--- parser3/src/main/pa_string.C 2003/03/27 10:53:01 1.172.2.21.2.33 +++ parser3/src/main/pa_string.C 2003/04/10 11:35:10 1.172.2.21.2.39 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/27 10:53:01 $"; +static const char* IDENT_STRING_C="$Date: 2003/04/10 11:35:10 $"; #include "pcre.h" @@ -28,8 +28,7 @@ public: *this+=new String("match"); *this+=new String("postmatch"); for(int i=0; i i(src); ; fragment_begin=fragment_end) { - const element_type fragment=i.next(); + const Fragment fragment=i.next(); fragment_end=fragment_begin+fragment.length; + //fprintf(err, "1end=%u\n", fragment_end);fflush(err); - if(substr_beginfragment_end) // not reached fragments which may include 'substr'? + // not reached fragments which may include 'substr'? + if(!(substr_begin>=fragment_begin && substr_begin<=fragment_end)) continue; // found first fragment including piece of 'substr' - size_t piece_end=min(substr_end, fragment_end); - *this+=Fragment(fragment.lang, piece_end-substr_begin); - - while(substr_end>fragment_end) { // are there more fragments including pieces of 'substr'? - fragment_begin=fragment_end; - - const element_type fragment=i.next(); - fragment_end=fragment_begin+fragment.length; - - if(substr_end>fragment_end) // are there still more? - append(Fragment(fragment.lang, fragment.length)); // appending whole fragment - else { // no, it was last - append(Fragment(fragment.lang, substr_end-fragment_begin)); - return; + if(substr_end<=fragment_end) // fits into first fragment? + *this+=Fragment(fragment.lang, substr_end-substr_begin); + else { // spans more then one fragment + *this+=Fragment(fragment.lang, fragment_end-substr_begin); + while(true) { + const Fragment fragment=i.next(); + fragment_end=(fragment_begin=fragment_end)+fragment.length; + //fprintf(err, "2end=%u\n", fragment_end);fflush(err); + + if(substr_end>fragment_end) // are there still more? + append(Fragment(fragment.lang, fragment.length)); // appending whole fragment + else { // no, it was last + append(Fragment(fragment.lang, substr_end-fragment_begin)); + //fclose(err); + return; + } } } break; - } + } + //fclose(err); } // StringBody methods @@ -105,24 +108,46 @@ uint StringBody::hash_code() const { // String methods -String::String(const char* cstr, size_t length, bool tainted): body(CORD_EMPTY) { - append(cstr, length, tainted?L_TAINTED:L_CLEAN); +String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { + append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); +} +String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) { + append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN); } String::String(const String& src): body(src.body) { fragments.append(src.fragments); + ASSERT_STRING_INVARIANT(*this); } -String& String::append(const char* str, size_t length, Language lang) { +String& String::append_know_length(const char* str, size_t known_length, Language lang) { + if(!known_length || !*str) + return *this; + + body.append_know_length(str, known_length); + fragments+=Fragment(lang, known_length); + + ASSERT_STRING_INVARIANT(*this); + return *this; +} +String& String::append_help_length(const char* str, size_t helper_length, Language lang) { if(!str) return *this; - size_t use_length=length?length:strlen(str); - if(!use_length) + size_t known_length=helper_length?helper_length:strlen(str); + if(!known_length) + return *this; + + return append_know_length(str, known_length, lang); +} +String& String::append_strdup(const char* str, size_t helper_length, Language lang) { + size_t known_length=helper_length?helper_length:strlen(str); + if(!known_length) return *this; - body.append(str, use_length); - fragments+=Fragment(lang, use_length); + body.append_strdup_know_length(str, known_length); + fragments+=Fragment(lang, known_length); + ASSERT_STRING_INVARIANT(*this); return *this; } @@ -131,7 +156,7 @@ String& String::mid(size_t substr_begin, String& result=*new String; size_t self_length=length(); - substr_begin=max(min(substr_begin, self_length), (size_t)0); + substr_begin=min(substr_begin, self_length); substr_end=min(max(substr_end, substr_begin), self_length); if(substr_begin==substr_end) return result; @@ -144,6 +169,7 @@ String& String::mid(size_t substr_begin, // SAPI::log("piece of '%s' from %d to %d is '%s'", //cstr(), substr_begin, substr_end, result.cstr()); + ASSERT_STRING_INVARIANT(result); return result; } @@ -200,9 +226,9 @@ size_t String::pos(const String& substr, } void String::split(ArrayString& result, - size_t& pos_after, - const char* delim, - Language lang, int limit) const { + size_t& pos_after, + const char* delim, + Language lang, int limit) const { size_t self_length=length(); if(size_t delim_length=strlen(delim)) { int pos_before; @@ -223,9 +249,9 @@ void String::split(ArrayString& result, } void String::split(ArrayString& result, - size_t& pos_after, - const String& delim, Language lang, - int limit) const { + size_t& pos_after, + const String& delim, Language lang, + int limit) const { if(!delim.is_empty()) { int pos_before; // while we have 'delim'... @@ -277,10 +303,10 @@ static void regex_options(const String* } Table* String::match(Charset& source_charset, - const String& regexp, - const String* options, - Row_action row_action, void *info, - bool& just_matched) const { + const String& regexp, + const String* options, + Row_action row_action, void *info, + bool& just_matched) const { if(regexp.is_empty()) throw Exception(0, 0, @@ -432,7 +458,7 @@ const String& String::replace(const Dict if(Table::element_type row=dict.first_that_begins(current)) { // prematch if(size_t prematch_length=current-prematch_begin) { - result.body.append(prematch_begin, prematch_length); + result.body.append_strdup_know_length(prematch_begin, prematch_length); result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); } @@ -444,10 +470,7 @@ const String& String::replace(const Dict if(row->count()>1) { // are there any b? const String* b=row->get(1); - if(size_t b_length=b->length()) { - result.body<cstr_to_string_body(); - result.fragments.append(b->fragments); - } + result<<*b; } } else // simply advance current++; @@ -455,10 +478,11 @@ const String& String::replace(const Dict // postmatch if(size_t postmatch_length=current-prematch_begin) { - result.body.append(prematch_begin, postmatch_length); + result.body.append_strdup_know_length(prematch_begin, postmatch_length); result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); } + ASSERT_STRING_INVARIANT(result); return result; } @@ -535,12 +559,12 @@ static int serialize_body_piece(const ch memcpy(*cur, s, length); *cur+=length; return 0; }; -String::C String::serialize(size_t prolog_length) const { +String::Cm String::serialize(size_t prolog_length) const { size_t buf_length= prolog_length +fragments.count()*(sizeof(Language)+sizeof(size_t)) +length(); - C result(new(PointerFreeGC) char[buf_length], buf_length); + String::Cm result(new(PointerFreeGC) char[buf_length], buf_length); // 1: prolog char *cur=result.str+prolog_length; @@ -577,27 +601,30 @@ bool String::deserialize(size_t prolog_l return false; size_t fragments_count=*reinterpret_cast(cur); cur+=sizeof(size_t); buf_length-=sizeof(size_t); + + if(fragments_count) { + // 3: lang info + size_t total_length=0; + for(size_t f=0; f(cur); cur+=sizeof(Language); + size_t fragment_length=*reinterpret_cast(cur); cur+=sizeof(size_t); + fragments+=Fragment(lang, fragment_length); + total_length+=fragment_length; - // 3: lang info - size_t total_length=0; - for(size_t f=0; f(cur); cur+=sizeof(Language); - size_t fragment_length=*reinterpret_cast(cur); cur+=sizeof(size_t); - fragments+=Fragment(lang, fragment_length); - total_length+=fragment_length; + // 4: letters + if(buf_length!=total_length) + return false; - buf_length-=piece_length; + body=StringBody(cur, buf_length); } - // 4: letters - if(buf_length!=total_length) - return false; - - body=CORD_cat_char_star(CORD_EMPTY, cur, buf_length); - + ASSERT_STRING_INVARIANT(*this); return true; }