--- parser3/src/main/pa_string.C 2003/03/24 16:43:27 1.172.2.21.2.27 +++ parser3/src/main/pa_string.C 2003/04/08 11:08:12 1.172.2.21.2.36 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/24 16:43:27 $"; +static const char* IDENT_STRING_C="$Date: 2003/04/08 11:08:12 $"; #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); } -// StringSimple methods +// StringBody methods +StringBody StringBody::Format(int value) { + char local[MAX_NUMBER]; + size_t length=snprintf(local, MAX_NUMBER, "%d", value); + return StringBody(pa_strdup(local, length), length); +} static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) { uint& result=*static_cast(client_data); @@ -101,20 +109,32 @@ uint StringBody::hash_code() const { // String methods String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { - append(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); + append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); } -String::String(const String& src): body(src.body), fragments(src.fragments) {} +String::String(const String& src): body(src.body) { + fragments.append(src.fragments); +} -String& String::append(const char* str, size_t length, Language lang) { +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; - body.append(str, use_length); - fragments+=Fragment(lang, use_length); + body.append_know_length(str, known_length); + fragments+=Fragment(lang, known_length); + + return *this; +} +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_strdup_know_length(str, known_length); + fragments+=Fragment(lang, known_length); return *this; } @@ -124,7 +144,7 @@ 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; @@ -157,7 +177,7 @@ size_t String::pos(const StringBody subs size_t fragment_begin=0; size_t fragment_end; for(Array_iterator i(fragments); i.has_next(); fragment_begin=fragment_end) { - const Fragment& fragment=i.next(); + const Fragment fragment=i.next(); fragment_end=fragment_begin+fragment.length; if(substr_begincount()>1) { // are there any b? const String* b=row->get(1); - if(size_t b_length=b->length()) { - result.body.append(b->cstr_to_string_body()); - result.fragments.append(b->fragments); - } + result<<*b; } } else // simply advance current++; @@ -445,7 +465,7 @@ 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); } @@ -542,7 +562,7 @@ String::C String::serialize(size_t prolo // 3: lang info for(Array_iterator i(fragments); i.has_next(); ) { - const Fragment& fragment=i.next(); + const Fragment fragment=i.next(); // lang memcpy(cur, &fragment.lang, sizeof(fragment.lang)); cur+=sizeof(fragment.lang); // length @@ -567,27 +587,29 @@ 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); - return true; }