--- parser3/src/main/pa_charset.C 2003/04/10 09:59:16 1.33.2.19.2.24 +++ parser3/src/main/pa_charset.C 2005/08/09 08:14:51 1.52 @@ -1,11 +1,11 @@ /** @file Parser: Charset connection implementation. - Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexander Petrosyan(http://paf.design.ru) */ -static const char* IDENT_CHARSET_C="$Date: 2003/04/10 09:59:16 $"; +static const char * const IDENT_CHARSET_C="$Date: 2005/08/09 08:14:51 $"; #include "pa_charset.h" #include "pa_charsets.h" @@ -14,6 +14,24 @@ static const char* IDENT_CHARSET_C="$Dat #include "libxml/encoding.h" #endif +//#define PA_PATCHED_LIBXML_BACKWARD + +// globals + +Charset::UTF8CaseTable::Rec UTF8CaseToUpperRecords[]={ +#include "utf8-to-upper.inc" +}; +Charset::UTF8CaseTable UTF8CaseToUpper={ + sizeof(UTF8CaseToUpperRecords)/sizeof(Charset::UTF8CaseTable::Rec), + UTF8CaseToUpperRecords}; + +Charset::UTF8CaseTable::Rec UTF8CaseToLowerRecords[]={ +#include "utf8-to-lower.inc" +}; +Charset::UTF8CaseTable UTF8CaseToLower={ + sizeof(UTF8CaseToLowerRecords)/sizeof(Charset::UTF8CaseTable::Rec), + UTF8CaseToLowerRecords}; + // helpers inline void prepare_case_tables(unsigned char *tables) { @@ -68,7 +86,7 @@ static void element2case(unsigned char f // methods extern "C" unsigned char pcre_default_tables[]; // pcre/chartables.c -Charset::Charset(Request_charsets* charsets, const StringBody ANAME, const String* afile_spec): +Charset::Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec): FNAME(ANAME), FNAME_CSTR(ANAME.cstrm()) { @@ -101,10 +119,6 @@ void Charset::load_definition(Request_ch // charset memset(&tables, 0, sizeof(tables)); - // strangly vital - tables.toTable[tables.toTableSize].intCh=0; - tables.toTable[tables.toTableSize].extCh=(XMLByte)0; - tables.toTableSize++; // loading text char *data=file_read_text(charsets, afile_spec); @@ -114,7 +128,7 @@ void Charset::load_definition(Request_ch // parse cells char *row; - while(row=getrow(&data)) { + while((row=getrow(&data))) { // remove empty&comment lines if(!*row || *row=='#') continue; @@ -122,7 +136,7 @@ void Charset::load_definition(Request_ch // char white-space digit hex-digit letter word lowercase unicode1 unicode2 unsigned int c=0; char *cell; - for(int column=0; cell=lsplit(&row, '\t'); column++) { + for(int column=0; (cell=lsplit(&row, '\t')); column++) { switch(column) { case 0: c=to_wchar_code(cell); break; // pcre_tables @@ -161,8 +175,8 @@ void Charset::load_definition(Request_ch static int sort_cmp_Trans_rec_intCh(const void *a, const void *b) { return - static_cast(a)->intCh- - static_cast(b)->intCh; + static_cast(a)->intCh- + static_cast(b)->intCh; } void Charset::sort_ToTable() { @@ -176,23 +190,20 @@ void Charset::sort_ToTable() { static XMLByte xlatOneTo(const XMLCh toXlat, const Charset::Tables& tables, XMLByte not_found) { - unsigned int lowOfs = 0; - unsigned int hiOfs = tables.toTableSize - 1; - XMLByte curByte = 0; - do { + int lo = 0; + int hi = tables.toTableSize - 1; + while(lo<=hi) { // Calc the mid point of the low and high offset. - const unsigned int midOfs =((hiOfs - lowOfs) / 2)+lowOfs; - - // If our test char is greater than the mid point char, then - // we move up to the upper half. Else we move to the lower - // half. If its equal, then its our guy. - if(toXlat>tables.toTable[midOfs].intCh) - lowOfs = midOfs; - else if(toXlatcur) + lo = i+1; else - return tables.toTable[midOfs].extCh; - } while(lowOfs+1cur) + lo = i+1; + else + hi = i-1; + } + + // not found + return src; +} + +static void store_UTF8(XMLCh src, XMLByte*& outPtr ) { + if(!src) { + // use the replacement character + *outPtr++= '?'; + return; + } + + // Figure out how many bytes we need + unsigned int encodedBytes; + if(src<0x80) + encodedBytes = 1; + else if(src<0x800) + encodedBytes = 2; + else if(src<0x10000) + encodedBytes = 3; + else if(src<0x200000) + encodedBytes = 4; + else if(src<0x4000000) + encodedBytes = 5; + else if(src<= 0x7FFFFFFF) + encodedBytes = 6; + else { + // use the replacement character + *outPtr++= '?'; + return; + } + + // And spit out the bytes. We spit them out in reverse order + // here, so bump up the output pointer and work down as we go. + outPtr+= encodedBytes; + switch(encodedBytes) { + case 6: *--outPtr = XMLByte((src | 0x80UL) & 0xBFUL); + src>>= 6; + case 5: *--outPtr = XMLByte((src | 0x80UL) & 0xBFUL); + src>>= 6; + case 4: *--outPtr = XMLByte((src | 0x80UL) & 0xBFUL); + src>>= 6; + case 3: *--outPtr = XMLByte((src | 0x80UL) & 0xBFUL); + src>>= 6; + case 2: *--outPtr = XMLByte((src | 0x80UL) & 0xBFUL); + src>>= 6; + case 1: *--outPtr = XMLByte(src | gFirstByteMark[encodedBytes]); + } + + // Add the encoded bytes back in again to indicate we've eaten them + outPtr+= encodedBytes; +} + +static void change_case_UTF8(XMLCh src, XMLByte*& outPtr, + const Charset::UTF8CaseTable& table) { + store_UTF8(change_case_UTF8(src, table), outPtr); +}; +void change_case_UTF8(const XMLByte* srcData, size_t srcLen, + XMLByte* toFill, size_t toFillLen, + const Charset::UTF8CaseTable& table) { + const XMLByte* srcPtr=srcData; + const XMLByte* srcEnd=srcData+srcLen; + XMLByte* outPtr=toFill; + XMLByte* outEnd=toFill+toFillLen; + + // We now loop until we either run out of input data, or room to store + while ((srcPtr < srcEnd) && (outPtr < outEnd)) { + // Get the next leading byte out + const XMLByte firstByte =* srcPtr; + + if(firstByte<= 127) { + change_case_UTF8(firstByte, outPtr, table); + srcPtr++; + continue; + } + + // See how many trailing src bytes this sequence is going to require + const unsigned int trailingBytes = gUTFBytes[firstByte]; + + // Looks ok, so lets build up the value + uint tmpVal=0; + switch(trailingBytes) { + case 5: tmpVal+=*srcPtr++; tmpVal<<=6; + case 4: tmpVal+=*srcPtr++; tmpVal<<=6; + case 3: tmpVal+=*srcPtr++; tmpVal<<=6; + case 2: tmpVal+=*srcPtr++; tmpVal<<=6; + case 1: tmpVal+=*srcPtr++; tmpVal<<=6; + case 0: tmpVal+=*srcPtr++; + break; + + default: + throw Exception(0, + 0, + "change_case_UTF8 error: wrong trailingBytes value(%d)", trailingBytes); + } + tmpVal-=gUTFOffsets[trailingBytes]; + + // If it will fit into a single char, then put it in. Otherwise + // fail [*encode it as a surrogate pair. If its not valid, use the + // replacement char.*] + if(!(tmpVal & 0xFFFF0000)) + change_case_UTF8(tmpVal, outPtr, table); + else + throw Exception(0, + 0, + "change_case_UTF8 error: too big tmpVal(0x%08X)", tmpVal); + } + + if(srcPtr!=outPtr) + throw Exception(0, + 0, + "change_case_UTF8 error: end pointers do not match"); +} + + const String::C Charset::transcodeFromUTF8(const String::C src) const { size_t src_length=src.length; size_t dest_length=src.length*6/*so that surly enough, "ÿ" has max ratio */; @@ -449,7 +593,7 @@ const String::C Charset::transcodeFromUT (XMLByte *)src.str, src_length, dest_body, dest_length, tables)<0) - throw(0, 0, + throw Exception(0, 0, "Charset::transcodeFromUTF8 buffer overflow"); @@ -484,6 +628,28 @@ const String::C Charset::transcodeToChar static const Charset::Tables* tables[MAX_CHARSETS]; +#ifdef PA_PATCHED_LIBXML_BACKWARD + +#define declareXml256ioFuncs(i) \ + static int xml256CharEncodingInputFunc##i( \ + unsigned char *out, int *outlen, \ + const unsigned char *in, int *inlen, void*) { \ + return transcodeToUTF8( \ + in, *(size_t*)inlen, \ + out, *(size_t*)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, \ + *tables[i]); \ + } + +#else + #define declareXml256ioFuncs(i) \ static int xml256CharEncodingInputFunc##i( \ unsigned char *out, int *outlen, \ @@ -502,6 +668,9 @@ static const Charset::Tables* tables[MAX *tables[i]); \ } +#endif + + declareXml256ioFuncs(0) declareXml256ioFuncs(1) declareXml256ioFuncs(2) declareXml256ioFuncs(3) declareXml256ioFuncs(4) declareXml256ioFuncs(5) @@ -531,7 +700,7 @@ void Charset::addEncoding(char *name_cst "already allocated %d handlers, no space for new encoding '%s'", MAX_CHARSETS, name_cstr); - xmlCharEncodingHandler* handler=new(PointerFreeGC) xmlCharEncodingHandler; + xmlCharEncodingHandler* handler=new(UseGC) xmlCharEncodingHandler; { handler->name=name_cstr; handler->input=inputFuncs[handlers_count]; @@ -544,12 +713,12 @@ void Charset::addEncoding(char *name_cst } -void Charset::initTranscoder(const StringBody NAME, const char* name_cstr) { +void Charset::initTranscoder(const String::Body NAME, const char* name_cstr) { ftranscoder=xmlFindCharEncodingHandler(name_cstr); transcoder(NAME); // check right way } -xmlCharEncodingHandler& Charset::transcoder(const StringBody NAME) { +xmlCharEncodingHandler& Charset::transcoder(const String::Body NAME) { if(!ftranscoder) throw Exception("parser.runtime", new String(NAME, String::L_TAINTED), @@ -562,7 +731,7 @@ String::C Charset::transcode_cstr(xmlCha return String::C("", 0); int inlen=strlen((const char*)s); - int outlen=inlen; // max + int outlen=inlen*6/*strlen("ÿ")*/; // max #ifndef NDEBUG int saved_outlen=outlen; #endif @@ -572,7 +741,11 @@ String::C Charset::transcode_cstr(xmlCha if(xmlCharEncodingOutputFunc output=transcoder(FNAME).output) { error=output( (unsigned char*)out, &outlen, - (const unsigned char*)s, &inlen); + (const unsigned char*)s, &inlen +#ifdef PA_PATCHED_LIBXML_BACKWARD + ,0 +#endif + ); } else { memcpy(out, s, outlen=inlen); error=0; @@ -599,29 +772,32 @@ const String& Charset::transcode(GdomeDO } /// @test less memory using -maybe- xmlParserInputBufferCreateMem -void* Charset::transcode_buf2mchar(transcode_buf_malloc_func malloc_func, - const char* buf, size_t buf_size) { - unsigned char* out; +xmlChar* Charset::transcode_buf2xchar(const char* buf, size_t buf_size) { + xmlChar* out; int outlen; int error; #ifndef NDEBUG int saved_outlen; #endif if(xmlCharEncodingInputFunc input=transcoder(FNAME).input) { - outlen=buf_size*6/*max*/; + outlen=buf_size*6/*max UTF8 bytes per char*/; #ifndef NDEBUG saved_outlen=outlen; #endif - out=(unsigned char*)malloc_func(outlen+1); + out=(xmlChar*)xmlMalloc(outlen+1); error=input( out, &outlen, - (const unsigned char*)buf, (int*)&buf_size); + (const unsigned char*)buf, (int*)&buf_size +#ifdef PA_PATCHED_LIBXML_BACKWARD + ,0 +#endif + ); } else { outlen=buf_size; #ifndef NDEBUG saved_outlen=outlen; #endif - out=(unsigned char*)malloc_func(outlen+1); + out=(xmlChar*)xmlMalloc(outlen+1); memcpy(out, buf, outlen); error=0; } @@ -634,30 +810,67 @@ void* Charset::transcode_buf2mchar(trans assert(outlen<=saved_outlen); out[outlen]=0; return out; } - -xmlChar* Charset::transcode_buf2xchar(const char* buf, size_t buf_size) { - return static_cast(transcode_buf2mchar(xmlMalloc, buf, buf_size)); -} -static void* g_malloc_wrapper(size_t size) { - if(void* out=g_malloc(size)) - return out; - - return pa_fail_alloc("g_malloc_wrapper", size); -} -gchar* Charset::transcode_buf2gchar(const char* buf, size_t buf_size) { - return static_cast(transcode_buf2mchar(g_malloc_wrapper, buf, buf_size)); -} GdomeDOMString_auto_ptr Charset::transcode_buf2dom(const char* buf, size_t buf_size) { - return GdomeDOMString_auto_ptr(transcode_buf2gchar(buf, buf_size)); + return GdomeDOMString_auto_ptr(transcode_buf2xchar(buf, buf_size)); } GdomeDOMString_auto_ptr Charset::transcode(const String& s) { const char* cstr=s.cstr(String::L_UNSPECIFIED); return transcode_buf2dom(cstr, strlen(cstr)); } -GdomeDOMString_auto_ptr Charset::transcode(const StringBody s) { +GdomeDOMString_auto_ptr Charset::transcode(const String::Body s) { const char* cstr=s.cstr(); return transcode_buf2dom(cstr, s.length()); } #endif + +String::Body Charset::transcode(const String::Body src, + const Charset& source_transcoder, + const Charset& dest_transcoder) { + + 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); + + return String::Body(dest.str, dest.length); +} + +String& Charset::transcode(const String& src, + const Charset& source_transcoder, + const Charset& dest_transcoder) { + if(!src.length()) + return *new String("", 0, false); + + return *new String(transcode((String::Body)src, source_transcoder, dest_transcoder), String::L_CLEAN); +} + +void Charset::transcode(ArrayString& src, + const Charset& source_transcoder, + const Charset& dest_transcoder) { + for(size_t i=0; isource_transcoder, + *info->dest_transcoder); +} +void Charset::transcode(HashStringString& src, + const Charset& source_transcoder, + const Charset& dest_transcoder) { + Transcode_pair_info info={&source_transcoder, &dest_transcoder}; + src.for_each_ref(transcode_pair, &info); +}