--- parser3/src/main/pa_string.C	2001/11/16 13:58:28	1.121
+++ parser3/src/main/pa_string.C	2001/12/25 09:00:28	1.136
@@ -4,13 +4,10 @@
 	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexander Petrosyan <paf@design.ru> (http://paf.design.ru)
 
-	$Id: pa_string.C,v 1.121 2001/11/16 13:58:28 paf Exp $
+	$Id: pa_string.C,v 1.136 2001/12/25 09:00:28 paf Exp $
 */
 
-#include "pa_config_includes.h"
-
 #include "pcre.h"
-#include "internal.h"
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -21,13 +18,14 @@
 #include "pa_globals.h"
 #include "pa_table.h"
 #include "pa_dictionary.h"
+#include "pa_charset.h"
 
 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.preallocated_link=0;
+	head_link=0;
 	link_row=&head.rows[head.count];
 
 	if(src)
@@ -66,10 +64,10 @@ String::String(const String& src) :
 		last_chunk=static_cast<Chunk *>(
 			malloc(sizeof(uint)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *), 9));
 		last_chunk->count=curr_chunk_rows;
-		head.preallocated_link=last_chunk;
+		head_link=last_chunk;
 		append_here=link_row=&last_chunk->rows[last_chunk->count];
 
-		Chunk *old_chunk=src.head.preallocated_link; 
+		Chunk *old_chunk=src.head_link; 
 		Chunk::Row *new_rows=last_chunk->rows;
 		uint rows_left_to_copy=last_chunk->count;
 		while(true) {
@@ -97,18 +95,9 @@ String::String(const String& src) :
 
 size_t  String::size() const {
 	size_t result=0;
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(uint i=0; i<chunk->count; i++, row++) {
-			if(row==append_here)
-				goto break2;
-
+	STRING_FOREACH_ROW(
 			result+=row->item.size;
-		}
-		chunk=row->link;
-	} while(chunk);
-
+	);
 break2:
 	return result;
 }
@@ -116,23 +105,17 @@ break2:
 /// @todo not very optimal
 uint String::used_rows() const {
 	uint result=0;
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(uint i=0; i<chunk->count; i++, row++) {
-			if(row==append_here)
-				goto break2;
-
-			result++;
-		}
-		chunk=row->link;
-	} while(chunk);
-
+	STRING_FOREACH_ROW(
+		result++;
+	);
 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<Chunk *>(
 		malloc(sizeof(uint)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10));
 	last_chunk->count=new_chunk_count;
@@ -142,24 +125,6 @@ void String::expand() {
 	link_row->link=0;
 }
 
-String& String::append(const String& src, Untaint_lang lang, bool forced) {
-	const Chunk *chunk=&src.head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(uint 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:(Untaint_lang)row->item.lang,
-				row->item.origin.file, row->item.origin.line);
-		}
-		chunk=row->link;
-	} while(chunk);
-break2:
-	return *this;
-}
-
 String& String::real_append(STRING_APPEND_PARAMS) {
 	if(!src)
 		return *this;
@@ -168,6 +133,23 @@ String& String::real_append(STRING_APPEN
 	if(!size)
 		return *this;
 
+	while(size>MAX_USHORT) {
+		if(chunk_is_full())
+			expand();
+
+		append_here->item.ptr=src;
+		append_here->item.size=MAX_USHORT;
+		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_USHORT;
+		size-=MAX_USHORT;
+	}
+
 	if(chunk_is_full())
 		expand();
 
@@ -194,19 +176,9 @@ 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(uint 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;
 }
@@ -215,7 +187,8 @@ 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;
@@ -230,11 +203,11 @@ int String::cmp(int& partial, const Stri
 	int result;
 	size_t pos=0; 
 
-	bool a_break=size()==0;
+	bool a_break=a_size==0;
 	bool b_break=src.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=
@@ -303,8 +276,9 @@ 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;
@@ -314,11 +288,11 @@ int String::cmp(int& partial, const char
 	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=
@@ -397,29 +371,22 @@ String& String::mid(size_t start, size_t
 		return result;
 
 	size_t pos=0;
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(uint 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.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;
-
-			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, 
-					(Untaint_lang)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());
@@ -428,7 +395,8 @@ break2:
 
 int String::pos(const String& substr, 
 				int result, Untaint_lang lang) const {
-	for(; result<size(); result++) {
+	size_t self_size=size();
+	for(; size_t(result)<self_size; result++) {
 		int partial; cmp(partial, substr, result, lang);
 		if(
 			partial==0 || // full match
@@ -441,7 +409,8 @@ int String::pos(const String& substr,
 
 int String::pos(const char *substr, size_t substr_size, 
 				int result, Untaint_lang lang) const {
-	for(; result<size(); result++) {
+	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
@@ -456,6 +425,7 @@ 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;
@@ -465,16 +435,16 @@ 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;
 	}
 }
 
@@ -531,7 +501,7 @@ 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,
@@ -551,7 +521,7 @@ 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,
@@ -633,8 +603,9 @@ 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;
@@ -676,7 +647,7 @@ String& String::change_case(Pool& pool,
 			}
 			
 			result.APPEND(new_cstr, row->item.size, 
-				(Untaint_lang)row->item.lang,
+				row->item.lang,
 				row->item.origin.file, row->item.origin.line);
 		}
 		chunk=row->link;
@@ -688,8 +659,8 @@ break2:
 
 void String::join_chain(Pool& pool, 
 					   uint& ai, const Chunk*& achunk, const Chunk::Row*& arow,
-					   Untaint_lang& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
-	joined_lang=(Untaint_lang)arow->item.lang;
+					   uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
+	joined_lang=arow->item.lang;
 	
 	// calc size
 	joined_size=0;
@@ -706,7 +677,7 @@ void String::join_chain(Pool& pool,
 				if(row->item.lang==joined_lang)
 					joined_size+=row->item.size;
 				else
-					break;
+					goto break21;
 			}
 			if(chunk=row->link) {
 				start_i=0;
@@ -721,8 +692,12 @@ 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,13);
@@ -742,7 +717,7 @@ break21:;
 					memcpy(ptr, row->item.ptr, row->item.size);
 					ptr+=row->item.size;
 				} else
-					break;
+					goto break22;
 			}
 			if(chunk=row->link) {
 				start_i=0;
@@ -763,26 +738,27 @@ 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(uint 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;
 };
@@ -790,40 +766,31 @@ 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(uint 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()), 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--;
-				}
+	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, 
-				(Untaint_lang)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;
 }
@@ -889,6 +856,13 @@ 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
@@ -897,60 +871,47 @@ int String::as_int() const {
 void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const {
 	buf_size=
 		prolog_size
-		+used_rows()*(sizeof(Untaint_lang)+sizeof(size_t))
+		+used_rows()*(sizeof(uchar)+sizeof(ushort))
 		+size();
 	buf=malloc(buf_size,15);
 	char *cur=(char *)buf+prolog_size;
 
-	const Chunk *chunk=&head; 
-	do {
-		const Chunk::Row *row=chunk->rows;
-		for(uint i=0; i<chunk->count; i++) {
-			if(row==append_here)
-				goto break2;
-
-			// lang
-			memcpy(cur, &row->item.lang, sizeof(Untaint_lang));
-			cur+=sizeof(Untaint_lang);
-			// size
-			memcpy(cur, &row->item.size, sizeof(size_t));
-			cur+=sizeof(size_t);
-			// bytes
-			memcpy(cur, row->item.ptr, row->item.size);
-			cur+=row->item.size;
-
-			row++;
-		}
-		chunk=row->link;
-	} while(chunk);
+	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:
 	;
 }
-
-/* @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
-*/
-#ifndef DOXYGEN
-struct Serialized_piece {
-	String::Untaint_lang lang;
-	size_t size;
-	char ptr[1];
-};
-#endif
-
 void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) {
-	char *cur=((char *)buf)+prolog_size;
+	if(buf_size<=prolog_size)
+		return;
+
+	char *cur=(char *)buf+prolog_size;
 	buf_size-=prolog_size;
 
 	while(buf_size) {
-		Serialized_piece& p=*(Serialized_piece *)cur;
-		APPEND(p.ptr, p.size, p.lang, file, 0);
+		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(p.lang)+sizeof(p.size)+p.size;
+		size_t piece_size=sizeof(uchar)+sizeof(ushort)+size;
 		cur+=piece_size;
 		buf_size-=piece_size;
 	}
 }
-