and operators
! 132: // -----------------------------------------------------------------------
! 133: ENameMapFor2();
! 134: ENameMapFor2(const ENameMapFor2<TType>&);
! 135: void operator=(const ENameMapFor2<TType>&);
! 136: };
! 137:
! 138: class XML256TableTranscoder2 : public XML256TableTranscoder
! 139: {
! 140: public :
! 141: XML256TableTranscoder2(
! 142: const XMLCh* const encodingName
! 143: , const unsigned int blockSize
! 144: , const XMLCh* const fromTable
! 145: , const XMLTransService::TransRec* const toTable
! 146: , const unsigned int toTableSize
! 147: ) : XML256TableTranscoder(encodingName, blockSize, fromTable, toTable, toTableSize) {}
! 148:
! 149: private :
! 150: XML256TableTranscoder2();
! 151: XML256TableTranscoder2(const XML256TableTranscoder2&);
! 152: void operator=(const XML256TableTranscoder2&);
! 153: };
! 154: #endif
! 155:
! 156: #undef XALAN_HACK_DIGITAL_ENTITIES
! 157: #ifdef XALAN_HACK_DIGITAL_ENTITIES
! 158: static void hack_s_maximumCharacterValues(const XalanDOMString& encoding) {
! 159: /*
! 160: open:
! 161: xml-xalan/c/src/PlatformSupport/XalanTranscodingServices.hpp
! 162: find:
! 163: static const MaximumCharacterValueMapType& s_maximumCharacterValues;
! 164: paste to next line:
! 165: friend static void hack_s_maximumCharacterValues(const XalanDOMString& encoding); // hack by paf
! 166: */
! 167:
! 168: const_cast<XalanTranscodingServices::MaximumCharacterValueMapType &>(
! 169: XalanTranscodingServices::s_maximumCharacterValues).insert(
! 170: XalanTranscodingServices::MaximumCharacterValueMapType::value_type(encoding, 0xFFFF));
! 171: }
! 172: #endif
! 173:
! 174: void Charset_connection::load(Pool& pool, time_t new_disk_time) {
! 175: // pcre_tables
! 176: // lowcase, flipcase, bits digit+word+whitespace, masks
! 177: prepare_case_tables(fpcre_tables);
! 178: cstr2ctypes(fpcre_tables, (const unsigned char *)"*+?{^.$|()[", ctype_meta);
! 179:
! 180: #ifdef XML
! 181: // transcoder
! 182: XMLCh *fromTable=(XMLCh *)calloc(sizeof(XMLCh)*0x100);
! 183: XMLTransService::TransRec *toTable=(XMLTransService::TransRec *)calloc(
! 184: sizeof(XMLTransService::TransRec)*MAX_CHARSET_UNI_CODES);
! 185: unsigned int toTableSz=0;
! 186: #endif
! 187:
! 188: // loading text
! 189: char *data=file_read_text(pool, ffile_spec);
! 190:
! 191: // ignore header
! 192: getrow(&data);
! 193:
! 194: // parse cells
! 195: char *row_chars;
! 196: while(row_chars=getrow(&data)) {
! 197: if(!*row_chars) // remove empty lines
! 198: continue;
! 199:
! 200: // char white-space digit hex-digit letter word lowercase unicode1 unicode2
! 201: unsigned int c=0;
! 202: char *cell;
! 203: for(int column=0; cell=lsplit(&row_chars, '\t'); column++) {
! 204: switch(column) {
! 205: case 0: c=to_wchar_code(cell); break;
! 206: // fpcre_tables
! 207: case 1: element2ctypes(c, to_bool(cell), fpcre_tables, ctype_space, cbit_space); break;
! 208: case 2: element2ctypes(c, to_bool(cell), fpcre_tables, ctype_digit, cbit_digit); break;
! 209: case 3: element2ctypes(c, to_bool(cell), fpcre_tables, ctype_xdigit); break;
! 210: case 4: element2ctypes(c, to_bool(cell), fpcre_tables, ctype_letter); break;
! 211: case 5: element2ctypes(c, to_bool(cell), fpcre_tables, ctype_word, cbit_word); break;
! 212: case 6: element2case(c, to_wchar_code(cell), fpcre_tables); break;
! 213: #ifdef XML
! 214: case 7:
! 215: case 8:
! 216: // transcoder
! 217: if(toTableSz>MAX_CHARSET_UNI_CODES)
! 218: PTHROW(0, 0,
! 219: &ffile_spec,
! 220: "charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES);
! 221:
! 222: XMLCh unicode=(XMLCh)to_wchar_code(cell);
! 223: if(!unicode && column==7/*unicode1 column*/)
! 224: unicode=(XMLCh)c;
! 225: if(unicode) {
! 226: if(!fromTable[c])
! 227: fromTable[c]=unicode;
! 228: toTable[toTableSz].intCh=unicode;
! 229: toTable[toTableSz].extCh=(XMLByte)c;
! 230: toTableSz++;
! 231: }
! 232: break;
! 233: #endif
! 234: }
! 235: }
! 236: };
! 237:
! 238: #ifdef XML
! 239: // sort by the Unicode code point
! 240: _qsort(toTable, toTableSz, sizeof(*toTable),
! 241: sort_cmp_Trans_rec_intCh);
! 242:
! 243: // addEncoding
! 244: XalanDOMString sencoding(fname.cstr());
! 245: #ifdef XALAN_HACK_DIGITAL_ENTITIES
! 246: hack_s_maximumCharacterValues(sencoding);
! 247: #endif
! 248: const XMLCh* const auto_encoding_cstr=sencoding.c_str();
! 249: int size=sizeof(XMLCh)*(sencoding.size()+1);
! 250: XMLCh* pool_encoding_cstr=(XMLCh*)malloc(size);
! 251: memcpy(pool_encoding_cstr, auto_encoding_cstr, size);
! 252: XMLString::upperCase(pool_encoding_cstr);
! 253:
! 254: /// @todo delete prev encoding with same name
! 255: XMLPlatformUtils::fgTransService->addEncoding(
! 256: pool_encoding_cstr,
! 257: new ENameMapFor2<XML256TableTranscoder2>(
! 258: pool_encoding_cstr
! 259: , fromTable
! 260: , toTable
! 261: , toTableSz
! 262: ));
! 263: #endif
! 264:
! 265: prev_disk_time=new_disk_time;
! 266: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_charset_connection.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main