|
|
| version 1.33.2.19.2.1, 2003/03/18 15:14:18 | version 1.33.2.19.2.6, 2003/03/24 13:43:41 |
|---|---|
| Line 68 static void element2case(unsigned char f | Line 68 static void element2case(unsigned char f |
| // methods | // methods |
| extern "C" unsigned char pcre_default_tables[]; // pcre/chartables.c | extern "C" unsigned char pcre_default_tables[]; // pcre/chartables.c |
| Charset::Charset(const String& aname, const String& afile_spec): | Charset::Charset(const String& aname, const String* afile_spec): |
| fname(aname), | fname(aname), |
| fname_cstr(aname->cstr()) { | fname_cstr(aname.cstrm()) { |
| for(char *c=fname_cstr; *c; c++) | for(char *c=fname_cstr; *c; c++) |
| *c = toupper(*c); | *c = toupper(*c); |
| if(afile_spec) { | if(afile_spec) { |
| fisUTF8=false; | fisUTF8=false; |
| load_definition(afile_spec); | load_definition(*afile_spec); |
| #ifdef XML | #ifdef XML |
| addEncoding(fname_cstr); | addEncoding(fname_cstr); |
| #endif | #endif |
| Line 117 void Charset::load_definition(const Stri | Line 117 void Charset::load_definition(const Stri |
| tables.toTableSize++; | tables.toTableSize++; |
| // loading text | // loading text |
| Pool pool_for_load_only; | char *data=file_read_text(UTF8_charset, afile_spec); |
| char *data=file_read_text(pool_for_load_only, *UTF8_charset, afile_spec); | |
| // ignore header | // ignore header |
| getrow(&data); | getrow(&data); |
| Line 148 void Charset::load_definition(const Stri | Line 147 void Charset::load_definition(const Stri |
| // charset | // charset |
| if(tables.toTableSize>MAX_CHARSET_UNI_CODES) | if(tables.toTableSize>MAX_CHARSET_UNI_CODES) |
| throw Exception("parser.runtime", | throw Exception("parser.runtime", |
| afile_spec, | &afile_spec, |
| "charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES); | "charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES); |
| XMLCh unicode=(XMLCh)to_wchar_code(cell); | XMLCh unicode=(XMLCh)to_wchar_code(cell); |
| Line 220 void Charset::transcode( | Line 219 void Charset::transcode( |
| switch((source_charset.isUTF8()?0x10:0x00)|(dest_charset.isUTF8()?0x01:0x00)) { | switch((source_charset.isUTF8()?0x10:0x00)|(dest_charset.isUTF8()?0x01:0x00)) { |
| default: // 0x00 | default: // 0x00 |
| source_charset.transcodeToCharset(pool, dest_charset, | source_charset.transcodeToCharset(dest_charset, |
| source_body, source_content_length, | source_body, source_content_length, |
| dest_body, dest_content_length); | dest_body, dest_content_length); |
| break; | break; |
| case 0x01: | case 0x01: |
| source_charset.transcodeToUTF8(pool, | source_charset.transcodeToUTF8( |
| source_body, source_content_length, | source_body, source_content_length, |
| dest_body, dest_content_length); | dest_body, dest_content_length); |
| break; | break; |
| case 0x10: | case 0x10: |
| dest_charset.transcodeFromUTF8(pool, | dest_charset.transcodeFromUTF8( |
| source_body, source_content_length, | source_body, source_content_length, |
| dest_body, dest_content_length); | dest_body, dest_content_length); |
| break; | break; |
| case 0x11: | case 0x11: |
| dest_body=pool.copy((char*)source_body, dest_content_length=source_content_length); | dest_body=source_body; |
| dest_content_length=source_content_length; | |
| break; | break; |
| } | } |
| } | } |
| Line 408 static int transcodeFromUTF8( | Line 408 static int transcodeFromUTF8( |
| default: | default: |
| throw Exception(0, | throw Exception(0, |
| Exception::undefined_source, | 0, |
| "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); | "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); |
| } | } |
| tmpVal-=gUTFOffsets[trailingBytes]; | tmpVal-=gUTFOffsets[trailingBytes]; |
| Line 423 static int transcodeFromUTF8( | Line 423 static int transcodeFromUTF8( |
| outPtr+=sprintf((char *)outPtr, "&#%d;", tmpVal); // &#decimal; | outPtr+=sprintf((char *)outPtr, "&#%d;", tmpVal); // &#decimal; |
| } else | } else |
| throw Exception(0, | throw Exception(0, |
| Exception::undefined_source, | 0, |
| "transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal); | "transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal); |
| } | } |
| Line 486 void Charset::transcodeToCharset( | Line 486 void Charset::transcodeToCharset( |
| const void* source_body, size_t source_content_length, | const void* source_body, size_t source_content_length, |
| const void *& adest_body, size_t& adest_content_length) const { | const void *& adest_body, size_t& adest_content_length) const { |
| if(&dest_charset==this) { | if(&dest_charset==this) { |
| adest_body=pool.copy((char*)source_body, adest_content_length=source_content_length); | adest_body=source_body; |
| adest_content_length=source_content_length; | |
| } else { | } else { |
| size_t dest_content_length=source_content_length; | size_t dest_content_length=source_content_length; |
| unsigned char *dest_body=new unsigned char[dest_content_length]; | unsigned char *dest_body=new unsigned char[dest_content_length]; |
| Line 565 const char* Charset::transcode_cstrxmlCh | Line 566 const char* Charset::transcode_cstrxmlCh |
| int inlen=strlen((const char* )s); | int inlen=strlen((const char* )s); |
| int outlen=inlen+1; // max | int outlen=inlen+1; // max |
| char *out=new char[outlen]; | char *out=new(PointerFreeGC) char[outlen]; |
| int error; | int error; |
| if(xmlCharEncodingOutputFunc output=transcoder(Exception::undefined_source).output) { | if(xmlCharEncodingOutputFunc output=transcoder(0).output) { |
| error=output( | error=output( |
| (unsigned char*)out, &outlen, | (unsigned char*)out, &outlen, |
| (const unsigned char*)s, &inlen, | (const unsigned char*)s, &inlen, |
| transcoder(Exception::undefined_source).outputInfo); | transcoder(0).outputInfo); |
| } else { | } else { |
| memcpy(out, s, outlen=inlen); | memcpy(out, s, outlen=inlen); |
| error=0; | error=0; |
| } | } |
| if(error<0) | if(error<0) |
| throw Exception(Exception::undefined_type, | throw Exception(0, |
| Exception::undefined_source, | 0, |
| "transcode_cstr failed (%d)", error); | "transcode_cstr failed (%d)", error); |
| out[outlen/*surely would be less then on input*/]=0; | out[outlen/*surely would be less then on input*/]=0; |
| Line 592 const String& Charset::transcodexmlChar* | Line 593 const String& Charset::transcodexmlChar* |
| ) { | ) { |
| const String& result(new String()); | const String& result(new String()); |
| result->APPEND_CLEAN( | result->APPEND_CLEAN( |
| transcode_cstr(pool, s), 0/*auto-size*/, | transcode_cstr(s), 0/*auto-size*/, |
| origin->origin().file, origin->origin().line); | origin->origin().file, origin->origin().line); |
| return result; | return result; |
| } | } |
| const char* Charset::transcode_cstrGdomeDOMString* s) { | const char* Charset::transcode_cstrGdomeDOMString* s) { |
| return s?transcode_cstr(pool, BAD_CAST s->str):""; | return s?transcode_cstr(BAD_CAST s->str):""; |
| } | } |
| const String& Charset::transcodeGdomeDOMString* s | const String& Charset::transcodeGdomeDOMString* s |
| #ifndef NO_STRING_ORIGIN | #ifndef NO_STRING_ORIGIN |
| Line 605 const String& Charset::transcodeGdomeDOM | Line 606 const String& Charset::transcodeGdomeDOM |
| #endif | #endif |
| ) { | ) { |
| const String& result(new String()); | const String& result(new String()); |
| result->APPEND_CLEAN(transcode_cstr(pool, s), 0/*auto-size*/, origin->origin().file, origin->origin().line); | result->APPEND_CLEAN(transcode_cstr(s), 0/*auto-size*/, origin->origin().file, origin->origin().line); |
| return result; | return result; |
| } | } |
| Line 615 void* Charset::transcode_buf2mchar(trans | Line 616 void* Charset::transcode_buf2mchar(trans |
| unsigned char* out; | unsigned char* out; |
| int outlen; | int outlen; |
| int error; | int error; |
| if(xmlCharEncodingInputFunc input=transcoder(Exception::undefined_source).input) { | if(xmlCharEncodingInputFunc input=transcoder(0).input) { |
| outlen=buf_size*6/*max*/; | outlen=buf_size*6/*max*/; |
| out=(unsigned char*)malloc_func(outlen+1); | out=(unsigned char*)malloc_func(outlen+1); |
| error=input( | error=input( |
| out, &outlen, | out, &outlen, |
| (const unsigned char *)buf, (int *)&buf_size, | (const unsigned char *)buf, (int *)&buf_size, |
| transcoder(Exception::undefined_source).inputInfo); | transcoder(0).inputInfo); |
| } else { | } else { |
| outlen=buf_size; | outlen=buf_size; |
| out=(unsigned char*)malloc_func(outlen+1); | out=(unsigned char*)malloc_func(outlen+1); |
| Line 630 void* Charset::transcode_buf2mchar(trans | Line 631 void* Charset::transcode_buf2mchar(trans |
| } | } |
| if(error<0) | if(error<0) |
| throw Exception(Exception::undefined_type, | throw Exception(0, |
| Exception::undefined_source, | 0, |
| "transcode_buf failed (%d)", error); | "transcode_buf failed (%d)", error); |
| out[outlen/*surely would be less then on input*/]=0; | out[outlen/*surely would be less then on input*/]=0; |
| Line 651 GdomeDOMString_auto_ptr Charset::transco | Line 652 GdomeDOMString_auto_ptr Charset::transco |
| return GdomeDOMString_auto_ptr(transcode_buf2gchar(buf, buf_size)); | return GdomeDOMString_auto_ptr(transcode_buf2gchar(buf, buf_size)); |
| } | } |
| GdomeDOMString_auto_ptr Charset::transcode(const String& s) { | GdomeDOMString_auto_ptr Charset::transcode(const String& s) { |
| const char* cstr=s->cstr(String::UL_UNSPECIFIED); | const char* cstr=s->cstr(String::L_UNSPECIFIED); |
| return transcode_buf2dom(cstr, strlen(cstr)); | return transcode_buf2dom(cstr, strlen(cstr)); |
| } | } |