|
|
| version 1.5, 2001/01/26 18:34:02 | version 1.161, 2002/06/25 16:05:22 |
|---|---|
| Line 1 | Line 1 |
| /* | /** @file |
| $Id$ | Parser: string class. @see untasize_t.C. |
| Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) | |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| $Id$ | |
| */ | */ |
| #include <string.h> | #include "pcre.h" |
| #include "pa_pool.h" | #include "pa_pool.h" |
| #include "pa_string.h" | |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_exception.h" | |
| #include "pa_common.h" | |
| #include "pa_array.h" | |
| #include "pa_globals.h" | |
| #include "pa_table.h" | |
| #include "pa_dictionary.h" | |
| #include "pa_charset.h" | |
| #define DEBUG_STRING_APPENDS_VS_EXPANDS | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | |
| ulong string_piece_appends=0; | |
| #endif | |
| String& String::OnPool(Pool& apool, const char *local_src, size_t src_size, bool tainted) { | |
| if(local_src && *local_src) { | |
| if(src_size==0) | |
| src_size=strlen(local_src); | |
| char *pooled_src=(char *)apool.malloc(src_size); | |
| memcpy(pooled_src, local_src, src_size); | |
| return *new(apool) String(apool, pooled_src, src_size, tainted); | |
| } else | |
| 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); | |
| } | |
| String::String(const String& src) : | |
| Pooled(src.pool()) { | |
| last_chunk=&head.chunk; | |
| head.chunk.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.chunk.rows; | |
| void *String::operator new(size_t size, Pool *apool) { | append(src, UL_UNSPECIFIED); |
| return apool->malloc(size); | |
| } | } |
| void String::construct(Pool *apool) { | size_t String::size() const { |
| pool=apool; | size_t result=0; |
| head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; | STRING_FOREACH_ROW( |
| append_here=head.rows; | result+=row->item.size; |
| head.preallocated_link=0; | ); |
| link_row=&head.rows[curr_chunk_rows]; | return result; |
| } | } |
| /// @todo not very optimal | |
| uint String::used_rows() const { | |
| uint result=0; | |
| STRING_FOREACH_ROW( | |
| result++; | |
| ); | |
| return result; | |
| } | |
| void String::expand() { | void String::expand() { |
| curr_chunk_rows=curr_chunk_rows*100/CR_GROW_PERCENT; | uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; |
| Chunk *chunk=static_cast<Chunk *>( | if(new_chunk_count>max_integral(Chunk::count_type)) |
| pool->calloc(sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | new_chunk_count=max_integral(Chunk::count_type); |
| chunk->count=curr_chunk_rows; | |
| link_row->link=chunk; | Chunk *new_chunk=static_cast<Chunk *>(malloc( |
| append_here=chunk->rows; | sizeof(Chunk)// count+interpadding(?)+rows[CR_PREALLOCATED_COUNT]+tailpadding(??) |
| link_row=&chunk->rows[curr_chunk_rows]; | -sizeof(Chunk::rows_type) // PREALLOCATED rows |
| } | +sizeof(Chunk::Row)*new_chunk_count // neaded rows |
| +sizeof(Chunk *) // link size | |
| String::String(String& src) {/* | , 10)); |
| int src_used_rows=src.used_rows(); { | new_chunk->rows[new_chunk->count=new_chunk_count].link=0; |
| curr_chunk_rows=src_total_rows | last_chunk->rows[last_chunk->count].link=new_chunk; |
| Chunk *chunk=static_cast<Chunk *>( | |
| pool->calloc(sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | last_chunk=new_chunk; |
| chunk->count=curr_chunk_rows; | append_here=last_chunk->rows; |
| link_row->link=chunk; | |
| append_here=chunk->rows; | |
| link_row=&chunk->rows[curr_chunk_rows]; | |
| } | |
| Chunk *chunk=&head; | |
| do { | |
| result+=chunk->count; | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result;*/ | |
| } | } |
| 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)"); | |
| if(!src) | |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | |
| 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++; | |
| src+=max_integral(Chunk::Row::item_size_type); | |
| size-=max_integral(Chunk::Row::item_size_type); | |
| } | |
| String& String::operator += (char *src) { | |
| if(chunk_is_full()) | if(chunk_is_full()) |
| expand(); | expand(); |
| append_here->item.ptr=src; | append_here->item.ptr=src; |
| append_here->item.size=strlen(src); | 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++; | append_here++; |
| return *this; | return *this; |
| } | } |
| size_t String::size() { | char String::first_char() const { |
| int result=0; | if(is_empty()) |
| throw Exception(0, | |
| this, | |
| "getting first char of empty string"); | |
| Chunk *chunk=&head; | return *head.chunk.rows[0].item.ptr; |
| do { | } |
| Chunk::Row *row=chunk->rows; | |
| for(int i=0; i<chunk->count; i++) { | |
| if(row==append_here) | |
| goto break2; | |
| result+=row->item.size; | uint String::hash_code() const { |
| row++; | uint result=0; |
| STRING_FOREACH_ROW( | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| ); | |
| return result; | |
| } | |
| /// @todo move 'lang' skipping to pos | |
| int String::cmp(int& partial, const String& src, | |
| size_t this_offset, Untaint_lang lang) const { | |
| partial=-1; | |
| size_t a_size=size(); | |
| this_offset=min(this_offset, a_size-1); | |
| const Chunk *a_chunk=&head.chunk; | |
| const Chunk *b_chunk=&src.head.chunk; | |
| const Chunk::Row *a_row=a_chunk->rows; | |
| const Chunk::Row *b_row=b_chunk->rows; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| Chunk::Row *a_end=append_here; | |
| Chunk::Row *b_end=src.append_here; | |
| uint a_countdown=a_chunk->count; | |
| uint b_countdown=b_chunk->count; | |
| int result; | |
| size_t pos=0; | |
| bool a_break=a_size==0; | |
| bool b_break=src.is_empty(); | |
| 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_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(a_break=a_row==a_end) { | |
| b_break=b_row==b_end; | |
| break; | |
| } | |
| if(!a_countdown) { | |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | } |
| chunk=row->link; | } |
| } while(chunk); | 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; | |
| } | |
| } | |
| /// @todo move 'lang' skipping to pos | |
| 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--; | |
| } | |
| 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=max(size_t(0), start); | |
| finish=min(size(), 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: | break2: |
| // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", | |
| //cstr(), start, finish, result.cstr()); | |
| return result; | return result; |
| } | } |
| char *String::c_str() { | int String::pos(const String& substr, |
| char *result=static_cast<char *>(pool->malloc(size()+1)); | 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; | |
| } | |
| char *copy_here=result; | int String::pos(const char *substr, size_t substr_size, |
| Chunk *chunk=&head; | int result, Untaint_lang lang) const { |
| do { | size_t self_size=size(); |
| Chunk::Row *row=chunk->rows; | for(; size_t(result)<self_size; result++) { |
| for(int i=0; i<chunk->count; i++) { | int partial; cmp(partial, substr, substr_size, result, lang); |
| if(row==append_here) | if( |
| goto break2; | partial==0 || // full match |
| partial==2) // 'substr' starts 'this'+'result' | |
| return result; | |
| } | |
| return -1; | |
| } | |
| memcpy(copy_here, row->item.ptr, row->item.size); | void String::split(Array& result, |
| copy_here+=row->item.size; | size_t* pos_after_ref, |
| row++; | const char *delim, size_t delim_size, |
| Untaint_lang lang, int limit) const { | |
| size_t self_size=size(); | |
| if(delim_size) { | |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | |
| int pos_before; | |
| // while we have 'delim'... | |
| for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { | |
| result+=&mid(pos_after, pos_before); | |
| pos_after=pos_before+delim_size; | |
| } | } |
| chunk=row->link; | // last piece |
| } while(chunk); | if(pos_after<self_size && limit) { |
| break2: | result+=&mid(pos_after, self_size); |
| *copy_here=0; | pos_after=self_size; |
| } | |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | |
| result+=this; | |
| if(pos_after_ref) | |
| *pos_after_ref+=self_size; | |
| } | |
| } | |
| void String::split(Array& result, | |
| size_t* pos_after_ref, | |
| const String& delim, Untaint_lang lang, | |
| int limit) const { | |
| if(!delim.is_empty()) { | |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | |
| 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(); | |
| } | |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | |
| result+=this; | |
| if(pos_after_ref) | |
| *pos_after_ref+=size(); | |
| } | |
| } | |
| static void regex_options(const String *options, int *result, bool& need_pre_post_match){ | |
| struct Regex_option { | |
| const char *keyL; | |
| const char *keyU; | |
| int clear, set; | |
| int *result; | |
| 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; | |
| 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; | |
| } | |
| } | |
| } | |
| /// @todo make replacement Table stacked | |
| bool String::match( | |
| const String *aorigin, | |
| const String& regexp, | |
| const String *options, | |
| Table **table, | |
| Row_action row_action, void *info, | |
| bool *was_global) const { | |
| if(regexp.is_empty()) | |
| throw Exception(0, | |
| aorigin, | |
| "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); | |
| if(was_global) | |
| *was_global=option_bits[1]!=0; | |
| pcre *code=pcre_compile(pattern, option_bits[0], | |
| &errptr, &erroffset, | |
| pool().get_source_charset().pcre_tables); | |
| if(!code) | |
| throw Exception(0, | |
| ®exp.mid(erroffset, regexp.size()), | |
| "regular expression syntax error - %s", errptr); | |
| int info_substrings=pcre_info(code, 0, 0); | |
| if(info_substrings<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| aorigin, | |
| "pcre_info error (%d)", | |
| info_substrings); | |
| } | |
| const char *subject=cstr(); | |
| int length=strlen(subject); | |
| const int ovecsize=(1/*match*/+MAX_STRING_MATCH_TABLE_COLUMNS)*3; | |
| int ovector[ovecsize]; | |
| // create table | |
| *table=NEW Table(pool(), *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, length, prestart, | |
| exec_option_bits, ovector, ovecsize); | |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | |
| pcre_free(code); | |
| row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return option_bits[1]!=0; // global=true+table, not global=false | |
| } | |
| if(exec_substrings<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| aorigin, | |
| "regular expression execute error (%d)", | |
| exec_substrings); | |
| } | |
| int prefinish=ovector[0]; | |
| poststart=ovector[1]; | |
| Array& row=*NEW Array(pool()); | |
| row+=need_pre_post_match?&mid(0, prefinish):0; // .prematch column value | |
| row+=need_pre_post_match?&mid(prefinish, poststart):0; // .match | |
| row+=need_pre_post_match?&mid(poststart, postfinish):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(!option_bits[1] || prestart==poststart) { // not global | going to hang | |
| pcre_free(code); | |
| row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return true; | |
| } | |
| prestart=poststart; | |
| /* | |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | |
| } | |
| String& String::change_case(Pool& pool, | |
| Change_case_kind kind) const { | |
| const unsigned char *tables=pool.get_source_charset().pcre_tables; | |
| String& result=*new(pool) String(pool); | |
| 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 | |
| } | |
| STRING_FOREACH_ROW( | |
| char *new_cstr=(char *)pool.malloc(row->item.size, 12); | |
| char *dest=new_cstr; | |
| const char *src=row->item.ptr; | |
| for(int size=row->item.size; size--; src++) { | |
| unsigned char c=a[(unsigned char)*src]; | |
| if(b) | |
| c=b[c]; | |
| *dest++=(char)c; | |
| } | |
| result.APPEND(new_cstr, row->item.size, | |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| ); | |
| return result; | return result; |
| } | } |
| int String::used_rows() { | /// @test if in some piece were found no dict words, append it, not it's duplicate |
| int result=0; | String& String::replace(Pool& pool, Dictionary& dict) const { |
| // return reconstruct(pool).replace_in_reconstructed(pool, dict); | |
| String& result=*new(pool) String(pool); | |
| STRING_FOREACH_ROW( | |
| const char *src=row->item.ptr; | |
| size_t src_size=row->item.size; | |
| char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14); | |
| char *dest=new_cstr; | |
| while(src_size) { | |
| // there is a row where first column starts 'src' | |
| if(Table::Item *item=dict.first_that_starts(src, src_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 'src' && reduce work size | |
| src+=a.size(); src_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++=*src++; src_size--; | |
| } | |
| } | |
| Chunk *chunk=&head; | result.APPEND(new_cstr, dest-new_cstr, row->item.lang, |
| do { | row->item.origin.file, row->item.origin.line); |
| int count=chunk->count; | ); |
| result+=count; | return result; |
| chunk=chunk->rows[count].link; | } |
| } while(chunk); | |
| result-=link_row-append_here; | String& String::join_chains(Pool& pool, char** acstr) const { |
| char *lcstr=cstr(); | |
| const char *current=lcstr; | |
| 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; |
| } | } |
| unsigned int String::hash_code() { | double String::as_double() const { |
| unsigned int result=0; | double result; |
| const char *cstr; | |
| char buf[MAX_NUMBER]; | |
| if(head.chunk.rows+1==append_here) { | |
| int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); | |
| 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) | |
| throw Exception("number.format", | |
| this, | |
| "blank string is invalid number (double)"); | |
| 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) | const char *cstr; |
| goto break2; | char buf[MAX_NUMBER]; |
| if(head.chunk.rows+1==append_here) { | |
| int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); | |
| 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) | |
| throw Exception("number.format", | |
| this, | |
| "blank string is invalid number (int)"); | |
| 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<String, String>::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 ushort2uchars(ushort word, uchar& byte1, uchar& byte2) { |
| return false; | byte1=word&0xFF; |
| byte2=word>>8; | |
| } | |
| inline ushort uchars2ushort(uchar byte1, uchar byte2) { | |
| return (byte2<<8) | byte1; | |
| } | |
| /* @todo maybe network order worth spending some effort? | |
| don't bothering myself with network byte order, | |
| am not planning to be able to move resulting file across platforms | |
| for now | |
| */ | |
| void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const { | |
| buf_size= | |
| prolog_size | |
| +used_rows()*(sizeof(uchar)+sizeof(ushort)) | |
| +size(); | |
| buf=malloc(buf_size,15); | |
| char *cur=(char *)buf+prolog_size; | |
| 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_size, void *buf, size_t buf_size, const char *file) { | |
| if(buf_size<=prolog_size) | |
| return false; | |
| char *cur=(char *)buf+prolog_size; | |
| buf_size-=prolog_size; | |
| while(buf_size) { | |
| if(sizeof(uchar)+sizeof(ushort)>buf_size) // lang+size | |
| return false; | |
| uchar lang=*(uchar *)(cur); | |
| ushort size=uchars2ushort( | |
| *(uchar*)(cur+sizeof(uchar)*1), | |
| *(uchar*)(cur+sizeof(uchar)*2) | |
| ); | |
| size_t piece_size=sizeof(uchar)+sizeof(ushort)+size; | |
| if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files | |
| return false; | |
| const char *ptr=(const char*)(cur+sizeof(uchar)*3); | |
| APPEND(ptr, size, lang, file, 0); | |
| cur+=piece_size; | |
| buf_size-=piece_size; | |
| } | |
| return true; | |
| } | } |