--- parser3/src/main/pa_string.C 2001/03/10 12:12:51 1.35 +++ parser3/src/main/pa_string.C 2001/03/25 08:52:36 1.52 @@ -1,8 +1,14 @@ -/* - $Id: pa_string.C,v 1.35 2001/03/10 12:12:51 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 1.52 2001/03/25 08:52:36 paf Exp $ */ -#include +#include "pa_config_includes.h" #include "pa_pool.h" #include "pa_string.h" @@ -11,7 +17,7 @@ // String -String::String(Pool& apool) : +String::String(Pool& apool, const char *src, bool tainted) : Pooled(apool) { last_chunk=&head; head.count=CR_PREALLOCATED_COUNT; @@ -19,6 +25,12 @@ String::String(Pool& apool) : head.preallocated_link=0; 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() { @@ -32,11 +44,10 @@ void String::expand() { 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 int curr_chunk_rows=head.count; @@ -90,14 +101,14 @@ String::String(const String& src) : fsize=src.fsize; } -String& String::append(const String& src, Untaint_lang lang) { - int src_used_rows=src.used_rows(); +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, 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, @@ -122,7 +133,7 @@ String& String::append(const String& src // not last source chunk // taking it all memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); - set_lang(new_rows, lang, old_count); + set_lang(new_rows, lang, forced, old_count); new_rows+=old_count; rows_left_to_copy-=old_count; @@ -131,7 +142,7 @@ String& String::append(const String& src // 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, rows_left_to_copy); + set_lang(new_rows, lang, forced, rows_left_to_copy); break; } } @@ -142,17 +153,34 @@ String& String::append(const String& src return *this; } -void String::set_lang(Chunk::Row *row, Untaint_lang lang, size_t size) { - if(lang==PASS_APPENDED) +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==YES) // tainted? need untaint language assignment + if(item_lang==UL_YES || forced) // tainted? need untaint language assignment item_lang=lang; // assign untaint language } } +/*void String::change_lang(Untaint_lang lang) { + Chunk *chunk=&head; + do { + Chunk::Row *row=chunk->rows; + for(int i=0; icount; i++) { + if(row==append_here) + goto break2; + + row->item.lang=lang; + row++; + } + chunk=row->link; + } while(chunk); +break2: + return; +} +*/ String& String::real_append(STRING_APPEND_PARAMS) { if(!src) return *this; @@ -166,7 +194,7 @@ String& String::real_append(STRING_APPEN append_here->item.ptr=src; fsize+=append_here->item.size=size; - append_here->item.lang=tainted?/*Untaint_lang::*/YES:Untaint_lang::NO; + append_here->item.lang=lang; #ifndef NO_STRING_ORIGIN append_here->item.origin.file=file; append_here->item.origin.line=line; @@ -176,43 +204,6 @@ String& String::real_append(STRING_APPEN return *this; } -char *String::cstr() const { - char *result=static_cast(malloc(size()+1)); - - char *copy_here=result; - const Chunk *chunk=&head; - do { - const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { - if(row==append_here) - goto break2; - - switch(row->item.lang) { - case NO: - case YES: // for VString.get_double of tainted values - case AS_IS: - memcpy(copy_here, row->item.ptr, row->item.size); - break; - case HTML_TYPO: - memset(copy_here, '?', row->item.size); - break; - default: - THROW(0,0, - this, - "unknown untaint language #%d of %d piece", - static_cast(row->item.lang), - i); - } - copy_here+=row->item.size; - row++; - } - chunk=row->link; - } while(chunk); -break2: - *copy_here=0; - return result; -} - uint String::hash_code() const { uint result=0; @@ -296,11 +287,10 @@ int String::cmp(const String& src) const return result; } -bool String::operator == (char* b_ptr) const { - size_t b_size=b_ptr?strlen(b_ptr):0; - if(size() != b_size) - return false; +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; + partial=-1; const Chunk *a_chunk=&head; const Chunk::Row *a_row=a_chunk->rows; int a_offset=0; @@ -310,36 +300,55 @@ bool String::operator == (char* b_ptr) c bool a_break=false; bool b_break=false; while(true) { + a_break=a_row==a_end; + if(a_break || b_break) + break; + 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; + 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(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, b_size-b_offset)!=0) - return false; + 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(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) - return false; + 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; } - 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; + 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; +} + +#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