|
|
| version 1.45, 2001/03/19 16:44:02 | version 1.172.2.21.2.5, 2003/03/19 16:28:09 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser | Parser: string class. @see untasize_t.C. |
| Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) | |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | |
| $Id$ | 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_pool.h" | |
| #include "pa_string.h" | #include "pa_string.h" |
| #include "pa_hash.h" | |
| #include "pa_exception.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)); | |
| } | |
| } | |
| }; | |
| // String | Table string_match_table_template(new String_match_table_template_columns); |
| String::String(Pool& apool, const char *src, bool tainted) : | |
| Pooled(apool) { | |
| last_chunk=&head; | |
| head.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.rows; | |
| head.preallocated_link=0; | |
| link_row=&head.rows[head.count]; | |
| fused_rows=fsize=0; | |
| if(src) | // String::ArrayFragment methods |
| if(tainted) | |
| APPEND_TAINTED(src, 0, 0, 0); | |
| else | |
| APPEND(src, 0, 0, 0); | |
| } | |
| void String::expand() { | String::ArrayFragment& String::ArrayFragment::append(const ArrayFragment& src, |
| int new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; | size_t substr_start, size_t substr_end) { |
| last_chunk=static_cast<Chunk *>( | if(substr_start==substr_end) |
| malloc(sizeof(int)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); | return *this; |
| last_chunk->count=new_chunk_count; | |
| link_row->link=last_chunk; | |
| append_here=last_chunk->rows; | |
| link_row=&last_chunk->rows[last_chunk->count]; | |
| link_row->link=0; | |
| } | |
| String::String(const String& src) : Pooled(src.pool()) { | size_t fragment_start=0; |
| head.count=CR_PREALLOCATED_COUNT; | size_t fragment_end; |
| for(Array_iterator<element_type> i(src); ; fragment_start=fragment_end) { | |
| int src_used_rows=src.fused_rows; | const element_type& fragment=i.next(); |
| if(src_used_rows<=head.count) { | fragment_end=fragment_start+fragment.size; |
| // all new rows fit into preallocated area | |
| int curr_chunk_rows=head.count; | if(substr_start<fragment_start) // not reached fragments which may include 'substr'? |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | continue; |
| append_here=&head.rows[src_used_rows]; | |
| link_row=&head.rows[curr_chunk_rows]; | // found first fragment including piece of 'substr' |
| } else { | size_t piece_end=min(substr_end, fragment_end); |
| // warning: | *this+=Fragment(fragment.lang, piece_end-substr_start); |
| // heavily relies on the fact | |
| // "preallocated area is the same for all strings" | while(substr_end>fragment_end) { // are there more fragments including pieces of 'substr'? |
| // | const element_type& fragment=i.next(); |
| // info: | fragment_end=fragment_start+fragment.size; |
| // allocating only enough mem to fit src string rows | |
| // next append would allocate a new chunk | if(substr_end>fragment_end) // are there still more? |
| // | simple_append(Fragment(fragment.lang, fragment.size)); // appending whole fragment |
| // new rows don't fit into preallocated area: splitting into two chunks | else { // no, it was last |
| // preallocated chunk src to constructing head | simple_append(Fragment(fragment.lang, substr_end-fragment_start)); |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | goto break2; |
| // remaining rows into new_chunk | |
| int curr_chunk_rows=src_used_rows-head.count; | |
| Chunk *new_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | |
| new_chunk->count=curr_chunk_rows; | |
| head.preallocated_link=new_chunk; | |
| append_here=link_row=&new_chunk->rows[new_chunk->count]; | |
| Chunk *old_chunk=src.head.preallocated_link; | |
| Chunk::Row *new_rows=new_chunk->rows; | |
| int rows_left_to_copy=new_chunk->count; | |
| while(true) { | |
| int old_count=old_chunk->count; | |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | |
| if(next_chunk) { | |
| // not last source chunk | |
| // taking it all | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | |
| old_chunk=next_chunk; | |
| } else { | |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | |
| break; | |
| } | } |
| } | } |
| } | |
| link_row->link=0; | |
| fused_rows=src_used_rows; | |
| fsize=src.fsize; | |
| } | |
| String& String::append(const String& src, Untaint_lang lang, bool forced) { | break; |
| int src_used_rows=src.fused_rows; | } |
| int dst_free_rows=link_row-append_here; | break2: |
| if(src_used_rows<=dst_free_rows) { | |
| // all new rows fit into last chunk | |
| memcpy(append_here, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | |
| set_lang(append_here, lang, forced, src_used_rows); | |
| append_here+=src_used_rows; | |
| } else { | |
| // not all new rows fit into last chunk: shrinking it to used part, | |
| int used_rows=last_chunk->count-dst_free_rows; | |
| //int *countp=append_here | |
| link_row=&last_chunk->rows[last_chunk->count=used_rows]; | |
| // allocating only enough mem to fit src string rows | |
| // next append would allocate a new chunk | |
| last_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(int)+sizeof(Chunk::Row)*src_used_rows+sizeof(Chunk *))); | |
| last_chunk->count=src_used_rows; | |
| link_row->link=last_chunk; | |
| append_here=link_row=&last_chunk->rows[src_used_rows]; | |
| const Chunk *old_chunk=&src.head; | |
| Chunk::Row *new_rows=last_chunk->rows; | |
| int rows_left_to_copy=src_used_rows; | |
| while(true) { | |
| int old_count=old_chunk->count; | |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | |
| if(next_chunk) { | |
| // not last source chunk | |
| // taking it all | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | |
| set_lang(new_rows, lang, forced, old_count); | |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | |
| old_chunk=next_chunk; | |
| } else { | |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | |
| set_lang(new_rows, lang, forced, rows_left_to_copy); | |
| break; | |
| } | |
| } | |
| link_row->link=0; | |
| } | |
| fused_rows+=src_used_rows; | |
| fsize+=src.fsize; | |
| return *this; | return *this; |
| } | } |
| void String::set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size) { | /* |
| if(lang==PASS_APPENDED) | void String::ArrayFragment::mid(ArrayFragment& result, size_t substr_start, size_t substr_end) { |
| return; | |
| while(size--) { | |
| Untaint_lang& item_lang=(row++)->item.lang; | |
| if(item_lang==YES || forced) // tainted? need untaint language assignment | |
| item_lang=lang; // assign untaint language | |
| } | |
| } | } |
| */ | |
| /*void String::change_lang(Untaint_lang lang) { | // String methods |
| Chunk *chunk=&head; | |
| do { | |
| Chunk::Row *row=chunk->rows; | |
| for(int i=0; i<chunk->count; i++) { | |
| if(row==append_here) | |
| goto break2; | |
| row->item.lang=lang; | String::String(const char* src, size_t src_size, bool tainted): body(CORD_EMPTY) { |
| row++; | if(src) |
| } | if(tainted) |
| chunk=row->link; | APPEND_TAINTED(src, src_size, 0, 0); |
| } while(chunk); | else |
| break2: | APPEND_CLEAN(src, src_size, 0, 0); |
| return; | |
| } | } |
| */ | |
| String::String(const String& src): body(src.body), fragments(src.fragments) {} | |
| String& String::real_append(STRING_APPEND_PARAMS) { | String& String::real_append(STRING_APPEND_PARAMS) { |
| if(!src) | if(!src) |
| return *this; | return *this; |
| Line 187 String& String::real_append(STRING_APPEN | Line 99 String& String::real_append(STRING_APPEN |
| if(!size) | if(!size) |
| return *this; | return *this; |
| if(chunk_is_full()) | body=CORD_cat_char_star(body, src, size); |
| expand(); | fragments+=Fragment(lang, size); |
| append_here->item.ptr=src; | |
| fsize+=append_here->item.size=size; | |
| append_here->item.lang=tainted?YES:NO; | |
| #ifndef NO_STRING_ORIGIN | |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; fused_rows++; | |
| return *this; | return *this; |
| } | } |
| static int CORD_batched_iter_fn_generic_hash_code(const char * s, void * client_data) { | |
| uint& result=*static_cast<uint*>(client_data); | |
| generic_hash_code(result, s); | |
| return 0; | |
| }; | |
| uint String::hash_code() const { | uint String::hash_code() const { |
| uint result=0; | uint result=0; |
| CORD_iter5(body, 0, 0, CORD_batched_iter_fn_generic_hash_code, &result); | |
| return result; | |
| } | |
| const Chunk *chunk=&head; | const String& String::mid(size_t substr_start, size_t substr_end) const { |
| do { | String& result=*new String; |
| const Chunk::Row *row=chunk->rows; | |
| for(int i=0; i<chunk->count; i++) { | |
| if(row==append_here) | |
| goto break2; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | substr_start=min(substr_start, size()); |
| row++; | substr_end=max(substr_start, substr_end); |
| } | if(substr_start==substr_end) |
| chunk=row->link; | return result; |
| } while(chunk); | |
| break2: | // 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()); | |
| return result; | return result; |
| } | } |
| int String::cmp(const String& src) const { | size_t String::pos(CORD substr, |
| const Chunk *a_chunk=&head; | size_t this_offset, Untaint_lang lang) const { |
| const Chunk *b_chunk=&src.head; | // first: letters themselves |
| const Chunk::Row *a_row=a_chunk->rows; | size_t substr_start=CORD_str(body, this_offset, substr); |
| const Chunk::Row *b_row=b_chunk->rows; | if(substr_start==CORD_NOT_FOUND) |
| int a_offset=0; | return STRING_NOT_FOUND; |
| int b_offset=0; | |
| Chunk::Row *a_end=append_here; | // next: check the lang when specified |
| Chunk::Row *b_end=src.append_here; | |
| int a_countdown=a_chunk->count; | if(lang==UL_UNSPECIFIED) // ignore lang? |
| int b_countdown=b_chunk->count; | return substr_start; |
| bool a_break=false; | |
| bool b_break=false; | // substr must be in one fragment, and fragments' lang must = lang |
| int result; | size_t substr_end=substr_start+CORD_len(substr); |
| while(true) { | size_t fragment_start=0; |
| a_break=a_row==a_end; | size_t fragment_end; |
| b_break=b_row==b_end; | for(Array_iterator<ArrayFragment::element_type> i(fragments); i.has_next(); fragment_start=fragment_end) { |
| if(a_break || b_break) | 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; | break; |
| int size_diff= | return fragment.lang==lang?substr_end:STRING_NOT_FOUND; |
| (a_row->item.size-a_offset)- | } |
| (b_row->item.size-b_offset); | |
| return STRING_NOT_FOUND; | |
| if(size_diff==0) { // a has same size as b | } |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset); | |
| if(result) | size_t String::pos(const String& substr, |
| return result; | size_t this_offset, Untaint_lang lang) const { |
| a_row++; a_countdown--; a_offset=0; | return pos(substr.body, this_offset, lang); |
| b_row++; b_countdown--; b_offset=0; | } |
| } else if (size_diff>0) { // a longer | |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset); | void String::split(ArrayString& result, |
| if(result) | size_t& pos_after, |
| return result; | const char* delim, |
| a_offset+=b_row->item.size-b_offset; | Untaint_lang lang, int limit) { |
| b_row++; b_countdown--; b_offset=0; | size_t self_size=size(); |
| } else { // b longer | if(size_t delim_size=strlen(delim)) { |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset); | int pos_before; |
| if(result) | // while we have 'delim'... |
| return result; | for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { |
| b_offset+=a_row->item.size-a_offset; | result+=&mid(pos_after, pos_before); |
| a_row++; a_countdown--; a_offset=0; | pos_after=pos_before+delim_size; |
| } | |
| if(!a_countdown) { | |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | |
| if(!b_countdown) { | |
| b_chunk=b_row->link; | |
| b_row=b_chunk->rows; | |
| b_countdown=b_chunk->count; | |
| } | } |
| // 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; | |
| } | } |
| if(a_break==b_break) // ended simultaneously | |
| result=0; | |
| else if(a_break) // first bytes equal, but a ended before b | |
| result=-1; | |
| else | |
| result=+1; | |
| return result; | |
| } | } |
| bool String::operator == (const char* b_ptr) const { | void String::split(ArrayString& result, |
| size_t b_size=b_ptr?strlen(b_ptr):0; | size_t& pos_after, |
| if(size() != b_size) | const String& delim, Untaint_lang lang, |
| return false; | 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(); | |
| } | |
| } | |
| const Chunk *a_chunk=&head; | static void regex_options(const String& options, int *result, bool& need_pre_post_match){ |
| const Chunk::Row *a_row=a_chunk->rows; | struct Regex_option { |
| int a_offset=0; | const char* keyL; |
| int b_offset=0; | const char* keyU; |
| Chunk::Row *a_end=append_here; | int clear, set; |
| int a_countdown=a_chunk->count; | int *result; |
| bool a_break=false; | bool *flag; |
| bool b_break=false; | } 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); | |
| } | |
| 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) { | while(true) { |
| int size_diff= | int exec_substrings=pcre_exec(code, 0, |
| (a_row->item.size-a_offset)- | subject, subject.size, prestart, |
| (b_size-b_offset); | exec_option_bits, ovector, ovecsize); |
| if(size_diff==0) { // a has same size as b | if(exec_substrings==PCRE_ERROR_NOMATCH) { |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) | pcre_free(code); |
| return false; | row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); |
| a_row++; a_countdown--; a_offset=0; | if(global || subpatterns) |
| b_break=true; | return table; // global or with subpatterns=true+result |
| } else if (size_diff>0) { // a longer | else { |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, b_size-b_offset)!=0) | just_matched=false; return 0; // not global=no result |
| return false; | } |
| a_offset+=b_size-b_offset; | |
| b_break=true; | |
| } else { // b longer | |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) | |
| return false; | |
| b_offset+=a_row->item.size-a_offset; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | } |
| a_break=a_row==a_end; | if(exec_substrings<0) { |
| if(a_break || b_break) | pcre_free(code); |
| break; | throw Exception(0, |
| ®exp, | |
| "regular expression execute error (%d)", | |
| exec_substrings); | |
| } | |
| int prefinish=ovector[0]; | |
| poststart=ovector[1]; | |
| ArrayString* row=new ArrayString; | |
| if(need_pre_post_match) { | |
| *row+=&mid(0, prefinish); // .prematch column value | |
| *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); | |
| if(!a_countdown) { | if(!global || prestart==poststart) { // not global | going to hang |
| a_chunk=a_row->link; | pcre_free(code); |
| a_row=a_chunk->rows; | row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); |
| a_countdown=a_chunk->count; | return table; |
| } | } |
| prestart=poststart; | |
| /* | |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | |
| } | |
| String& String::change_case(Charset& source_charset, Change_case_kind kind) { | |
| String& result=*new String(); | |
| const unsigned char *tables=source_charset.pcre_tables; | |
| const unsigned char *a; | |
| const unsigned char *b; | |
| switch(kind) { | |
| case CC_UPPER: | |
| 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()); | |
| return result; | |
| } | |
| const String& String::replace(const Dictionary& dict) const { | |
| String& result=*new String(); | |
| C old_cstr=cstr(); | |
| 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 { | |
| if(Table::element_type row=dict.first_that_starts(current, remaining_size)) { | |
| { // prematch | |
| size_t now_pos=current-old_cstr; | |
| 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); | |
| { // 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; | |
| } | |
| double String::as_double() const { | |
| 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)"); | |
| return result; | |
| } | |
| int String::as_int() const { | |
| int 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=(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)"); | |
| 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<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); | |
| } | } |
| return a_break==b_break; | |
| // 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; | |
| } | } |