--- parser3/src/main/pa_string.C 2009/07/14 11:15:29 1.232 +++ parser3/src/main/pa_string.C 2012/03/12 22:29:41 1.239 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_STRING_C="$Date: 2009/07/14 11:15:29 $"; +static const char * const IDENT_STRING_C="$Date: 2012/03/12 22:29:41 $"; #include "pa_string.h" #include "pa_exception.h" @@ -218,7 +218,7 @@ String::Body String::Body::trim(String:: } } } else { - const XMLByte* src_begin=(const XMLByte*)cstrm(); + const XMLByte* src_begin=(const XMLByte*)cstr(); const XMLByte* src_end=src_begin+our_length; // from left... @@ -351,22 +351,31 @@ String& String::append_strdup(const char return *this; } -int CORD_batched_len(const char* s, size_t* len){ - (*len) += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); - return 0; +struct CORD_length_info { + size_t len; + size_t skip; +}; + +int CORD_batched_len(const char* s, CORD_length_info* info){ + info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0; } -// can be called only for IS_FUNCTION(CORD) which are used only in Lang -int CORD_batched_len(const char, size_t *len){ - (*len)++; +// can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid +int CORD_batched_len(const char c, CORD_length_info* info){ + if (info->skip==0){ + info->len++; + info->skip = lengthUTF8Char(c)-1; + } else { + info->skip--; + } return 0; } size_t String::length(Charset& charset) const { if(charset.isUTF8()){ - size_t len=0; - body.for_each(CORD_batched_len, CORD_batched_len, &len); - return len; + CORD_length_info info = {0, 0}; + body.for_each(CORD_batched_len, CORD_batched_len, &info); + return info.len; } else return body.length(); } @@ -409,7 +418,7 @@ String& String::mid(Charset& charset, si return result; if(charset.isUTF8()){ - const XMLByte* src_begin=(const XMLByte*)cstrm(); + const XMLByte* src_begin=(const XMLByte*)cstr(); const XMLByte* src_end=src_begin+body.length(); // convert 'from' and 'substr_length' from 'characters' to 'bytes' @@ -452,7 +461,7 @@ size_t String::pos(Charset& charset, con size_t this_offset, Language lang) const { if(charset.isUTF8()){ - const XMLByte* srcPtr=(const XMLByte*)cstrm(); + const XMLByte* srcPtr=(const XMLByte*)cstr(); const XMLByte* srcEnd=srcPtr+body.length(); // convert 'this_offset' from 'characters' to 'bytes' @@ -474,6 +483,8 @@ void String::split(ArrayString& result, size_t& pos_after, const char* delim, Language lang, int limit) const { + if(is_empty()) + return; size_t self_length=length(); if(size_t delim_length=strlen(delim)) { size_t pos_before; @@ -497,6 +508,8 @@ void String::split(ArrayString& result, size_t& pos_after, const String& delim, Language lang, int limit) const { + if(is_empty()) + return; if(!delim.is_empty()) { size_t pos_before; // while we have 'delim'... @@ -525,7 +538,7 @@ Table* String::match(VRegex* vregex, bool global=vregex->is_global_search(); const char* subject=cstr(); - size_t subject_length=strlen(subject); + size_t subject_length=length(); const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; int ovector[ovector_size]; @@ -543,6 +556,12 @@ Table* String::match(VRegex* vregex, int prefinish=ovector[0]; poststart=ovector[1]; + + if (prestart==poststart && subject[poststart]=='\n'){ + prestart++; + continue; + } + ArrayString* row=new ArrayString(3); if(need_pre_post_match) { *row+=&mid(0, prefinish); // .prematch column value @@ -636,34 +655,63 @@ const String& String::escape(Charset& so return Charset::escape(*this, source_charset); } +#define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \ + result.langs.append(result.body, langs, langs_offset, length); \ + result.body.append_strdup_know_length(from_cstr, length); + const String& String::replace(const Dictionary& dict) const { + if(!dict.count() || is_empty()) + return *this; + String& result=*new String(); const char* old_cstr=cstr(); const char* prematch_begin=old_cstr; - const char* current=old_cstr; - while(*current) { - if(Dictionary::Subst subst=dict.first_that_begins(current)) { + if(dict.count()==1) { + // optimized simple case + + Dictionary::Subst subst=dict.get(0); + while(const char* p=strstr(prematch_begin, subst.from)) { // prematch - if(size_t prematch_length=current-prematch_begin) { - result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); - result.body.append_strdup_know_length(prematch_begin, prematch_length); + if(size_t prematch_length=p-prematch_begin) { + STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length) } // match - // skip 'a' in 'current'; move prematch_begin - current+=subst.from_length; prematch_begin=current; + prematch_begin=p+subst.from_length; if(const String* b=subst.to) // are there any b? result<<*b; - } else // simply advance - current++; + } + + } else { + + const char* current=old_cstr; + while(*current) { + if(Dictionary::Subst subst=dict.first_that_begins(current)) { + // prematch + if(size_t prematch_length=current-prematch_begin) { + STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length) + } + + // match + // skip 'a' in 'current'; move prematch_begin + current+=subst.from_length; prematch_begin=current; + + if(const String* b=subst.to) // are there any b? + result<<*b; + } else // simply advance + current++; + } + } + if(prematch_begin==old_cstr) // not modified + return *this; + // postmatch - if(size_t postmatch_length=current-prematch_begin) { - result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); - result.body.append_strdup_know_length(prematch_begin, postmatch_length); + if(size_t postmatch_length=old_cstr+length()-prematch_begin) { + STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length) } ASSERT_STRING_INVARIANT(result); @@ -788,7 +836,7 @@ bool String::deserialize(size_t prolog_s } const char* String::Body::v() const { - return CORD_to_const_char_star(body); + return CORD_to_const_char_star(body, length()); } void String::Body::dump() const { CORD_dump(body); @@ -796,7 +844,7 @@ void String::Body::dump() const { const char* String::Languages::v() const { if(opt.is_not_just_lang) - return CORD_to_const_char_star(langs); + return CORD_to_const_char_star(langs, 0); else return (const char*)&langs; }