--- parser3/src/main/pa_string.C 2003/03/20 14:19:10 1.172.2.21.2.9 +++ parser3/src/main/pa_string.C 2003/03/20 15:39:03 1.172.2.21.2.12 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/20 14:19:10 $"; +static const char* IDENT_STRING_C="$Date: 2003/03/20 15:39:03 $"; #include "pcre.h" @@ -39,31 +39,33 @@ Table string_match_table_template(new St // String::ArrayFragment methods String::ArrayFragment& String::ArrayFragment::append(const ArrayFragment& src, - size_t substr_start, size_t substr_end) { - if(substr_start==substr_end) + size_t substr_begin, size_t substr_end) { + if(substr_begin==substr_end) return *this; - size_t fragment_start=0; + size_t fragment_begin=0; size_t fragment_end; - for(Array_iterator i(src); ; fragment_start=fragment_end) { + for(Array_iterator i(src); ; fragment_begin=fragment_end) { const element_type& fragment=i.next(); - fragment_end=fragment_start+fragment.length; + fragment_end=fragment_begin+fragment.length; - if(substr_startfragment_end) { // are there more fragments including pieces of 'substr'? + fragment_begin=fragment_end; + const element_type& fragment=i.next(); - fragment_end=fragment_start+fragment.length; + fragment_end=fragment_begin+fragment.length; if(substr_end>fragment_end) // are there still more? simple_append(Fragment(fragment.lang, fragment.length)); // appending whole fragment else { // no, it was last - simple_append(Fragment(fragment.lang, substr_end-fragment_start)); + simple_append(Fragment(fragment.lang, substr_end-fragment_begin)); goto break2; } } @@ -75,7 +77,7 @@ break2: return *this; } /* -void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_start, size_t substr_end) { +void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_begin, size_t substr_end) { } */ @@ -90,6 +92,19 @@ String::String(const String& src): body( String& String::append_cstr(const char* heap_variable, size_t helper_length, Untaint_lang lang) { if(!heap_variable) return *this; + size_t use_length; + if(helper_length) { +#ifdef _DEBUG + size_t real_length=strlen(heap_variable); + if(helper_length!=real_length) + throw Exception(0, + this, + "bad String::String(real=%d, helper%d", + real_length, helper_length); +#endif + use_length=helper_length; + } else + use_length=strlen(heap_variable); size_t length=helper_length?helper_length:strlen(heap_variable); if(!length) return *this; @@ -103,12 +118,12 @@ String& String::append_cstr(const char* String& String::append_copy(const char* auto_variable, size_t helper_length, Untaint_lang lang) { if(!auto_variable) return *this; - size_t length=helper_length?helper_length:strlen(auto_variable); - if(!length) + size_t use_length=helper_length?helper_length:strlen(auto_variable); + if(!use_length) return *this; - body=pa_strdup(auto_variable, length); - fragments+=Fragment(lang, length); + body=pa_strdup(auto_variable, use_length); + fragments+=Fragment(lang, use_length); return *this; } @@ -124,52 +139,69 @@ uint String::hash_code() const { return result; } -const String& String::mid(size_t substr_start, size_t substr_end) const { +/// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" +const String& String::mid(size_t substr_begin, size_t substr_end) const { String& result=*new String; - substr_start=min(substr_start, length()); - substr_end=max(substr_start, substr_end); - if(substr_start==substr_end) + size_t self_length=length(); + substr_begin=max(min(substr_begin, self_length), (size_t)0); + substr_end=min(max(substr_end, substr_begin), self_length); + if(substr_begin==substr_end) return result; // first: letters themselves - result.body=CORD_substr(body, substr_start, substr_end-substr_start); + result.body=CORD_substr(body, substr_begin, substr_end-substr_begin); // next: their langs - result.fragments.append(fragments, substr_start, substr_end); + result.fragments.append(fragments, substr_begin, substr_end); // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", - //cstr(), substr_start, substr_end, result.cstr()); + //cstr(), substr_begin, substr_end, result.cstr()); return result; } size_t String::pos(CORD substr, size_t this_offset, Untaint_lang lang) const { // first: letters themselves - size_t substr_start=CORD_str(body, this_offset, substr); - if(substr_start==CORD_NOT_FOUND) + size_t substr_begin=CORD_str(body, this_offset, substr); + if(substr_begin==CORD_NOT_FOUND) return STRING_NOT_FOUND; // next: check the lang when specified if(lang==UL_UNSPECIFIED) // ignore lang? - return substr_start; + return substr_begin; // substr must be in one fragment, and fragments' lang must = lang - size_t substr_end=substr_start+CORD_len(substr); - size_t fragment_start=0; + size_t substr_end=substr_begin+CORD_len(substr); + size_t fragment_begin=0; size_t fragment_end; - for(Array_iterator i(fragments); i.has_next(); fragment_start=fragment_end) { + for(Array_iterator i(fragments); i.has_next(); fragment_begin=fragment_end) { const Fragment& fragment=i.next(); - fragment_end=fragment_start+fragment.length; + fragment_end=fragment_begin+fragment.length; - if(substr_start=fragment_end) // begin of substr OUT of current fragment? continue; if(substr_end>fragment_end) // end of substr OUT of current fragment? - break; + throw Exception(0, // (*) see below + this, + "searching for '%s' starting from %ud problem: found begin in one fragment, but end in another", + CORD_to_const_char_star(substr), this_offset); + + if(fragment.lang==lang) + return substr_begin; + else { // bad lang... + /// WARNING: this possibly skips assert (*), but it's fast + substr_begin=CORD_str(body, fragment_end/*...search AFTER for more*/, substr); + if(substr_begin==CORD_NOT_FOUND) + return STRING_NOT_FOUND; - return fragment.lang==lang?substr_end:STRING_NOT_FOUND; + size_t substr_end=substr_begin+CORD_len(substr); + // and continuing with next fragment + } } return STRING_NOT_FOUND; @@ -404,21 +436,21 @@ String& String::change_case(Charset& sou const String& String::replace(const Dictionary& dict) const { String& result=*new String(); const char* old_cstr=cstr(); - const char* prematch_start=old_cstr; + const char* prematch_begin=old_cstr; for(const char* current=old_cstr; *current; ) { - if(Table::element_type row=dict.first_that_starts(current)) { + if(Table::element_type row=dict.first_that_begins(current)) { // prematch - if(size_t prematch_length=current-prematch_start) { - result.body=CORD_cat_char_star(result.body, prematch_start, prematch_length); - result.fragments.append(fragments, prematch_start-old_cstr, prematch_length); + if(size_t prematch_length=current-prematch_begin) { + result.body=CORD_cat_char_star(result.body, prematch_begin, prematch_length); + result.fragments.append(fragments, prematch_begin-old_cstr, prematch_length); } // match const String* a=row->get(0); - // skip 'a' in 'current'; move prematch_start - current+=a->length(); prematch_start=current; + // skip 'a' in 'current'; move prematch_begin + current+=a->length(); prematch_begin=current; if(row->count()>1) { // are there any b? const String* b=row->get(1); @@ -432,9 +464,9 @@ const String& String::replace(const Dict } // postmatch - if(size_t prematch_length=current-prematch_start) { - result.body=CORD_cat_char_star(result.body, prematch_start, prematch_length); - result.fragments.append(fragments, prematch_start-old_cstr, prematch_length); + if(size_t prematch_length=current-prematch_begin) { + result.body=CORD_cat_char_star(result.body, prematch_begin, prematch_length); + result.fragments.append(fragments, prematch_begin-old_cstr, prematch_length); } return result;