--- parser3/src/main/pa_string.C	2001/10/19 12:43:30	1.112
+++ parser3/src/main/pa_string.C	2002/02/20 12:35:29	1.142
@@ -1,16 +1,13 @@
 /** @file
 	Parser: string class. @see untasize_t.C.
 
-	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
-	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
+	Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 
-	$Id: pa_string.C,v 1.112 2001/10/19 12:43:30 parser Exp $
+	$Id: pa_string.C,v 1.142 2002/02/20 12:35:29 paf Exp $
 */
 
-#include "pa_config_includes.h"
-
 #include "pcre.h"
-#include "internal.h"
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -21,16 +18,21 @@
 #include "pa_globals.h"
 #include "pa_table.h"
 #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),
-	forigins_mode(false) {
+	Pooled(apool) {
 	last_chunk=&head;
 	head.count=CR_PREALLOCATED_COUNT;
 	append_here=head.rows;
-	head.preallocated_link=0;
-	link_row=&head.rows[head.count];
-	fused_rows=fsize=0;
+	initial_head_link=0;
 
 	if(src)
 		if(tainted)
@@ -40,95 +42,53 @@ String::String(Pool& apool, const char *
 }
 
 String::String(const String& src) :	
-	Pooled(src.pool()),
-	forigins_mode(false) {
+	Pooled(src.pool()) {
+	last_chunk=&head;
 	head.count=CR_PREALLOCATED_COUNT;
-	
-	size_t src_used_rows=src.fused_rows;
-	if(src_used_rows<=head.count) {
-		// all new rows fit size_to preallocated area
-		last_chunk=&head;
-		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];
-	} 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
-		size_t curr_chunk_rows=src_used_rows-head.count;
-		last_chunk=static_cast<Chunk *>(
-			malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *)));
-		last_chunk->count=curr_chunk_rows;
-		head.preallocated_link=last_chunk;
-		append_here=link_row=&last_chunk->rows[last_chunk->count];
-
-		Chunk *old_chunk=src.head.preallocated_link; 
-		Chunk::Row *new_rows=last_chunk->rows;
-		size_t rows_left_to_copy=last_chunk->count;
-		while(true) {
-			size_t 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;
+	append_here=head.rows;
+	initial_head_link=0;
 
-				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(src, UL_UNSPECIFIED);
 }
 
-void String::expand() {
-	size_t new_chunk_count=last_chunk->count+CR_GROW_COUNT;
-	last_chunk=static_cast<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;
-	link_row=&last_chunk->rows[last_chunk->count];
-	link_row->link=0;
+size_t  String::size() const {
+	size_t result=0;
+	STRING_FOREACH_ROW(
+			result+=row->item.size;
+	);
+break2:
+	return result;
 }
 
-String& String::append(const String& src, Untaint_lang lang, bool forced) {
-	const Chunk *chunk=&src.head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; i++, row++) {
-			if(row==src.append_here)
-				goto break2;
-			
-			APPEND(row->item.ptr, row->item.size, 
-				(lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang,
-				row->item.origin.file, row->item.origin.line);
-		}
-		chunk=row->link;
-	} while(chunk);
+/// @todo not very optimal
+uint String::used_rows() const {
+	uint result=0;
+	STRING_FOREACH_ROW(
+		result++;
+	);
 break2:
-	return *this;
+	return result;
+}
+void String::expand() {
+	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<Chunk *>(
+		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;
 }
 
 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)
@@ -136,23 +96,45 @@ String& String::real_append(STRING_APPEN
 	if(!size)
 		return *this;
 
+#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_integral(Chunk::Row::item_size_type);
+		append_here->item.lang=lang;
+#ifndef NO_STRING_ORIGIN
+		append_here->item.origin.file=file;
+		append_here->item.origin.line=line;
+#endif
+		append_here++;
+
+		src+=max_integral(Chunk::Row::item_size_type);
+		size-=max_integral(Chunk::Row::item_size_type);
+	}
+
 	if(chunk_is_full())
 		expand();
 
 	append_here->item.ptr=src;
-	fsize+=append_here->item.size=size;
+	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++;
+	append_here++;
 
 	return *this;
 }
 
 char String::first_char() const {
-	if(!fused_rows)
+	if(is_empty())
 		throw Exception(0, 0,
 			this,
 			"getting first char of empty string");
@@ -162,19 +144,9 @@ char String::first_char() const {
 
 uint String::hash_code() const {
 	uint result=0;
-
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; i++) {
-			if(row==append_here)
-				goto break2;
-
+	STRING_FOREACH_ROW(
 			result=Hash::generic_code(result, row->item.ptr, row->item.size);
-			row++;
-		}
-		chunk=row->link;
-	} while(chunk);
+	);
 break2:
 	return result;
 }
@@ -183,7 +155,8 @@ break2:
 int String::cmp(int& partial, const String& src, 
 				size_t this_offset, Untaint_lang lang) const {
 	partial=-1;
-	this_offset=min(this_offset, size()-1);
+	size_t a_size=size();
+	this_offset=min(this_offset, a_size-1);
 
 	const Chunk *a_chunk=&head;
 	const Chunk *b_chunk=&src.head;
@@ -193,16 +166,16 @@ int String::cmp(int& partial, const Stri
 	size_t b_offset=0;
 	Chunk::Row *a_end=append_here;
 	Chunk::Row *b_end=src.append_here;
-	size_t a_countdown=a_chunk->count;
-	size_t b_countdown=b_chunk->count;
-	size_t result;
+	uint a_countdown=a_chunk->count;
+	uint b_countdown=b_chunk->count;
+	int result;
 	size_t pos=0; 
 
-	bool a_break=size()==0;
-	bool b_break=src.size()==0;
+	bool a_break=a_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) 
+			if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) 
 				return -1; // wrong lang -- bail out
 
 			int size_diff=
@@ -271,22 +244,23 @@ int String::cmp(int& partial, const Stri
 int String::cmp(int& partial, const char* b_ptr, size_t src_size, 
 				size_t this_offset, Untaint_lang lang) const {
 	partial=-1;
+	size_t a_size=size();
 	size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
-	this_offset=min(this_offset, size()-1);
+	this_offset=min(this_offset, a_size-1);
 
 	const Chunk *a_chunk=&head;
 	const Chunk::Row *a_row=a_chunk->rows;
 	size_t a_offset=this_offset;
 	size_t b_offset=0;
 	Chunk::Row *a_end=append_here;
-	size_t a_countdown=a_chunk->count;
+	uint a_countdown=a_chunk->count;
 	size_t pos=0;
 
-	bool a_break=size()==0;
+	bool a_break=a_size==0;
 	bool b_break=b_size==0;
 	if(!(a_break || b_break)) while(true) {
 		if(pos+a_row->item.size > this_offset) {
-			if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) 
+			if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) 
 				return -1; // wrong lang -- bail out
 
 			int size_diff=
@@ -294,20 +268,20 @@ int String::cmp(int& partial, const char
 				(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, 
+				if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, 
 					a_row->item.size-a_offset)!=0)
 					return result;
 				pos+=a_row->item.size;
 				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, 
+				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(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, 
+				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;
@@ -340,7 +314,7 @@ int String::cmp(int& partial, const char
 
 #ifndef NO_STRING_ORIGIN
 const Origin& String::origin() const { 
-	if(!fused_rows) {
+	if(is_empty()) {
 		static const Origin empty_origin={"empty string"};
 		return empty_origin;
 	}
@@ -359,35 +333,28 @@ 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;
 
 	size_t pos=0;
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) {
-			if(row==append_here)
+	STRING_FOREACH_ROW(
+		size_t item_finish=pos+row->item.size;
+		if(item_finish > start) { // started now or already?
+			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;
+			result.APPEND(
+				row->item.ptr+offset, size-offset, 
+				row->item.lang,
+				row->item.origin.file, row->item.origin.line);
+			if(finished)
 				goto break2;
-
-			size_t item_finish=pos+row->item.size;
-			if(item_finish > start) { // started now or already?
-				bool started=result.size()==0; // 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);
+		pos+=row->item.size;
+	);
 break2:
 //	SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
 		//cstr(), start, finish, result.cstr());
@@ -395,8 +362,9 @@ break2:
 }
 
 int String::pos(const String& substr, 
-				size_t result, Untaint_lang lang) const {
-	for(; result<size(); result++) {
+				int result, Untaint_lang lang) const {
+	size_t self_size=size();
+	for(; size_t(result)<self_size; result++) {
 		int partial; cmp(partial, substr, result, lang);
 		if(
 			partial==0 || // full match
@@ -408,8 +376,9 @@ int String::pos(const String& substr,
 }
 
 int String::pos(const char *substr, size_t substr_size, 
-				size_t result, Untaint_lang lang) const {
-	for(; result<size(); result++) {
+				int result, Untaint_lang lang) const {
+	size_t self_size=size();
+	for(; size_t(result)<self_size; result++) {
 		int partial; cmp(partial, substr, substr_size, result, lang);
 		if(
 			partial==0 || // full match
@@ -424,6 +393,7 @@ void String::split(Array& result,
 				   size_t* pos_after_ref, 
 				   const char *delim, size_t delim_size, 
 				   Untaint_lang lang, int limit) const {
+	size_t self_size=size();
 	if(delim_size) {
 		size_t pos_after=pos_after_ref?*pos_after_ref:0;
 		int pos_before;
@@ -433,16 +403,16 @@ void String::split(Array& result,
 			pos_after=pos_before+delim_size;
 		}
 		// last piece
-		if(pos_after<size() && limit) {
-			result+=&mid(pos_after, size());
-			pos_after=size();
+		if(pos_after<self_size && limit) {
+			result+=&mid(pos_after, self_size);
+			pos_after=self_size;
 		}
 		if(pos_after_ref)
 			*pos_after_ref=pos_after;
 	} else { // empty delim
 		result+=this;
 		if(pos_after_ref)
-			*pos_after_ref+=size();
+			*pos_after_ref+=self_size;
 	}
 }
 
@@ -450,7 +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'...
@@ -499,7 +469,7 @@ static void regex_options(char *options,
 }
 
 /// @todo maybe need speedup: some option to remove pre/match/post string generation
-bool String::match(const unsigned char *pcre_tables,
+bool String::match(
 				   const String *aorigin,
 				   const String& regexp, 
 				   const String *options,
@@ -507,11 +477,11 @@ bool String::match(const unsigned char *
 				   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");
-	const char *pattern=regexp.cstr(UL_AS_IS);
+	const char *pattern=regexp.cstr();
 	const char *errptr;
 	int erroffset;
     int option_bits[2];  regex_options(options?options->cstr():0, option_bits);
@@ -519,7 +489,7 @@ bool String::match(const unsigned char *
 		*was_global=option_bits[1]!=0;
 	pcre *code=pcre_compile(pattern, option_bits[0], 
 		&errptr, &erroffset,
-		pcre_tables);
+		pool().get_source_charset().pcre_tables);
 
 	if(!code)
 		throw Exception(0, 0,
@@ -536,11 +506,11 @@ bool String::match(const unsigned char *
 	}
 
 	int startoffset=0;
-	const char *subject=cstr(UL_AS_IS);
+	const char *subject=cstr();
 	int length=strlen(subject);
 	int ovecsize;
 	int *ovector=(int *)malloc(sizeof(int)*
-		(ovecsize=(1/*match*/+info_substrings)*3));
+		(ovecsize=(1/*match*/+info_substrings)*3), 11);
 
 	{ // create table
 		Array& columns=*NEW Array(pool());
@@ -601,8 +571,9 @@ bool String::match(const unsigned char *
 	}
 }
 
-String& String::change_case(Pool& pool, const unsigned char *tables, 
+String& String::change_case(Pool& pool, 
 							Change_case_kind kind) const {
+	const unsigned char *tables=pool.get_source_charset().pcre_tables;
 	String& result=*new(pool) String(pool);
 
 	const unsigned char *a;
@@ -628,11 +599,11 @@ String& String::change_case(Pool& pool,
 	const Chunk *chunk=&head; 
 	do {
 		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; i++, row++) {
+		for(Chunk::count_type i=0; i<chunk->count; i++, row++) {
 			if(row==append_here)
 				goto break2;
 
-			char *new_cstr=(char *)pool.malloc(row->item.size);
+			char *new_cstr=(char *)pool.malloc(row->item.size, 12);
 			char *dest=new_cstr;
 			const char *src=row->item.ptr; 
 			for(int size=row->item.size; size--; src++) {
@@ -655,26 +626,26 @@ break2:
 }
 
 void String::join_chain(Pool& pool, 
-					   size_t& ai, const Chunk*& achunk, const Chunk::Row*& arow,
-					   Untaint_lang& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
+					   uint& ai, const Chunk*& achunk, const Chunk::Row*& arow,
+					   uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
 	joined_lang=arow->item.lang;
 	
 	// calc size
 	joined_size=0;
 	{
-		size_t start_i=ai;
+		uint start_i=ai;
 		const Chunk::Row *start_row=arow;
 		const Chunk *chunk=achunk;
 		do {
 			const Chunk::Row *row=start_row;
-			for(size_t i=start_i; i<chunk->count; i++, row++) {
+			for(uint i=start_i; i<chunk->count; i++, row++) {
 				if(row==append_here)
 					goto break21;
 				
 				if(row->item.lang==joined_lang)
 					joined_size+=row->item.size;
 				else
-					break;
+					goto break21;
 			}
 			if(chunk=row->link) {
 				start_i=0;
@@ -689,16 +660,20 @@ break21:;
 	if(joined_size==arow->item.size) {
 		joined_ptr=arow->item.ptr;
 		ai++; arow++;
-		if(ai==achunk->count)
-			achunk=arow->link;		
+		if(ai==achunk->count) {
+			if(achunk=arow->link) {
+				ai=0;
+				arow=achunk->rows;
+			}
+		}
 	} else {
 		// join adjacent rows
-		char *ptr=(char *)pool.malloc(joined_size);
+		char *ptr=(char *)pool.malloc(joined_size,13);
 		joined_ptr=ptr;
-		size_t start_i=ai;
+		uint start_i=ai;
 		const Chunk::Row *start_row=arow;
 		const Chunk *chunk=achunk;
-		size_t i;
+		uint i;
 		const Chunk::Row *row;
 		do {
 			row=start_row;
@@ -710,7 +685,7 @@ break21:;
 					memcpy(ptr, row->item.ptr, row->item.size);
 					ptr+=row->item.size;
 				} else
-					break;
+					goto break22;
 			}
 			if(chunk=row->link) {
 				start_i=0;
@@ -731,26 +706,27 @@ String& String::reconstruct(Pool& pool)
 	//_asm int 3;
 	String& result=*new(pool) String(pool);
 	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; ) {
-			if(row==append_here)
-				goto break2;
+	const Chunk::Row *row=chunk->rows;
+	for(uint i=0; i<chunk->count; ) {
+		if(row==append_here)
+			break;
 
-			Untaint_lang joined_lang;
-			const char *joined_ptr;
-			size_t joined_size;
-			join_chain(pool, i, chunk, row,
-				joined_lang, joined_ptr, joined_size);
+		uchar joined_lang;
+		const char *joined_ptr;
+		size_t joined_size;
+#ifndef NO_STRING_ORIGIN
+		const char *joined_origin_file=row->item.origin.file;
+		const size_t joined_origin_line=row->item.origin.line;
+#endif
+		join_chain(pool, i, chunk, row,
+			joined_lang, joined_ptr, joined_size);
 
-			result.APPEND(joined_ptr, joined_size, 
-				joined_lang,
-				row->item.origin.file, row->item.origin.line);
-			if(!chunk)
-				goto break2;
-		}
-	} while(true);
-break2:
+		result.APPEND(joined_ptr, joined_size, joined_lang,
+			joined_origin_file, joined_origin_line);
+
+		if(!chunk)
+			break;
+	}
 
 	return result;
 };
@@ -758,40 +734,31 @@ break2:
 String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const {
 	//_asm int 3;
 	String& result=*new(pool) String(pool);
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; i++, row++) {
-			if(row==append_here)
-				goto break2;
-
-			const char *src=row->item.ptr; 
-			size_t src_size=row->item.size;
-			char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()));
-			char *dest=new_cstr;
-			while(src_size) {
-				// there is a row where first column starts 'src'
-				if(Table::Item *item=dict.first_that_starts(src, src_size)) {
-					// get a=>b values
-					const String& a=*static_cast<Array *>(item)->get_string(0);
-					const String& b=*static_cast<Array *>(item)->get_string(1);
-					// skip 'a' in 'src' && reduce work size
-					src+=a.size();  src_size-=a.size();
-					// write 'b' to 'dest' && skip 'b' in 'dest'
-					b.store_to(dest);  dest+=b.size();
-				} else {
-					// write a char to b && reduce work size
-					*dest++=*src++;  src_size--;
-				}
+	STRING_FOREACH_ROW(
+		const char *src=row->item.ptr; 
+		size_t src_size=row->item.size;
+		char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14);
+		char *dest=new_cstr;
+		while(src_size) {
+			// there is a row where first column starts 'src'
+			if(Table::Item *item=dict.first_that_starts(src, src_size)) {
+				// get a=>b values
+				const String& a=*static_cast<Array *>(item)->get_string(0);
+				const String& b=*static_cast<Array *>(item)->get_string(1);
+				// skip 'a' in 'src' && reduce work size
+				src+=a.size();  src_size-=a.size();
+				// write 'b' to 'dest' && skip 'b' in 'dest'
+				b.store_to(dest);  dest+=b.size();
+			} else {
+				// write a char to b && reduce work size
+				*dest++=*src++;  src_size--;
 			}
-
-			result.APPEND(new_cstr, dest-new_cstr, 
-				row->item.lang,
-				row->item.origin.file, row->item.origin.line);
 		}
-		chunk=row->link;
-	} while(chunk);
 
+		result.APPEND(new_cstr, dest-new_cstr, 
+			row->item.lang,
+			row->item.origin.file, row->item.origin.line);
+	);
 break2:
 	return result;
 }
@@ -802,7 +769,15 @@ String& String::replace(Pool& pool, Dict
 
 double String::as_double() const { 
 	double result;
-	const char *cstr=this->cstr();
+	const char *cstr;
+	char buf[MAX_NUMBER];
+	if(head.rows+1==append_here) {
+		int size=min(head.rows[0].item.size, MAX_NUMBER-1);
+		memcpy(buf, head.rows[0].item.ptr, size);
+		buf[size]=0;
+		cstr=buf;
+	} else
+		cstr=this->cstr();
 	char *error_pos;
 	// 0xABC
 	if(cstr[0]=='0')
@@ -822,7 +797,15 @@ double String::as_double() const {
 }
 int String::as_int() const { 
 	int result;
-	const char *cstr=this->cstr();
+	const char *cstr;
+	char buf[MAX_NUMBER];
+	if(head.rows+1==append_here) {
+		int size=min(head.rows[0].item.size, MAX_NUMBER-1);
+		memcpy(buf, head.rows[0].item.ptr, size);
+		buf[size]=0;
+		cstr=buf;
+	} else
+		cstr=this->cstr();
 	char *error_pos;
 	// 0xABC
 	if(cstr[0]=='0')
@@ -840,3 +823,63 @@ int String::as_int() const {
 
 	return result;
 }
+
+inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) {
+	byte1=word&0xFF;
+	byte2=word>>8;
+}
+inline ushort uchars2ushort(uchar byte1, uchar byte2) {
+	return (byte2<<8) | byte1;
+}
+/* @todo maybe network order worth spending some effort?
+	don't bothering myself with network byte order,
+	am not planning to be able to move resulting file across platforms
+	for now
+*/
+void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const {
+	buf_size=
+		prolog_size
+		+used_rows()*(sizeof(uchar)+sizeof(ushort))
+		+size();
+	buf=malloc(buf_size,15);
+	char *cur=(char *)buf+prolog_size;
+
+	STRING_FOREACH_ROW(
+		// lang
+		memcpy(cur, &row->item.lang, sizeof(uchar));
+		cur+=sizeof(uchar);
+		// size
+		uchar byte1; uchar byte2;
+		ushort2uchars(row->item.size, byte1, byte2);
+		memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar);
+		memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar);
+		// bytes
+		memcpy(cur, row->item.ptr, row->item.size);
+		cur+=row->item.size;
+	);
+break2:
+	;
+}
+void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) {
+	if(buf_size<=prolog_size)
+		return;
+
+	char *cur=(char *)buf+prolog_size;
+	buf_size-=prolog_size;
+
+	while(buf_size) {
+		uchar lang=*(uchar *)(cur);
+		
+		ushort size=uchars2ushort(
+			*(uchar*)(cur+sizeof(uchar)*1),
+			*(uchar*)(cur+sizeof(uchar)*2)
+		);
+
+		const char *ptr=(const char*)(cur+sizeof(uchar)*3); 
+		APPEND(ptr, size, lang, file, 0);
+
+		size_t piece_size=sizeof(uchar)+sizeof(ushort)+size;
+		cur+=piece_size;
+		buf_size-=piece_size;
+	}
+}