--- parser3/src/main/pa_string.C	2001/11/21 08:26:55	1.124
+++ parser3/src/main/pa_string.C	2001/12/24 07:25:43	1.133
@@ -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 1.124 2001/11/21 08:26:55 paf Exp $
+	$Id: pa_string.C,v 1.133 2001/12/24 07:25:43 paf Exp $
 */
 
-#include "pa_config_includes.h"
-
 #include "pcre.h"
-#include "internal.h"
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -21,6 +18,7 @@
 #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) {
@@ -189,7 +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;
@@ -204,7 +203,7 @@ 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) {
@@ -277,8 +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;
@@ -288,7 +288,7 @@ 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) {
@@ -395,7 +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
@@ -408,7 +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
@@ -423,7 +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 {
-	int self_size=size();
+	size_t self_size=size();
 	if(delim_size) {
 		size_t pos_after=pos_after_ref?*pos_after_ref:0;
 		int pos_before;
@@ -499,7 +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,
@@ -519,7 +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,
@@ -601,8 +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;
@@ -674,7 +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;
@@ -689,8 +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=chunk->rows;
+			}
+		}
 	} else {
 		// join adjacent rows
 		char *ptr=(char *)pool.malloc(joined_size,13);
@@ -710,7 +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;
@@ -731,26 +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;
 
-			uchar 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;
 };
@@ -848,6 +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
@@ -856,18 +871,20 @@ 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;
 
 	STRING_FOREACH_ROW(
 		// lang
-		memcpy(cur, &row->item.lang, sizeof(Untaint_lang));
-		cur+=sizeof(Untaint_lang);
+		memcpy(cur, &row->item.lang, sizeof(uchar));
+		cur+=sizeof(uchar);
 		// size
-		memcpy(cur, &row->item.size, sizeof(size_t));
-		cur+=sizeof(size_t);
+		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;
@@ -875,31 +892,23 @@ void String::serialize(size_t prolog_siz
 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;
+	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)
+		);
 
-		size_t piece_size=sizeof(p.lang)+sizeof(p.size)+p.size;
+		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;
 	}
 }
-