|
|
| version 1.42, 2001/03/18 17:18:36 | version 1.172.2.14, 2003/02/04 15:59:05 |
|---|---|
| Line 1 | Line 1 |
| /* | /** @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 { | |
| Pool pool; | |
| public: | |
| String_match_table_template_columns() { | |
| *this+=StringPtr(new String("prematch")); | |
| *this+=StringPtr(new String("match")); | |
| *this+=StringPtr(new String("postmatch")); | |
| for(int i=0; i<MAX_MATCH_GROUPS; i++) { | |
| char *cname=new(pool) char[3/*strlen("100")*/+1/*terminating 0*/]; | |
| StringPtr sname(new String(cname, sprintf(cname, "%d", 1+i))); | |
| *this+=sname; | |
| } | |
| } | |
| }; | |
| // String | Table string_match_table_template( |
| Exception::undefined_source, | |
| Table::columns_type(new String_match_table_template_columns)); | |
| String::String(Pool& apool, const char *src, bool tainted) : | // methods |
| 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; | |
| String::String(const char* src, size_t src_size, bool tainted): Array<String_fragment>(1), fsize(0) { | |
| if(src) | if(src) |
| if(tainted) | if(tainted) |
| APPEND_TAINTED(src, 0, 0, 0); | APPEND_TAINTED(src, src_size, 0, 0); |
| else | else |
| APPEND(src, 0, 0, 0); | APPEND_CLEAN(src, src_size, 0, 0); |
| } | } |
| void String::expand() { | String::String(const String& src): Array<String_fragment>(src.count()) { |
| int new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; | append(src, UL_PASS_APPENDED); |
| last_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(int)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); | |
| 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()) { | String& String::real_append(STRING_APPEND_PARAMS) { |
| head.count=CR_PREALLOCATED_COUNT; | if(!src) |
| return *this; | |
| int src_used_rows=src.used_rows(); | if(!size) |
| if(src_used_rows<=head.count) { | size=strlen(src); |
| // all new rows fit into preallocated area | if(!size) |
| int curr_chunk_rows=head.count; | return *this; |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | |
| append_here=&head.rows[src_used_rows]; | if(is_full()) |
| link_row=&head.rows[curr_chunk_rows]; | expand(fdelta); |
| String_fragment *fragment=&felements[fused++]; | |
| fragment->ptr=src; | |
| fragment->size=size; | |
| fragment->lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| fragment->origin.file=file; | |
| fragment->origin.line=line; | |
| #endif | |
| return *this; | |
| } | |
| uint String::hash_code() const { | |
| uint result=0; | |
| STRING_FOREACH_FRAGMENT( | |
| result=generic_hash_code(result, fragment->ptr, fragment->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 String_fragment *a_current=felements; | |
| const String_fragment *b_current=src.felements; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| String_fragment *a_end=felements+fused; | |
| String_fragment *b_end=src.felements+src.fused; | |
| int result; | |
| size_t pos=0; | |
| bool a_break; | |
| bool b_break; | |
| while(!( | |
| (a_break=(a_current==a_end)) | |
| || (b_break=(b_current==b_end)))) { | |
| if(pos+a_current->size > this_offset) { | |
| if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang) | |
| return -1; // wrong lang -- bail out | |
| int size_diff= | |
| (a_current->size-a_offset)- | |
| (b_current->size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| a_current->size-a_offset); | |
| if(result) | |
| return result; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| b_current++; b_offset=0; | |
| } else if(size_diff>0) { // a longer | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| b_current->size-b_offset); | |
| if(result) | |
| return result; | |
| a_offset+=b_current->size-b_offset; | |
| b_current++; b_offset=0; | |
| } else { // b longer | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| a_current->size-a_offset); | |
| if(result) | |
| return result; | |
| b_offset+=a_current->size-a_offset; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| } | |
| } else { | |
| a_offset-=a_current->size; | |
| pos+=a_current->size; | |
| a_current++; | |
| } | |
| } | |
| 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 { | } else { |
| // warning: | partial=2; return +1; |
| // heavily relies on the fact | } |
| // "preallocated area is the same for all strings" | } |
| // | |
| // info: | |
| // allocating only enough mem to fit src string rows | |
| // next append would allocate a new chunk | |
| // | |
| // new rows don't fit into preallocated area: splitting into two chunks | |
| // preallocated chunk src to constructing head | |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | |
| // 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; | /// @todo move 'lang' skipping to pos |
| } else { | int String::cmp(int& partial, const char* b_ptr, size_t src_size, |
| // the last source chunk | size_t this_offset, Untaint_lang lang) const { |
| // taking only those rows of chunk that _left_to_copy | partial=-1; |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | size_t a_size=size(); |
| break; | size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; |
| this_offset=min(this_offset, a_size-1); | |
| const String_fragment *a_current=felements; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| String_fragment *a_end=felements+fused; | |
| size_t pos=0; | |
| bool a_break; | |
| bool b_break=b_size==0; | |
| while(!( | |
| (a_break=(a_current==a_end)) | |
| || b_break)) { | |
| if(pos+a_current->size > this_offset) { | |
| if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang) | |
| return -1; // wrong lang -- bail out | |
| int size_diff= | |
| (a_current->size-a_offset)- | |
| (b_size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset, | |
| a_current->size-a_offset)!=0) | |
| return result; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| b_break=true; | |
| } else if (size_diff>0) { // a longer | |
| if(int result=memcmp(a_current->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_current->ptr+a_offset, b_ptr+b_offset, | |
| a_current->size-a_offset)!=0) | |
| return result; | |
| b_offset+=a_current->size-a_offset; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| } | } |
| } else { | |
| a_offset-=a_current->size; | |
| pos+=a_current->size; | |
| a_current++; | |
| } | } |
| } | } |
| link_row->link=0; | if(a_break==b_break) { // ended simultaneously |
| fused_rows=src_used_rows; | partial=0; return 0; |
| fsize=src.fsize; | } else if(a_break) { // first bytes equal, but a ended before b |
| partial=1; return -1; | |
| } else { | |
| partial=2; return +1; | |
| } | |
| } | } |
| String& String::append(const String& src, Untaint_lang lang, bool forced) { | #ifndef NO_STRING_ORIGIN |
| int src_used_rows=src.used_rows(); | const String_fragment::Origin& String::origin() const { |
| int dst_free_rows=link_row-append_here; | if(is_empty()) { |
| static const String_fragment::Origin empty_origin={"empty string"}; | |
| return empty_origin; | |
| } | |
| if(src_used_rows<=dst_free_rows) { | // determining origin by first piece or last appended piece |
| // all new rows fit into last chunk | // because any of them can be constant=without origin: |
| memcpy(append_here, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | // ex: ^load[/file] "document_root" + "/file" |
| set_lang(append_here, lang, forced, src_used_rows); | // when last peice is constant, |
| append_here+=src_used_rows; | // ex: parser_root_auto_path{dynamic} / auto.p{const} |
| } else { | // using first piece |
| // not all new rows fit into last chunk: shrinking it to used part, | String_fragment::Origin& first_origin=felements[0].origin; |
| int used_rows=last_chunk->count-dst_free_rows; | return first_origin.file ? first_origin : felements[fused-1].origin; |
| //int *countp=append_here | } |
| link_row=&last_chunk->rows[last_chunk->count=used_rows]; | #endif |
| // 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; | StringPtr String::mid(size_t start, size_t finish) const { |
| } else { | StringPtr result(new String()); |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | start=min(start, size()); |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | finish=max(start, finish); |
| set_lang(new_rows, lang, forced, rows_left_to_copy); | if(start==finish) |
| break; | return result; |
| } | |
| size_t pos=0; | |
| STRING_FOREACH_FRAGMENT( | |
| size_t item_finish=pos+fragment->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:fragment->size; | |
| result->APPEND( | |
| fragment->ptr+offset, size-offset, | |
| fragment->lang, | |
| fragment->origin.file, fragment->origin.line); | |
| if(finished) | |
| goto break2; | |
| } | } |
| link_row->link=0; | pos+=fragment->size; |
| } | ); |
| fused_rows+=src_used_rows; | break2: |
| fsize+=src.fsize; | // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", |
| //cstr(), start, finish, result.cstr()); | |
| return result; | |
| } | |
| return *this; | 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; | |
| } | } |
| void String::set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size) { | |
| if(lang==PASS_APPEND) | int String::pos(const char* substr, size_t substr_size, |
| return; | int result, Untaint_lang lang) const { |
| size_t self_size=size(); | |
| while(size--) { | for(; size_t(result)<self_size; result++) { |
| Untaint_lang& item_lang=(row++)->item.lang; | int partial; cmp(partial, substr, substr_size, result, lang); |
| if(item_lang==YES || forced) // tainted? need untaint language assignment | if( |
| item_lang=lang; // assign untaint language | partial==0 || // full match |
| partial==2) // 'substr' starts 'this'+'result' | |
| return result; | |
| } | } |
| return -1; | |
| } | } |
| /*void String::change_lang(Untaint_lang lang) { | void String::split(ArrayString& result, |
| Chunk *chunk=&head; | size_t* pos_after_ref, |
| do { | const char* delim, size_t delim_size, |
| Chunk::Row *row=chunk->rows; | Untaint_lang lang, int limit) { |
| for(int i=0; i<chunk->count; i++) { | size_t self_size=size(); |
| if(row==append_here) | if(delim_size) { |
| goto break2; | 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; | |
| } | |
| // last piece | |
| if(pos_after<self_size && limit) { | |
| result+=mid(pos_after, self_size); | |
| pos_after=self_size; | |
| } | |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | |
| result+=StringPtr(this); | |
| if(pos_after_ref) | |
| *pos_after_ref+=self_size; | |
| } | |
| } | |
| row->item.lang=lang; | void String::split(ArrayString& result, |
| row++; | size_t* pos_after_ref, |
| const String& delim, Untaint_lang lang, | |
| int limit) { | |
| 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(); | |
| } | } |
| chunk=row->link; | // last piece |
| } while(chunk); | if(pos_after<size() && limit) { |
| break2: | result+=mid(pos_after, size()); |
| return; | pos_after=size(); |
| } | |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | |
| result+=StringPtr(this); | |
| if(pos_after_ref) | |
| *pos_after_ref+=size(); | |
| } | |
| } | } |
| */ | |
| String& String::real_append(STRING_APPEND_PARAMS) { | |
| if(!src) | |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| if(chunk_is_full()) | static void regex_options(StringPtr options, int *result, bool& need_pre_post_match){ |
| expand(); | 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 | 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; | |
| } | |
| } | |
| } | |
| append_here->item.ptr=src; | TablePtr String::match(Charset& source_charset, |
| fsize+=append_here->item.size=size; | StringPtr aorigin, |
| append_here->item.lang=tainted?YES:NO; | const String& regexp, |
| #ifndef NO_STRING_ORIGIN | StringPtr options, |
| append_here->item.origin.file=file; | Row_action row_action, void *info, |
| append_here->item.origin.line=line; | bool *was_global) const { |
| #endif | |
| append_here++; fused_rows++; | if(regexp.is_empty()) |
| throw Exception(0, | |
| aorigin, | |
| "regexp is empty"); | |
| CharPtr 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.get(), option_bits[0], | |
| &errptr, &erroffset, | |
| source_charset.pcre_tables); | |
| if(!code) | |
| throw Exception(0, | |
| regexp.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); | |
| } | |
| return *this; | CharPtr subject=cstr(); |
| int length=size(); | |
| const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3; | |
| int ovector[ovecsize]; | |
| // create table | |
| TablePtr result(new Table(string_match_table_template)); | |
| int exec_option_bits=0; | |
| int prestart=0; | |
| int poststart=0; | |
| int postfinish=size(); | |
| while(true) { | |
| int exec_substrings=pcre_exec(code, 0, | |
| subject.get(), length, prestart, | |
| exec_option_bits, ovector, ovecsize); | |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | |
| pcre_free(code); | |
| row_action(*result.get(), 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); | |
| if(option_bits[1]) | |
| return result; // global=true+result | |
| else | |
| return TablePtr(0);// not global=no result | |
| } | |
| 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]; | |
| object_ptr<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+=StringPtr(0); // .prematch column value | |
| *row+=StringPtr(0); // .match | |
| *row+=StringPtr(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(*result, row.get(), prestart, prefinish, poststart, postfinish, info); | |
| if(!option_bits[1] || prestart==poststart) { // not global | going to hang | |
| pcre_free(code); | |
| row_action(*result, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return result; | |
| } | |
| prestart=poststart; | |
| /* | |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | |
| } | } |
| uint String::hash_code() const { | StringPtr String::change_case(Pool& pool, Charset& source_charset, Change_case_kind kind) { |
| uint result=0; | StringPtr result(new String()); |
| const Chunk *chunk=&head; | const unsigned char *tables=source_charset.pcre_tables; |
| do { | |
| const Chunk::Row *row=chunk->rows; | const unsigned char *a; |
| for(int i=0; i<chunk->count; i++) { | const unsigned char *b; |
| if(row==append_here) | switch(kind) { |
| goto break2; | 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, | |
| StringPtr(this), | |
| "unknown change case kind #%d", | |
| static_cast<int>(kind)); // never | |
| a=b=0; // calm, compiler | |
| break; // never | |
| } | |
| STRING_FOREACH_FRAGMENT( | |
| char *new_cstr=new(pool) char[fragment->size]; | |
| char *dest=new_cstr; | |
| const char* end=fragment->ptr+fragment->size; | |
| for(const char* current=fragment->ptr; current<end; current++) { | |
| unsigned char c=a[(unsigned char)*current]; | |
| if(b) | |
| c=b[c]; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | *dest++=(char)c; |
| row++; | |
| } | } |
| chunk=row->link; | |
| } while(chunk); | result->APPEND(new_cstr, fragment->size, |
| break2: | fragment->lang, |
| fragment->origin.file, fragment->origin.line); | |
| ); | |
| return result; | |
| } | |
| /// @test if in some piece were found no dict words, append it, not it's duplicate | |
| StringPtr String::replace(Pool& pool, Dictionary& dict) const { | |
| StringPtr result(new String()); | |
| CharPtr lcstr(cstr()); | |
| const char* current=lcstr.get(); | |
| STRING_FOREACH_FRAGMENT( | |
| IFNDEF_NO_STRING_ORIGIN( | |
| const char* joined_origin_file=fragment->origin.file; | |
| const size_t joined_origin_line=fragment->origin.line; | |
| ); | |
| uchar joined_lang=fragment->lang; | |
| const char* joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_FRAGMENT( | |
| if(fragment->lang==joined_lang) | |
| joined_size+=fragment->size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT | |
| --fragment; | |
| char *new_cstr=new(pool) char[(size_t)ceil(joined_size*dict.max_ratio())]; | |
| char *dest=new_cstr; | |
| while(joined_size) { | |
| // there is a fragment where first column starts 'joined_ptr' | |
| if(Table::element_type row=dict.first_that_starts(joined_ptr, joined_size)) { | |
| // get a=>b values | |
| StringPtr a=row->get(0); | |
| StringPtr b=row->get(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, String::UL_AS_IS); dest+=b->size(); | |
| } else { | |
| // write a char to b && reduce work size | |
| *dest++=*joined_ptr++; joined_size--; | |
| } | |
| } | |
| result->APPEND(new_cstr, dest-new_cstr, joined_lang, | |
| joined_origin_file, joined_origin_line); | |
| ); | |
| return result; | |
| } | |
| StringPtr String::join_chains(Pool& pool) const { | |
| StringPtr result(new String()); | |
| char *pooled_buf=new(pool) char[cstr_bufsize()]; | |
| store_to(pooled_buf, String::UL_AS_IS); | |
| const char* current=pooled_buf; | |
| STRING_FOREACH_FRAGMENT( | |
| IFNDEF_NO_STRING_ORIGIN( | |
| const char* joined_origin_file=fragment->origin.file; | |
| const size_t joined_origin_line=fragment->origin.line; | |
| ); | |
| uchar joined_lang=fragment->lang; | |
| const char* joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_FRAGMENT( | |
| if(fragment->lang==joined_lang) | |
| joined_size+=fragment->size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT | |
| --fragment; | |
| result->APPEND(joined_ptr, joined_size, joined_lang, | |
| joined_origin_file, joined_origin_line); | |
| ); | |
| return result; | return result; |
| } | } |
| int String::cmp(const String& src) const { | double String::as_double() { |
| const Chunk *a_chunk=&head; | double result; |
| const Chunk *b_chunk=&src.head; | char buf[MAX_STRING]; |
| const Chunk::Row *a_row=a_chunk->rows; | if(size()>MAX_STRING-1) |
| const Chunk::Row *b_row=b_chunk->rows; | throw Exception("number.format", |
| int a_offset=0; | StringPtr(this), |
| int b_offset=0; | "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); |
| Chunk::Row *a_end=append_here; | char *eol=store_to(buf, String::UL_AS_IS); *eol=0; |
| Chunk::Row *b_end=src.append_here; | const char* cstr=buf; |
| int a_countdown=a_chunk->count; | |
| int b_countdown=b_chunk->count; | while(*cstr && isspace(*cstr)) |
| bool a_break=false; | cstr++; |
| bool b_break=false; | 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", | |
| StringPtr(this), | |
| "invalid number (double)"); | |
| return result; | |
| } | |
| int String::as_int() { | |
| int result; | int result; |
| while(true) { | char buf[MAX_STRING]; |
| a_break=a_row==a_end; | if(size()>MAX_STRING-1) |
| b_break=b_row==b_end; | throw Exception("number.format", |
| if(a_break || b_break) | StringPtr(this), |
| break; | "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); |
| char *eol=store_to(buf, String::UL_AS_IS); *eol=0; | |
| int size_diff= | const char* cstr=buf; |
| (a_row->item.size-a_offset)- | |
| (b_row->item.size-b_offset); | while(*cstr && isspace(*cstr)) |
| cstr++; | |
| if(size_diff==0) { // a has same size as b | if(!*cstr) |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset); | return 0; |
| if(result) | |
| return result; | char *error_pos; |
| a_row++; a_countdown--; a_offset=0; | // 0xABC |
| b_row++; b_countdown--; b_offset=0; | if(cstr[0]=='0') |
| } else if (size_diff>0) { // a longer | if(cstr[1]=='x' || cstr[1]=='X') |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset); | result=(int)(unsigned long)strtol(cstr, &error_pos, 0); |
| if(result) | else |
| return result; | result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0); |
| 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; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | |
| 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; | |
| } | |
| } | |
| if(a_break==b_break) // ended simultaneously | |
| result=0; | |
| else if(a_break) // first bytes equal, but a ended before b | |
| result=-1; | |
| else | else |
| result=+1; | result=(int)strtol(cstr, &error_pos, 0); |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| StringPtr(this), | |
| "invalid number (int)"); | |
| return result; | return result; |
| } | } |
| bool String::operator == (const char* b_ptr) const { | inline void uint2uchars(uint word, uchar *bytes) { |
| size_t b_size=b_ptr?strlen(b_ptr):0; | bytes[0]=word&0xFF; |
| if(size() != b_size) | 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]; | |
| } | |
| void String::serialize(Pool& pool, size_t prolog_size, char *& buf, size_t& buf_size) const { | |
| buf_size= | |
| prolog_size | |
| +fused*(sizeof(uchar)+sizeof(size_t)) | |
| +size(); | |
| buf=new(pool) char[buf_size]; | |
| char *cur=buf+prolog_size; | |
| STRING_FOREACH_FRAGMENT( | |
| // lang | |
| memcpy(cur, &fragment->lang, sizeof(fragment->lang)); | |
| cur+=sizeof(fragment->lang); | |
| // size | |
| // bug on some sparc platform [you can't work with integers on odd pointers] | |
| // forces us to use byte array instead | |
| uchar bytes[4]; | |
| uint2uchars(fragment->size, bytes); | |
| memcpy(cur, &bytes, sizeof(bytes)); cur+=sizeof(bytes); | |
| // bytes | |
| memcpy(cur, fragment->ptr, fragment->size); | |
| cur+=fragment->size; | |
| ); | |
| } | |
| bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) { | |
| if(buf_size<=prolog_size) | |
| return false; | return false; |
| const Chunk *a_chunk=&head; | char *cur=(char *)buf+prolog_size; |
| const Chunk::Row *a_row=a_chunk->rows; | buf_size-=prolog_size; |
| int a_offset=0; | |
| int b_offset=0; | while(buf_size) { |
| Chunk::Row *a_end=append_here; | if(sizeof(uchar)+sizeof(size_t)>buf_size) // lang+size |
| int a_countdown=a_chunk->count; | return false; |
| bool a_break=false; | |
| bool b_break=false; | String_UL lang=*(String_UL *)(cur); |
| while(true) { | size_t size=uchars2uint((uchar *)cur); |
| int size_diff= | |
| (a_row->item.size-a_offset)- | size_t piece_size=sizeof(uchar)+sizeof(size_t)+size; |
| (b_size-b_offset); | if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files |
| return false; | |
| if(size_diff==0) { // a has same size as b | |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) | const char* ptr=(const char*)(cur+sizeof(uchar)+sizeof(size_t)); |
| return false; | APPEND(ptr, size, lang, file, 0); |
| a_row++; a_countdown--; a_offset=0; | |
| b_break=true; | cur+=piece_size; |
| } else if (size_diff>0) { // a longer | buf_size-=piece_size; |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, b_size-b_offset)!=0) | |
| 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(a_break || b_break) | |
| break; | |
| if(!a_countdown) { | |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | |
| } | } |
| return a_break==b_break; | return true; |
| } | } |