--- parser3/src/main/pa_charset.C 2009/04/17 09:04:49 1.70 +++ parser3/src/main/pa_charset.C 2009/05/26 21:33:05 1.77 @@ -5,7 +5,7 @@ Author: Alexander Petrosyan(http://paf.design.ru) */ -static const char * const IDENT_CHARSET_C="$Date: 2009/04/17 09:04:49 $"; +static const char * const IDENT_CHARSET_C="$Date: 2009/05/26 21:33:05 $"; #include "pa_charset.h" #include "pa_charsets.h" @@ -270,9 +270,9 @@ static const XMLByte gFirstByteMark[7] = 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC }; -static int transcodeToUTF8(const XMLByte* srcData, size_t& srcLen, - XMLByte *toFill, size_t& toFillLen, - const Charset::Tables& tables) { +static int transcodeToUTF8(const XMLByte* srcData, int& srcLen, + XMLByte *toFill, int& toFillLen, + const Charset::Tables& tables) { const XMLByte* srcPtr=srcData; const XMLByte* srcEnd=srcData+srcLen; XMLByte* outPtr=toFill; @@ -353,9 +353,9 @@ 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 int transcodeFromUTF8(const XMLByte* srcData, size_t& srcLen, - XMLByte* toFill, size_t& toFillLen, - const Charset::Tables& tables) { +static int transcodeFromUTF8(const XMLByte* srcData, int& srcLen, + XMLByte* toFill, int& toFillLen, + const Charset::Tables& tables) { const XMLByte* srcPtr=srcData; const XMLByte* srcEnd=srcData+srcLen; XMLByte* outPtr=toFill; @@ -642,21 +642,21 @@ String::Body Charset::escape(const Strin String::C dest=Charset::escape(String::C(src_ptr, src_size), source_charset); - return String::Body(dest.str, dest.length); + return String::Body(dest.length ? dest.str:0); } String& Charset::escape(const String& src, const Charset& source_charset) { - if(!src.length()) - return *new String("", 0, false); + if(src.is_empty()) + return *new String(); return *new String(escape((String::Body)src, source_charset), String::L_CLEAN); } const String::C Charset::transcodeToUTF8(const String::C src) const { - size_t src_length=src.length; + int src_length=src.length; #ifdef PRECALCULATE_DEST_LENGTH - size_t dest_length=0; + int dest_length=0; const XMLByte* srcPtr=(XMLByte*)src.str; const XMLByte* srcEnd=srcPtr+src_length; XMLByte firstByte; @@ -664,13 +664,13 @@ const String::C Charset::transcodeToUTF8 while(uint charSize=readChar(srcPtr, srcEnd, firstByte, UTF8Char, tables)) dest_length+=charSize; #else - size_t dest_length=src_length*6; // so that surly enough (max utf8 seq len=6) but too memory-hyngry + int dest_length=src_length*6; // so that surly enough (max utf8 seq len=6) but too memory-hyngry #endif //throw Exception(0,0,"%u",dest_length); #ifndef NDEBUG - size_t saved_dest_length=dest_length; + int saved_dest_length=dest_length; #endif XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_length+1/*for terminator*/]; @@ -828,10 +828,10 @@ static size_t getDecNumLength(XMLCh UTF8 } const String::C Charset::transcodeFromUTF8(const String::C src) const { - size_t src_length=src.length; + int src_length=src.length; #ifdef PRECALCULATE_DEST_LENGTH - size_t dest_length=0; + int dest_length=0; const XMLByte* srcPtr=(XMLByte*)src.str; const XMLByte* srcEnd=srcPtr+src_length; XMLByte firstByte; @@ -848,13 +848,11 @@ const String::C Charset::transcodeFromUT } #else // so that surly enough, "&#XXX;" has max ratio (huh? 8 bytes needed for '&#XXXXX;') - size_t dest_length=src_length*6; + int dest_length=src_length*6; #endif - //throw Exception(0,0,"%u",dest_length); - #ifndef NDEBUG - size_t saved_dest_length=dest_length; + int saved_dest_length=dest_length; #endif XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_length+1/*for terminator*/]; @@ -912,16 +910,16 @@ static const Charset::Tables* tables[MAX unsigned char *out, int *outlen, \ const unsigned char *in, int *inlen, void*) { \ return transcodeToUTF8( \ - in, *(size_t*)inlen, \ - out, *(size_t*)outlen, \ + in, *inlen, \ + out, *outlen, \ *tables[i]); \ } \ static int xml256CharEncodingOutputFunc##i( \ unsigned char *out, int *outlen, \ const unsigned char *in, int *inlen, void*) { \ return transcodeFromUTF8( \ - in, *(size_t*)inlen, \ - out, *(size_t*)outlen, \ + in, *inlen, \ + out, *outlen, \ *tables[i]); \ } @@ -932,16 +930,16 @@ static const Charset::Tables* tables[MAX unsigned char *out, int *outlen, \ const unsigned char *in, int *inlen) { \ return transcodeToUTF8( \ - in, *(size_t*)inlen, \ - out, *(size_t*)outlen, \ + in, *inlen, \ + out, *outlen, \ *tables[i]); \ } \ static int xml256CharEncodingOutputFunc##i( \ unsigned char *out, int *outlen, \ const unsigned char *in, int *inlen) { \ return transcodeFromUTF8( \ - in, *(size_t*)inlen, \ - out, *(size_t*)outlen, \ + in, *inlen, \ + out, *outlen, \ *tables[i]); \ } @@ -1037,7 +1035,7 @@ String::C Charset::transcode_cstr(const } const String& Charset::transcode(const xmlChar* s) { String::C cstr=transcode_cstr(s); - return *new String(cstr.str, cstr.length, true); + return *new String(cstr.str, String::L_TAINTED); } /// @test less memory using -maybe- xmlParserInputBufferCreateMem @@ -1098,18 +1096,16 @@ String::Body Charset::transcode(const St const char *src_ptr=src.cstr(); size_t src_size=strlen(src_ptr); - String::C dest=Charset::transcode(String::C(src_ptr, src_size), - source_transcoder, - dest_transcoder); + String::C dest=Charset::transcode(String::C(src_ptr, src_size), source_transcoder, dest_transcoder); - return String::Body(dest.str, dest.length); + return String::Body(dest.length ? dest.str:0); } String& Charset::transcode(const String& src, const Charset& source_transcoder, const Charset& dest_transcoder) { - if(!src.length()) - return *new String("", 0, false); + if(src.is_empty()) + return *new String(); return *new String(transcode((String::Body)src, source_transcoder, dest_transcoder), String::L_CLEAN); } @@ -1127,7 +1123,7 @@ struct Transcode_pair_info { const Charset* dest_transcoder; }; #endif -static void transcode_pair(const String::Body /*akey*/, +static void transcode_pair(HashStringValue::key_type /*akey*/, String::Body& avalue, Transcode_pair_info* info) { avalue=Charset::transcode(avalue,