|
|
| version 1.14, 2002/01/14 17:48:57 | version 1.52, 2005/08/09 08:14:51 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: Charset connection implementation. | Parser: Charset connection implementation. |
| Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) | Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexander Petrosyan<paf@design.ru>(http://paf.design.ru) | Author: Alexander Petrosyan<paf@design.ru>(http://paf.design.ru) |
| $Id$ | |
| */ | */ |
| static const char * const IDENT_CHARSET_C="$Date$"; | |
| #include "pa_charset.h" | #include "pa_charset.h" |
| #include "pa_charsets.h" | |
| #ifdef XML | #ifdef XML |
| #include "libxml/encoding.h" | #include "libxml/encoding.h" |
| #endif | #endif |
| // globals | //#define PA_PATCHED_LIBXML_BACKWARD |
| // globals | |
| // consts | Charset::UTF8CaseTable::Rec UTF8CaseToUpperRecords[]={ |
| #include "utf8-to-upper.inc" | |
| }; | |
| Charset::UTF8CaseTable UTF8CaseToUpper={ | |
| sizeof(UTF8CaseToUpperRecords)/sizeof(Charset::UTF8CaseTable::Rec), | |
| UTF8CaseToUpperRecords}; | |
| #define MAX_CHARSET_UNI_CODES 500 | Charset::UTF8CaseTable::Rec UTF8CaseToLowerRecords[]={ |
| #include "utf8-to-lower.inc" | |
| }; | |
| Charset::UTF8CaseTable UTF8CaseToLower={ | |
| sizeof(UTF8CaseToLowerRecords)/sizeof(Charset::UTF8CaseTable::Rec), | |
| UTF8CaseToLowerRecords}; | |
| // helpers | // helpers |
| Line 37 inline void cstr2ctypes(unsigned char *t | Line 49 inline void cstr2ctypes(unsigned char *t |
| ctypes_table[c]|=bit; | ctypes_table[c]|=bit; |
| } | } |
| } | } |
| inline unsigned int to_wchar_code(const char *cstr) { | inline unsigned int to_wchar_code(const char* cstr) { |
| if(!cstr || !*cstr) | if(!cstr || !*cstr) |
| return 0; | return 0; |
| if(cstr[1]==0) | if(cstr[1]==0) |
| Line 46 inline unsigned int to_wchar_code(const | Line 58 inline unsigned int to_wchar_code(const |
| char *error_pos; | char *error_pos; |
| return(unsigned int)strtol(cstr, &error_pos, 0); | return(unsigned int)strtol(cstr, &error_pos, 0); |
| } | } |
| inline bool to_bool(const char *cstr) { | inline bool to_bool(const char* cstr) { |
| return cstr && *cstr!=0; | return cstr && *cstr!=0; |
| } | } |
| static void element2ctypes(unsigned char c, bool belongs, | static void element2ctypes(unsigned char c, bool belongs, |
| Line 74 static void element2case(unsigned char f | Line 86 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(Pool& apool, const String& aname, const String *request_file_spec): Pooled(apool), | Charset::Charset(Request_charsets* charsets, const String::Body ANAME, const String* afile_spec): |
| fname(aname) { | FNAME(ANAME), |
| FNAME_CSTR(ANAME.cstrm()) { | |
| char *name_cstr=fname.cstr(); | |
| for(char *c=name_cstr; *c; c++) | |
| *c = toupper(*c); | |
| if(request_file_spec) { | if(afile_spec) { |
| fisUTF8=false; | fisUTF8=false; |
| loadDefinition(*request_file_spec); | load_definition(*charsets, *afile_spec); |
| #ifdef XML | #ifdef XML |
| addEncoding(name_cstr); | addEncoding(FNAME_CSTR); |
| #endif | #endif |
| } else { | } else { |
| fisUTF8=true; | fisUTF8=true; |
| Line 94 Charset::Charset(Pool& apool, const Stri | Line 103 Charset::Charset(Pool& apool, const Stri |
| } | } |
| #ifdef XML | #ifdef XML |
| initTranscoder(&aname, name_cstr); | initTranscoder(FNAME, FNAME_CSTR); |
| #endif | |
| } | |
| Charset::~Charset() { | |
| #ifdef XML | |
| // not deleting transcoder, that's not our business | |
| #endif | #endif |
| } | } |
| void Charset::loadDefinition(const String& request_file_spec) { | void Charset::load_definition(Request_charsets& charsets, const String& afile_spec) { |
| // pcre_tables | // pcre_tables |
| // lowcase, flipcase, bits digit+word+whitespace, masks | // lowcase, flipcase, bits digit+word+whitespace, masks |
| Line 115 void Charset::loadDefinition(const Strin | Line 118 void Charset::loadDefinition(const Strin |
| cstr2ctypes(pcre_tables,(const unsigned char *)"*+?{^.$|()[", ctype_meta); | cstr2ctypes(pcre_tables,(const unsigned char *)"*+?{^.$|()[", ctype_meta); |
| // charset | // charset |
| memset(tables.fromTable, 0, sizeof(tables.fromTable)); | memset(&tables, 0, sizeof(tables)); |
| tables.toTable=(Charset_TransRec *)calloc(sizeof(Charset_TransRec)*MAX_CHARSET_UNI_CODES); | |
| tables.toTableSize=0; | |
| // strangly vital | |
| tables.toTable[tables.toTableSize].intCh=0; | |
| tables.toTable[tables.toTableSize].extCh=(XMLByte)0; | |
| tables.toTableSize++; | |
| // loading text | // loading text |
| char *data=file_read_text(pool(), request_file_spec); | char *data=file_read_text(charsets, afile_spec); |
| // ignore header | // ignore header |
| getrow(&data); | getrow(&data); |
| // parse cells | // parse cells |
| char *row; | char *row; |
| while(row=getrow(&data)) { | while((row=getrow(&data))) { |
| // remove empty&comment lines | // remove empty&comment lines |
| if(!*row || *row=='#') | if(!*row || *row=='#') |
| continue; | continue; |
| Line 139 void Charset::loadDefinition(const Strin | Line 136 void Charset::loadDefinition(const Strin |
| // char white-space digit hex-digit letter word lowercase unicode1 unicode2 | // char white-space digit hex-digit letter word lowercase unicode1 unicode2 |
| unsigned int c=0; | unsigned int c=0; |
| char *cell; | char *cell; |
| for(int column=0; cell=lsplit(&row, '\t'); column++) { | for(int column=0; (cell=lsplit(&row, '\t')); column++) { |
| switch(column) { | switch(column) { |
| case 0: c=to_wchar_code(cell); break; | case 0: c=to_wchar_code(cell); break; |
| // pcre_tables | // pcre_tables |
| Line 153 void Charset::loadDefinition(const Strin | Line 150 void Charset::loadDefinition(const Strin |
| case 8: | case 8: |
| // charset | // charset |
| if(tables.toTableSize>MAX_CHARSET_UNI_CODES) | if(tables.toTableSize>MAX_CHARSET_UNI_CODES) |
| throw Exception(0, 0, | throw Exception("parser.runtime", |
| &request_file_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 178 void Charset::loadDefinition(const Strin | Line 175 void Charset::loadDefinition(const Strin |
| static int sort_cmp_Trans_rec_intCh(const void *a, const void *b) { | static int sort_cmp_Trans_rec_intCh(const void *a, const void *b) { |
| return | return |
| static_cast<const Charset_TransRec *>(a)->intCh- | static_cast<const Charset::Tables::Rec *>(a)->intCh- |
| static_cast<const Charset_TransRec *>(b)->intCh; | static_cast<const Charset::Tables::Rec *>(b)->intCh; |
| } | } |
| void Charset::sort_ToTable() { | void Charset::sort_ToTable() { |
| Line 191 void Charset::sort_ToTable() { | Line 188 void Charset::sort_ToTable() { |
| } | } |
| static XMLByte xlatOneTo(const XMLCh toXlat, | static XMLByte xlatOneTo(const XMLCh toXlat, |
| const Charset::Tables& tables) { | const Charset::Tables& tables, |
| unsigned int lowOfs = 0; | XMLByte not_found) { |
| unsigned int hiOfs = tables.toTableSize - 1; | int lo = 0; |
| XMLByte curByte = 0; | int hi = tables.toTableSize - 1; |
| do { | while(lo<=hi) { |
| // Calc the mid point of the low and high offset. | // Calc the mid point of the low and high offset. |
| const unsigned int midOfs =((hiOfs - lowOfs) / 2)+lowOfs; | const unsigned int i = (lo + hi) / 2; |
| // If our test char is greater than the mid point char, then | XMLCh cur=tables.toTable[i].intCh; |
| // we move up to the upper half. Else we move to the lower | if(toXlat==cur) |
| // half. If its equal, then its our guy. | return tables.toTable[i].extCh; |
| if(toXlat>tables.toTable[midOfs].intCh) | if(toXlat>cur) |
| lowOfs = midOfs; | lo = i+1; |
| else if(toXlat<tables.toTable[midOfs].intCh) | |
| hiOfs = midOfs; | |
| else | else |
| return tables.toTable[midOfs].extCh; | hi = i-1; |
| } while(lowOfs+1<hiOfs); | } |
| return '?'; | return not_found; |
| } | } |
| void Charset::transcode(Pool& pool, | String::C Charset::transcode(const String::C src, |
| const Charset& source_charset, const void *source_body, size_t source_content_length, | const Charset& source_charset, |
| const Charset& dest_charset, const void *& dest_body, size_t& dest_content_length | const Charset& dest_charset) { |
| ) { | if(!src.length) |
| if(!source_content_length) { | return String::C("", 0); |
| dest_body=0; | |
| dest_content_length=0; | |
| return; | |
| } | |
| 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, | return source_charset.transcodeToCharset(src, dest_charset); |
| source_body, source_content_length, | |
| dest_body, dest_content_length); | |
| break; | |
| case 0x01: | case 0x01: |
| source_charset.transcodeToUTF8(pool, | return source_charset.transcodeToUTF8(src); |
| source_body, source_content_length, | |
| dest_body, dest_content_length); | |
| break; | |
| case 0x10: | case 0x10: |
| dest_charset.transcodeFromUTF8(pool, | return dest_charset.transcodeFromUTF8(src); |
| source_body, source_content_length, | |
| dest_body, dest_content_length); | |
| break; | |
| case 0x11: | case 0x11: |
| dest_body=source_body; | return src; |
| dest_content_length=source_content_length; | |
| break; | |
| } | } |
| } | } |
| Line 287 static const XMLByte gFirstByteMark[7] = | Line 267 static const XMLByte gFirstByteMark[7] = |
| 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC | 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC |
| }; | }; |
| static int transcodeToUTF8( | static int transcodeToUTF8(const XMLByte* srcData, size_t& srcLen, |
| const XMLByte* srcData, size_t& srcLen, | XMLByte *toFill, size_t& toFillLen, |
| XMLByte *toFill, size_t& toFillLen, | const Charset::Tables& tables) { |
| const Charset::Tables& tables) { | |
| const XMLByte* srcPtr=srcData; | const XMLByte* srcPtr=srcData; |
| const XMLByte* srcEnd=srcData+srcLen; | const XMLByte* srcEnd=srcData+srcLen; |
| XMLByte* outPtr=toFill; | XMLByte* outPtr=toFill; |
| XMLByte* outEnd=toFill+toFillLen; | XMLByte* outEnd=toFill+toFillLen; |
| while(srcPtr<srcEnd) { | while(srcPtr<srcEnd) { |
| uint curVal = tables.fromTable[*srcPtr]; | uint curVal = tables.fromTable[*srcPtr]; |
| if(!curVal) { | if(!curVal) { |
| // use the replacement character | // use the replacement character |
| *outPtr++= '?'; | *outPtr++= '?'; |
| srcPtr++; | srcPtr++; |
| continue; | continue; |
| } | } |
| // Figure out how many bytes we need | // Figure out how many bytes we need |
| unsigned int encodedBytes; | unsigned int encodedBytes; |
| if(curVal<0x80) | if(curVal<0x80) |
| encodedBytes = 1; | encodedBytes = 1; |
| else if(curVal<0x800) | else if(curVal<0x800) |
| encodedBytes = 2; | encodedBytes = 2; |
| else if(curVal<0x10000) | else if(curVal<0x10000) |
| encodedBytes = 3; | encodedBytes = 3; |
| else if(curVal<0x200000) | else if(curVal<0x200000) |
| encodedBytes = 4; | encodedBytes = 4; |
| else if(curVal<0x4000000) | else if(curVal<0x4000000) |
| encodedBytes = 5; | encodedBytes = 5; |
| else if(curVal<= 0x7FFFFFFF) | else if(curVal<= 0x7FFFFFFF) |
| encodedBytes = 6; | encodedBytes = 6; |
| else { | else { |
| // use the replacement character | // use the replacement character |
| *outPtr++= '?'; | *outPtr++= '?'; |
| srcPtr++; | srcPtr++; |
| continue; | continue; |
| } | } |
| // If we cannot fully get this char into the output buffer | // If we cannot fully get this char into the output buffer |
| if (outPtr + encodedBytes > outEnd) | if (outPtr + encodedBytes > outEnd) |
| break; | break; |
| // We can do it, so update the source index | // We can do it, so update the source index |
| srcPtr++; | srcPtr++; |
| // And spit out the bytes. We spit them out in reverse order | // 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. | // here, so bump up the output pointer and work down as we go. |
| outPtr+= encodedBytes; | outPtr+= encodedBytes; |
| switch(encodedBytes) { | switch(encodedBytes) { |
| case 6: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); | case 6: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); |
| curVal>>= 6; | curVal>>= 6; |
| case 5: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); | case 5: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); |
| curVal>>= 6; | curVal>>= 6; |
| case 4: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); | case 4: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); |
| curVal>>= 6; | curVal>>= 6; |
| case 3: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); | case 3: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); |
| curVal>>= 6; | curVal>>= 6; |
| case 2: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); | case 2: *--outPtr = XMLByte((curVal | 0x80UL) & 0xBFUL); |
| curVal>>= 6; | curVal>>= 6; |
| case 1: *--outPtr = XMLByte(curVal | gFirstByteMark[encodedBytes]); | case 1: *--outPtr = XMLByte(curVal | gFirstByteMark[encodedBytes]); |
| } | } |
| // Add the encoded bytes back in again to indicate we've eaten them | // Add the encoded bytes back in again to indicate we've eaten them |
| outPtr+= encodedBytes; | outPtr+= encodedBytes; |
| } | } |
| // Update the bytes eaten | // Update the bytes eaten |
| srcLen = srcPtr - srcData; | srcLen = srcPtr - srcData; |
| // Return the characters read | // Return the characters read |
| toFillLen = outPtr - toFill; | toFillLen = outPtr - toFill; |
| return srcPtr==srcEnd?toFillLen:-1; | //return srcPtr==srcEnd?(int)toFillLen:-1; |
| } | /* |
| static size_t transcodeFromUTF8( | xmlCharEncodingInputFunc |
| const XMLByte *srcData, size_t& srcLen, | Returns : |
| XMLByte* toFill, size_t& toFillLen, | the number of byte written, or -1 by lack of space, or -2 if the transcoding failed. The value of inlen after return is the |
| const Charset::Tables& tables) { | number of octets consumed as the return value is positive, else unpredictiable. The value of outlen after return is the number |
| 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) { | |
| const XMLByte* srcPtr=srcData; | const XMLByte* srcPtr=srcData; |
| const XMLByte* srcEnd=srcData+srcLen; | const XMLByte* srcEnd=srcData+srcLen; |
| XMLByte* outPtr=toFill; | XMLByte* outPtr=toFill; |
| XMLByte* outEnd=toFill+toFillLen; | XMLByte* outEnd=toFill+toFillLen; |
| // We now loop until we either run out of input data, or room to store | // We now loop until we either run out of input data, or room to store |
| while ((srcPtr < srcEnd) && (outPtr < outEnd)) { | while ((srcPtr < srcEnd) && (outPtr < outEnd)) { |
| // Get the next leading byte out | // Get the next leading byte out |
| const XMLByte firstByte = *srcPtr; | const XMLByte firstByte =* srcPtr; |
| // Special-case ASCII, which is a leading byte value of<= 127 | // Special-case ASCII, which is a leading byte value of<= 127 |
| if(firstByte<= 127) { | if(firstByte<= 127) { |
| *outPtr++= firstByte; | *outPtr++= firstByte; |
| srcPtr++; | srcPtr++; |
| continue; | continue; |
| } | } |
| // See how many trailing src bytes this sequence is going to require | // See how many trailing src bytes this sequence is going to require |
| const unsigned int trailingBytes = gUTFBytes[firstByte]; | const unsigned int trailingBytes = gUTFBytes[firstByte]; |
| // If there are not enough source bytes to do this one, then we | // If there are not enough source bytes to do this one, then we |
| // are done. Note that we done>= here because we are implicitly | // are done. Note that we done>= here because we are implicitly |
| // counting the 1 byte we get no matter what. | // counting the 1 byte we get no matter what. |
| if(srcPtr+trailingBytes>= srcEnd) | if(srcPtr+trailingBytes>= srcEnd) |
| break; | break; |
| // Looks ok, so lets build up the value | // Looks ok, so lets build up the value |
| uint tmpVal=0; | uint tmpVal=0; |
| switch(trailingBytes) { | switch(trailingBytes) { |
| case 5: tmpVal+=*srcPtr++; tmpVal<<=6; | case 5: tmpVal+=*srcPtr++; tmpVal<<=6; |
| case 4: tmpVal+=*srcPtr++; tmpVal<<=6; | case 4: tmpVal+=*srcPtr++; tmpVal<<=6; |
| case 3: tmpVal+=*srcPtr++; tmpVal<<=6; | case 3: tmpVal+=*srcPtr++; tmpVal<<=6; |
| case 2: tmpVal+=*srcPtr++; tmpVal<<=6; | case 2: tmpVal+=*srcPtr++; tmpVal<<=6; |
| case 1: tmpVal+=*srcPtr++; tmpVal<<=6; | case 1: tmpVal+=*srcPtr++; tmpVal<<=6; |
| case 0: tmpVal+=*srcPtr++; | case 0: tmpVal+=*srcPtr++; |
| break; | break; |
| default: | default: |
| throw Exception(0, 0, | throw Exception(0, |
| 0, | |
| "transcodeFromUTF8 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)) | |
| *outPtr++= xlatOneTo(tmpVal, tables); | |
| else | |
| throw Exception(0, 0, | |
| 0, | 0, |
| "transcodeFromUTF8 error: too big tmpVal(0x%08X)", tmpVal); | "transcodeFromUTF8 error: wrong trailingBytes value(%d)", trailingBytes); // never |
| } | |
| 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)) { | |
| if(XMLByte xlat=xlatOneTo(tmpVal, tables, 0)) | |
| *outPtr++=xlat; | |
| 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 | |
| srcLen = srcPtr - srcData; | |
| // Return the characters read | |
| toFillLen = outPtr - toFill; | |
| // Update the bytes eaten | //return srcPtr==srcEnd?(int)toFillLen:-1; |
| srcLen = srcPtr - srcData; | /* |
| xmlCharEncodingOutputFunc | |
| // Return the characters read | Returns : |
| toFillLen = outPtr - toFill; | the number of byte written, or -1 by lack of space, or -2 if the transcoding failed. The value of inlen after return is the |
| number of octets consumed as the return value is positive, else unpredictiable. The value of outlen after return is the number | |
| return srcPtr==srcEnd?toFillLen:-1; | of ocetes consumed. |
| */ | |
| return 0; | |
| } | } |
| /// @todo not so memory-hungry with prescan | /// @todo not so memory-hungry with prescan |
| void Charset::transcodeToUTF8(Pool& pool, | const String::C Charset::transcodeToUTF8(const String::C src) const { |
| const void *source_body, size_t source_content_length, | size_t src_length=src.length; |
| const void *& adest_body, size_t& dest_content_length) const { | size_t dest_length=src.length*6/*so that surly enough, max utf8 seq len=6*/; |
| dest_content_length=source_content_length*6/*so that surly enough*/; | #ifndef NDEBUG |
| XMLByte *dest_body=(XMLByte*)pool.malloc(dest_content_length); | size_t saved_dest_length=dest_length; |
| #endif | |
| XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_length+1/*for terminator*/]; | |
| if(::transcodeToUTF8( | if(::transcodeToUTF8( |
| (XMLByte *)source_body, source_content_length, | (XMLByte *)src.str, src_length, |
| dest_body, dest_content_length, | dest_body, dest_length, |
| tables)<0) | tables)<0) |
| throw(0, 0, | throw Exception(0, |
| 0, | 0, |
| "Charset::transcodeToUTF8 buffer overflow"); | "Charset::transcodeToUTF8 buffer overflow"); |
| // return | assert(dest_length<=saved_dest_length); dest_body[dest_length]=0; // terminator |
| adest_body=dest_body; | 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; | |
| } | } |
| /// @test buf overflow | |
| void Charset::transcodeFromUTF8(Pool& pool, | static void store_UTF8(XMLCh src, XMLByte*& outPtr ) { |
| const void *source_body, size_t source_content_length, | if(!src) { |
| const void *& adest_body, size_t& dest_content_length) const { | // use the replacement character |
| dest_content_length=source_content_length/*surly enough*/; | *outPtr++= '?'; |
| XMLByte *dest_body=(XMLByte*)pool.malloc(dest_content_length); | 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 */; | |
| #ifndef NDEBUG | |
| size_t saved_dest_length=dest_length; | |
| #endif | |
| XMLByte *dest_body=new(PointerFreeGC) XMLByte[dest_length+1/*for terminator*/]; | |
| if(::transcodeFromUTF8( | if(::transcodeFromUTF8( |
| (XMLByte *)source_body, source_content_length, | (XMLByte *)src.str, src_length, |
| dest_body, dest_content_length, | dest_body, dest_length, |
| tables)<0) | tables)<0) |
| throw(0, 0, | throw Exception(0, |
| 0, | 0, |
| "Charset::transcodeToUTF8 buffer overflow"); | "Charset::transcodeFromUTF8 buffer overflow"); |
| // return | assert(dest_length<=saved_dest_length); dest_body[dest_length]=0; // terminator |
| adest_body=dest_body; | return String::C((char*)dest_body, dest_length); |
| } | } |
| /// transcode using both charsets | /// transcode using both charsets |
| void Charset::transcodeToCharset(Pool& pool, | const String::C Charset::transcodeToCharset(const String::C src, |
| const Charset& dest_charset, | const Charset& dest_charset) const { |
| const void *source_body, size_t source_content_length, | if(&dest_charset==this) |
| const void *& adest_body, size_t& adest_content_length) const { | return src; |
| if(&dest_charset==this) { | else { |
| adest_body=source_body; | size_t dest_length=src.length; |
| adest_content_length=source_content_length; | XMLByte* dest_body=new(PointerFreeGC) XMLByte[dest_length+1/*for terminator*/]; |
| } else { | |
| size_t dest_content_length=source_content_length; | XMLByte* output=dest_body; |
| unsigned char *dest_body=(unsigned char *)pool.malloc(dest_content_length); | const XMLByte* input=(XMLByte *)src.str; |
| while(XMLCh c=*input++) { | |
| const XMLByte* srcPtr=(XMLByte *)source_body; | XMLCh curVal = tables.fromTable[c]; |
| const XMLByte* srcEnd=(XMLByte *)source_body+source_content_length; | *output++=curVal? |
| xlatOneTo(curVal, dest_charset.tables, '?') // OK | |
| for(XMLByte* outPtr=dest_body; srcPtr<srcEnd; srcPtr++) { | :'?'; // use the replacement character |
| XMLCh curVal = tables.fromTable[*srcPtr]; | |
| if(curVal) | |
| *outPtr++=xlatOneTo(curVal, dest_charset.tables); | |
| else { | |
| // use the replacement character | |
| *outPtr++= '?'; | |
| } | |
| } | } |
| adest_body=dest_body; | dest_body[dest_length]=0; // terminator |
| adest_content_length=dest_content_length; | return String::C((char*)dest_body, dest_length); |
| } | } |
| } | } |
| #ifdef XML | #ifdef XML |
| static int xml256CharEncodingInputFunc ( | |
| unsigned char *out, | static const Charset::Tables* tables[MAX_CHARSETS]; |
| int *outlen, | |
| const unsigned char *in, | #ifdef PA_PATCHED_LIBXML_BACKWARD |
| int *inlen, | |
| void *info) { | #define declareXml256ioFuncs(i) \ |
| return transcodeToUTF8( | static int xml256CharEncodingInputFunc##i( \ |
| in, *(unsigned int*)inlen, | unsigned char *out, int *outlen, \ |
| out, *(unsigned int*)outlen, | const unsigned char *in, int *inlen, void*) { \ |
| *(const Charset::Tables *)info); | return transcodeToUTF8( \ |
| } | in, *(size_t*)inlen, \ |
| out, *(size_t*)outlen, \ | |
| static int xml256CharEncodingOutputFunc ( | *tables[i]); \ |
| unsigned char *out, | } \ |
| int *outlen, | static int xml256CharEncodingOutputFunc##i( \ |
| const unsigned char *in, | unsigned char *out, int *outlen, \ |
| int *inlen, | const unsigned char *in, int *inlen, void*) { \ |
| void *info) { | return transcodeFromUTF8( \ |
| return transcodeFromUTF8( | in, *(size_t*)inlen, \ |
| in, *(unsigned int*)inlen, | out, *(size_t*)outlen, \ |
| out, *(unsigned int*)outlen, | *tables[i]); \ |
| *(const Charset::Tables *)info); | } |
| } | |
| #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, \ | |
| *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, \ | |
| *tables[i]); \ | |
| } | |
| #endif | |
| declareXml256ioFuncs(0) declareXml256ioFuncs(1) | |
| declareXml256ioFuncs(2) declareXml256ioFuncs(3) | |
| declareXml256ioFuncs(4) declareXml256ioFuncs(5) | |
| declareXml256ioFuncs(6) declareXml256ioFuncs(7) | |
| declareXml256ioFuncs(8) declareXml256ioFuncs(9) | |
| static xmlCharEncodingInputFunc inputFuncs[MAX_CHARSETS]={ | |
| xml256CharEncodingInputFunc0, xml256CharEncodingInputFunc1, | |
| xml256CharEncodingInputFunc2, xml256CharEncodingInputFunc3, | |
| xml256CharEncodingInputFunc4, xml256CharEncodingInputFunc5, | |
| xml256CharEncodingInputFunc6, xml256CharEncodingInputFunc7, | |
| xml256CharEncodingInputFunc8, xml256CharEncodingInputFunc9 | |
| }; | |
| static xmlCharEncodingOutputFunc outputFuncs[MAX_CHARSETS]={ | |
| xml256CharEncodingOutputFunc0, xml256CharEncodingOutputFunc1, | |
| xml256CharEncodingOutputFunc2, xml256CharEncodingOutputFunc3, | |
| xml256CharEncodingOutputFunc4, xml256CharEncodingOutputFunc5, | |
| xml256CharEncodingOutputFunc6, xml256CharEncodingOutputFunc7, | |
| xml256CharEncodingOutputFunc8, xml256CharEncodingOutputFunc9 | |
| }; | |
| static size_t handlers_count=0; | |
| void Charset::addEncoding(char *name_cstr) { | void Charset::addEncoding(char *name_cstr) { |
| xmlCharEncodingHandler *handler= | if(handlers_count==MAX_CHARSETS) |
| (xmlCharEncodingHandler *)malloc(sizeof(xmlCharEncodingHandler)); | throw Exception(0, |
| handler->name=name_cstr; | 0, |
| handler->input=xml256CharEncodingInputFunc; handler->inputInfo=&tables; | "already allocated %d handlers, no space for new encoding '%s'", |
| handler->output=xml256CharEncodingOutputFunc; handler->outputInfo=&tables; | MAX_CHARSETS, name_cstr); |
| xmlCharEncodingHandler* handler=new(UseGC) xmlCharEncodingHandler; | |
| { | |
| handler->name=name_cstr; | |
| handler->input=inputFuncs[handlers_count]; | |
| handler->output=outputFuncs[handlers_count]; | |
| ::tables[handlers_count]=&tables; | |
| handlers_count++; | |
| } | |
| xmlRegisterCharEncodingHandler(handler); | xmlRegisterCharEncodingHandler(handler); |
| } | } |
| void Charset::initTranscoder(const String *source, const char *name_cstr) { | void Charset::initTranscoder(const String::Body NAME, const char* name_cstr) { |
| transcoder=xmlFindCharEncodingHandler(name_cstr); | ftranscoder=xmlFindCharEncodingHandler(name_cstr); |
| if(!transcoder) | transcoder(NAME); // check right way |
| throw Exception(0, 0, | } |
| source, | |
| xmlCharEncodingHandler& Charset::transcoder(const String::Body NAME) { | |
| if(!ftranscoder) | |
| throw Exception("parser.runtime", | |
| new String(NAME, String::L_TAINTED), | |
| "unsupported encoding"); | "unsupported encoding"); |
| return *ftranscoder; | |
| } | } |
| const char *Charset::transcode_cstr(xmlChar *s) { | String::C Charset::transcode_cstr(xmlChar* s) { |
| if(!transcoder) | |
| throw Exception(0, 0, | |
| 0, | |
| "transcode_cstr no transcoder"); | |
| if(!s) | if(!s) |
| return ""; | return String::C("", 0); |
| int inlen=strlen((const char *)s); | int inlen=strlen((const char*)s); |
| int outlen=inlen+1; // max | int outlen=inlen*6/*strlen("ÿ")*/; // max |
| char *out=(char *)malloc(outlen*sizeof(char)); | #ifndef NDEBUG |
| int saved_outlen=outlen; | |
| int size=transcoder->output( | #endif |
| (unsigned char*)out, &outlen, | char *out=new(PointerFreeGC) char[outlen+1]; |
| (const unsigned char*)s, &inlen, | |
| transcoder->outputInfo); | int error; |
| if(size<0) | if(xmlCharEncodingOutputFunc output=transcoder(FNAME).output) { |
| throw Exception(0, 0, | error=output( |
| (unsigned char*)out, &outlen, | |
| (const unsigned char*)s, &inlen | |
| #ifdef PA_PATCHED_LIBXML_BACKWARD | |
| ,0 | |
| #endif | |
| ); | |
| } else { | |
| memcpy(out, s, outlen=inlen); | |
| error=0; | |
| } | |
| if(error<0) | |
| throw Exception(0, | |
| 0, | 0, |
| "transcode_cstr failed (%d)", size); | "transcode_cstr failed (%d)", error); |
| 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) { | |
| String::C cstr=transcode_cstr(s); | |
| return *new String(cstr.str, cstr.length, true); | |
| } | |
| out[size]=0; | /// @test less memory using -maybe- xmlParserInputBufferCreateMem |
| 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 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 | |
| #ifdef PA_PATCHED_LIBXML_BACKWARD | |
| ,0 | |
| #endif | |
| ); | |
| } else { | |
| outlen=buf_size; | |
| #ifndef NDEBUG | |
| saved_outlen=outlen; | |
| #endif | |
| out=(xmlChar*)xmlMalloc(outlen+1); | |
| memcpy(out, buf, outlen); | |
| error=0; | |
| } | |
| if(error<0) | |
| throw Exception(0, | |
| 0, | |
| "transcode_buf failed (%d)", error); | |
| assert(outlen<=saved_outlen); out[outlen]=0; | |
| return out; | return out; |
| } | } |
| String& Charset::transcode(xmlChar *s) { | GdomeDOMString_auto_ptr Charset::transcode_buf2dom(const char* buf, size_t buf_size) { |
| return *NEW String(pool(), transcode_cstr(s)); | return GdomeDOMString_auto_ptr(transcode_buf2xchar(buf, buf_size)); |
| } | } |
| const char *Charset::transcode_cstr(GdomeDOMString *s) { | GdomeDOMString_auto_ptr Charset::transcode(const String& s) { |
| return s?transcode_cstr(BAD_CAST s->str):""; | const char* cstr=s.cstr(String::L_UNSPECIFIED); |
| return transcode_buf2dom(cstr, strlen(cstr)); | |
| } | } |
| String& Charset::transcode(GdomeDOMString *s) { | GdomeDOMString_auto_ptr Charset::transcode(const String::Body s) { |
| return *NEW String(pool(), transcode_cstr(s)); | const char* cstr=s.cstr(); |
| return transcode_buf2dom(cstr, s.length()); | |
| } | } |
| #endif | |
| /// @test less memory using -maybe- xmlParserInputBufferCreateMem | String::Body Charset::transcode(const String::Body src, |
| GdomeDOMString_auto_ptr Charset::transcode_buf(const char *buf, size_t buf_size) { | const Charset& source_transcoder, |
| if(!transcoder) | const Charset& dest_transcoder) { |
| throw Exception(0, 0, | |
| 0, | |
| "transcode_buf no transcoder"); | |
| int outlen=buf_size*6/*max*/+1; | const char *src_ptr=src.cstr(); |
| unsigned char *out=(unsigned char*)malloc(outlen*sizeof(unsigned char)); | size_t src_size=strlen(src_ptr); |
| int size=transcoder->input( | |
| out, &outlen, | |
| (const unsigned char *)buf, (int *)&buf_size, | |
| transcoder->inputInfo); | |
| if(size<0) | |
| throw Exception(0, 0, | |
| 0, | |
| "transcode_buf failed (%d)", size); | |
| out[size]=0; | String::C dest=Charset::transcode(String::C(src_ptr, src_size), |
| return (gchar*)out; | 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); | |
| } | } |
| GdomeDOMString_auto_ptr Charset::transcode(const String& s) { | |
| const char *cstr=s.cstr(String::UL_UNSPECIFIED); | |
| return transcode_buf(cstr, strlen(cstr)); | void Charset::transcode(ArrayString& src, |
| const Charset& source_transcoder, | |
| const Charset& dest_transcoder) { | |
| for(size_t i=0; i<src.count(); i++) | |
| src.put(i, &transcode(*src[i], source_transcoder, dest_transcoder)); | |
| } | } |
| #ifndef DOXYGEN | |
| struct Transcode_pair_info { | |
| const Charset* source_transcoder; | |
| const Charset* dest_transcoder; | |
| }; | |
| #endif | #endif |
| static void transcode_pair(const String::Body /*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); | |
| } |