|
|
| version 1.116, 2001/10/29 16:29:08 | version 1.149, 2002/03/27 15:30:37 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class. @see untasize_t.C. | Parser: string class. @see untasize_t.C. |
| Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| $Id$ | $Id$ |
| */ | */ |
| #include "pa_config_includes.h" | |
| #include "pcre.h" | #include "pcre.h" |
| #include "internal.h" | |
| #include "pa_pool.h" | #include "pa_pool.h" |
| #include "pa_string.h" | #include "pa_string.h" |
| Line 21 | Line 18 |
| #include "pa_globals.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" | |
| #define DEBUG_STRING_APPENDS_VS_EXPANDS | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | |
| ulong string_piece_appends=0; | |
| #endif | |
| String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : | String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : |
| Pooled(apool), | Pooled(apool) { |
| forigins_mode(false) { | |
| last_chunk=&head; | last_chunk=&head; |
| head.count=CR_PREALLOCATED_COUNT; | head.count=CR_PREALLOCATED_COUNT; |
| append_here=head.rows; | append_here=head.rows; |
| head.preallocated_link=0; | |
| link_row=&head.rows[head.count]; | |
| fsize=0; | |
| if(src) | if(src) |
| if(tainted) | if(tainted) |
| Line 40 String::String(Pool& apool, const char * | Line 41 String::String(Pool& apool, const char * |
| } | } |
| String::String(const String& src) : | String::String(const String& src) : |
| Pooled(src.pool()), | Pooled(src.pool()) { |
| forigins_mode(false) { | last_chunk=&head; |
| head.count=CR_PREALLOCATED_COUNT; | head.count=CR_PREALLOCATED_COUNT; |
| append_here=head.rows; | |
| uint src_used_rows=src.used_rows(); | |
| if(src_used_rows<=head.count) { | |
| // all new rows fit size_to preallocated area | |
| last_chunk=&head; | |
| uint curr_chunk_rows=head.count; | |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | |
| append_here=&head.rows[src_used_rows]; | |
| link_row=&head.rows[curr_chunk_rows]; | |
| } else { | |
| // warning: | |
| // 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 size_to preallocated area: splitting size_to two chunks | |
| // preallocated chunk src to constructing head | |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | |
| // remaining rows size_to new_chunk | |
| uint curr_chunk_rows=src_used_rows-head.count; | |
| last_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(uint)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *), 9)); | |
| last_chunk->count=curr_chunk_rows; | |
| head.preallocated_link=last_chunk; | |
| append_here=link_row=&last_chunk->rows[last_chunk->count]; | |
| Chunk *old_chunk=src.head.preallocated_link; | |
| Chunk::Row *new_rows=last_chunk->rows; | |
| uint rows_left_to_copy=last_chunk->count; | |
| while(true) { | |
| uint 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; | append(src, UL_UNSPECIFIED); |
| } else { | } |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | size_t String::size() const { |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | size_t result=0; |
| break; | STRING_FOREACH_ROW( |
| } | result+=row->item.size; |
| } | ); |
| } | return result; |
| link_row->link=0; | |
| src_used_rows; | |
| fsize=src.fsize; | |
| } | } |
| /// @todo not very optimal | /// @todo not very optimal |
| uint String::used_rows() const { | uint String::used_rows() const { |
| uint result=0; | uint result=0; |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | result++; |
| const Chunk::Row *row=chunk->rows; | ); |
| for(uint i=0; i<chunk->count; i++, row++) { | |
| if(row==append_here) | |
| goto break2; | |
| result++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result; | return result; |
| } | } |
| void String::expand() { | void String::expand() { |
| uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; | uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; |
| last_chunk=static_cast<Chunk *>( | if(new_chunk_count>max_integral(Chunk::count_type)) |
| malloc(sizeof(uint)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10)); | new_chunk_count=max_integral(Chunk::count_type); |
| 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::append(const String& src, Untaint_lang lang, bool forced) { | Chunk *new_chunk=static_cast<Chunk *>( |
| const Chunk *chunk=&src.head; | malloc(sizeof(Chunk::count_type)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10)); |
| do { | new_chunk->rows[new_chunk->count=new_chunk_count].link=0; |
| const Chunk::Row *row=chunk->rows; | last_chunk->rows[last_chunk->count].link=new_chunk; |
| for(uint i=0; i<chunk->count; i++, row++) { | |
| if(row==src.append_here) | last_chunk=new_chunk; |
| goto break2; | append_here=last_chunk->rows; |
| APPEND(row->item.ptr, row->item.size, | |
| (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return *this; | |
| } | } |
| String& String::real_append(STRING_APPEND_PARAMS) { | 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) | if(!src) |
| return *this; | return *this; |
| if(!size) | if(!size) |
| Line 154 String& String::real_append(STRING_APPEN | Line 92 String& String::real_append(STRING_APPEN |
| if(!size) | if(!size) |
| return *this; | 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); | |
| } | |
| if(chunk_is_full()) | if(chunk_is_full()) |
| expand(); | expand(); |
| append_here->item.ptr=src; | append_here->item.ptr=src; |
| fsize+=append_here->item.size=size; | append_here->item.size=size; |
| append_here->item.lang=lang; | append_here->item.lang=lang; |
| #ifndef NO_STRING_ORIGIN | #ifndef NO_STRING_ORIGIN |
| append_here->item.origin.file=file; | append_here->item.origin.file=file; |
| Line 170 String& String::real_append(STRING_APPEN | Line 130 String& String::real_append(STRING_APPEN |
| } | } |
| char String::first_char() const { | char String::first_char() const { |
| if(!fsize) | if(is_empty()) |
| throw Exception(0, 0, | throw Exception(0, |
| this, | this, |
| "getting first char of empty string"); | "getting first char of empty string"); |
| Line 180 char String::first_char() const { | Line 140 char String::first_char() const { |
| uint String::hash_code() const { | uint String::hash_code() const { |
| uint result=0; | uint result=0; |
| STRING_FOREACH_ROW( | |
| const Chunk *chunk=&head; | |
| do { | |
| const Chunk::Row *row=chunk->rows; | |
| for(uint i=0; i<chunk->count; i++) { | |
| if(row==append_here) | |
| goto break2; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | result=Hash::generic_code(result, row->item.ptr, row->item.size); |
| row++; | ); |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result; | return result; |
| } | } |
| Line 201 break2: | Line 150 break2: |
| int String::cmp(int& partial, const String& src, | int String::cmp(int& partial, const String& src, |
| size_t this_offset, Untaint_lang lang) const { | size_t this_offset, Untaint_lang lang) const { |
| partial=-1; | partial=-1; |
| this_offset=min(this_offset, size()-1); | size_t a_size=size(); |
| this_offset=min(this_offset, a_size-1); | |
| const Chunk *a_chunk=&head; | const Chunk *a_chunk=&head; |
| const Chunk *b_chunk=&src.head; | const Chunk *b_chunk=&src.head; |
| Line 216 int String::cmp(int& partial, const Stri | Line 166 int String::cmp(int& partial, const Stri |
| int result; | int result; |
| size_t pos=0; | size_t pos=0; |
| bool a_break=size()==0; | bool a_break=a_size==0; |
| bool b_break=src.size()==0; | bool b_break=src.is_empty(); |
| if(!(a_break || b_break)) while(true) { | if(!(a_break || b_break)) while(true) { |
| if(pos+a_row->item.size > this_offset) { | if(pos+a_row->item.size > this_offset) { |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) | if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) |
| return -1; // wrong lang -- bail out | return -1; // wrong lang -- bail out |
| int size_diff= | int size_diff= |
| Line 289 int String::cmp(int& partial, const Stri | Line 239 int String::cmp(int& partial, const Stri |
| int String::cmp(int& partial, const char* b_ptr, size_t src_size, | int String::cmp(int& partial, const char* b_ptr, size_t src_size, |
| size_t this_offset, Untaint_lang lang) const { | size_t this_offset, Untaint_lang lang) const { |
| partial=-1; | partial=-1; |
| size_t a_size=size(); | |
| size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; | size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; |
| this_offset=min(this_offset, size()-1); | this_offset=min(this_offset, a_size-1); |
| const Chunk *a_chunk=&head; | const Chunk *a_chunk=&head; |
| const Chunk::Row *a_row=a_chunk->rows; | const Chunk::Row *a_row=a_chunk->rows; |
| Line 300 int String::cmp(int& partial, const char | Line 251 int String::cmp(int& partial, const char |
| uint a_countdown=a_chunk->count; | uint a_countdown=a_chunk->count; |
| size_t pos=0; | size_t pos=0; |
| bool a_break=size()==0; | bool a_break=a_size==0; |
| bool b_break=b_size==0; | bool b_break=b_size==0; |
| if(!(a_break || b_break)) while(true) { | if(!(a_break || b_break)) while(true) { |
| if(pos+a_row->item.size > this_offset) { | if(pos+a_row->item.size > this_offset) { |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) | if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) |
| return -1; // wrong lang -- bail out | return -1; // wrong lang -- bail out |
| int size_diff= | int size_diff= |
| Line 358 int String::cmp(int& partial, const char | Line 309 int String::cmp(int& partial, const char |
| #ifndef NO_STRING_ORIGIN | #ifndef NO_STRING_ORIGIN |
| const Origin& String::origin() const { | const Origin& String::origin() const { |
| if(!fsize) { | if(is_empty()) { |
| static const Origin empty_origin={"empty string"}; | static const Origin empty_origin={"empty string"}; |
| return empty_origin; | return empty_origin; |
| } | } |
| // determining origin by last appended piece | // determining origin by first piece or last appended piece |
| // because first one frequently constant. | // because any of them can be constant=without origin: |
| // ex: ^load[/file] "document_root" + "/file" | // ex: ^load[/file] "document_root" + "/file" |
| // when last peice is constant, | // when last peice is constant, |
| // ex: parser_root_auto_path{dynamic} / auto.p{const} | // ex: parser_root_auto_path{dynamic} / auto.p{const} |
| // using first piece | // using first piece |
| Origin& last_origin=append_here[-1].item.origin; | Origin& first_origin=head.rows[0].item.origin; |
| return last_origin.file ? last_origin : head.rows[0].item.origin; | return first_origin.file ? first_origin : append_here[-1].item.origin; |
| } | } |
| #endif | #endif |
| String& String::mid(size_t start, size_t finish) const { | String& String::mid(size_t start, size_t finish) const { |
| String& result=*NEW String(pool()); | String& result=*NEW String(pool()); |
| start=max(0, start); | start=max(size_t(0), start); |
| finish=min(size(), finish); | finish=min(size(), finish); |
| if(start==finish) | if(start==finish) |
| return result; | return result; |
| size_t pos=0; | size_t pos=0; |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | size_t item_finish=pos+row->item.size; |
| const Chunk::Row *row=chunk->rows; | if(item_finish > start) { // started now or already? |
| for(uint i=0; i<chunk->count; pos+=row->item.size, i++, row++) { | bool started=result.is_empty(); // started now? |
| if(row==append_here) | 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; | goto break2; |
| size_t item_finish=pos+row->item.size; | |
| if(item_finish > start) { // started now or already? | |
| bool started=result.size()==0; // 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; | |
| } | |
| } | } |
| chunk=row->link; | pos+=row->item.size; |
| } while(chunk); | ); |
| break2: | break2: |
| // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", | // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", |
| //cstr(), start, finish, result.cstr()); | //cstr(), start, finish, result.cstr()); |
| Line 414 break2: | Line 358 break2: |
| int String::pos(const String& substr, | int String::pos(const String& substr, |
| int result, Untaint_lang lang) const { | int result, Untaint_lang lang) const { |
| for(; result<size(); result++) { | size_t self_size=size(); |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, result, lang); | int partial; cmp(partial, substr, result, lang); |
| if( | if( |
| partial==0 || // full match | partial==0 || // full match |
| Line 427 int String::pos(const String& substr, | Line 372 int String::pos(const String& substr, |
| int String::pos(const char *substr, size_t substr_size, | int String::pos(const char *substr, size_t substr_size, |
| int result, Untaint_lang lang) const { | int result, Untaint_lang lang) const { |
| for(; result<size(); result++) { | size_t self_size=size(); |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, substr_size, result, lang); | int partial; cmp(partial, substr, substr_size, result, lang); |
| if( | if( |
| partial==0 || // full match | partial==0 || // full match |
| Line 442 void String::split(Array& result, | Line 388 void String::split(Array& result, |
| size_t* pos_after_ref, | size_t* pos_after_ref, |
| const char *delim, size_t delim_size, | const char *delim, size_t delim_size, |
| Untaint_lang lang, int limit) const { | Untaint_lang lang, int limit) const { |
| size_t self_size=size(); | |
| if(delim_size) { | if(delim_size) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | size_t pos_after=pos_after_ref?*pos_after_ref:0; |
| int pos_before; | int pos_before; |
| Line 451 void String::split(Array& result, | Line 398 void String::split(Array& result, |
| pos_after=pos_before+delim_size; | pos_after=pos_before+delim_size; |
| } | } |
| // last piece | // last piece |
| if(pos_after<size() && limit) { | if(pos_after<self_size && limit) { |
| result+=&mid(pos_after, size()); | result+=&mid(pos_after, self_size); |
| pos_after=size(); | pos_after=self_size; |
| } | } |
| if(pos_after_ref) | if(pos_after_ref) |
| *pos_after_ref=pos_after; | *pos_after_ref=pos_after; |
| } else { // empty delim | } else { // empty delim |
| result+=this; | result+=this; |
| if(pos_after_ref) | if(pos_after_ref) |
| *pos_after_ref+=size(); | *pos_after_ref+=self_size; |
| } | } |
| } | } |
| Line 468 void String::split(Array& result, | Line 415 void String::split(Array& result, |
| size_t* pos_after_ref, | size_t* pos_after_ref, |
| const String& delim, Untaint_lang lang, | const String& delim, Untaint_lang lang, |
| int limit) const { | int limit) const { |
| if(delim.size()) { | if(!delim.is_empty()) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | 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'... |
| Line 517 static void regex_options(char *options, | Line 464 static void regex_options(char *options, |
| } | } |
| /// @todo maybe need speedup: some option to remove pre/match/post string generation | /// @todo maybe need speedup: some option to remove pre/match/post string generation |
| bool String::match(const unsigned char *pcre_tables, | bool String::match( |
| const String *aorigin, | const String *aorigin, |
| const String& regexp, | const String& regexp, |
| const String *options, | const String *options, |
| Line 525 bool String::match(const unsigned char * | Line 472 bool String::match(const unsigned char * |
| Row_action row_action, void *info, | Row_action row_action, void *info, |
| bool *was_global) const { | bool *was_global) const { |
| if(!regexp.size()) | if(regexp.is_empty()) |
| throw Exception(0, 0, | throw Exception(0, |
| aorigin, | aorigin, |
| "regexp is empty"); | "regexp is empty"); |
| const char *pattern=regexp.cstr(UL_AS_IS); | const char *pattern=regexp.cstr(); |
| const char *errptr; | const char *errptr; |
| int erroffset; | int erroffset; |
| int option_bits[2]; regex_options(options?options->cstr():0, option_bits); | int option_bits[2]; regex_options(options?options->cstr():0, option_bits); |
| Line 537 bool String::match(const unsigned char * | Line 484 bool String::match(const unsigned char * |
| *was_global=option_bits[1]!=0; | *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, |
| pcre_tables); | pool().get_source_charset().pcre_tables); |
| if(!code) | if(!code) |
| throw Exception(0, 0, | throw Exception(0, |
| ®exp.mid(erroffset, regexp.size()), | ®exp.mid(erroffset, regexp.size()), |
| "regular expression syntax error - %s", errptr); | "regular expression syntax error - %s", errptr); |
| int info_substrings=pcre_info(code, 0, 0); | int info_substrings=pcre_info(code, 0, 0); |
| if(info_substrings<0) { | if(info_substrings<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, 0, | throw Exception(0, |
| aorigin, | aorigin, |
| "pcre_info error (%d)", | "pcre_info error (%d)", |
| info_substrings); | info_substrings); |
| } | } |
| int startoffset=0; | int startoffset=0; |
| const char *subject=cstr(UL_AS_IS); | const char *subject=cstr(); |
| int length=strlen(subject); | int length=strlen(subject); |
| int ovecsize; | int ovecsize; |
| int *ovector=(int *)malloc(sizeof(int)* | int *ovector=(int *)malloc(sizeof(int)* |
| Line 587 bool String::match(const unsigned char * | Line 534 bool String::match(const unsigned char * |
| if(exec_substrings<0) { | if(exec_substrings<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, 0, | throw Exception(0, |
| aorigin, | aorigin, |
| "regular expression execute error (%d)", | "regular expression execute error (%d)", |
| exec_substrings); | exec_substrings); |
| Line 619 bool String::match(const unsigned char * | Line 566 bool String::match(const unsigned char * |
| } | } |
| } | } |
| String& String::change_case(Pool& pool, const unsigned char *tables, | String& String::change_case(Pool& pool, |
| Change_case_kind kind) const { | Change_case_kind kind) const { |
| const unsigned char *tables=pool.get_source_charset().pcre_tables; | |
| String& result=*new(pool) String(pool); | String& result=*new(pool) String(pool); |
| const unsigned char *a; | const unsigned char *a; |
| Line 635 String& String::change_case(Pool& pool, | Line 583 String& String::change_case(Pool& pool, |
| b=0; | b=0; |
| break; | break; |
| default: | default: |
| throw Exception(0, 0, | throw Exception(0, |
| this, | this, |
| "unknown change case kind #%d", | "unknown change case kind #%d", |
| static_cast<int>(kind)); // never | static_cast<int>(kind)); // never |
| Line 643 String& String::change_case(Pool& pool, | Line 591 String& String::change_case(Pool& pool, |
| break; // never | break; // never |
| } | } |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | char *new_cstr=(char *)pool.malloc(row->item.size, 12); |
| const Chunk::Row *row=chunk->rows; | char *dest=new_cstr; |
| for(uint i=0; i<chunk->count; i++, row++) { | const char *src=row->item.ptr; |
| if(row==append_here) | for(int size=row->item.size; size--; src++) { |
| goto break2; | unsigned char c=a[(unsigned char)*src]; |
| if(b) | |
| c=b[c]; | |
| char *new_cstr=(char *)pool.malloc(row->item.size, 12); | *dest++=(char)c; |
| 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); | |
| } | } |
| chunk=row->link; | |
| } while(chunk); | result.APPEND(new_cstr, row->item.size, |
| break2: | row->item.lang, |
| row->item.origin.file, row->item.origin.line); | |
| ); | |
| return result; | return result; |
| } | } |
| void String::join_chain(Pool& pool, | void String::join_chain(Pool& pool, |
| uint& ai, const Chunk*& achunk, const Chunk::Row*& arow, | const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, |
| Untaint_lang& joined_lang, const char *& joined_ptr, size_t& joined_size) const { | uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const { |
| joined_lang=arow->item.lang; | joined_lang=arow->item.lang; |
| // calc size | // calc size |
| joined_size=0; | joined_size=0; |
| { | { |
| uint start_i=ai; | const Chunk* chunk=achunk; |
| const Chunk::Row *start_row=arow; | const Chunk::Row* row=arow; |
| const Chunk *chunk=achunk; | uint countdown=acountdown; |
| do { | STRING_PREPARED_FOREACH_ROW(*this, |
| const Chunk::Row *row=start_row; | if(row->item.lang==joined_lang) |
| for(uint i=start_i; i<chunk->count; i++, row++) { | joined_size+=row->item.size; |
| if(row==append_here) | else |
| goto break21; | |
| if(row->item.lang==joined_lang) | |
| joined_size+=row->item.size; | |
| else | |
| break; | |
| } | |
| if(chunk=row->link) { | |
| start_i=0; | |
| start_row=chunk->rows; | |
| } else | |
| break; | break; |
| } while(true); | ); |
| break21:; | |
| } | } |
| // if one row, return simply itself | // if one row, return simply itself |
| if(joined_size==arow->item.size) { | if(joined_size==arow->item.size) { |
| joined_ptr=arow->item.ptr; | joined_ptr=arow->item.ptr; |
| ai++; arow++; | |
| if(ai==achunk->count) | |
| achunk=arow->link; | |
| } else { | } else { |
| // join adjacent rows | // join adjacent rows |
| char *ptr=(char *)pool.malloc(joined_size,13); | char *ptr=(char *)pool.malloc(joined_size,13); |
| joined_ptr=ptr; | joined_ptr=ptr; |
| uint start_i=ai; | |
| const Chunk::Row *start_row=arow; | const Chunk* chunk=achunk; |
| const Chunk *chunk=achunk; | const Chunk::Row* row=arow; |
| uint i; | uint countdown=acountdown; |
| const Chunk::Row *row; | STRING_PREPARED_FOREACH_ROW(*this, |
| do { | if(row->item.lang==joined_lang) { |
| row=start_row; | memcpy(ptr, row->item.ptr, row->item.size); ptr+=row->item.size; |
| for(i=start_i; i<chunk->count; i++, row++) { | |
| if(row==append_here) | |
| goto break22; | |
| if(row->item.lang==joined_lang) { | |
| memcpy(ptr, row->item.ptr, row->item.size); | |
| ptr+=row->item.size; | |
| } else | |
| break; | |
| } | |
| if(chunk=row->link) { | |
| start_i=0; | |
| start_row=chunk->rows; | |
| } else | } else |
| break; | break; // before non-ours |
| } while(true); | ); |
| break22:; | |
| // return joined rows | // set pointers after joined piece |
| ai=i; | achunk=chunk; arow=row; acountdown=countdown; |
| arow=row; | // & one step back, see String::reconstruct |
| achunk=chunk; | --arow; ++acountdown; |
| } | } |
| } | } |
| String& String::reconstruct(Pool& pool) const { | String& String::reconstruct(Pool& pool) const { |
| //_asm int 3; | //_asm int 3; |
| String& result=*new(pool) String(pool); | String& result=*new(pool) String(pool); |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | uchar joined_lang; |
| const Chunk::Row *row=chunk->rows; | const char *joined_ptr; |
| for(uint i=0; i<chunk->count; ) { | size_t joined_size; |
| if(row==append_here) | IFNDEF_NO_STRING_ORIGIN( |
| goto break2; | const char *joined_origin_file=row->item.origin.file; |
| const size_t joined_origin_line=row->item.origin.line; | |
| Untaint_lang joined_lang; | ); |
| const char *joined_ptr; | join_chain(pool, chunk, row, countdown, |
| size_t joined_size; | joined_lang, joined_ptr, joined_size); |
| join_chain(pool, i, chunk, row, | |
| joined_lang, joined_ptr, joined_size); | result.APPEND(joined_ptr, joined_size, joined_lang, |
| joined_origin_file, joined_origin_line); | |
| result.APPEND(joined_ptr, joined_size, | ); |
| joined_lang, | |
| row->item.origin.file, row->item.origin.line); | |
| if(!chunk) | |
| goto break2; | |
| } | |
| } while(true); | |
| break2: | |
| return result; | return result; |
| }; | }; |
| Line 776 break2: | Line 679 break2: |
| String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const { | String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const { |
| //_asm int 3; | //_asm int 3; |
| String& result=*new(pool) String(pool); | String& result=*new(pool) String(pool); |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | const char *src=row->item.ptr; |
| const Chunk::Row *row=chunk->rows; | size_t src_size=row->item.size; |
| for(uint i=0; i<chunk->count; i++, row++) { | char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14); |
| if(row==append_here) | char *dest=new_cstr; |
| goto break2; | while(src_size) { |
| // there is a row where first column starts 'src' | |
| const char *src=row->item.ptr; | if(Table::Item *item=dict.first_that_starts(src, src_size)) { |
| size_t src_size=row->item.size; | // get a=>b values |
| char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14); | const String& a=*static_cast<Array *>(item)->get_string(0); |
| char *dest=new_cstr; | const String& b=*static_cast<Array *>(item)->get_string(1); |
| while(src_size) { | // skip 'a' in 'src' && reduce work size |
| // there is a row where first column starts 'src' | src+=a.size(); src_size-=a.size(); |
| if(Table::Item *item=dict.first_that_starts(src, src_size)) { | // write 'b' to 'dest' && skip 'b' in 'dest' |
| // get a=>b values | b.store_to(dest); dest+=b.size(); |
| const String& a=*static_cast<Array *>(item)->get_string(0); | } else { |
| const String& b=*static_cast<Array *>(item)->get_string(1); | // write a char to b && reduce work size |
| // skip 'a' in 'src' && reduce work size | *dest++=*src++; src_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--; | |
| } | |
| } | } |
| result.APPEND(new_cstr, dest-new_cstr, | |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| } | } |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | result.APPEND(new_cstr, dest-new_cstr, |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | |
| ); | |
| return result; | return result; |
| } | } |
| Line 840 double String::as_double() const { | Line 733 double String::as_double() const { |
| result=(double)strtod(cstr, &error_pos); | result=(double)strtod(cstr, &error_pos); |
| if(*error_pos/*not EOS*/) | if(*error_pos/*not EOS*/) |
| throw Exception(0, 0, | throw Exception("number.format", |
| this, | this, |
| "invalid number (double)"); | "invalid number (double)"); |
| Line 868 int String::as_int() const { | Line 761 int String::as_int() const { |
| result=(int)strtol(cstr, &error_pos, 0); | result=(int)strtol(cstr, &error_pos, 0); |
| if(*error_pos/*not EOS*/) | if(*error_pos/*not EOS*/) |
| throw Exception(0, 0, | throw Exception("number.format", |
| this, | this, |
| "invalid number (int)"); | "invalid number (int)"); |
| return result; | return result; |
| } | } |
| inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) { | |
| 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? | /* @todo maybe network order worth spending some effort? |
| don't bothering myself with network byte order, | don't bothering myself with network byte order, |
| am not planning to be able to move resulting file across platforms | am not planning to be able to move resulting file across platforms |
| Line 883 int String::as_int() const { | Line 783 int String::as_int() const { |
| void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const { | void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const { |
| buf_size= | buf_size= |
| prolog_size | prolog_size |
| +used_rows()*(sizeof(Untaint_lang)+sizeof(size_t)) | +used_rows()*(sizeof(uchar)+sizeof(ushort)) |
| +size(); | +size(); |
| buf=malloc(buf_size,15); | buf=malloc(buf_size,15); |
| char *cur=(char *)buf+prolog_size; | char *cur=(char *)buf+prolog_size; |
| const Chunk *chunk=&head; | STRING_FOREACH_ROW( |
| do { | // lang |
| const Chunk::Row *row=chunk->rows; | memcpy(cur, &row->item.lang, sizeof(uchar)); |
| for(uint i=0; i<chunk->count; i++) { | cur+=sizeof(uchar); |
| if(row==append_here) | // size |
| goto break2; | uchar byte1; uchar byte2; |
| ushort2uchars(row->item.size, byte1, byte2); | |
| // lang | memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar); |
| memcpy(cur, &row->item.lang, sizeof(Untaint_lang)); | memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar); |
| cur+=sizeof(Untaint_lang); | // bytes |
| // size | memcpy(cur, row->item.ptr, row->item.size); |
| memcpy(cur, &row->item.size, sizeof(size_t)); | cur+=row->item.size; |
| cur+=sizeof(size_t); | ); |
| // bytes | } |
| memcpy(cur, row->item.ptr, row->item.size); | bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { |
| cur+=row->item.size; | if(buf_size<=prolog_size) |
| return false; | |
| row++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| ; | |
| } | |
| /* @todo maybe network order worth spending some effort? | char *cur=(char *)buf+prolog_size; |
| don't bothering myself with network byte order, | |
| am not planning to be able to move resulting file across platforms | |
| for now | |
| */ | |
| #ifndef DOXYGEN | |
| struct Serialized_piece { | |
| String::Untaint_lang lang; | |
| size_t size; | |
| char ptr[1]; | |
| }; | |
| #endif | |
| void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { | |
| char *cur=((char *)buf)+prolog_size; | |
| buf_size-=prolog_size; | buf_size-=prolog_size; |
| while(buf_size) { | while(buf_size) { |
| Serialized_piece& p=*(Serialized_piece *)cur; | if(sizeof(uchar)+sizeof(ushort)>buf_size) // lang+size |
| APPEND(p.ptr, p.size, p.lang, file, 0); | 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); | |
| size_t piece_size=sizeof(p.lang)+sizeof(p.size)+p.size; | |
| cur+=piece_size; | cur+=piece_size; |
| buf_size-=piece_size; | buf_size-=piece_size; |
| } | } |
| return true; | |
| } | } |