--- parser3/src/main/pa_string.C 2003/03/20 13:09:54 1.172.2.21.2.8 +++ parser3/src/main/pa_string.C 2003/09/26 13:36:35 1.180 @@ -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 13:09:54 $"; +static const char* IDENT_STRING_C="$Date: 2003/09/26 13:36:35 $"; #include "pcre.h" @@ -15,6 +15,45 @@ static const char* IDENT_STRING_C="$Date #include "pa_dictionary.h" #include "pa_charset.h" +// cord lib extension + +#ifndef DOXYGEN +typedef struct { + ssize_t countdown; + char target; /* Character we're looking for */ +} chr_data; +#endif +static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data) +{ + register chr_data * d = (chr_data *)client_data; + + if (d -> countdown<=0) return(2); + d -> countdown -= size; + if (c > d -> target) return(1); + return(0); +} +int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c) +{ + chr_data d; + + d.countdown = n; + d.target = c; + return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/); +} + +static int CORD_block_count_proc(char c, size_t size, void* client_data) +{ + int* result=(int*)client_data; + (*result)++; + return(0); // 0=continue +} +size_t CORD_block_count(CORD x) +{ + size_t result=0; + CORD_block_iter(x, 0, CORD_block_count_proc, &result); + return result; +} + // helpers /// String::match uses this as replace & global search table columns @@ -28,196 +67,164 @@ public: *this+=new String("match"); *this+=new String("postmatch"); for(int i=0; i i(src); ; fragment_start=fragment_end) { - const element_type& fragment=i.next(); - fragment_end=fragment_start+fragment.size; - - if(substr_startfragment_end) { // are there more fragments including pieces of 'substr'? - const element_type& fragment=i.next(); - fragment_end=fragment_start+fragment.size; - - if(substr_end>fragment_end) // are there still more? - simple_append(Fragment(fragment.lang, fragment.size)); // appending whole fragment - else { // no, it was last - simple_append(Fragment(fragment.lang, substr_end-fragment_start)); - goto break2; - } - } +String::Body String::Body::Format(int value) { + char local[MAX_NUMBER]; + size_t length=snprintf(local, MAX_NUMBER, "%d", value); + return String::Body(pa_strdup(local, length), length); +} - break; - } -break2: - - return *this; +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; } -/* -void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_start, size_t substr_end) { +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); + return 0; +}; +uint String::Body::hash_code() const { + uint result=0; + 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* src, size_t src_size, bool tainted): body(CORD_EMPTY) { - if(src) - if(tainted) - APPEND_TAINTED(src, src_size); - else - APPEND_CLEAN(src, src_size); +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(src.fragments) {} - -String& String::append(const char* src, size_t size, Untaint_lang lang) { - if(!src) - return *this; - if(!size) - size=strlen(src); - if(!size) +String& String::append_know_length(const char* str, size_t known_length, Language lang) { + if(!known_length) return *this; - body=CORD_cat_char_star(body, src, size); - fragments+=Fragment(lang, size); + // first: langs + langs.append(body, lang, known_length); + // next: letters themselves + body.append_know_length(str, 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 known_length=helper_length?helper_length:strlen(str); + if(!known_length) + return *this; -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); - return 0; -}; -uint String::hash_code() const { - uint result=0; - CORD_iter5(body, 0, 0, CORD_batched_iter_fn_generic_hash_code, &result); - return result; + 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; + + // first: langs + langs.append(body, lang, known_length); + // next: letters themselves + body.append_strdup_know_length(str, known_length); + + ASSERT_STRING_INVARIANT(*this); + return *this; } -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" +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=min(substr_begin, self_length); + substr_end=min(max(substr_end, substr_begin), self_length); + size_t substr_length=substr_end-substr_begin; + if(!substr_length) return result; - // first: letters themselves - result.body=CORD_substr(body, substr_start, substr_end-substr_start); - - // next: their langs - result.fragments.append(fragments, substr_start, substr_end); - -// SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", - //cstr(), substr_start, substr_end, result.cstr()); + // first: their langs + result.langs.append(result.body, langs, substr_begin, substr_length); + // next: letters themselves + result.body=body.mid(substr_begin, substr_length); + +// SAPI::log("piece of '%s' from %d to %d is '%s'", + //cstr(), substr_begin, substr_end, result.cstr()); + ASSERT_STRING_INVARIANT(result); 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 String::pos(const String::Body substr, size_t this_offset, Language lang) const { + size_t substr_begin=body.pos(substr, this_offset); + if(substr_begin==CORD_NOT_FOUND || !langs.check_lang(lang, substr_begin, substr.length())) return STRING_NOT_FOUND; - // next: check the lang when specified - - if(lang==UL_UNSPECIFIED) // ignore lang? - return substr_start; - - // 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 fragment_end; - for(Array_iterator i(fragments); i.has_next(); fragment_start=fragment_end) { - const Fragment& fragment=i.next(); - fragment_end=fragment_start+fragment.size; - - if(substr_startfragment_end) // end of substr OUT of current fragment? - break; - - return fragment.lang==lang?substr_end:STRING_NOT_FOUND; - } - - return STRING_NOT_FOUND; + return substr_begin; } size_t String::pos(const String& substr, - size_t this_offset, Untaint_lang lang) const { + size_t this_offset, Language lang) const { return pos(substr.body, this_offset, lang); } 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& 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; // 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_size; + pos_after=pos_before+delim_length; } // last piece - if(pos_after=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.size(); + pos_after=pos_before+delim.length(); } // last piece - if(pos_afteris_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 @@ -250,11 +257,11 @@ 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 { - if(!regexp) + const String& regexp, + const String* options, + Row_action row_action, void *info, + bool& just_matched) const { + if(regexp.is_empty()) throw Exception(0, 0, "regexp is empty"); @@ -263,7 +270,7 @@ 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, @@ -271,7 +278,7 @@ Table* String::match(Charset& source_cha if(!code) throw Exception(0, - ®exp.mid(erroffset, regexp.size()), + ®exp.mid(erroffset, regexp.length()), "regular expression syntax error - %s", errptr); int subpatterns=pcre_info(code, 0, 0); @@ -284,27 +291,28 @@ Table* String::match(Charset& source_cha } const char* subject=cstr(); - size_t subject_size=strlen(subject); - const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3; - int ovector[ovecsize]; + size_t subject_length=strlen(subject); + const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3; + 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; int poststart=0; - int postfinish=size(); + int postfinish=length(); while(true) { int exec_substrings=pcre_exec(code, 0, - subject, subject_size, prestart, - exec_option_bits, ovector, ovecsize); + subject, subject_length, prestart, + exec_option_bits, ovector, oveclength); if(exec_substrings==PCRE_ERROR_NOMATCH) { 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 } @@ -341,7 +349,7 @@ 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; @@ -352,8 +360,10 @@ 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; @@ -387,8 +397,8 @@ String& String::change_case(Charset& sou *dest++=(char)c; } + result.langs=langs; result.body=new_cstr; - result.fragments.append(fragments, 0, fragments.count()); return result; } @@ -396,39 +406,38 @@ String& String::change_case(Charset& sou const String& String::replace(const Dictionary& dict) const { String& result=*new String(); const char* old_cstr=cstr(); - const char* prematch_start=old_cstr; + const char* prematch_begin=old_cstr; - for(const char* current=old_cstr; *current; ) { - if(Table::element_type row=dict.first_that_starts(current)) { + const char* current=old_cstr; + while(*current) { + if(Table::element_type row=dict.first_that_begins(current)) { // prematch - 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.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); + result.body.append_strdup_know_length(prematch_begin, prematch_length); } // match const String* a=row->get(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()) { - result.body=CORD_cat(result.body, b->cstr_to_cord()); - result.fragments.append(b->fragments, 0, b->fragments.count()); - } + result<<*b; } } else // simply advance current++; } // 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 postmatch_length=current-prematch_begin) { + result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); + result.body.append_strdup_know_length(prematch_begin, postmatch_length); } + ASSERT_STRING_INVARIANT(result); return result; } @@ -487,89 +496,109 @@ int String::as_int() const { return result; } -inline void uint2uchars(uint word, uchar *bytes) { - bytes[0]=word&0xFF; - bytes[1]=(word>>8)&0xFF; - bytes[2]=(word>>16)&0xFF; - bytes[3]=(word>>24)&0xFF; -} -inline uint uchars2uint(uchar *bytes) { - return bytes[3]<<24 - | bytes[2]<<16 - | bytes[1]<<8 - | bytes[0]; -} - -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; - return 0; +static int serialize_body_char(char c, char** cur) { + *((*cur)++)=c; + return 0; // 0=continue }; -String::C String::serialize(size_t prolog_size) const { - size_t buf_size= - prolog_size - +fragments.count()*(sizeof(Untaint_lang)+sizeof(size_t)) - +size(); - C result(new(PointerFreeGC) char[buf_size], buf_size); +static int serialize_body_piece(const char* s, char** cur) { + size_t length=strlen(s); + memcpy(*cur, s, length); *cur+=length; + return 0; // 0=continue +}; +static int serialize_lang_piece(char alang, size_t asize, char** cur) { + // lang + memcpy(*cur, &alang, sizeof(alang)); *cur+=sizeof(alang); + // length + memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize); + + return 0; // 0=continue +} +String::Cm String::serialize(size_t prolog_length) const { + size_t fragments_count=langs.count(); + size_t buf_length= + prolog_length //1 + +sizeof(size_t) //2 + +fragments_count*(sizeof(char)+sizeof(size_t)) //3 + +body.length() //4 + +1; // for zero terminator used in deserialize + String::Cm result(new(PointerFreeGC) char[buf_length], buf_length); // 1: prolog - char *cur=result.str+prolog_size; - - - // 2: fragments.count - size_t fragments_count=fragments.count(); + char *cur=result.str+prolog_length; + // 2: langs.count memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count); - // 3: lang info - for(Array_iterator i(fragments); i.has_next(); ) { - 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); - } - + langs.for_each(body, serialize_lang_piece, &cur); // 4: letters - CORD_iter5(body, 0, 0, CORD_batched_iter_fn_append, &cur); + body.for_each(serialize_body_char, serialize_body_piece, &cur); + // 5: zero terminator already there put by new(PointerFreeGC) 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) { + if(buf_length<=prolog_length) return false; - buf_size-=prolog_size; + buf_length-=prolog_length; + buf_length-=1; // 5: zero terminator // 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); + + if(fragments_count) { + // 3: lang info + size_t total_length=0; + for(size_t f=0; f(cur); cur+=sizeof(char); + size_t fragment_length=*reinterpret_cast(cur); cur+=sizeof(size_t); + langs.append(total_length, lang, fragment_length); + total_length+=fragment_length; - // 3: lang info - size_t total_size=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; + // 4: letters + if(buf_length!=total_length) + return false; - buf_size-=piece_size; + // serialize wrote extra zero byte there, we can rely on that + body=String::Body(cur, buf_length); } - // 4: letters - if(buf_size!=total_size) - return false; + ASSERT_STRING_INVARIANT(*this); + return true; +} - body=CORD_cat_char_star(CORD_EMPTY, cur, buf_size); +const char* String::Body::v() const { + return CORD_to_const_char_star(body); +} +const char* String::Languages::v() const { + if(opt.is_not_just_lang) + return CORD_to_const_char_star(langs); + else + return (const char*)&langs; +} +const char* String::v() const { +#define LIMIT_VIEW 20 + char* buf=(char*)malloc(MAX_STRING); + const char*body_view=body.v(); + const char*langs_view=langs.v(); + snprintf(buf, MAX_STRING, + "%d:%.*s%s} " + "{%d:%s", + langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"", + strlen(body_view), body_view + ); - return true; + return buf; +#undef LIMIT_VIEW }