|
|
| version 1.172.2.21.2.26, 2003/03/24 16:08:35 | version 1.173, 2003/04/11 15:00:05 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class. @see untalength_t.C. | Parser: string class. @see untasize_t.C. |
| Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| */ | */ |
| Line 9 static const char* IDENT_STRING_C="$Date | Line 9 static const char* IDENT_STRING_C="$Date |
| #include "pcre.h" | #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_common.h" | |
| #include "pa_array.h" | |
| #include "pa_globals.h" | |
| #include "pa_table.h" | #include "pa_table.h" |
| #include "pa_dictionary.h" | #include "pa_dictionary.h" |
| #include "pa_charset.h" | #include "pa_charset.h" |
| // helpers | #define DEBUG_STRING_APPENDS_VS_EXPANDS |
| /// String::match uses this as replace & global search table columns | |
| const int MAX_MATCH_GROUPS=100; | #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS |
| ulong string_piece_appends=0; | |
| #endif | |
| class String_match_table_template_columns: public ArrayString { | String& String::OnPool(Pool& apool, const char *local_src, size_t src_size, bool tainted) { |
| public: | if(local_src && *local_src) { |
| String_match_table_template_columns() { | if(src_size==0) |
| *this+=new String("prematch"); | src_size=strlen(local_src); |
| *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 | |
| void String::ArrayFragment::append_positions(const ArrayFragment& src, | |
| size_t substr_begin, size_t substr_end) { | |
| if(substr_begin==substr_end) | |
| return; | |
| size_t fragment_begin=0; | |
| size_t fragment_end; | |
| for(Array_iterator<element_type> i(src); ; fragment_begin=fragment_end) { | |
| const element_type& fragment=i.next(); | |
| fragment_end=fragment_begin+fragment.length; | |
| if(substr_begin<fragment_begin) // not reached fragments which may include 'substr'? | |
| continue; | |
| if(substr_begin>fragment_end) // 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_begin); | |
| while(substr_end>fragment_end) { // are there more fragments including pieces of 'substr'? | char *pooled_src=(char *)apool.malloc(src_size); |
| fragment_begin=fragment_end; | memcpy(pooled_src, local_src, src_size); |
| return *new(apool) String(apool, pooled_src, src_size, tainted); | |
| const element_type& fragment=i.next(); | } else |
| fragment_end=fragment_begin+fragment.length; | return *new(apool) String(apool); |
| } | |
| String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : | |
| Pooled(apool) { | |
| last_chunk=&head.chunk; | |
| head.chunk.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.chunk.rows; | |
| if(src) | |
| if(tainted) | |
| APPEND_TAINTED(src, src_size, 0, 0); | |
| else | |
| APPEND_CLEAN(src, src_size, 0, 0); | |
| } | |
| if(substr_end>fragment_end) // are there still more? | String::String(const String& src) : |
| append(Fragment(fragment.lang, fragment.length)); // appending whole fragment | Pooled(src.pool()) { |
| else { // no, it was last | last_chunk=&head.chunk; |
| append(Fragment(fragment.lang, substr_end-fragment_begin)); | head.chunk.count=CR_PREALLOCATED_COUNT; |
| return; | append_here=head.chunk.rows; |
| } | |
| } | |
| break; | append(src, UL_PASS_APPENDED); |
| } | |
| } | } |
| // StringSimple methods | size_t String::size() const { |
| size_t result=0; | |
| STRING_FOREACH_ROW( | |
| result+=row->item.size; | |
| ); | |
| return result; | |
| } | |
| static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) { | /// @todo not very optimal |
| uint& result=*static_cast<uint*>(client_data); | uint String::used_rows() const { |
| 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<uint*>(client_data); | |
| generic_hash_code(result, s); | |
| return 0; | |
| }; | |
| uint StringBody::hash_code() const { | |
| uint result=0; | uint result=0; |
| CORD_iter5(body, 0, | STRING_FOREACH_ROW( |
| CORD_batched_iter_fn_generic_hash_code, | result++; |
| CORD_batched_iter_fn_generic_hash_code, &result); | ); |
| return result; | return result; |
| } | } |
| void String::expand() { | |
| // String methods | uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; |
| if(new_chunk_count>max_integral(Chunk::count_type)) | |
| String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { | new_chunk_count=max_integral(Chunk::count_type); |
| append(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); | |
| Chunk *new_chunk=static_cast<Chunk *>(malloc( | |
| sizeof(Chunk)// count+interpadding(?)+rows[CR_PREALLOCATED_COUNT]+tailpadding(??) | |
| -sizeof(Chunk::rows_type) // PREALLOCATED rows | |
| +sizeof(Chunk::Row)*new_chunk_count // neaded rows | |
| +sizeof(Chunk *) // link size | |
| , 10)); | |
| new_chunk->rows[new_chunk->count=new_chunk_count].link=0; | |
| last_chunk->rows[last_chunk->count].link=new_chunk; | |
| last_chunk=new_chunk; | |
| append_here=last_chunk->rows; | |
| } | } |
| String::String(const String& src): body(src.body), fragments(src.fragments) {} | String& String::real_append(STRING_APPEND_PARAMS) { |
| if(!last_chunk) // growth stopped [we're appended as string to somebody] | |
| throw Exception(0, | |
| this, | |
| "string growth stopped (append cstr)"); | |
| String& String::append(const char* str, size_t length, Language lang) { | if(!src) |
| if(!str) | |
| return *this; | return *this; |
| size_t use_length=length?length:strlen(str); | if(!size) |
| if(!use_length) | size=strlen(src); |
| if(!size) | |
| return *this; | return *this; |
| body.append(str, use_length); | #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS |
| fragments+=Fragment(lang, use_length); | string_piece_appends++; |
| #endif | |
| // manually unrolled to avoid extra check | |
| while(size>max_integral(Chunk::Row::item_size_type)) { | |
| if(chunk_is_full()) | |
| expand(); | |
| append_here->item.ptr=src; | |
| append_here->item.size=max_integral(Chunk::Row::item_size_type); | |
| append_here->item.lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; | |
| return *this; | src+=max_integral(Chunk::Row::item_size_type); |
| } | size-=max_integral(Chunk::Row::item_size_type); |
| } | |
| /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" | if(chunk_is_full()) |
| String& String::mid(size_t substr_begin, size_t substr_end) const { | expand(); |
| String& result=*new String; | |
| size_t self_length=length(); | |
| substr_begin=max(min(substr_begin, self_length), (size_t)0); | |
| substr_end=min(max(substr_end, substr_begin), self_length); | |
| if(substr_begin==substr_end) | |
| return result; | |
| // first: letters themselves | append_here->item.ptr=src; |
| result.body=body.mid(substr_begin, substr_end-substr_begin); | append_here->item.size=size; |
| append_here->item.lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; | |
| // next: their langs | return *this; |
| result.fragments.append_positions(fragments, substr_begin, substr_end); | } |
| // SAPI::log("piece of '%s' from %d to %d is '%s'", | uint String::hash_code() const { |
| //cstr(), substr_begin, substr_end, result.cstr()); | uint result=0; |
| STRING_FOREACH_ROW( | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| ); | |
| return result; | return result; |
| } | } |
| size_t String::pos(const StringBody substr, | /// @todo move 'lang' skipping to pos |
| size_t this_offset, Language lang) const { | int String::cmp(int& partial, const String& src, |
| // first: letters themselves | size_t this_offset, Untaint_lang lang) const { |
| size_t substr_begin=body.pos(substr, this_offset); | partial=-1; |
| if(substr_begin==CORD_NOT_FOUND) | size_t a_size=size(); |
| return STRING_NOT_FOUND; | this_offset=min(this_offset, a_size-1); |
| // next: check the lang when specified | const Chunk *a_chunk=&head.chunk; |
| const Chunk *b_chunk=&src.head.chunk; | |
| if(lang==L_UNSPECIFIED) // ignore lang? | const Chunk::Row *a_row=a_chunk->rows; |
| return substr_begin; | const Chunk::Row *b_row=b_chunk->rows; |
| size_t a_offset=this_offset; | |
| // substr must be in one fragment, and fragments' lang must = lang | size_t b_offset=0; |
| size_t substr_end=substr_begin+substr.length(); | Chunk::Row *a_end=append_here; |
| size_t fragment_begin=0; | Chunk::Row *b_end=src.append_here; |
| size_t fragment_end; | uint a_countdown=a_chunk->count; |
| for(Array_iterator<ArrayFragment::element_type> i(fragments); i.has_next(); fragment_begin=fragment_end) { | uint b_countdown=b_chunk->count; |
| const Fragment& fragment=i.next(); | int result; |
| fragment_end=fragment_begin+fragment.length; | size_t pos=0; |
| if(substr_begin<fragment_begin) // not reached fragments which may include 'result'? | bool a_break=a_size==0; |
| continue; | bool b_break=src.is_empty(); |
| if(substr_begin>=fragment_end) // begin of substr OUT of current fragment? | if(!(a_break || b_break)) while(true) { |
| continue; | if(pos+a_row->item.size > this_offset) { |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) | |
| if(substr_end>fragment_end) // end of substr OUT of current fragment? | return -1; // wrong lang -- bail out |
| throw Exception(0, // (*) see below | |
| this, | int size_diff= |
| "searching for '%s' starting from %ud problem: found begin in one fragment, but end in another", | (a_row->item.size-a_offset)- |
| substr.cstr(), this_offset); | (b_row->item.size-b_offset); |
| 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) | |
| return result; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; a_offset=0; | |
| 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); | |
| if(result) | |
| return result; | |
| a_offset+=b_row->item.size-b_offset; | |
| b_row++; b_countdown--; b_offset=0; | |
| } else { // b longer | |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, | |
| a_row->item.size-a_offset); | |
| if(result) | |
| return result; | |
| b_offset+=a_row->item.size-a_offset; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | |
| if(b_break=b_row==b_end) { | |
| a_break=a_row==a_end; | |
| break; | |
| } | |
| if(!b_countdown) { | |
| b_chunk=b_row->link; | |
| b_row=b_chunk->rows; | |
| b_countdown=b_chunk->count; | |
| } | |
| } else { | |
| a_offset-=a_row->item.size; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; | |
| } | |
| if(fragment.lang<=lang) | if(a_break=a_row==a_end) { |
| return substr_begin; | b_break=b_row==b_end; |
| else { // bad lang... | break; |
| /// WARNING: this possibly skips assert (*), but it's fast | } |
| substr_begin=body.pos(substr, fragment_end/*...search AFTER for more*/); | if(!a_countdown) { |
| if(substr_begin==CORD_NOT_FOUND) | a_chunk=a_row->link; |
| return STRING_NOT_FOUND; | a_row=a_chunk->rows; |
| a_countdown=a_chunk->count; | |
| } | |
| } | |
| if(a_break==b_break) { // ended simultaneously | |
| partial=0; return 0; | |
| } else if(a_break) { // first bytes equal, but a ended before b | |
| partial=1; return -1; | |
| } else { | |
| partial=2; return +1; | |
| } | |
| } | |
| size_t substr_end=substr_begin+substr.length(); | /// @todo move 'lang' skipping to pos |
| // and continuing with next fragment | int String::cmp(int& partial, const char* b_ptr, size_t src_size, |
| size_t this_offset, Untaint_lang lang) const { | |
| partial=-1; | |
| size_t a_size=size(); | |
| size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; | |
| this_offset=min(this_offset, a_size-1); | |
| const Chunk *a_chunk=&head.chunk; | |
| const Chunk::Row *a_row=a_chunk->rows; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| Chunk::Row *a_end=append_here; | |
| uint a_countdown=a_chunk->count; | |
| size_t pos=0; | |
| bool a_break=a_size==0; | |
| bool b_break=b_size==0; | |
| if(!(a_break || b_break)) while(true) { | |
| if(pos+a_row->item.size > this_offset) { | |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) | |
| return -1; // wrong lang -- bail out | |
| int size_diff= | |
| (a_row->item.size-a_offset)- | |
| (b_size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, | |
| a_row->item.size-a_offset)!=0) | |
| return result; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; a_offset=0; | |
| b_break=true; | |
| } else if (size_diff>0) { // a longer | |
| if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, | |
| b_size-b_offset)!=0) | |
| return result; | |
| a_offset+=b_size-b_offset; | |
| b_break=true; | |
| } else { // b longer | |
| if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, | |
| a_row->item.size-a_offset)!=0) | |
| return result; | |
| b_offset+=a_row->item.size-a_offset; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | |
| } else { | |
| a_offset-=a_row->item.size; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; | |
| } | } |
| } | |
| return STRING_NOT_FOUND; | a_break=a_row==a_end; |
| if(a_break || b_break) | |
| break; | |
| if(!a_countdown) { | |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | |
| } | |
| if(a_break==b_break) { // ended simultaneously | |
| partial=0; return 0; | |
| } else if(a_break) { // first bytes equal, but a ended before b | |
| partial=1; return -1; | |
| } else { | |
| partial=2; return +1; | |
| } | |
| } | |
| #ifndef NO_STRING_ORIGIN | |
| const Origin& String::origin() const { | |
| if(is_empty()) { | |
| static const Origin empty_origin={"empty string"}; | |
| return empty_origin; | |
| } | |
| // determining origin by first piece or last appended piece | |
| // because any of them can be constant=without origin: | |
| // ex: ^load[/file] "document_root" + "/file" | |
| // when last peice is constant, | |
| // ex: parser_root_auto_path{dynamic} / auto.p{const} | |
| // using first piece | |
| Origin& first_origin=head.chunk.rows[0].item.origin; | |
| return first_origin.file ? first_origin : append_here[-1].item.origin; | |
| } | |
| #endif | |
| String& String::mid(size_t start, size_t finish) const { | |
| String& result=*NEW String(pool()); | |
| start=min(start, size()); | |
| finish=max(start, finish); | |
| if(start==finish) | |
| return result; | |
| size_t pos=0; | |
| STRING_FOREACH_ROW( | |
| size_t item_finish=pos+row->item.size; | |
| if(item_finish > start) { // started now or already? | |
| bool started=result.is_empty(); // started now? | |
| bool finished=finish <= item_finish; // finished now? | |
| size_t offset=started?start-pos:0; | |
| size_t size=finished?finish-pos:row->item.size; | |
| result.APPEND( | |
| row->item.ptr+offset, size-offset, | |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| if(finished) | |
| goto break2; | |
| } | |
| pos+=row->item.size; | |
| ); | |
| break2: | |
| // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", | |
| //cstr(), start, finish, result.cstr()); | |
| return result; | |
| } | |
| int String::pos(const String& substr, | |
| int result, Untaint_lang lang) const { | |
| size_t self_size=size(); | |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, result, lang); | |
| if( | |
| partial==0 || // full match | |
| partial==2) // 'substr' starts 'this'+'result' | |
| return result; | |
| } | |
| return -1; | |
| } | } |
| size_t String::pos(const String& substr, | int String::pos(const char *substr, size_t substr_size, |
| size_t this_offset, Language lang) const { | int result, Untaint_lang lang) const { |
| return pos(substr.body, this_offset, lang); | size_t self_size=size(); |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, substr_size, result, lang); | |
| if( | |
| partial==0 || // full match | |
| partial==2) // 'substr' starts 'this'+'result' | |
| return result; | |
| } | |
| return -1; | |
| } | } |
| void String::split(ArrayString& result, | void String::split(Array& result, |
| size_t& pos_after, | size_t* pos_after_ref, |
| const char* delim, | const char *delim, size_t delim_size, |
| Language lang, int limit) const { | Untaint_lang lang, int limit) const { |
| size_t self_length=length(); | size_t self_size=size(); |
| if(size_t delim_length=strlen(delim)) { | if(delim_size) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | |
| int pos_before; | int pos_before; |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { |
| result+=&mid(pos_after, pos_before); | result+=&mid(pos_after, pos_before); |
| pos_after=pos_before+delim_length; | pos_after=pos_before+delim_size; |
| } | } |
| // last piece | // last piece |
| if(pos_after<self_length && limit) { | if(pos_after<self_size && limit) { |
| result+=&mid(pos_after, self_length); | result+=&mid(pos_after, self_size); |
| pos_after=self_length; | pos_after=self_size; |
| } | } |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | } else { // empty delim |
| result+=this; | result+=this; |
| pos_after+=self_length; | if(pos_after_ref) |
| *pos_after_ref+=self_size; | |
| } | } |
| } | } |
| void String::split(ArrayString& result, | void String::split(Array& result, |
| size_t& pos_after, | size_t* pos_after_ref, |
| const String& delim, Language lang, | const String& delim, Untaint_lang lang, |
| int limit) const { | int limit) const { |
| if(!delim.is_empty()) { | if(!delim.is_empty()) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | |
| int pos_before; | int pos_before; |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { |
| result+=&mid(pos_after, pos_before); | result+=&mid(pos_after, pos_before); |
| pos_after=pos_before+delim.length(); | pos_after=pos_before+delim.size(); |
| } | } |
| // last piece | // last piece |
| if(pos_after<length() && limit) { | if(pos_after<size() && limit) { |
| result+=&mid(pos_after, length()); | result+=&mid(pos_after, size()); |
| pos_after=length(); | pos_after=size(); |
| } | } |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | } else { // empty delim |
| result+=this; | result+=this; |
| pos_after+=length(); | if(pos_after_ref) |
| *pos_after_ref+=size(); | |
| } | } |
| } | } |
| static void regex_options(const String* options, int *result, bool& need_pre_post_match){ | static void regex_options(const String *options, int *result, bool& need_pre_post_match){ |
| struct Regex_option { | struct Regex_option { |
| const char* keyL; | const char *keyL; |
| const char* keyU; | const char *keyU; |
| int clear, set; | int clear, set; |
| int *result; | int *result; |
| bool *flag; | bool *flag; |
| Line 256 static void regex_options(const String* | Line 462 static void regex_options(const String* |
| result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; | result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; |
| result[1]=0; | result[1]=0; |
| if(options && !options->is_empty()) | if(options) |
| for(Regex_option *o=regex_option; o->keyL; o++) | for(Regex_option *o=regex_option; o->keyL; o++) |
| if(options->pos(o->keyL)>=0 | if(options->pos(o->keyL)>=0 |
| || (o->keyU && options->pos(o->keyU)>=0)) { | || (o->keyU && options->pos(o->keyU)>=0)) { |
| Line 269 static void regex_options(const String* | Line 475 static void regex_options(const String* |
| } | } |
| } | } |
| Table* String::match(Charset& source_charset, | /// @todo make replacement Table stacked |
| const String& regexp, | bool String::match( |
| const String* options, | const String *aorigin, |
| Row_action row_action, void *info, | const String& regexp, |
| bool& just_matched) const { | const String *options, |
| Table **table, | |
| Row_action row_action, void *info, | |
| bool *was_global) const { | |
| if(regexp.is_empty()) | if(regexp.is_empty()) |
| throw Exception(0, | throw Exception(0, |
| 0, | aorigin, |
| "regexp is empty"); | "regexp is empty"); |
| const char* pattern=regexp.cstr(); | const char *pattern=regexp.cstr(); |
| const char* errptr; | const char *errptr; |
| int erroffset; | int erroffset; |
| bool need_pre_post_match=false; | bool need_pre_post_match=false; |
| int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); | int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); |
| bool global=option_bits[1]!=0; | if(was_global) |
| *was_global=option_bits[1]!=0; | |
| pcre *code=pcre_compile(pattern, option_bits[0], | pcre *code=pcre_compile(pattern, option_bits[0], |
| &errptr, &erroffset, | &errptr, &erroffset, |
| source_charset.pcre_tables); | pool().get_source_charset().pcre_tables); |
| if(!code) | if(!code) |
| throw Exception(0, | throw Exception(0, |
| ®exp.mid(erroffset, regexp.length()), | ®exp.mid(erroffset, regexp.size()), |
| "regular expression syntax error - %s", errptr); | "regular expression syntax error - %s", errptr); |
| int subpatterns=pcre_info(code, 0, 0); | int info_substrings=pcre_info(code, 0, 0); |
| if(subpatterns<0) { | if(info_substrings<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, | throw Exception(0, |
| ®exp, | aorigin, |
| "pcre_info error (%d)", | "pcre_info error (%d)", |
| subpatterns); | info_substrings); |
| } | } |
| const char* subject=cstr(); | const char *subject=cstr(); |
| size_t subject_length=strlen(subject); | int length=strlen(subject); |
| const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3; | const int ovecsize=(1/*match*/+MAX_STRING_MATCH_TABLE_COLUMNS)*3; |
| int ovector[oveclength]; | int ovector[ovecsize]; |
| // create table | // create table |
| Table& table=*new Table(string_match_table_template); | Table::Action_options table_options; |
| *table=NEW Table(pool(), *string_match_table_template, table_options); | |
| int exec_option_bits=0; | int exec_option_bits=0; |
| int prestart=0; | int prestart=0; |
| int poststart=0; | int poststart=0; |
| int postfinish=length(); | int postfinish=size(); |
| while(true) { | while(true) { |
| int exec_substrings=pcre_exec(code, 0, | int exec_substrings=pcre_exec(code, 0, |
| subject, subject_length, prestart, | subject, length, prestart, |
| exec_option_bits, ovector, oveclength); | exec_option_bits, ovector, ovecsize); |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | if(exec_substrings==PCRE_ERROR_NOMATCH) { |
| pcre_free(code); | pcre_free(code); |
| row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); | row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); |
| if(global || subpatterns) | return option_bits[1]!=0; // global=true+table, not global=false |
| return &table; // global or with subpatterns=true+result | |
| else { | |
| just_matched=false; return 0; // not global=no result | |
| } | |
| } | } |
| if(exec_substrings<0) { | if(exec_substrings<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, | throw Exception(0, |
| ®exp, | aorigin, |
| "regular expression execute error (%d)", | "regular expression execute error (%d)", |
| exec_substrings); | exec_substrings); |
| } | } |
| int prefinish=ovector[0]; | int prefinish=ovector[0]; |
| poststart=ovector[1]; | poststart=ovector[1]; |
| ArrayString* row=new ArrayString; | Array& row=*NEW Array(pool()); |
| if(need_pre_post_match) { | row+=need_pre_post_match?&mid(0, prefinish):0; // .prematch column value |
| *row+=&mid(0, prefinish); // .prematch column value | row+=need_pre_post_match?&mid(prefinish, poststart):0; // .match |
| *row+=&mid(prefinish, poststart); // .match | row+=need_pre_post_match?&mid(poststart, postfinish):0; // .postmatch |
| *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++) { | for(int i=1; i<exec_substrings; i++) { |
| // -1:-1 case handled peacefully by mid() itself | // -1:-1 case handled peacefully by mid() itself |
| *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value | row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value |
| } | } |
| row_action(table, row, prestart, prefinish, poststart, postfinish, info); | row_action(**table, &row, prestart, prefinish, poststart, postfinish, info); |
| if(!global || prestart==poststart) { // not global | going to hang | if(!option_bits[1] || prestart==poststart) { // not global | going to hang |
| pcre_free(code); | pcre_free(code); |
| row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); |
| return &table; | return true; |
| } | } |
| prestart=poststart; | prestart=poststart; |
| Line 372 Table* String::match(Charset& source_cha | Line 574 Table* String::match(Charset& source_cha |
| } | } |
| } | } |
| String& String::change_case(Charset& source_charset, Change_case_kind kind) const { | String& String::change_case(Pool& pool, |
| String& result=*new String(); | Change_case_kind kind) const { |
| const unsigned char *tables=pool.get_source_charset().pcre_tables; | |
| const unsigned char *tables=source_charset.pcre_tables; | String& result=*new(pool) String(pool); |
| const unsigned char *a; | const unsigned char *a; |
| const unsigned char *b; | const unsigned char *b; |
| Line 397 String& String::change_case(Charset& sou | Line 599 String& String::change_case(Charset& sou |
| break; // never | break; // never |
| } | } |
| char* new_cstr=cstrm(); | STRING_FOREACH_ROW( |
| char *dest=new_cstr; | char *new_cstr=(char *)pool.malloc(row->item.size, 12); |
| unsigned char index; | char *dest=new_cstr; |
| for(const char* current=new_cstr; index=(unsigned char)*current; current++) { | const char *src=row->item.ptr; |
| unsigned char c=a[index]; | for(int size=row->item.size; size--; src++) { |
| if(b) | unsigned char c=a[(unsigned char)*src]; |
| c=b[c]; | if(b) |
| c=b[c]; | |
| *dest++=(char)c; | *dest++=(char)c; |
| } | } |
| result.body=new_cstr; | |
| result.fragments.append(fragments); | result.APPEND(new_cstr, row->item.size, |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| ); | |
| return result; | return result; |
| } | } |
| const String& String::replace(const Dictionary& dict) const { | /// @test if in some piece were found no dict words, append it, not it's duplicate |
| String& result=*new String(); | String& String::replace(Pool& pool, Dictionary& dict) const { |
| const char* old_cstr=cstr(); | char *lcstr=cstr(); |
| const char* prematch_begin=old_cstr; | const char *current=lcstr; |
| for(const char* current=old_cstr; *current; ) { | String& result=*new(pool) String(pool); |
| if(Table::element_type row=dict.first_that_begins(current)) { | STRING_FOREACH_ROW( |
| // prematch | IFNDEF_NO_STRING_ORIGIN( |
| if(size_t prematch_length=current-prematch_begin) { | const char *joined_origin_file=row->item.origin.file; |
| result.body.append(prematch_begin, prematch_length); | const size_t joined_origin_line=row->item.origin.line; |
| result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); | ); |
| uchar joined_lang=row->item.lang; | |
| const char *joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_ROW(*this, | |
| if(row->item.lang==joined_lang) | |
| joined_size+=row->item.size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_FOREACH_ROW | |
| --row; ++countdown; | |
| char *new_cstr=(char *)pool.malloc((size_t)ceil(joined_size*dict.max_ratio()), 14); | |
| char *dest=new_cstr; | |
| while(joined_size) { | |
| // there is a row where first column starts 'joined_ptr' | |
| if(Table::Item *item=dict.first_that_starts(joined_ptr, joined_size)) { | |
| // get a=>b values | |
| const String& a=*static_cast<Array *>(item)->get_string(0); | |
| const String& b=*static_cast<Array *>(item)->get_string(1); | |
| // skip 'a' in 'joined_ptr' && reduce work size | |
| joined_ptr+=a.size(); joined_size-=a.size(); | |
| // write 'b' to 'dest' && skip 'b' in 'dest' | |
| b.store_to(dest); dest+=b.size(); | |
| } else { | |
| // write a char to b && reduce work size | |
| *dest++=*joined_ptr++; joined_size--; | |
| } | } |
| } | |
| // match | result.APPEND(new_cstr, dest-new_cstr, joined_lang, |
| joined_origin_file, joined_origin_line); | |
| ); | |
| const String* a=row->get(0); | return result; |
| // 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_length=b->length()) { | |
| result.body.append(b->cstr_to_string_body()); | |
| result.fragments.append(b->fragments); | |
| } | |
| } | |
| } else // simply advance | |
| current++; | |
| } | |
| // postmatch | String& String::join_chains(Pool& pool, char** acstr) const { |
| if(size_t postmatch_length=current-prematch_begin) { | char *lcstr=cstr(); |
| result.body.append(prematch_begin, postmatch_length); | const char *current=lcstr; |
| result.fragments.append_positions(fragments, prematch_begin-old_cstr, current-old_cstr); | |
| } | String& result=*new(pool) String(pool); |
| STRING_FOREACH_ROW( | |
| IFNDEF_NO_STRING_ORIGIN( | |
| const char *joined_origin_file=row->item.origin.file; | |
| const size_t joined_origin_line=row->item.origin.line; | |
| ); | |
| uchar joined_lang=row->item.lang; | |
| const char *joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_ROW(*this, | |
| if(row->item.lang==joined_lang) | |
| joined_size+=row->item.size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_FOREACH_ROW | |
| --row; ++countdown; | |
| result.APPEND(joined_ptr, joined_size, joined_lang, | |
| joined_origin_file, joined_origin_line); | |
| ); | |
| if(acstr) | |
| *acstr=lcstr; | |
| return result; | return result; |
| } | } |
| double String::as_double() const { | double String::as_double() const { |
| double result; | double result; |
| const char *str=cstr(); | const char *cstr; |
| char buf[MAX_NUMBER]; | |
| while(*str && isspace(*str)) | if(head.chunk.rows+1==append_here) { |
| str++; | int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); |
| if(!*str) | memcpy(buf, head.chunk.rows[0].item.ptr, size); |
| buf[size]=0; | |
| cstr=buf; | |
| } else | |
| cstr=this->cstr(); | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | return 0; |
| char *error_pos; | char *error_pos; |
| // 0xABC | // 0xABC |
| if(str[0]=='0') | if(cstr[0]=='0') |
| if(str[1]=='x' || str[1]=='X') | if(cstr[1]=='x' || cstr[1]=='X') |
| result=(double)(unsigned long)strtol(str, &error_pos, 0); | result=(double)(unsigned long)strtol(cstr, &error_pos, 0); |
| else | else |
| result=(double)strtod(str+1/*skip leading 0*/, &error_pos); | result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos); |
| else | else |
| result=(double)strtod(str, &error_pos); | result=(double)strtod(cstr, &error_pos); |
| while(char c=*error_pos++) | while(char c=*error_pos++) |
| if(!isspace(c)) | if(!isspace(c)) |
| Line 481 double String::as_double() const { | Line 742 double String::as_double() const { |
| } | } |
| int String::as_int() const { | int String::as_int() const { |
| int result; | int result; |
| const char *str=cstr(); | const char *cstr; |
| char buf[MAX_NUMBER]; | |
| while(*str && isspace(*str)) | if(head.chunk.rows+1==append_here) { |
| str++; | size_t size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); |
| if(!*str) | memcpy(buf, head.chunk.rows[0].item.ptr, size); |
| buf[size]=0; | |
| cstr=buf; | |
| } else | |
| cstr=this->cstr(); | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | return 0; |
| char *error_pos; | char *error_pos; |
| // 0xABC | // 0xABC |
| if(str[0]=='0') | if(cstr[0]=='0') |
| if(str[1]=='x' || str[1]=='X') | if(cstr[1]=='x' || cstr[1]=='X') |
| result=(int)(unsigned long)strtol(str, &error_pos, 0); | result=(int)(unsigned long)strtol(cstr, &error_pos, 0); |
| else | else |
| result=(int)strtol(str+1/*skip leading 0*/, &error_pos, 0); | result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0); |
| else | else |
| result=(int)strtol(str, &error_pos, 0); | result=(int)strtol(cstr, &error_pos, 0); |
| while(char c=*error_pos++) | while(char c=*error_pos++) |
| if(!isspace(c)) | if(!isspace(c)) |
| Line 507 int String::as_int() const { | Line 775 int String::as_int() const { |
| return result; | return result; |
| } | } |
| inline void uint2uchars(uint word, uchar *bytes) { | inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) { |
| bytes[0]=word&0xFF; | byte1=word&0xFF; |
| bytes[1]=(word>>8)&0xFF; | byte2=word>>8; |
| bytes[2]=(word>>16)&0xFF; | } |
| bytes[3]=(word>>24)&0xFF; | inline ushort uchars2ushort(uchar byte1, uchar byte2) { |
| } | return (byte2<<8) | byte1; |
| inline uint uchars2uint(uchar *bytes) { | } |
| return bytes[3]<<24 | /* @todo maybe network order worth spending some effort? |
| | bytes[2]<<16 | don't bothering myself with network byte order, |
| | bytes[1]<<8 | am not planning to be able to move resulting file across platforms |
| | bytes[0]; | for now |
| } | */ |
| void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const { | |
| static int serialize_body_piece(const char* s, char** cur) { | buf_size= |
| size_t length=strlen(s); | prolog_size |
| memcpy(*cur, s, length); *cur+=length; | +used_rows()*(sizeof(uchar)+sizeof(ushort)) |
| return 0; | +size(); |
| }; | buf=malloc(buf_size,15); |
| String::C String::serialize(size_t prolog_length) const { | char *cur=(char *)buf+prolog_size; |
| size_t buf_length= | |
| prolog_length | |
| +fragments.count()*(sizeof(Language)+sizeof(size_t)) | |
| +length(); | |
| C result(new(PointerFreeGC) char[buf_length], buf_length); | |
| // 1: prolog | |
| char *cur=result.str+prolog_length; | |
| // 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); | |
| // length | |
| memcpy(cur, &fragment.length, sizeof(fragment.length)); cur+=sizeof(fragment.length); | |
| } | |
| // 4: letters | |
| body.for_each(serialize_body_piece, &cur); | |
| return result; | STRING_FOREACH_ROW( |
| // lang | |
| memcpy(cur, &row->item.lang, sizeof(uchar)); | |
| cur+=sizeof(uchar); | |
| // size | |
| uchar byte1; uchar byte2; | |
| ushort2uchars(row->item.size, byte1, byte2); | |
| memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar); | |
| memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar); | |
| // bytes | |
| memcpy(cur, row->item.ptr, row->item.size); | |
| cur+=row->item.size; | |
| ); | |
| } | } |
| bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length) { | bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { |
| if(buf_length<=prolog_length) | if(buf_size<=prolog_size) |
| return false; | return false; |
| buf_length-=prolog_length; | |
| // 1: prolog | |
| const char* cur=(const char* )buf+prolog_length; | |
| // 2: fragments.count | char *cur=(char *)buf+prolog_size; |
| if(buf_length<sizeof(size_t)) // fragments.count don't fit? | buf_size-=prolog_size; |
| return false; | |
| size_t fragments_count=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); | |
| buf_length-=sizeof(size_t); | |
| // 3: lang info | while(buf_size) { |
| size_t total_length=0; | if(sizeof(uchar)+sizeof(ushort)>buf_size) // lang+size |
| for(size_t f=0; f<fragments_count; f++) { | |
| size_t piece_length=sizeof(Language)+sizeof(size_t); | |
| if(buf_length<piece_length) // lang+length | |
| return false; | return false; |
| Language lang=*reinterpret_cast<const Language *>(cur); cur+=sizeof(Language); | uchar lang=*(uchar *)(cur); |
| size_t fragment_length=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); | ushort size=uchars2ushort( |
| fragments+=Fragment(lang, fragment_length); | *(uchar*)(cur+sizeof(uchar)*1), |
| total_length+=fragment_length; | *(uchar*)(cur+sizeof(uchar)*2) |
| ); | |
| buf_length-=piece_length; | size_t piece_size=sizeof(uchar)+sizeof(ushort)+size; |
| } | if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files |
| return false; | |
| // 4: letters | |
| if(buf_length!=total_length) | |
| return false; | |
| body=CORD_cat_char_star(CORD_EMPTY, cur, buf_length); | const char *ptr=(const char*)(cur+sizeof(uchar)*3); |
| APPEND(ptr, size, lang, file, 0); | |
| cur+=piece_size; | |
| buf_size-=piece_size; | |
| } | |
| return true; | return true; |
| } | } |