--- parser3/src/main/pa_string.C 2003/03/20 11:11:55 1.172.2.21.2.7 +++ parser3/src/main/pa_string.C 2003/03/20 15:39:03 1.172.2.21.2.12 @@ -1,11 +1,11 @@ /** @file - Parser: string class. @see untasize_t.C. + Parser: string class. @see untalength_t.C. Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/20 11:11:55 $"; +static const char* IDENT_STRING_C="$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.size; + 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.size; + fragment_end=fragment_begin+fragment.length; if(substr_end>fragment_end) // are there still more? - simple_append(Fragment(fragment.lang, fragment.size)); // appending whole fragment + 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; } } @@ -75,32 +77,53 @@ 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) { } */ // String methods -String::String(const char* src, size_t src_size, bool tainted): body(CORD_EMPTY) { - if(src) - if(tainted) - APPEND_TAINTED(src, src_size, 0, 0); - else - APPEND_CLEAN(src, src_size, 0, 0); +String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { + append_cstr(cstr, helper_length, tainted?UL_TAINTED:UL_CLEAN); } String::String(const String& src): body(src.body), fragments(src.fragments) {} -String& String::real_append(STRING_APPEND_PARAMS) { - if(!src) +String& String::append_cstr(const char* heap_variable, size_t helper_length, Untaint_lang lang) { + if(!heap_variable) return *this; - if(!size) - size=strlen(src); - if(!size) + 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; - body=CORD_cat_char_star(body, src, size); - fragments+=Fragment(lang, size); + body=CORD_cat_char_star(body, heap_variable, length); + fragments+=Fragment(lang, length); + + return *this; +} + +String& String::append_copy(const char* auto_variable, size_t helper_length, Untaint_lang lang) { + if(!auto_variable) + return *this; + size_t use_length=helper_length?helper_length:strlen(auto_variable); + if(!use_length) + return *this; + + body=pa_strdup(auto_variable, use_length); + fragments+=Fragment(lang, use_length); return *this; } @@ -116,52 +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, size()); - 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.size; + 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; @@ -176,22 +216,22 @@ void String::split(ArrayString& result, size_t& pos_after, const char* delim, Untaint_lang lang, int limit) { - size_t self_size=size(); - if(size_t delim_size=strlen(delim)) { + size_t self_length=length(); + if(size_t delim_length=strlen(delim)) { int pos_before; // while we have 'delim'... for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { result+=&mid(pos_after, pos_before); - pos_after=pos_before+delim_size; + pos_after=pos_before+delim_length; } // last piece - if(pos_after=0 && limit; limit--) { result+=&mid(pos_after, pos_before); - pos_after=pos_before+delim.size(); + pos_after=pos_before+delim.length(); } // last piece - if(pos_afterget(0); - // skip 'a' in 'current'; move prematch_start - current+=a->size(); 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); - if(size_t b_size=b->size()) { + if(size_t b_length=b->length()) { result.body=CORD_cat(result.body, b->cstr_to_cord()); result.fragments.append(b->fragments, 0, b->fragments.count()); } @@ -424,9 +464,9 @@ const String& String::replace(const Dict } // postmatch - if(size_t prematch_size=current-prematch_start) { - result.body=CORD_cat_char_star(result.body, prematch_start, prematch_size); - result.fragments.append(fragments, prematch_start-old_cstr, prematch_size); + 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; @@ -503,19 +543,19 @@ inline uint uchars2uint(uchar *bytes) { static int CORD_batched_iter_fn_append(const char* s, void* client_data) { char*& cur=*static_cast(client_data); - size_t size=strlen(s); - memcpy(cur, s, size); cur+=size; + size_t length=strlen(s); + memcpy(cur, s, length); cur+=length; return 0; }; -String::C String::serialize(size_t prolog_size) const { - size_t buf_size= - prolog_size +String::C String::serialize(size_t prolog_length) const { + size_t buf_length= + prolog_length +fragments.count()*(sizeof(Untaint_lang)+sizeof(size_t)) - +size(); - C result(new(PointerFreeGC) char[buf_size], buf_size); + +length(); + C result(new(PointerFreeGC) char[buf_length], buf_length); // 1: prolog - char *cur=result.str+prolog_size; + char *cur=result.str+prolog_length; // 2: fragments.count @@ -527,8 +567,8 @@ String::C String::serialize(size_t prolo const Fragment& fragment=i.next(); // lang memcpy(cur, &fragment.lang, sizeof(fragment.lang)); cur+=sizeof(fragment.lang); - // size - memcpy(cur, &fragment.size, sizeof(fragment.size)); cur+=sizeof(fragment.size); + // length + memcpy(cur, &fragment.length, sizeof(fragment.length)); cur+=sizeof(fragment.length); } // 4: letters @@ -536,40 +576,40 @@ String::C String::serialize(size_t prolo return result; } -bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) { - if(buf_size<=prolog_size) +bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length, const char* file) { + if(buf_length<=prolog_length) return false; - buf_size-=prolog_size; + buf_length-=prolog_length; // 1: prolog - const char* cur=(const char* )buf+prolog_size; + const char* cur=(const char* )buf+prolog_length; // 2: fragments.count - if(buf_size(cur); cur+=sizeof(size_t); - buf_size-=sizeof(size_t); + buf_length-=sizeof(size_t); // 3: lang info - size_t total_size=0; + size_t total_length=0; for(size_t f=0; f(cur); cur+=sizeof(Untaint_lang); - size_t fragment_size=*reinterpret_cast(cur); cur+=sizeof(size_t); - fragments+=Fragment(lang, fragment_size); - total_size+=fragment_size; + size_t fragment_length=*reinterpret_cast(cur); cur+=sizeof(size_t); + fragments+=Fragment(lang, fragment_length); + total_length+=fragment_length; - buf_size-=piece_size; + buf_length-=piece_length; } // 4: letters - if(buf_size!=total_size) + if(buf_length!=total_length) return false; - body=CORD_cat_char_star(CORD_EMPTY, cur, buf_size); + body=CORD_cat_char_star(CORD_EMPTY, cur, buf_length); return true; }