--- parser3/src/main/pa_string.C 2003/03/24 11:05:28 1.172.2.21.2.19 +++ parser3/src/main/pa_string.C 2003/03/27 14:51:29 1.172.2.21.2.34 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/24 11:05:28 $"; +static const char* IDENT_STRING_C="$Date: 2003/03/27 14:51:29 $"; #include "pcre.h" @@ -46,10 +46,13 @@ void String::ArrayFragment::append_posit size_t fragment_begin=0; size_t fragment_end; for(Array_iterator i(src); ; fragment_begin=fragment_end) { - const element_type& fragment=i.next(); + const element_type fragment=i.next(); fragment_end=fragment_begin+fragment.length; - if(fragment_beginfragment_end) // not reached fragments which may include 'substr'? continue; // found first fragment including piece of 'substr' @@ -59,7 +62,7 @@ void String::ArrayFragment::append_posit while(substr_end>fragment_end) { // are there more fragments including pieces of 'substr'? fragment_begin=fragment_end; - const element_type& fragment=i.next(); + const element_type fragment=i.next(); fragment_end=fragment_begin+fragment.length; if(substr_end>fragment_end) // are there still more? @@ -73,14 +76,20 @@ void String::ArrayFragment::append_posit break; } } -/* -void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_begin, size_t substr_end) { -} -*/ -// 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); + generic_hash_code(result, c); + return 0; +} static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) { uint& result=*static_cast(client_data); generic_hash_code(result, s); @@ -88,30 +97,40 @@ static int CORD_batched_iter_fn_generic_ }; uint StringBody::hash_code() const { uint result=0; - CORD_iter5(body, 0, 0, CORD_batched_iter_fn_generic_hash_code, &result); + CORD_iter5(body, 0, + CORD_batched_iter_fn_generic_hash_code, + CORD_batched_iter_fn_generic_hash_code, &result); return result; } // 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); +String::String(const char* cstr, size_t help_length, bool tainted): body(CORD_EMPTY) { + append_help_length(cstr, help_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); + size_t use_length=helper_length?helper_length:strlen(str); if(!use_length) return *this; - body.append(str, use_length); + body.append_help_length(str, use_length); fragments+=Fragment(lang, use_length); return *this; } +String& String::append_strdup(const char* str, size_t length, Language lang) { + body.append_strdup(str, length); + fragments+=Fragment(lang, length); + + return *this; +} /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" String& String::mid(size_t substr_begin, size_t substr_end) const { @@ -137,7 +156,7 @@ String& String::mid(size_t substr_begin, size_t String::pos(const StringBody substr, size_t this_offset, Language lang) const { // first: letters themselves - size_t substr_begin=body.pos(this_offset, substr); + size_t substr_begin=body.pos(substr, this_offset); if(substr_begin==CORD_NOT_FOUND) return STRING_NOT_FOUND; @@ -151,7 +170,7 @@ 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_beginis_empty()) for(Regex_option *o=regex_option; o->keyL; o++) - if(options.pos(o->keyL)>=0 - || (o->keyU && options.pos(o->keyU)>=0)) { + if(options->pos(o->keyL)>=0 + || (o->keyU && options->pos(o->keyU)>=0)) { if(o->flag) *o->flag=true; else { // result @@ -265,7 +284,7 @@ static void regex_options(const String& Table* String::match(Charset& source_charset, const String& regexp, - const String& options, + const String* options, Row_action row_action, void *info, bool& just_matched) const { if(regexp.is_empty()) @@ -277,7 +296,7 @@ Table* String::match(Charset& source_cha const char* errptr; int erroffset; bool need_pre_post_match=false; - int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); + int option_bits[2]={0}; regex_options(options, option_bits, need_pre_post_match); bool global=option_bits[1]!=0; pcre *code=pcre_compile(pattern, option_bits[0], &errptr, &erroffset, @@ -303,7 +322,7 @@ Table* String::match(Charset& source_cha int ovector[oveclength]; // create table - Table* table=new Table(string_match_table_template); + Table& table=*new Table(string_match_table_template); int exec_option_bits=0; int prestart=0; @@ -318,7 +337,7 @@ Table* String::match(Charset& source_cha pcre_free(code); row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); if(global || subpatterns) - return table; // global or with subpatterns=true+result + return &table; // global or with subpatterns=true+result else { just_matched=false; return 0; // not global=no result } @@ -355,7 +374,7 @@ Table* String::match(Charset& source_cha if(!global || prestart==poststart) { // not global | going to hang pcre_free(code); row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); - return table; + return &table; } prestart=poststart; @@ -366,8 +385,10 @@ Table* String::match(Charset& source_cha } } -String& String::change_case(Charset& source_charset, Change_case_kind kind) { +String& String::change_case(Charset& source_charset, Change_case_kind kind) const { String& result=*new String(); + if(is_empty()) + return result; const unsigned char *tables=source_charset.pcre_tables; @@ -412,11 +433,12 @@ const String& String::replace(const Dict const char* old_cstr=cstr(); const char* prematch_begin=old_cstr; - for(const char* current=old_cstr; *current; ) { + const char* current=old_cstr; + while(*current) { 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(prematch_begin, prematch_length); result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); } @@ -429,7 +451,7 @@ 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.append(b->cstr_to_string_body()); + result.body<cstr_to_string_body(); result.fragments.append(b->fragments); } } @@ -439,7 +461,7 @@ 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(prematch_begin, postmatch_length); result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); } @@ -536,7 +558,7 @@ 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