--- parser3/src/main/pa_charset.C 2003/07/24 15:10:42 1.35.2.1 +++ parser3/src/main/pa_charset.C 2013/10/28 21:59:31 1.95 @@ -1,26 +1,47 @@ /** @file Parser: Charset connection implementation. - Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexander Petrosyan(http://paf.design.ru) */ -static const char* IDENT_CHARSET_C="$Date: 2003/07/24 15:10:42 $"; - #include "pa_charset.h" #include "pa_charsets.h" +volatile const char * IDENT_PA_CHARSET_C="$Id: pa_charset.C,v 1.95 2013/10/28 21:59:31 moko Exp $" IDENT_PA_CHARSET_H; + #ifdef XML #include "libxml/encoding.h" #endif +//#define PA_PATCHED_LIBXML_BACKWARD + +// reduce memory usage by pre-calculation utf-8 string length +#define PRECALCULATE_DEST_LENGTH + +// 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) { unsigned char *lcc_table=tables+lcc_offset; unsigned char *fcc_table=tables+fcc_offset; for(int i=0; i<0x100; i++) - lcc_table[i]=fcc_table[i]=i; + lcc_table[i]=fcc_table[i]=(unsigned char)i; } inline void cstr2ctypes(unsigned char *tables, const unsigned char *cstr, unsigned char bit) { @@ -37,8 +58,7 @@ inline unsigned int to_wchar_code(const if(cstr[1]==0) return(unsigned int)(unsigned char)cstr[0]; - char *error_pos; - return(unsigned int)strtol(cstr, &error_pos, 0); + return pa_atoui(cstr,0); } inline bool to_bool(const char* cstr) { return cstr && *cstr!=0; @@ -65,10 +85,31 @@ static void element2case(unsigned char f fcc_table[from]=to; fcc_table[to]=from; } +inline XMLByte *append_hex_8(XMLByte *dest, unsigned char c, const char* prefix=0) { + if(prefix) { + strcpy((char *)dest, prefix); + dest+=strlen(prefix); + } + *dest++=hex_digits[c >> 4]; + *dest++=hex_digits[c & 0x0F]; + return dest; +} + +inline XMLByte *append_hex_16(XMLByte *dest, unsigned int c, const char* prefix=0) { + if(prefix) { + strcpy((char *)dest, prefix); + dest+=strlen(prefix); + } + *dest++=hex_digits[(c >> 12) & 0x0F]; + *dest++=hex_digits[(c >> 8) & 0x0F]; + *dest++=hex_digits[(c >> 4) & 0x0F]; + *dest++=hex_digits[(c) & 0x0F]; + return dest; +} + // 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()) { @@ -81,7 +122,7 @@ Charset::Charset(Request_charsets* chars } else { fisUTF8=true; // grab default onces [for UTF-8 so to be able to make a-z =>A-Z - memcpy(pcre_tables, pcre_default_tables, sizeof(pcre_tables)); + memcpy(pcre_tables, _pcre_default_tables, sizeof(pcre_tables)); } #ifdef XML @@ -101,10 +142,6 @@ 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); @@ -114,29 +151,29 @@ 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; // char white-space digit hex-digit letter word lowercase unicode1 unicode2 - unsigned int c=0; + unsigned char 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; + case 0: c=(unsigned char)to_wchar_code(cell); break; // pcre_tables case 1: element2ctypes(c, to_bool(cell), pcre_tables, ctype_space, cbit_space); break; case 2: element2ctypes(c, to_bool(cell), pcre_tables, ctype_digit, cbit_digit); break; case 3: element2ctypes(c, to_bool(cell), pcre_tables, ctype_xdigit); break; case 4: element2ctypes(c, to_bool(cell), pcre_tables, ctype_letter); break; case 5: element2ctypes(c, to_bool(cell), pcre_tables, ctype_word, cbit_word); break; - case 6: element2case(c, to_wchar_code(cell), pcre_tables); break; + case 6: element2case(c, (unsigned char)to_wchar_code(cell), pcre_tables); break; case 7: case 8: // charset if(tables.toTableSize>MAX_CHARSET_UNI_CODES) - throw Exception("parser.runtime", + throw Exception(PARSER_RUNTIME, &afile_spec, "charset must contain not more then %d unicode values", MAX_CHARSET_UNI_CODES); @@ -155,44 +192,50 @@ void Charset::load_definition(Request_ch } }; + // parser charset tables declare only white-space before 0x20, thus adding the missing chars + for(uint i=0; i<0x20; i++) + if(!tables.fromTable[i]){ + tables.fromTable[i]=i; + tables.toTable[tables.toTableSize].intCh=i; + tables.toTable[tables.toTableSize].extCh=(XMLByte)i; + tables.toTableSize++; + } + // sort by the Unicode code point sort_ToTable(); } 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() { - _qsort(tables.toTable, tables.toTableSize, sizeof(*tables.toTable), - sort_cmp_Trans_rec_intCh); + qsort(tables.toTable, tables.toTableSize, sizeof(*tables.toTable), sort_cmp_Trans_rec_intCh); //FILE *f=fopen("c:\\temp\\a", "wb"); //fwrite(tables.toTable, tables.toTableSize, sizeof(*tables.toTable), f); //fclose(f); } +// @todo: precache for spedup searching 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+1>= 6; - case 5: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); - curVal>>= 6; - case 4: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); - curVal>>= 6; - case 3: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); - curVal>>= 6; - case 2: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); - curVal>>= 6; - case 1: *--outPtr = XMLByte(curVal | gFirstByteMark[encodedBytes]); + case 6: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); + curVal>>= 6; + case 5: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); + curVal>>= 6; + case 4: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); + curVal>>= 6; + case 3: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); + curVal>>= 6; + case 2: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); + curVal>>= 6; + case 1: *--outPtr = XMLByte(curVal | gFirstByteMark[encodedBytes]); } // Add the encoded bytes back in again to indicate we've eaten them @@ -339,9 +382,9 @@ 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; @@ -353,7 +396,7 @@ static int transcodeFromUTF8(const XMLBy const XMLByte firstByte =* srcPtr; // Special-case ASCII, which is a leading byte value of<= 127 - if(firstByte<= 127) { + if(firstByte<=127) { *outPtr++= firstByte; srcPtr++; continue; @@ -382,7 +425,7 @@ static int transcodeFromUTF8(const XMLBy default: throw Exception(0, 0, - "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); + "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); // never } tmpVal-=gUTFOffsets[trailingBytes]; @@ -392,12 +435,14 @@ static int transcodeFromUTF8(const XMLBy if(!(tmpVal & 0xFFFF0000)) { if(XMLByte xlat=xlatOneTo(tmpVal, tables, 0)) *outPtr++=xlat; - else - outPtr+=sprintf((char *)outPtr, "&#%d;", tmpVal); // &#decimal; - } else - throw Exception(0, - 0, - "transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal); + else { + outPtr+=sprintf((char *)outPtr, "&#%u;", tmpVal); // &#decimal; + } + } else { + const XMLByte* recoverPtr=srcPtr-trailingBytes-1; + for(uint i=0; i<=trailingBytes; i++) + outPtr+=sprintf((char*)outPtr, "%%%02X", *recoverPtr++); + } } // Update the bytes eaten @@ -417,12 +462,341 @@ of ocetes consumed. return 0; } -/// @todo not so memory-hungry with prescan +static bool need_escape(XMLByte c){ + return + !( + (c<=127) + && ( + pa_isalnum((unsigned char)c) + || strchr("*@-_+./", c)!=0 + ) + ); +} + +// read one UTF8 char and return length of this char (in bytes) +static unsigned int readUTF8Char(const XMLByte*& srcPtr, const XMLByte* srcEnd, XMLByte& firstByte, XMLCh& UTF8Char){ + if(!srcPtr || !*srcPtr || srcPtr>=srcEnd) + return 0; + + firstByte=*srcPtr; + + if(firstByte<=127){ + UTF8Char=firstByte; + srcPtr++; + return 1; + } + + unsigned int trailingBytes=gUTFBytes[firstByte]; + + if(srcPtr+trailingBytes>=srcEnd){ + return 0; // not enough bytes in source string for reading + } + + 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++; + } + + tmpVal-=gUTFOffsets[trailingBytes]; + UTF8Char=tmpVal; + + return trailingBytes+1; +} + +// skip UTF8 char and return length of this char (in bytes) +static unsigned int skipUTF8Char(const XMLByte*& srcPtr, const XMLByte* srcEnd){ + if(!srcPtr || !*srcPtr || srcPtr>=srcEnd) + return 0; + + unsigned int trailingBytes=gUTFBytes[*srcPtr]+1; + srcPtr+=trailingBytes; + + return trailingBytes; +} + +// read non-UTF8 char, and return number of bytes needed for storing this char in UTF8 +static unsigned int readChar(const XMLByte*& srcPtr, const XMLByte* srcEnd, XMLByte& firstByte, XMLCh& UTF8Char, const Charset::Tables& tables){ + if(!srcPtr || !*srcPtr || srcPtr>=srcEnd) + return 0; + + firstByte=*srcPtr++; + UTF8Char=tables.fromTable[firstByte]; + + if(UTF8Char<0x80) + return 1; + else if(UTF8Char<0x800) + return 2; + else if(UTF8Char<0x10000) + return 3; + else if(UTF8Char<0x200000) + return 4; + else if(UTF8Char<0x4000000) + return 5; + else if(UTF8Char<= 0x7FFFFFFF) + return 6; + + // will use the replacement character '?' + firstByte=0; + return 1; +} + +size_t Charset::calc_escaped_length_UTF8(XMLByte* src, size_t src_length){ + size_t dest_length=0; + + for(UTF8_string_iterator i(src, src_length); i.has_next(); ){ + if(i.getCharSize()==1) + dest_length+=!need_escape(i.getFirstByte())?1/*as-is*/:3/*%XX*/; + else + dest_length+=6; // %uXXXX + } + + return dest_length; +} + +size_t Charset::calc_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables){ + const XMLByte* src_end=src+src_length; + XMLByte first_byte; + XMLCh UTF8_char; + size_t dest_length=0; + + while(uint char_size=readChar(src, src_end, first_byte, UTF8_char, tables)){ + if(char_size==1) + dest_length+=(!first_byte/*replacement char '?'*/ || !need_escape(first_byte))?1:3/*'%XX'*/; + else + dest_length+=6; // %uXXXX + } + + return dest_length; +} + +size_t Charset::calc_escaped_length(const String::C src, const Charset& source_charset){ + if(!src.length) + return 0; + +#ifdef PRECALCULATE_DEST_LENGTH + if(source_charset.isUTF8()) + return calc_escaped_length_UTF8((XMLByte *)src.str, src.length); + else + return calc_escaped_length((XMLByte *)src.str, src.length, source_charset.tables); +#else + return src_length*6; // enough for %uXXXX but too memory-hungry +#endif +} + +#define escape_char(dest_ptr, char_size, first_byte, UTF8_char) \ + if(char_size==1) \ + if(first_byte){ \ + if(need_escape(first_byte)) \ + dest_ptr=append_hex_8(dest_ptr, first_byte, "%"); /* %XX */ \ + else \ + *dest_ptr++=first_byte; /*as is*/ \ + } else \ + *dest_ptr++='?'; /* replacement char '?' */ \ + else \ + dest_ptr=append_hex_16(dest_ptr, UTF8_char, "%u"); /* %uXXXX */ + + +size_t Charset::escape_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest) { + XMLByte* dest_ptr=dest; + + // loop until we either run out of input data + for(UTF8_string_iterator i((XMLByte *)src, src_length); i.has_next(); ) + escape_char(dest_ptr, i.getCharSize(), i.getFirstByte(), i.next()) + + return dest_ptr - dest; +} + +size_t Charset::escape(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables) { + const XMLByte* src_end=src+src_length; + XMLByte* dest_ptr=dest; + + XMLByte first_byte; + XMLCh UTF8_char; + uint char_size; + + while(char_size=readChar(src, src_end, first_byte, UTF8_char, tables)) + escape_char(dest_ptr, char_size, first_byte, UTF8_char) + + return dest_ptr - dest; +} + +String::C Charset::escape(const String::C src, const Charset& source_charset){ + if(!src.length) + return String::C("", 0); + + size_t dest_calculated_length=calc_escaped_length(src, source_charset); + XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_calculated_length+1/*terminator*/]; + + size_t dest_length; + if(source_charset.isUTF8()) + dest_length=escape_UTF8((XMLByte *)src.str, src.length, dest_body); + else + dest_length=escape((XMLByte *)src.str, src.length, dest_body, source_charset.tables); + + if(dest_length>dest_calculated_length) + throw Exception(0, 0, "Charset::escape buffer overflow"); + + dest_body[dest_length]=0; // terminator + return String::C((char*)dest_body, dest_length); +} + +String::Body Charset::escape(const String::Body src, const Charset& source_charset) { + String::C dest=Charset::escape(String::C(src.cstr(), src.length()), source_charset); + return String::Body(dest.length ? dest.str:0); +} + +String& Charset::escape(const String& src, const Charset& source_charset) { + if(src.is_empty()) + return *new String(); + + return *new String(escape((String::Body)src, source_charset), String::L_CLEAN); +} + +inline bool need_json_escape(unsigned char c){ + return strchr("\n\"\\/\t\r\b\f", c)!=0; +} + +size_t Charset::calc_JSON_escaped_length_UTF8(XMLByte* src, size_t src_length){ + size_t dest_length=0; + + for(UTF8_string_iterator i(src, src_length); i.has_next(); ){ + if(i.getCharSize()==1){ + XMLByte first_byte=i.getFirstByte(); + dest_length+=need_json_escape(first_byte) ? 2 : (first_byte < 0x20 && first_byte /* 0 replacement char is '?' */) ? 6 : 1; + } else + dest_length+=6; // \uXXXX + } + + return dest_length; +} + +size_t Charset::calc_JSON_escaped_length(const XMLByte* src, size_t src_length, const Charset::Tables& tables){ + const XMLByte* src_end=src+src_length; + XMLByte first_byte; + XMLCh UTF8_char; + size_t dest_length=0; + + while(uint char_size=readChar(src, src_end, first_byte, UTF8_char, tables)){ + if(char_size==1) + dest_length+=need_json_escape(first_byte) ? 2 : (first_byte < 0x20 && first_byte /* 0 replacement char is '?' */) ? 6 : 1; + else + dest_length+=6; // \uXXXX + } + + return dest_length; +} + +size_t Charset::calc_JSON_escaped_length(const String::C src, const Charset& source_charset){ + if(!src.length) + return 0; + +#ifdef PRECALCULATE_DEST_LENGTH + if(source_charset.isUTF8()) + return calc_JSON_escaped_length_UTF8((XMLByte *)src.str, src.length); + else + return calc_JSON_escaped_length((XMLByte *)src.str, src.length, source_charset.tables); +#else + return src_length*6; // enough for \uXXXX but too memory-hungry +#endif +} + +#define escape_char_JSON(dest_ptr, char_size, first_byte, UTF8_char) \ + if(char_size==1) \ + switch(first_byte){ \ + case '\n': *dest_ptr++='\\'; *dest_ptr++='n'; break; \ + case '"' : *dest_ptr++='\\'; *dest_ptr++='"'; break; \ + case '\\': *dest_ptr++='\\'; *dest_ptr++='\\'; break; \ + case '/' : *dest_ptr++='\\'; *dest_ptr++='/'; break; \ + case '\t': *dest_ptr++='\\'; *dest_ptr++='t'; break; \ + case '\r': *dest_ptr++='\\'; *dest_ptr++='r'; break; \ + case '\b': *dest_ptr++='\\'; *dest_ptr++='b'; break; \ + case '\f': *dest_ptr++='\\'; *dest_ptr++='f'; break; \ + case 0 : *dest_ptr++='?'; break; /*replacement char*/ \ + default : if(first_byte < 0x20) dest_ptr=append_hex_16(dest_ptr, UTF8_char, "\\u"); \ + else *dest_ptr++=first_byte; \ + } \ + else \ + dest_ptr=append_hex_16(dest_ptr, UTF8_char, "\\u"); // \uXXXX + + +size_t Charset::escape_JSON_UTF8(const XMLByte* src, size_t src_length, XMLByte* dest) { + XMLByte* dest_ptr=dest; + + // loop until we either run out of input data + for(UTF8_string_iterator i((XMLByte *)src, src_length); i.has_next(); ) + escape_char_JSON(dest_ptr, i.getCharSize(), i.getFirstByte(), i.next()) + + return dest_ptr - dest; +} + +size_t Charset::escape_JSON(const XMLByte* src, size_t src_length, XMLByte* dest, const Charset::Tables& tables) { + const XMLByte* src_end=src+src_length; + XMLByte* dest_ptr=dest; + + XMLByte first_byte; + XMLCh UTF8_char; + uint char_size; + + while(char_size=readChar(src, src_end, first_byte, UTF8_char, tables)) + escape_char_JSON(dest_ptr, char_size, first_byte, UTF8_char) + + return dest_ptr - dest; +} + +String::C Charset::escape_JSON(const String::C src, const Charset& source_charset){ + if(!src.length) + return String::C("", 0); + + size_t dest_calculated_length=calc_JSON_escaped_length(src, source_charset); + XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_calculated_length+1/*terminator*/]; + + size_t dest_length; + if(source_charset.isUTF8()) + dest_length=escape_JSON_UTF8((XMLByte *)src.str, src.length, dest_body); + else + dest_length=escape_JSON((XMLByte *)src.str, src.length, dest_body, source_charset.tables); + + if(dest_length>dest_calculated_length) + throw Exception(0, 0, "Charset::escape_JSON buffer overflow"); + + dest_body[dest_length]=0; // terminator + return String::C((char*)dest_body, dest_length); +} + +String::Body Charset::escape_JSON(const String::Body src, const Charset& source_charset) { + String::C dest=Charset::escape_JSON(String::C(src.cstr(), src.length()), source_charset); + return String::Body(dest.length ? dest.str:0); +} + +String& Charset::escape_JSON(const String& src, const Charset& source_charset) { + if(src.is_empty()) + return *new String(); + + return *new String(escape_JSON((String::Body)src, source_charset), String::L_CLEAN); +} + const String::C Charset::transcodeToUTF8(const String::C src) const { - size_t src_length=src.length; - size_t dest_length=src.length*6/*so that surly enough, max utf8 seq len=6*/; + int src_length=src.length; + +#ifdef PRECALCULATE_DEST_LENGTH + int dest_length=0; + const XMLByte* srcPtr=(XMLByte*)src.str; + const XMLByte* srcEnd=srcPtr+src_length; + XMLByte firstByte; + XMLCh UTF8Char; + while(uint charSize=readChar(srcPtr, srcEnd, firstByte, UTF8Char, tables)) + dest_length+=charSize; +#else + int dest_length=src_length*6; // so that surly enough (max utf8 seq len=6) but too memory-hungry +#endif + #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*/]; @@ -430,18 +804,173 @@ const String::C Charset::transcodeToUTF8 (XMLByte *)src.str, src_length, dest_body, dest_length, tables)<0) - throw(0, 0, + throw Exception(0, 0, "Charset::transcodeToUTF8 buffer overflow"); - assert(dest_length<=saved_dest_length); dest_body[dest_length]=0; // terminator + assert(dest_length<=saved_dest_length); + dest_body[dest_length]=0; // terminator return String::C((char*)dest_body, dest_length); } + +static XMLCh change_case_UTF8(const XMLCh src, const Charset::UTF8CaseTable& table) { + int lo = 0; + int hi = table.size - 1; + while(lo<=hi) { + // Calc the mid point of the low and high offset. + const unsigned int i = (lo + hi) / 2; + + XMLCh cur=table.records[i].from; + if(src==cur) + return table.records[i].to; + if(src>cur) + 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"); +} + +static size_t getDecNumLength(XMLCh UTF8Char){ + return + (UTF8Char < 100) + ?2 + :(UTF8Char < 1000) + ?3 + :(UTF8Char < 10000) + ?4 + :5; +} + 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 */; + int src_length=src.length; +#ifdef PRECALCULATE_DEST_LENGTH + int dest_length=0; + for(UTF8_string_iterator i((XMLByte *)src.str, src_length); i.has_next(); ){ + dest_length += ( i.next() & 0xFFFF0000 ) + ? 3*i.getCharSize() // %XX for each byte + : ( xlatOneTo(i.next(), tables, 0) != 0 ) + ? 1 // can convert it to a single char + : 3+getDecNumLength( i.next() ); // print char as &#XX;, &#XXX;, &#XXXX; or &#XXXXX; + } +#else + // so that surly enough, "&#XXX;" has max ratio (huh? 8 bytes needed for '&#XXXXX;') + int dest_length=src_length*6; +#endif + #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*/]; @@ -449,17 +978,18 @@ 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"); - assert(dest_length<=saved_dest_length); dest_body[dest_length]=0; // terminator + assert(dest_length<=saved_dest_length); + dest_body[dest_length]=0; // terminator return String::C((char*)dest_body, dest_length); } /// transcode using both charsets const String::C Charset::transcodeToCharset(const String::C src, - const Charset& dest_charset) const { + const Charset& dest_charset) const { if(&dest_charset==this) return src; else { @@ -480,28 +1010,60 @@ const String::C Charset::transcodeToChar } } +void Charset::store_Char(XMLByte*& outPtr, XMLCh src, XMLByte not_found){ + if(isUTF8()) + store_UTF8(src, outPtr); + else if(char ch=xlatOneTo(src, tables, not_found)) + *outPtr++=ch; +} + #ifdef XML 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, *inlen, \ + out, *outlen, \ + *tables[i]); \ + } \ + static int xml256CharEncodingOutputFunc##i( \ + unsigned char *out, int *outlen, \ + const unsigned char *in, int *inlen, void*) { \ + return transcodeFromUTF8( \ + in, *inlen, \ + out, *outlen, \ + *tables[i]); \ + } + +#else + #define declareXml256ioFuncs(i) \ static int xml256CharEncodingInputFunc##i( \ 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]); \ } +#endif + + declareXml256ioFuncs(0) declareXml256ioFuncs(1) declareXml256ioFuncs(2) declareXml256ioFuncs(3) declareXml256ioFuncs(4) declareXml256ioFuncs(5) @@ -531,7 +1093,7 @@ 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]; @@ -544,25 +1106,25 @@ 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", + throw Exception(PARSER_RUNTIME, new String(NAME, String::L_TAINTED), "unsupported encoding"); return *ftranscoder; } -String::C Charset::transcode_cstr(xmlChar* s) { +String::C Charset::transcode_cstr(const xmlChar* s) { if(!s) 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 +1134,11 @@ 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; @@ -585,17 +1151,9 @@ String::C Charset::transcode_cstr(xmlCha assert(outlen<=saved_outlen); out[outlen]=0; return String::C(out, outlen); } -const String& Charset::transcode(xmlChar* s) { - String::C cstr=transcode_cstr(s); - return *new String(cstr.str, cstr.length, true); -} -String::C Charset::transcode_cstr(GdomeDOMString* s) { - return s?transcode_cstr(BAD_CAST s->str) - :String::C("", 0); -} -const String& Charset::transcode(GdomeDOMString* s) { +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 @@ -607,14 +1165,18 @@ xmlChar* Charset::transcode_buf2xchar(co 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=(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 @@ -633,42 +1195,36 @@ xmlChar* Charset::transcode_buf2xchar(co assert(outlen<=saved_outlen); out[outlen]=0; return out; } -GdomeDOMString_auto_ptr Charset::transcode_buf2dom(const char* buf, size_t 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)); +xmlChar* Charset::transcode(const String& s) { + String::Body sbody=s.cstr_to_string_body_untaint(String::L_AS_IS); + return transcode_buf2xchar(sbody.cstr(), sbody.length()); } -GdomeDOMString_auto_ptr Charset::transcode(const StringBody s) { - const char* cstr=s.cstr(); - return transcode_buf2dom(cstr, s.length()); +xmlChar* Charset::transcode(const String::Body s) { + return transcode_buf2xchar(s.cstr(), s.length()); } #endif -StringBody Charset::transcode(const StringBody src, +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); + size_t src_size=src.length(); - 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 StringBody(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((StringBody)src, source_transcoder, dest_transcoder), String::L_CLEAN); + return *new String(transcode((String::Body)src, source_transcoder, dest_transcoder), String::L_CLEAN); } void Charset::transcode(ArrayString& src, @@ -684,16 +1240,96 @@ struct Transcode_pair_info { const Charset* dest_transcoder; }; #endif -static void transcode_pair(const StringBody akey, - StringBody& avalue, +static void transcode_pair(HashStringValue::key_type /*akey*/, + String::Body& avalue, Transcode_pair_info* info) { avalue=Charset::transcode(avalue, *info->source_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); + src.for_each_ref(transcode_pair, &info); +} + +size_t getUTF8BytePos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t charPos){ + const XMLByte* ptr=srcBegin; + while(charPos-- && skipUTF8Char(ptr, srcEnd)); + + return ptr-srcBegin; +} + +size_t getUTF8CharPos(const XMLByte* srcBegin, const XMLByte* srcEnd, size_t bytePos){ + size_t charPos=0; + const XMLByte* ptr=srcBegin; + const XMLByte* ptrEnd=srcBegin+bytePos; + while(skipUTF8Char(ptr, srcEnd)){ + if(ptr>ptrEnd) + return charPos; + charPos++; + } + + // scan till end but position in bytes still too low + throw Exception(0, + 0, + "Error convertion byte pos to char pos"); +} + +size_t lengthUTF8(const XMLByte* srcBegin, const XMLByte* srcEnd){ + size_t size=0; + while(skipUTF8Char(srcBegin, srcEnd)) + size++; + + return size; +} + +unsigned int lengthUTF8Char(const XMLByte c){ + return gUTFBytes[c]+1; +} + +const char *fixUTF8(const char *src){ + if(src && *src){ + size_t length=strlen(src); + + int error_offset; + if(_pcre_valid_utf((unsigned char *)src, length, &error_offset)){ + + char *result=(char *)pa_malloc_atomic(length+1); + char *dst=result; + + do { + + if(error_offset){ + strncpy(dst, src, error_offset); + dst+=error_offset; + + src+=error_offset; + length-=error_offset; + + } + + *dst++='?'; + src++; + length--; + + } while (length && _pcre_valid_utf((unsigned char *)src, length, &error_offset)); + + if(length){ + strcpy(dst, src); + } else { + *dst='\0'; + } + + return result; + } + } + return src; +} + +bool UTF8_string_iterator::has_next(){ + fcharSize=readUTF8Char(fsrcPtr, fsrcEnd, ffirstByte, fUTF8Char); + return fcharSize!=0; }