--- parser3/src/main/pa_string.C 2003/03/24 15:26:03 1.172.2.21.2.23 +++ parser3/src/main/pa_string.C 2003/07/28 07:36:39 1.174.2.1 @@ -5,7 +5,7 @@ Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_STRING_C="$Date: 2003/03/24 15:26:03 $"; +static const char* IDENT_STRING_C="$Date: 2003/07/28 07:36:39 $"; #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(fragment_begin=fragment_begin && substr_beginfragment_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; - } -} -/* -void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_begin, size_t substr_end) { + } + //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); @@ -102,21 +109,45 @@ 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::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(src.fragments) {} +String::String(const String& src): body(src.body) { + fragments.append(src.fragments); + ASSERT_STRING_INVARIANT(*this); +} + +String& String::append_know_length(const char* str, size_t known_length, Language lang) { + if(!known_length) + return *this; + + body.append_know_length(str, known_length); + fragments+=Fragment(lang, known_length); -String& String::append(const char* str, size_t length, Language lang) { + 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; } @@ -125,7 +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; @@ -138,6 +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; } @@ -158,7 +190,7 @@ 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_begin=0 && limit; limit--) { + for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { result+=&mid(pos_after, pos_before); pos_after=pos_before+delim_length; } @@ -217,13 +249,13 @@ 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'... - for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { + for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { result+=&mid(pos_after, pos_before); pos_after=pos_before+delim.length(); } @@ -259,8 +291,8 @@ static void regex_options(const String* if(options && !options->is_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)!=STRING_NOT_FOUND + || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) { if(o->flag) *o->flag=true; else { // result @@ -271,10 +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, @@ -284,7 +316,7 @@ 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, @@ -310,7 +342,8 @@ Table* String::match(Charset& source_cha int ovector[oveclength]; // create table - Table& table=*new Table(string_match_table_template); + Table::Action_options table_options; + Table& table=*new Table(string_match_table_template, table_options); int exec_option_bits=0; int prestart=0; @@ -375,6 +408,8 @@ Table* String::match(Charset& source_cha 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; @@ -419,11 +454,12 @@ 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_know_length(prematch_begin, prematch_length); result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); } @@ -435,10 +471,7 @@ 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.fragments.append(b->fragments); - } + result<<*b; } } else // simply advance current++; @@ -446,10 +479,11 @@ 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; } @@ -526,12 +560,14 @@ 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 { + //_asm int 3; size_t buf_length= prolog_length + +sizeof(size_t) +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; @@ -543,7 +579,7 @@ 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 @@ -568,27 +604,30 @@ 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; }