--- parser3/src/main/pa_charset.C 2002/09/10 08:41:00 1.29 +++ parser3/src/main/pa_charset.C 2003/01/27 15:35:24 1.33.2.1 @@ -1,13 +1,14 @@ /** @file Parser: Charset connection implementation. - Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) Author: Alexander Petrosyan(http://paf.design.ru) */ -static const char* IDENT_CHARSET_C="$Date: 2002/09/10 08:41:00 $"; +static const char* IDENT_CHARSET_C="$Date: 2003/01/27 15:35:24 $"; #include "pa_charset.h" +#include "pa_globals.h" #ifdef XML #include "libxml/encoding.h" @@ -74,18 +75,18 @@ static void element2case(unsigned char f // methods extern "C" unsigned char pcre_default_tables[]; // pcre/chartables.c -Charset::Charset(Pool& apool, const String& aname, const String *request_file_spec): Pooled(apool), - fname(aname) { +Charset::Charset(Pool& pool, ConstStringPtr aname, ConstStringPtr afile_spec): + fname(aname), + fname_cstr(aname->cstr()) { - char *name_cstr=fname.cstr(); - for(char *c=name_cstr; *c; c++) + for(char *c=fname_cstr; *c; c++) *c = toupper(*c); - if(request_file_spec) { + if(afile_spec) { fisUTF8=false; - loadDefinition(*request_file_spec); + loadDefinition(pool, fname); #ifdef XML - addEncoding(name_cstr); + addEncoding(fname_cstr); #endif } else { fisUTF8=true; @@ -94,17 +95,18 @@ Charset::Charset(Pool& apool, const Stri } #ifdef XML - initTranscoder(&aname, name_cstr); + initTranscoder(fname, fname_cstr); #endif } Charset::~Charset() { + // @todonow unregister encodings #ifdef XML // not deleting transcoder, that's not our business #endif } -void Charset::loadDefinition(const String& request_file_spec) { +void Charset::loadDefinition(Pool& pool, ConstStringPtr afile_spec) { // pcre_tables // lowcase, flipcase, bits digit+word+whitespace, masks @@ -116,7 +118,7 @@ void Charset::loadDefinition(const Strin // charset memset(tables.fromTable, 0, sizeof(tables.fromTable)); - tables.toTable=(Charset_TransRec *)calloc(sizeof(Charset_TransRec)*MAX_CHARSET_UNI_CODES); + tables.toTable=(Charset_TransRec *)pool.calloc(sizeof(Charset_TransRec)*MAX_CHARSET_UNI_CODES); tables.toTableSize=0; // strangly vital tables.toTable[tables.toTableSize].intCh=0; @@ -124,7 +126,7 @@ void Charset::loadDefinition(const Strin tables.toTableSize++; // loading text - char *data=file_read_text(pool(), request_file_spec); + char *data=file_read_text(pool, utf8_charset, afile_spec); // ignore header getrow(&data); @@ -154,7 +156,7 @@ void Charset::loadDefinition(const Strin // charset if(tables.toTableSize>MAX_CHARSET_UNI_CODES) throw Exception("parser.runtime", - &request_file_spec, + afile_spec, "charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES); XMLCh unicode=(XMLCh)to_wchar_code(cell); @@ -372,7 +374,7 @@ of ocetes consumed. return 0; } /// @todo digital entites only when xml/html output [at output in html/xml mode, in html part of a letter] -static size_t transcodeFromUTF8( +static int transcodeFromUTF8( const XMLByte *srcData, size_t& srcLen, XMLByte* toFill, size_t& toFillLen, const Charset::Tables& tables) { @@ -415,7 +417,7 @@ static size_t transcodeFromUTF8( default: throw Exception(0, - 0, + Exception::undefined_source, "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); } tmpVal-=gUTFOffsets[trailingBytes]; @@ -430,7 +432,7 @@ static size_t transcodeFromUTF8( outPtr+=sprintf((char *)outPtr, "&#%d;", tmpVal); // &#decimal; } else throw Exception(0, - 0, + Exception::undefined_source, "transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal); } @@ -574,52 +576,71 @@ const char *Charset::transcode_cstr(xmlC int outlen=inlen+1; // max char *out=(char *)malloc(outlen*sizeof(char)); - int size; + int error; if(xmlCharEncodingOutputFunc output=transcoder(0)->output) { - size=output( + error=output( (unsigned char*)out, &outlen, (const unsigned char*)s, &inlen, transcoder(0)->outputInfo); - } else - memcpy(out, s, size=inlen); - if(size<0) + } else { + memcpy(out, s, outlen=inlen); + error=0; + } + if(error<0) throw Exception(0, 0, - "transcode_cstr failed (%d)", size); + "transcode_cstr failed (%d)", error); - out[size]=0; + out[outlen/*surely would be less then on input*/]=0; return out; } -String& Charset::transcode(xmlChar *s) { - return *NEW String(pool(), transcode_cstr(s)); +String& Charset::transcode(xmlChar *s +#ifndef NO_STRING_ORIGIN + , const String *origin +#endif + ) { + String& result=*NEW String(pool()); + result.APPEND_CLEAN(transcode_cstr(s), 0/*auto-size*/, origin->origin().file, origin->origin().line); + return result; } const char *Charset::transcode_cstr(GdomeDOMString *s) { return s?transcode_cstr(BAD_CAST s->str):""; } -String& Charset::transcode(GdomeDOMString *s) { - return *NEW String(pool(), transcode_cstr(s)); +String& Charset::transcode(GdomeDOMString *s +#ifndef NO_STRING_ORIGIN + , const String *origin +#endif + ) { + String& result=*NEW String(pool()); + result.APPEND_CLEAN(transcode_cstr(s), 0/*auto-size*/, origin->origin().file, origin->origin().line); + return result; } /// @test less memory using -maybe- xmlParserInputBufferCreateMem xmlChar *Charset::transcode_buf2xchar(const char *buf, size_t buf_size) { - int outlen=buf_size*6/*max*/+1; - unsigned char *out=(unsigned char*)malloc(outlen*sizeof(unsigned char)); - - int size; + unsigned char *out; + int outlen; + int error; if(xmlCharEncodingInputFunc input=transcoder(0)->input) { - size=input( + outlen=buf_size*6/*max*/; + out=(unsigned char*)malloc((outlen+1)*sizeof(unsigned char)); + error=input( out, &outlen, (const unsigned char *)buf, (int *)&buf_size, transcoder(0)->inputInfo); - } else - memcpy(out, buf, size=buf_size); + } else { + outlen=buf_size; + out=(unsigned char*)malloc((outlen+1)*sizeof(unsigned char)); + memcpy(out, buf, outlen); + error=0; + } - if(size<0) + if(error<0) throw Exception(0, 0, - "transcode_buf failed (%d)", size); + "transcode_buf failed (%d)", error); - out[size]=0; + out[outlen/*surely would be less then on input*/]=0; return (xmlChar *)out; } GdomeDOMString_auto_ptr Charset::transcode_buf2dom(const char *buf, size_t buf_size) {