--- parser3/src/main/pa_string.C 2001/01/26 15:28:39 1.2 +++ parser3/src/main/pa_string.C 2001/03/23 13:08:11 1.48 @@ -1,77 +1,347 @@ -#include +/** @file + Parser: string class. @see untaint.C. -#include "pa_pool.h" + Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) -void *String::operator new(size_t size, Pool *apool) { - return apool->alloc(size); -} + Author: Alexander Petrosyan (http://design.ru/paf) + + $Id: pa_string.C,v 1.48 2001/03/23 13:08:11 paf Exp $ +*/ -void String::construct(Pool *apool) { - pool=apool; - head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; - append_here=head.first; +#include "pa_config_includes.h" + +#include "pa_pool.h" +#include "pa_string.h" +#include "pa_hash.h" +#include "pa_exception.h" + +// String + +String::String(Pool& apool, const char *src, bool tainted) : + Pooled(apool) { + last_chunk=&head; + head.count=CR_PREALLOCATED_COUNT; + append_here=head.rows; head.preallocated_link=0; - link_row=&head.first[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(src, 0, 0, 0); } void String::expand() { - curr_chunk_rows=curr_chunk_rows*100/CR_GROW_PERCENT; - Chunk *chunk=static_cast( - pool->calloc(sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); - chunk->count=curr_chunk_rows; - link_row->link=chunk; - append_here=chunk->first; - link_row=&chunk->first[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::operator += (char *src) { - if(chunk_is_full()) - expand(); +String::String(const String& src) : Pooled(src.pool()) { + head.count=CR_PREALLOCATED_COUNT; + + int src_used_rows=src.fused_rows; + if(src_used_rows<=head.count) { + // all new rows fit into preallocated area + 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]; + } 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 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( + 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; + } 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); + break; + } + } + } + link_row->link=0; + fused_rows=src_used_rows; + fsize=src.fsize; +} - append_here->item.ptr=src; - append_here->item.size=strlen(src); - append_here++; +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; + + 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; + + 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 + } +} -size_t String::size() { - int result=0; +/*void String::change_lang(Untaint_lang lang) { Chunk *chunk=&head; do { - Chunk::Row *row=chunk->first; + Chunk::Row *row=chunk->rows; for(int i=0; icount; i++) { if(row==append_here) goto break2; - result+=row->item.size; + row->item.lang=lang; row++; } chunk=row->link; } while(chunk); break2: - return result; + return; } +*/ +String& String::real_append(STRING_APPEND_PARAMS) { + if(!src) + return *this; + if(!size) + size=strlen(src); + if(!size) + return *this; -char *String::c_str() { - char *result=static_cast(pool->alloc(size()+1)); + if(chunk_is_full()) + expand(); - char *copy_here=result; - Chunk *chunk=&head; + append_here->item.ptr=src; + fsize+=append_here->item.size=size; + append_here->item.lang=tainted?UL_YES:UL_NO; +#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 { + uint result=0; + + const Chunk *chunk=&head; do { - Chunk::Row *row=chunk->first; + const 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; + result=Hash::generic_code(result, row->item.ptr, row->item.size); row++; } chunk=row->link; } while(chunk); break2: - *copy_here=0; return result; } +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; + const Chunk::Row *b_row=b_chunk->rows; + int a_offset=0; + int b_offset=0; + Chunk::Row *a_end=append_here; + Chunk::Row *b_end=src.append_here; + int a_countdown=a_chunk->count; + 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 + 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 + 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; + 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 + result=+1; + return result; +} + +bool String::operator == (const char* b_ptr) const { + size_t b_size=b_ptr?strlen(b_ptr):0; + if(size() != b_size) + return false; + + 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) { + int size_diff= + (a_row->item.size-a_offset)- + (b_size-b_offset); + + 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) + return false; + a_row++; a_countdown--; a_offset=0; + b_break=true; + } else if (size_diff>0) { // a longer + 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; +} + +#ifndef NO_STRING_ORIGIN +const Origin& String::origin() const { + if(!fused_rows) + THROW(0, 0, + 0, + "String::origin() of empty string called"); + + return head.rows[0].item.origin; +} +#endif