--- parser3/src/main/pa_string.C 2001/03/19 15:29:40 1.44 +++ parser3/src/main/pa_string.C 2001/04/02 15:59:56 1.59 @@ -1,21 +1,24 @@ -/* - Parser +/** @file + Parser: string class. @see untasize_t.C. + Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexander Petrosyan (http://design.ru/paf) - $Id: pa_string.C,v 1.44 2001/03/19 15:29:40 paf Exp $ + $Id: pa_string.C,v 1.59 2001/04/02 15:59:56 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, const char *src, bool tainted) : +String::String(Pool& apool, const char *src, bool tasize_ted) : Pooled(apool) { last_chunk=&head; head.count=CR_PREALLOCATED_COUNT; @@ -25,16 +28,16 @@ String::String(Pool& apool, const char * fused_rows=fsize=0; if(src) - if(tainted) + if(tasize_ted) APPEND_TAINTED(src, 0, 0, 0); else - APPEND(src, 0, 0, 0); + APPEND_CONST(src); } void String::expand() { - int new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; + size_t 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 *))); + malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); last_chunk->count=new_chunk_count; link_row->link=last_chunk; append_here=last_chunk->rows; @@ -45,10 +48,10 @@ void String::expand() { String::String(const String& src) : Pooled(src.pool()) { head.count=CR_PREALLOCATED_COUNT; - int src_used_rows=src.fused_rows; + size_t 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; + // all new rows fit size_to preallocated area + size_t 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]; @@ -61,22 +64,22 @@ String::String(const String& src) : Pool // 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 + // 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 into new_chunk - int curr_chunk_rows=src_used_rows-head.count; + // remaining rows size_to new_chunk + size_t curr_chunk_rows=src_used_rows-head.count; Chunk *new_chunk=static_cast( - malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); + malloc(sizeof(size_t)+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; + size_t rows_left_to_copy=new_chunk->count; while(true) { - int old_count=old_chunk->count; + size_t old_count=old_chunk->count; Chunk *next_chunk=old_chunk->rows[old_count].link; if(next_chunk) { // not last source chunk @@ -100,32 +103,32 @@ String::String(const String& src) : Pool } 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; + size_t src_used_rows=src.fused_rows; + size_t dst_free_rows=link_row-append_here; if(src_used_rows<=dst_free_rows) { - // all new rows fit into last chunk + // all new rows fit size_to 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 + // not all new rows fit size_to last chunk: shrinking it to used part, + size_t used_rows=last_chunk->count-dst_free_rows; + //size_t *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 *))); + malloc(sizeof(size_t)+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; + size_t rows_left_to_copy=src_used_rows; while(true) { - int old_count=old_chunk->count; + size_t old_count=old_chunk->count; Chunk *next_chunk=old_chunk->rows[old_count].link; if(next_chunk) { // not last source chunk @@ -152,13 +155,13 @@ String& String::append(const String& src return *this; } void String::set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size) { - if(lang==PASS_APPENDED) + if(lang==UL_PASS_APPENDED) return; while(size--) { Untaint_lang& item_lang=(row++)->item.lang; - if(item_lang==YES || forced) // tainted? need untaint language assignment - item_lang=lang; // assign untaint language + if(item_lang==UL_YES || forced) // tasize_ted? need untasize_t language assignment + item_lang=lang; // assign untasize_t language } } @@ -166,7 +169,7 @@ void String::set_lang(Chunk::Row *row, U Chunk *chunk=&head; do { Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { + for(size_t i=0; icount; i++) { if(row==append_here) goto break2; @@ -192,7 +195,7 @@ String& String::real_append(STRING_APPEN append_here->item.ptr=src; fsize+=append_here->item.size=size; - append_here->item.lang=tainted?YES:NO; + append_here->item.lang=lang; #ifndef NO_STRING_ORIGIN append_here->item.origin.file=file; append_here->item.origin.line=line; @@ -208,7 +211,7 @@ uint String::hash_code() const { const Chunk *chunk=&head; do { const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { + for(size_t i=0; icount; i++) { if(row==append_here) goto break2; @@ -221,48 +224,61 @@ break2: return result; } -int String::cmp(const String& src) const { +int String::cmp(int& partial, const String& src, size_t this_offset) const { + partial=-1; + this_offset=min(this_offset, size()-1); + 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; + size_t a_offset=this_offset; + size_t 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; + size_t a_countdown=a_chunk->count; + size_t b_countdown=b_chunk->count; bool a_break=false; bool b_break=false; - int result; - while(true) { + size_t result; + for(size_t pos=0; true; pos+=a_row->item.size) { 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(pos+a_row->item.size > this_offset) { + 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(!b_countdown) { + b_chunk=b_row->link; + b_row=b_chunk->rows; + b_countdown=b_chunk->count; + } + } else { + a_row++; a_countdown--; a_offset-=a_row->item.size; } if(!a_countdown) { @@ -270,65 +286,146 @@ int String::cmp(const String& src) const 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; + 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 { + partial=2; return +1; + } } -bool String::operator == (const char* b_ptr) const { - size_t b_size=b_ptr?strlen(b_ptr):0; - if(size() != b_size) - return false; +int String::cmp(int& partial, const char* b_ptr, size_t src_size, + size_t this_offset) const { + partial=-1; + size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; + this_offset=min(this_offset, size()-1); const Chunk *a_chunk=&head; const Chunk::Row *a_row=a_chunk->rows; - int a_offset=0; - int b_offset=0; + size_t a_offset=this_offset; + size_t b_offset=0; Chunk::Row *a_end=append_here; - int a_countdown=a_chunk->count; + size_t 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; - } - + for(size_t pos=0; true; pos+=a_row->item.size) { a_break=a_row==a_end; if(a_break || b_break) break; + if(pos+a_row->item.size > this_offset) { + int size_diff= + (a_row->item.size-a_offset)- + (b_size-b_offset); + + if(size_diff==0) { // a has same size as b + if(size_t 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(size_t 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(size_t 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; + } + } else { + a_row++; a_countdown--; a_offset-=a_row->item.size; + } + 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 + partial=0; return 0; + } else if(a_break) { // first bytes equal, but a ended before b + partial=1; return -1; + } else { + partial=2; return +1; + } } + +#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 + +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(size_t 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; +} + +int String::pos(const String& substr, size_t result) const { + for(; result