|
|
| version 1.7, 2001/01/27 10:02:59 | version 1.172.2.21.2.5, 2003/03/19 16:28:09 |
|---|---|
| Line 1 | Line 1 |
| /* | /** @file |
| $Id$ | Parser: string class. @see untasize_t.C. |
| Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| */ | */ |
| #include <string.h> | static const char* IDENT_STRING_C="$Date$"; |
| #include "pcre.h" | |
| #include "pa_string.h" | |
| #include "pa_exception.h" | |
| #include "pa_table.h" | |
| #include "pa_dictionary.h" | |
| #include "pa_charset.h" | |
| // helpers | |
| /// String::match uses this as replace & global search table columns | |
| const int MAX_MATCH_GROUPS=100; | |
| class String_match_table_template_columns: public ArrayString { | |
| public: | |
| String_match_table_template_columns() { | |
| *this+=new String("prematch"); | |
| *this+=new String("match"); | |
| *this+=new String("postmatch"); | |
| for(int i=0; i<MAX_MATCH_GROUPS; i++) { | |
| char *cname=new(PointerFreeGC) char[3/*strlen("100")*/+1/*terminating 0*/]; | |
| *this+=new String(cname, sprintf(cname, "%d", 1+i)); | |
| } | |
| } | |
| }; | |
| Table string_match_table_template(new String_match_table_template_columns); | |
| // String::ArrayFragment methods | |
| #include "pa_pool.h" | String::ArrayFragment& String::ArrayFragment::append(const ArrayFragment& src, |
| #include "pa_hash.h" | size_t substr_start, size_t substr_end) { |
| if(substr_start==substr_end) | |
| return *this; | |
| size_t fragment_start=0; | |
| size_t fragment_end; | |
| for(Array_iterator<element_type> i(src); ; fragment_start=fragment_end) { | |
| const element_type& fragment=i.next(); | |
| fragment_end=fragment_start+fragment.size; | |
| if(substr_start<fragment_start) // not reached fragments which may include 'substr'? | |
| continue; | |
| // found first fragment including piece of 'substr' | |
| size_t piece_end=min(substr_end, fragment_end); | |
| *this+=Fragment(fragment.lang, piece_end-substr_start); | |
| while(substr_end>fragment_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; | |
| } | |
| } | |
| void *String::operator new(size_t size, Pool *apool) { | break; |
| return apool->malloc(size); | } |
| break2: | |
| return *this; | |
| } | } |
| /* | |
| void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_start, size_t substr_end) { | |
| } | |
| */ | |
| void String::construct(Pool *apool) { | // String methods |
| pool=apool; | |
| head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; | String::String(const char* src, size_t src_size, bool tainted): body(CORD_EMPTY) { |
| append_here=head.rows; | if(src) |
| head.preallocated_link=0; | if(tainted) |
| link_row=&head.rows[curr_chunk_rows]; | APPEND_TAINTED(src, src_size, 0, 0); |
| else | |
| APPEND_CLEAN(src, src_size, 0, 0); | |
| } | |
| String::String(const String& src): body(src.body), fragments(src.fragments) {} | |
| String& String::real_append(STRING_APPEND_PARAMS) { | |
| if(!src) | |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| body=CORD_cat_char_star(body, src, size); | |
| fragments+=Fragment(lang, size); | |
| return *this; | |
| } | } |
| void String::expand() { | static int CORD_batched_iter_fn_generic_hash_code(const char * s, void * client_data) { |
| curr_chunk_rows=curr_chunk_rows*100/CR_GROW_PERCENT; | uint& result=*static_cast<uint*>(client_data); |
| Chunk *chunk=static_cast<Chunk *>( | generic_hash_code(result, s); |
| pool->calloc(sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | return 0; |
| chunk->count=curr_chunk_rows; | }; |
| link_row->link=chunk; | uint String::hash_code() const { |
| append_here=chunk->rows; | uint result=0; |
| link_row=&chunk->rows[curr_chunk_rows]; | CORD_iter5(body, 0, 0, CORD_batched_iter_fn_generic_hash_code, &result); |
| return result; | |
| } | } |
| String::String(String& src) {/* | const String& String::mid(size_t substr_start, size_t substr_end) const { |
| int src_used_rows=src.used_rows(); { | String& result=*new String; |
| curr_chunk_rows=src_total_rows | |
| Chunk *chunk=static_cast<Chunk *>( | |
| pool->calloc(sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | |
| chunk->count=curr_chunk_rows; | |
| link_row->link=chunk; | |
| append_here=chunk->rows; | |
| link_row=&chunk->rows[curr_chunk_rows]; | |
| } | |
| Chunk *chunk=&head; | substr_start=min(substr_start, size()); |
| do { | substr_end=max(substr_start, substr_end); |
| result+=chunk->count; | if(substr_start==substr_end) |
| chunk=row->link; | return result; |
| } while(chunk); | |
| break2: | // first: letters themselves |
| return result;*/ | 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()); | |
| return result; | |
| } | } |
| String& String::operator = (String& src) { | size_t String::pos(CORD substr, |
| return *this; | 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) | |
| 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<ArrayFragment::element_type> i(fragments); i.has_next(); fragment_start=fragment_end) { | |
| const Fragment& fragment=i.next(); | |
| fragment_end=fragment_start+fragment.size; | |
| if(substr_start<fragment_start) // not reached fragments which may include 'result'? | |
| continue; | |
| if(substr_end>fragment_end) // end of substr OUT of current fragment? | |
| break; | |
| return fragment.lang==lang?substr_end:STRING_NOT_FOUND; | |
| } | |
| return STRING_NOT_FOUND; | |
| } | |
| size_t String::pos(const String& substr, | |
| size_t this_offset, Untaint_lang 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)) { | |
| int pos_before; | |
| // while we have 'delim'... | |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | |
| result+=&mid(pos_after, pos_before); | |
| pos_after=pos_before+delim_size; | |
| } | |
| // last piece | |
| if(pos_after<self_size && limit) { | |
| result+=&mid(pos_after, self_size); | |
| pos_after=self_size; | |
| } | |
| } else { // empty delim | |
| result+=this; | |
| pos_after+=self_size; | |
| } | |
| } | } |
| void String::split(ArrayString& result, | |
| size_t& pos_after, | |
| const String& delim, Untaint_lang lang, | |
| int limit) const { | |
| if(delim) { | |
| int pos_before; | |
| // while we have 'delim'... | |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | |
| result+=&mid(pos_after, pos_before); | |
| pos_after=pos_before+delim.size(); | |
| } | |
| // last piece | |
| if(pos_after<size() && limit) { | |
| result+=&mid(pos_after, size()); | |
| pos_after=size(); | |
| } | |
| } else { // empty delim | |
| result+=this; | |
| pos_after+=size(); | |
| } | |
| } | |
| String& String::operator += (char *src) { | static void regex_options(const String& options, int *result, bool& need_pre_post_match){ |
| if(chunk_is_full()) | struct Regex_option { |
| expand(); | const char* keyL; |
| const char* keyU; | |
| append_here->item.ptr=src; | int clear, set; |
| append_here->item.size=strlen(src); | int *result; |
| append_here++; | bool *flag; |
| } regex_option[]={ | |
| {"i", "I", 0, PCRE_CASELESS, result}, // a=A | |
| {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default] | |
| {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored | |
| {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ | |
| {"g", "G", 0, true, result+1}, // many rows | |
| {"'", 0, 0, 0, 0, &need_pre_post_match}, | |
| {0} | |
| }; | |
| result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; | |
| result[1]=0; | |
| if(options) | |
| for(Regex_option *o=regex_option; o->keyL; o++) | |
| if(options.pos(o->keyL)>=0 | |
| || (o->keyU && options.pos(o->keyU)>=0)) { | |
| if(o->flag) | |
| *o->flag=true; | |
| else { // result | |
| *o->result &= ~o->clear; | |
| *o->result |= o->set; | |
| } | |
| } | |
| } | |
| Table* String::match(Charset& source_charset, | |
| const String& regexp, | |
| const String& options, | |
| Row_action row_action, void *info, | |
| bool& just_matched) const { | |
| if(!regexp) | |
| throw Exception(0, | |
| 0, | |
| "regexp is empty"); | |
| const char* pattern=regexp.cstr(); | |
| const char* errptr; | |
| int erroffset; | |
| bool need_pre_post_match=false; | |
| int option_bits[2]; 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, | |
| source_charset.pcre_tables); | |
| if(!code) | |
| throw Exception(0, | |
| ®exp.mid(erroffset, regexp.size()), | |
| "regular expression syntax error - %s", errptr); | |
| int subpatterns=pcre_info(code, 0, 0); | |
| if(subpatterns<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| ®exp, | |
| "pcre_info error (%d)", | |
| subpatterns); | |
| } | |
| return *this; | C subject=cstr(); |
| } | const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3; |
| int ovector[ovecsize]; | |
| // create table | |
| Table* table=new Table(string_match_table_template); | |
| int exec_option_bits=0; | |
| int prestart=0; | |
| int poststart=0; | |
| int postfinish=size(); | |
| while(true) { | |
| int exec_substrings=pcre_exec(code, 0, | |
| subject, subject.size, prestart, | |
| exec_option_bits, ovector, ovecsize); | |
| 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 | |
| else { | |
| just_matched=false; return 0; // not global=no result | |
| } | |
| } | |
| size_t String::size() { | if(exec_substrings<0) { |
| int result=0; | pcre_free(code); |
| throw Exception(0, | |
| ®exp, | |
| "regular expression execute error (%d)", | |
| exec_substrings); | |
| } | |
| Chunk *chunk=&head; | int prefinish=ovector[0]; |
| do { | poststart=ovector[1]; |
| Chunk::Row *row=chunk->rows; | ArrayString* row=new ArrayString; |
| for(int i=0; i<chunk->count; i++) { | if(need_pre_post_match) { |
| if(row==append_here) | *row+=&mid(0, prefinish); // .prematch column value |
| goto break2; | *row+=&mid(prefinish, poststart); // .match |
| *row+=&mid(poststart, postfinish); // .postmatch | |
| } else { | |
| *row+=0; // .prematch column value | |
| *row+=0; // .match | |
| *row+=0; // .postmatch | |
| } | |
| for(int i=1; i<exec_substrings; i++) { | |
| // -1:-1 case handled peacefully by mid() itself | |
| *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value | |
| } | |
| row_action(table, row, prestart, prefinish, poststart, postfinish, info); | |
| result+=row->item.size; | if(!global || prestart==poststart) { // not global | going to hang |
| row++; | pcre_free(code); |
| row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return table; | |
| } | } |
| chunk=row->link; | prestart=poststart; |
| } while(chunk); | |
| break2: | /* |
| return result; | if(option_bits[0] & PCRE_MULTILINE) |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | |
| } | } |
| char *String::c_str() { | String& String::change_case(Charset& source_charset, Change_case_kind kind) { |
| char *result=static_cast<char *>(pool->malloc(size()+1)); | String& result=*new String(); |
| char *copy_here=result; | const unsigned char *tables=source_charset.pcre_tables; |
| Chunk *chunk=&head; | |
| do { | const unsigned char *a; |
| Chunk::Row *row=chunk->rows; | const unsigned char *b; |
| for(int i=0; i<chunk->count; i++) { | switch(kind) { |
| if(row==append_here) | case CC_UPPER: |
| goto break2; | a=tables+lcc_offset; |
| b=tables+fcc_offset; | |
| break; | |
| case CC_LOWER: | |
| a=tables+lcc_offset; | |
| b=0; | |
| break; | |
| default: | |
| throw Exception(0, | |
| this, | |
| "unknown change case kind #%d", | |
| static_cast<int>(kind)); // never | |
| a=b=0; // calm, compiler | |
| break; // never | |
| } | |
| C new_cstr=cstr(); | |
| const char *end=new_cstr+new_cstr.size; | |
| char *dest=new_cstr; | |
| for(const char* current=new_cstr; current<end; current++) { | |
| unsigned char c=a[(unsigned char)*current]; | |
| if(b) | |
| c=b[c]; | |
| *dest++=(char)c; | |
| } | |
| result.body=new_cstr; | |
| result.fragments.append(fragments, 0, fragments.count()); | |
| memcpy(copy_here, row->item.ptr, row->item.size); | |
| copy_here+=row->item.size; | |
| row++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| *copy_here=0; | |
| return result; | return result; |
| } | } |
| int String::used_rows() { | const String& String::replace(const Dictionary& dict) const { |
| int result=0; | String& result=*new String(); |
| C old_cstr=cstr(); | |
| Chunk *chunk=&head; | const char* current=old_cstr; |
| size_t old_pos=0; | |
| char *new_cstr=new(PointerFreeGC) char[(size_t)ceil(old_cstr.size*dict.max_ratio())]; | |
| char *dest=new_cstr; | |
| size_t remaining_size=old_cstr.size; | |
| do { | do { |
| int count=chunk->count; | if(Table::element_type row=dict.first_that_starts(current, remaining_size)) { |
| result+=count; | { // prematch |
| chunk=chunk->rows[count].link; | size_t now_pos=current-old_cstr; |
| } while(chunk); | if(now_pos!=old_pos) { |
| result.fragments.append(fragments, old_pos, now_pos); | |
| old_pos=now_pos; | |
| } | |
| } | |
| const String* a=row->get(0); size_t a_size=a->size(); | |
| // skip 'a' in 'current', in fragment lang index && reduce work size | |
| current+=a_size; old_pos+=a_size; remaining_size-=a_size; | |
| if(row->count()>1) { // are there any b? | |
| const String* b=row->get(1); size_t b_size=b->size(); | |
| if(b_size) { | |
| // write 'b' to 'dest' && skip 'b' in 'dest' | |
| b->store_to(dest, String::UL_AS_IS); dest+=b_size; | |
| // append match fragments | |
| result.fragments.append(b->fragments, 0, b->fragments.count()); | |
| } | |
| } | |
| } else { | |
| // write a char to b && reduce work size | |
| *dest++=*current++; remaining_size--; | |
| } | |
| } while(remaining_size); | |
| result-=link_row-append_here; | { // postmatch |
| size_t now_pos=current-old_cstr; | |
| if(now_pos!=old_pos) // have something at end? | |
| result.fragments.append(fragments, old_pos, now_pos-old_pos); | |
| } | |
| result.body=new_cstr; | |
| return result; | return result; |
| } | } |
| uint String::hash_code() { | double String::as_double() const { |
| uint result=0; | double result; |
| char buf[MAX_STRING]; | |
| if(size()>MAX_STRING-1) | |
| throw Exception("number.format", | |
| this, | |
| "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); | |
| char *eol=store_to(buf, String::UL_AS_IS); *eol=0; | |
| const char* cstr=buf; | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | |
| char *error_pos; | |
| // 0xABC | |
| if(cstr[0]=='0') | |
| if(cstr[1]=='x' || cstr[1]=='X') | |
| result=(double)(unsigned long)strtol(cstr, &error_pos, 0); | |
| else | |
| result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos); | |
| else | |
| result=(double)strtod(cstr, &error_pos); | |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| this, | |
| "invalid number (double)"); | |
| Chunk *chunk=&head; | return result; |
| do { | } |
| Chunk::Row *row=chunk->rows; | int String::as_int() const { |
| for(int i=0; i<chunk->count; i++) { | int result; |
| if(row==append_here) | char buf[MAX_STRING]; |
| goto break2; | if(size()>MAX_STRING-1) |
| throw Exception("number.format", | |
| this, | |
| "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); | |
| char *eol=store_to(buf, String::UL_AS_IS); *eol=0; | |
| const char* cstr=buf; | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | |
| char *error_pos; | |
| // 0xABC | |
| if(cstr[0]=='0') | |
| if(cstr[1]=='x' || cstr[1]=='X') | |
| result=(int)(unsigned long)strtol(cstr, &error_pos, 0); | |
| else | |
| result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0); | |
| else | |
| result=(int)strtol(cstr, &error_pos, 0); | |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| this, | |
| "invalid number (int)"); | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| row++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result; | return result; |
| } | } |
| bool String::operator == (String& src) { | inline void uint2uchars(uint word, uchar *bytes) { |
| return false; | 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<char**>(client_data); | |
| size_t size=strlen(s); | |
| memcpy(cur, s, size); cur+=size; | |
| return 0; | |
| }; | |
| 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); | |
| // 1: prolog | |
| char *cur=result.str+prolog_size; | |
| // 2: fragments.count | |
| size_t fragments_count=fragments.count(); | |
| memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count); | |
| // 3: lang info | |
| for(Array_iterator<ArrayFragment::element_type> 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); | |
| } | |
| // 4: letters | |
| CORD_iter5(body, 0, 0, CORD_batched_iter_fn_append, &cur); | |
| return result; | |
| } | |
| bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) { | |
| if(buf_size<=prolog_size) | |
| return false; | |
| buf_size-=prolog_size; | |
| // 1: prolog | |
| const char *cur=(const char *)buf+prolog_size; | |
| // 2: fragments.count | |
| if(buf_size<sizeof(size_t)) // fragments.count don't fit? | |
| return false; | |
| size_t fragments_count=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); | |
| buf_size-=sizeof(size_t); | |
| // 3: lang info | |
| size_t total_size=0; | |
| for(size_t f=0; f<fragments_count; f++) { | |
| size_t piece_size=sizeof(Untaint_lang)+sizeof(size_t); | |
| if(buf_size<piece_size) // lang+size | |
| return false; | |
| Untaint_lang lang=*reinterpret_cast<const Untaint_lang *>(cur); cur+=sizeof(Untaint_lang); | |
| size_t fragment_size=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); | |
| fragments+=Fragment(lang, fragment_size); | |
| total_size+=fragment_size; | |
| buf_size-=piece_size; | |
| } | |
| // 4: letters | |
| if(buf_size!=total_size) | |
| return false; | |
| body=CORD_cat_char_star(CORD_EMPTY, cur, buf_size); | |
| return true; | |
| } | } |