--- parser3/src/include/pa_string.h	2001/11/19 12:17:06	1.119
+++ parser3/src/include/pa_string.h	2002/02/20 12:40:24	1.129
@@ -1,16 +1,15 @@
 /** @file
 	Parser: string class decl.
 
-	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
-	Author: Alexander Petrosyan <paf@design.ru> (http://paf.design.ru)
+	Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 
-	$Id: pa_string.h,v 1.119 2001/11/19 12:17:06 paf Exp $
+	$Id: pa_string.h,v 1.129 2002/02/20 12:40:24 paf Exp $
 */
 
 #ifndef PA_STRING_H
 #define PA_STRING_H
 
-#include "pa_config_includes.h"
 #include "pa_pool.h"
 #include "pa_types.h"
 
@@ -75,7 +74,7 @@ public:
 
 	enum {
 		CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
-		CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()
+		CR_GROW_COUNT=1 ///< each time the String chunk_is_full() string expanded()
 	};
 
 	/** piece is tainted or not. the language to use when detaint
@@ -83,15 +82,17 @@ public:
 	*/
 	enum Untaint_lang {
 		UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
+		// these two must go before others, there are checks for >UL_AS_IS
 		UL_CLEAN, ///< clean
-		UL_TAINTED,  ///< tainted, untaint language as assigned later 
-		// untaint languages. assigned by ^untaint[lang]{...}
+		UL_AS_IS,     ///< leave all characters intact
+
 		UL_PASS_APPENDED,
 			/**<
 				leave language built into string being appended.
 				just a flag, that value not stored
 			*/
-		UL_AS_IS,     ///< leave all characters intact
+		UL_TAINTED,  ///< tainted, untaint language as assigned later 
+		// untaint languages. assigned by ^untaint[lang]{...}
 		UL_FILE_SPEC, ///< file specification
 		UL_HTTP_HEADER,    ///< text in HTTP response header
 		UL_MAIL_HEADER,    ///< text in mail header
@@ -108,14 +109,15 @@ public:
 
 	String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
 	String(const String& src);
+	bool is_empty() const { return append_here==head.rows; }
 	size_t size() const;
 	/// convert to C string. if 'lang' known, forcing 'lang' to it
 	char *cstr(Untaint_lang lang=UL_AS_IS, 
 		SQL_Connection *connection=0,
-		const char *charset=0) const {
+		Charset *cstr_charset=0) const {
 
-		char *result=(char *)malloc(cstr_bufsize(lang, connection, charset));
-		char *eol=store_to(result, lang, connection, charset);
+		char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
+		char *eol=store_to(result, lang, connection, cstr_charset);
 		*eol=0;
 		return result;
 	}
@@ -164,12 +166,6 @@ public:
 		return cmp(partial, src_ptr, 0)!=0; 
 	}
 
-	/** 
-		appends other String.
-
-		marking all tainted pieces of it with @a lang.
-		or marking ALL pieces of it with a @a lang when @a forced to.
-	*/
 	String& append(const String& src, uchar lang, bool forced=false);
 	String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
 	String& operator << (const char *src) { return APPEND_CONST(src); }
@@ -208,7 +204,7 @@ public:
 			prematch/match/postmatch/1/2/3/...
 		@endverbatim
 	*/
-	bool match(const unsigned char *pcre_tables,
+	bool match(
 		const String *aorigin,		
 		const String& regexp, 
 		const String *options,
@@ -219,7 +215,7 @@ public:
 		CC_UPPER,
 		CC_LOWER
 	};
-	String& change_case(Pool& pool, const unsigned char *pcre_tables, 
+	String& change_case(Pool& pool,
 		Change_case_kind kind) const;
 	String& replace(Pool& pool, Dictionary& dict) const;
 	double as_double() const;
@@ -232,16 +228,22 @@ public:
 
 private:
 
-	/// several String fragments
-	struct Chunk {
-		ushort count; ///< the number of rows in chunk
+	/** several String fragments
+		
+		'mutable' because can write after it's end, after it was appended to somebody 
+		@see String::append
+	*/
+	mutable struct Chunk {
+		typedef uchar count_type;
+		count_type count; ///< the number of rows in chunk
 		/// string fragment or a link to next chunk union
 		union Row {
+			typedef uchar item_size_type;
 			/// fragment
 			struct { 
 				const char *ptr;  ///< pointer to the start
-				ushort size;  ///< length
-				unsigned char/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
+				item_size_type size;  ///< length
+				uchar/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
 #ifndef NO_STRING_ORIGIN
 				Origin origin;  ///< origin
 #endif
@@ -250,35 +252,32 @@ private:
 		} rows[CR_PREALLOCATED_COUNT];
 	}
 		head;  ///< the head chunk of the chunk chain
-	Chunk *head_link; ///< next rows are here
+	Chunk *initial_head_link; ///< next rows initially are here[can shrink! @see String::append]. must go right after 'head'
 
 	/// next append would write to this record
 	Chunk::Row *append_here;
 	
-	/** the address of place where lies address 
-		of the link to the next chunk to allocate
-	*/
-	Chunk::Row *link_row;
-
 private:
 	/// last chunk
-	Chunk *last_chunk;
+	mutable Chunk *last_chunk;
 
 private:
 
 	bool chunk_is_full() {
-		return append_here == link_row;
+		return append_here == last_chunk->rows+last_chunk->count;
 	}
 	uint used_rows() const;
 	void expand();
+	
+	Untaint_lang lang_of(size_t offset) const;
 
 	size_t cstr_bufsize(Untaint_lang lang,
 		SQL_Connection *connection,
-		const char *charset) const;
+		Charset *buf_charset) const;
 	/// convert to C string, store to 'dest' which must be big enough for proper untaint
 	char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
 		SQL_Connection *connection=0,
-		const char *charset=0) const;
+		Charset *store_to_charset=0) const;
 
 	String& reconstruct(Pool& pool) const;
 	void join_chain(Pool& pool,