--- parser3/src/main/pa_string.C 2001/12/25 09:00:28 1.136 +++ parser3/src/main/pa_string.C 2002/02/20 12:35:29 1.142 @@ -1,10 +1,10 @@ /** @file Parser: string class. @see untasize_t.C. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) + Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) - $Id: pa_string.C,v 1.136 2001/12/25 09:00:28 paf Exp $ + $Id: pa_string.C,v 1.142 2002/02/20 12:35:29 paf Exp $ */ #include "pcre.h" @@ -20,13 +20,19 @@ #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) : Pooled(apool) { last_chunk=&head; head.count=CR_PREALLOCATED_COUNT; append_here=head.rows; - head_link=0; - link_row=&head.rows[head.count]; + initial_head_link=0; if(src) if(tainted) @@ -37,60 +43,12 @@ String::String(Pool& apool, const char * String::String(const String& src) : Pooled(src.pool()) { + last_chunk=&head; head.count=CR_PREALLOCATED_COUNT; - - 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( - malloc(sizeof(uint)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *), 9)); - last_chunk->count=curr_chunk_rows; - head_link=last_chunk; - append_here=link_row=&last_chunk->rows[last_chunk->count]; - - Chunk *old_chunk=src.head_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; - } 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; - src_used_rows; + append_here=head.rows; + initial_head_link=0; + + append(src, UL_UNSPECIFIED); } size_t String::size() const { @@ -112,20 +70,25 @@ break2: return result; } void String::expand() { - uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; - if(new_chunk_count>MAX_USHORT) - new_chunk_count=MAX_USHORT; - - last_chunk=static_cast( - malloc(sizeof(uint)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10)); - last_chunk->count=new_chunk_count; - link_row->link=last_chunk; + Chunk::count_type new_chunk_count=last_chunk->count+CR_GROW_COUNT; + if(new_chunk_count>max_integral(Chunk::count_type)) + new_chunk_count=max_integral(Chunk::count_type); + + Chunk *new_chunk=static_cast( + malloc(sizeof(Chunk::count_type)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10)); + new_chunk->rows[new_chunk->count=new_chunk_count].link=0; + last_chunk->rows[last_chunk->count].link=new_chunk; + + last_chunk=new_chunk; append_here=last_chunk->rows; - link_row=&last_chunk->rows[last_chunk->count]; - link_row->link=0; } String& String::real_append(STRING_APPEND_PARAMS) { + if(!last_chunk) // growth stopped [we're appended as string to somebody] + throw Exception(0, 0, + this, + "string growth stopped (append cstr)"); + if(!src) return *this; if(!size) @@ -133,12 +96,17 @@ String& String::real_append(STRING_APPEN if(!size) return *this; - while(size>MAX_USHORT) { +#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_USHORT; + 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; @@ -146,8 +114,8 @@ String& String::real_append(STRING_APPEN #endif append_here++; - src+=MAX_USHORT; - size-=MAX_USHORT; + src+=max_integral(Chunk::Row::item_size_type); + size-=max_integral(Chunk::Row::item_size_type); } if(chunk_is_full()) @@ -166,7 +134,7 @@ String& String::real_append(STRING_APPEN } char String::first_char() const { - if(!used_rows()) + if(is_empty()) throw Exception(0, 0, this, "getting first char of empty string"); @@ -204,7 +172,7 @@ int String::cmp(int& partial, const Stri size_t pos=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(pos+a_row->item.size > this_offset) { if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) @@ -346,7 +314,7 @@ int String::cmp(int& partial, const char #ifndef NO_STRING_ORIGIN const Origin& String::origin() const { - if(!used_rows()) { + if(is_empty()) { static const Origin empty_origin={"empty string"}; return empty_origin; } @@ -365,7 +333,7 @@ const Origin& String::origin() const { String& String::mid(size_t start, size_t finish) const { String& result=*NEW String(pool()); - start=max(0, start); + start=max(size_t(0), start); finish=min(size(), finish); if(start==finish) return result; @@ -374,7 +342,7 @@ String& String::mid(size_t start, size_t STRING_FOREACH_ROW( size_t item_finish=pos+row->item.size; if(item_finish > start) { // started now or already? - bool started=result.size()==0; // started now? + bool started=result.is_empty(); // 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; @@ -452,7 +420,7 @@ void String::split(Array& result, size_t* pos_after_ref, const String& delim, Untaint_lang lang, int limit) const { - if(delim.size()) { + if(!delim.is_empty()) { size_t pos_after=pos_after_ref?*pos_after_ref:0; int pos_before; // while we have 'delim'... @@ -509,7 +477,7 @@ bool String::match( Row_action row_action, void *info, bool *was_global) const { - if(!regexp.size()) + if(regexp.is_empty()) throw Exception(0, 0, aorigin, "regexp is empty"); @@ -631,7 +599,7 @@ String& String::change_case(Pool& pool, const Chunk *chunk=&head; do { const Chunk::Row *row=chunk->rows; - for(uint i=0; icount; i++, row++) { + for(Chunk::count_type i=0; icount; i++, row++) { if(row==append_here) goto break2;