--- parser3/src/main/pa_string.C 2001/02/13 10:00:27 1.19 +++ parser3/src/main/pa_string.C 2001/03/29 15:00:22 1.53 @@ -1,43 +1,57 @@ -/* - $Id: pa_string.C,v 1.19 2001/02/13 10:00:27 paf Exp $ +/** @file + Parser: string class. @see untaint.C. + + Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) + + Author: Alexander Petrosyan (http://design.ru/paf) + + $Id: pa_string.C,v 1.53 2001/03/29 15:00:22 paf Exp $ */ -#include +#include "pa_config_includes.h" #include "pa_pool.h" #include "pa_string.h" #include "pa_hash.h" +#include "pa_exception.h" +#include "pa_common.h" // String -String::String(Pool& apool) : +String::String(Pool& apool, const char *src, bool tainted) : Pooled(apool) { - head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; + last_chunk=&head; + head.count=CR_PREALLOCATED_COUNT; append_here=head.rows; head.preallocated_link=0; - link_row=&head.rows[curr_chunk_rows]; + link_row=&head.rows[head.count]; fused_rows=fsize=0; + + if(src) + if(tainted) + APPEND_TAINTED(src, 0, 0, 0); + else + APPEND_CONST(src); } void String::expand() { - curr_chunk_rows+=curr_chunk_rows*CR_GROW_PERCENT/100; - Chunk *chunk=static_cast( - pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); - chunk->count=curr_chunk_rows; - link_row->link=chunk; - append_here=chunk->rows; - link_row=&chunk->rows[curr_chunk_rows]; + int new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; + last_chunk=static_cast( + 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(const String& src) : Pooled(src.pool()) { head.count=CR_PREALLOCATED_COUNT; - int src_used_rows=src.used_rows(); + int src_used_rows=src.fused_rows; if(src_used_rows<=head.count) { // all new rows fit into preallocated area - curr_chunk_rows=head.count; + int 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]; @@ -54,16 +68,16 @@ String::String(const String& src) : // preallocated chunk src to constructing head memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); // remaining rows into new_chunk - curr_chunk_rows=src_used_rows-head.count; + int curr_chunk_rows=src_used_rows-head.count; Chunk *new_chunk=static_cast( - pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(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[curr_chunk_rows]; + 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=curr_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; @@ -88,47 +102,107 @@ String::String(const String& src) : fsize=src.fsize; } -String& String::real_append(STRING_APPEND_PARAMS) { - if(!src) - return *this; - int len=strlen(src); - if(!len) - return *this; - - if(chunk_is_full()) - expand(); +String& String::append(const String& src, Untaint_lang lang, bool forced) { + int src_used_rows=src.fused_rows; + int dst_free_rows=link_row-append_here; + + if(src_used_rows<=dst_free_rows) { + // all new rows fit into last chunk + memcpy(append_here, src.head.rows, sizeof(Chunk::Row)*src_used_rows); + set_lang(append_here, lang, forced, src_used_rows); + append_here+=src_used_rows; + } else { + // not all new rows fit into last chunk: shrinking it to used part, + int used_rows=last_chunk->count-dst_free_rows; + //int *countp=append_here + link_row=&last_chunk->rows[last_chunk->count=used_rows]; + // allocating only enough mem to fit src string rows + // next append would allocate a new chunk + last_chunk=static_cast( + 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; - append_here->item.ptr=src; - fsize+=append_here->item.size=len; -#ifndef NO_STRING_ORIGIN - append_here->item.origin.file=file; - append_here->item.origin.line=line; -#endif - append_here++; fused_rows++; + old_chunk=next_chunk; + } else { + // the last source chunk + // taking only those rows of chunk that _left_to_copy + memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); + set_lang(new_rows, lang, forced, rows_left_to_copy); + break; + } + } + link_row->link=0; + } + fused_rows+=src_used_rows; + fsize+=src.fsize; return *this; } +void String::set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size) { + if(lang==UL_PASS_APPENDED) + return; -char *String::cstr() const { - char *result=static_cast(pool.malloc(size()+1)); + while(size--) { + Untaint_lang& item_lang=(row++)->item.lang; + if(item_lang==UL_YES || forced) // tainted? need untaint language assignment + item_lang=lang; // assign untaint language + } +} - char *copy_here=result; - const Chunk *chunk=&head; +/*void String::change_lang(Untaint_lang lang) { + Chunk *chunk=&head; do { - const Chunk::Row *row=chunk->rows; + Chunk::Row *row=chunk->rows; for(int i=0; icount; i++) { if(row==append_here) goto break2; - memcpy(copy_here, row->item.ptr, row->item.size); - copy_here+=row->item.size; + row->item.lang=lang; row++; } chunk=row->link; } while(chunk); break2: - *copy_here=0; - return result; + return; +} +*/ +String& String::real_append(STRING_APPEND_PARAMS) { + if(!src) + return *this; + if(!size) + size=strlen(src); + if(!size) + return *this; + + if(chunk_is_full()) + expand(); + + append_here->item.ptr=src; + fsize+=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++; fused_rows++; + + return *this; } uint String::hash_code() const { @@ -150,10 +224,7 @@ break2: return result; } -bool String::operator == (const String& src) const { - if(size() != src.size()) - return false; - +int String::cmp(const String& src) const { const Chunk *a_chunk=&head; const Chunk *b_chunk=&src.head; const Chunk::Row *a_row=a_chunk->rows; @@ -166,33 +237,37 @@ bool String::operator == (const String& int b_countdown=b_chunk->count; bool a_break=false; bool b_break=false; + int result; while(true) { + a_break=a_row==a_end; + b_break=b_row==b_end; + if(a_break || b_break) + break; + 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 - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) - return false; + result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset); + if(result) + return result; a_row++; a_countdown--; a_offset=0; b_row++; b_countdown--; b_offset=0; } else if (size_diff>0) { // a longer - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset)!=0) - return false; + 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 - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) - return false; + 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; } - a_break=a_row==a_end; - b_break=b_row==b_end; - if(a_break || b_break) - break; - if(!a_countdown) { a_chunk=a_row->link; a_row=a_chunk->rows; @@ -204,77 +279,111 @@ bool String::operator == (const String& b_countdown=b_chunk->count; } } - return a_break==b_break; -} - -String& String::append(const String_iterator& begin, const String_iterator& end) { - return z; -} - -// Char_set - -Char_set::Char_set() { - memset(bools, 0, sizeof(bools)); + if(a_break==b_break) // ended simultaneously + result=0; + else if(a_break) // first bytes equal, but a ended before b + result=-1; + else + result=+1; + return result; } -// String_iterator +int String::cmp(const char* b_ptr, int& partial, size_t src_size) const { + size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; - // home string - String& string; - // the row in which we are - Chunk::Row *read_here; - // position in that row's string fragment - int offset; - // when read_here reaches this row, move to the next chunk - Chunk::Row *link_row; - - bool feof; - -String_iterator::String_iterator(String& astring) : - string(astring), - offset(0), - feof(astring.size()==0), - read_here(astring.head.rows), - link_row(astring.preallocated_link) { -} - -char String_iterator::operator() { - return feof?0:read_here->item.ptr[offset]; -} + partial=-1; + const Chunk *a_chunk=&head; + const Chunk::Row *a_row=a_chunk->rows; + int a_offset=0; + int b_offset=0; + Chunk::Row *a_end=append_here; + int a_countdown=a_chunk->count; + bool a_break=false; + bool b_break=false; + while(true) { + a_break=a_row==a_end; + if(a_break || b_break) + break; -void String_iterator::skip() { - if(feof) - return; + int size_diff= + (a_row->item.size-a_offset)- + (b_size-b_offset); - if(++offset==read_here->item.size) { - if(++read_here==string.append_here) { - feof=true; - return; + 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; + 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; + a_row++; a_countdown--; a_offset=0; } - offset=0; - if(read_here==link_row) { - Chunk *chunk=link_row->link; - if(!chunk) - string.pool.exception().raise( - "String_iterator::skip() missed " - "read_here==string.append_here check"); - read_here=chunk->rows; - link_row=chunk->rows[chunk->count]; + 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 + return partial=0; + else if(a_break) // first bytes equal, but a ended before b + return partial=1; + else + return partial=2; } -bool String_iterator::skip_to(char c) { - for(; !feof; skip()) - if(operator()==c) - return true; - return false; +#ifndef NO_STRING_ORIGIN +const Origin& String::origin() const { + if(!fused_rows) + THROW(0, 0, + 0, + "String::origin() of empty string called"); + + // determining origin by last appended piece + // because first one frequently constant. + // ex: ^load[/file] "document_root" + "/file" + return append_here[-1].item.origin; } +#endif -int String_iterator::skip_to(Char_type& types) { - for(; !feof; skip()) - if(char type=types.get(operator())) - return type; - return 0; -} +String& String::piece(size_t start, size_t finish) const { + start=max(0, start); + finish=min(size(), finish); + + String& result=*NEW String(pool()); + + size_t pos=0; + const Chunk *chunk=&head; + do { + const Chunk::Row *row=chunk->rows; + for(int i=0; icount; pos+=row->item.size, i++, row++) { + if(row==append_here) + goto break2; + + if(start>=pos) { // started now or already? + size_t item_finish=pos+row->item.size; + bool started=start < item_finish; // 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; + } while(chunk); +break2: + return result; +} \ No newline at end of file