--- parser3/src/main/pa_string.C 2009/06/20 02:16:18 1.228 +++ parser3/src/main/pa_string.C 2009/07/14 11:15:29 1.232 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_STRING_C="$Date: 2009/06/20 02:16:18 $"; +static const char * const IDENT_STRING_C="$Date: 2009/07/14 11:15:29 $"; #include "pa_string.h" #include "pa_exception.h" @@ -166,41 +166,108 @@ String::Body String::Body::Format(int va } String::Body String::Body::trim(String::Trim_kind kind, const char* chars, - size_t* out_start, size_t* out_length) const { + size_t* out_start, size_t* out_length, Charset* source_charset) const { size_t our_length=length(); if(!our_length) return *this; - if(!chars) - chars=" \t\n"; // white space + + // check if any UTF-8 in chars + bool fast=true; + if(chars && source_charset && source_charset->isUTF8()){ + const char* pos=chars; + while(unsigned char c=*pos++) + if(c>127){ + fast=false; + break; + } + } size_t start=0; size_t end=our_length; - // from left... - if(kind!=TRIM_END) { - CORD_pos pos; set_pos(pos, 0); - while(true) { - char c=CORD_pos_fetch(pos); - if(strchr(chars, c)) { - if(++start==our_length) - return 0; // all chars are empty, just return empty string - } else - break; - - CORD_next(pos); - } - } - // from right.. - if(kind!=TRIM_START) { - CORD_pos pos; set_pos(pos, end-1); - while(true) { - char c=CORD_pos_fetch(pos); - if(strchr(chars, c)) { - if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible - return 0; // all chars are empty, just return empty string - } else - break; + if(!chars) + chars=" \t\n"; // white space + + if(fast){ + // from left... + if(kind!=TRIM_END) { + CORD_pos pos; set_pos(pos, 0); + while(true) { + char c=CORD_pos_fetch(pos); + if(strchr(chars, c)) { + if(++start==our_length) + return 0; // all chars are empty, just return empty string + } else + break; + + CORD_next(pos); + } + } + + // from right.. + if(kind!=TRIM_START) { + CORD_pos pos; set_pos(pos, end-1); + while(true) { + char c=CORD_pos_fetch(pos); + if(strchr(chars, c)) { + if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible + return 0; // all chars are empty, just return empty string + } else + break; + + CORD_prev(pos); + } + } + } else { + const XMLByte* src_begin=(const XMLByte*)cstrm(); + const XMLByte* src_end=src_begin+our_length; + + // from left... + if(kind!=TRIM_END) { + while(src_begin127 && *src_begin<0xC0)) + char_length++; + + bool found=false; + for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++) + if(strncmp(chars_byte, (const char*)ptr, char_length)==0){ + found=true; + break; + } + + if(found){ + start+=char_length; + if(start==our_length) + return 0; // all chars are empty, just return empty string + } else + break; + } + } - CORD_prev(pos); + // from right.. + if(kind!=TRIM_START) { + while(src_begin127 && *src_end<0xC0)) + char_length++; + + bool found=false; + for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++) + if(strncmp(chars_byte, (const char*)src_end, char_length)==0){ + found=true; + break; + } + + if(found){ + end-=char_length; + if(end==0) + return 0; // all chars are empty, just return empty string + } else + break; + } } } @@ -342,12 +409,12 @@ String& String::mid(Charset& charset, si return result; if(charset.isUTF8()){ - const XMLByte* srcPtr=(const XMLByte*)cstrm(); - const XMLByte* srcEnd=srcPtr+body.length(); + const XMLByte* src_begin=(const XMLByte*)cstrm(); + const XMLByte* src_end=src_begin+body.length(); // convert 'from' and 'substr_length' from 'characters' to 'bytes' - from=getUTF8BytePos(srcPtr, srcEnd, from); - substr_length=getUTF8BytePos(srcPtr+from, srcEnd, substr_length); + from=getUTF8BytePos(src_begin, src_end, from); + substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length); if(!substr_length) return result; } @@ -384,7 +451,6 @@ size_t String::pos(const String& substr, size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const { - size_t result; if(charset.isUTF8()){ const XMLByte* srcPtr=(const XMLByte*)cstrm(); const XMLByte* srcEnd=srcPtr+body.length(); @@ -392,19 +458,16 @@ size_t String::pos(Charset& charset, con // convert 'this_offset' from 'characters' to 'bytes' this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset); - result=pos(substr.body, this_offset, lang); - if(result==CORD_NOT_FOUND) - return STRING_NOT_FOUND; - - // convert 'result' from 'bytes' to 'characters' - result=getUTF8CharPos(srcPtr, srcEnd, result); + size_t result=pos(substr.body, this_offset, lang); + return (result==CORD_NOT_FOUND) + ? STRING_NOT_FOUND + : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters' } else { - result=pos(substr.body, this_offset, lang); - if(result==CORD_NOT_FOUND) - return STRING_NOT_FOUND; + size_t result=pos(substr.body, this_offset, lang); + return (result==CORD_NOT_FOUND) + ? STRING_NOT_FOUND + : result; } - - return result; } void String::split(ArrayString& result, @@ -480,7 +543,7 @@ Table* String::match(VRegex* vregex, int prefinish=ovector[0]; poststart=ovector[1]; - ArrayString* row=new ArrayString; + ArrayString* row=new ArrayString(3); if(need_pre_post_match) { *row+=&mid(0, prefinish); // .prematch column value *row+=&mid(prefinish, poststart); // .match @@ -493,7 +556,7 @@ Table* String::match(VRegex* vregex, for(int i=1; i=0 && ovector[i*2+1]>0)?&mid(ovector[i*2+0], ovector[i*2+1]):new String; // .i column value } matches_count++; @@ -676,9 +739,9 @@ bool String::deserialize(size_t prolog_s return false; // 3: letters body=String::Body(*cur?cur:0); -#ifdef STRING_LENGTH_CACHING - body.set_length(body_length); -#endif +#ifdef STRING_LENGTH_CACHING + body.set_length(body_length); +#endif cur+=body_length+1; in_buf-=body_length+1; @@ -757,16 +820,18 @@ const char* String::v() const { return buf; } + void String::dump() const { body.dump(); langs.dump(); } -const String& String::trim(String::Trim_kind kind, const char* chars) const { + +const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const { if(is_empty()) return *this; size_t substr_begin, substr_length; - Body new_body=body.trim(kind, chars, &substr_begin, &substr_length); + Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset); if(new_body==body) // we received unchanged pointer, do likewise return *this; // new_body differs from body, adjust langs along