--- parser3/src/include/pa_string.h	2001/12/17 18:44:51	1.121
+++ parser3/src/include/pa_string.h	2002/04/19 08:28:35	1.136
@@ -1,10 +1,10 @@
 /** @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 1.121 2001/12/17 18:44:51 paf Exp $
+	$Id: pa_string.h,v 1.136 2002/04/19 08:28:35 paf Exp $
 */
 
 #ifndef PA_STRING_H
@@ -13,8 +13,6 @@
 #include "pa_pool.h"
 #include "pa_types.h"
 
-class Table;
-
 #ifndef NO_STRING_ORIGIN
 #	define STRING_APPEND_PARAMS \
 		const char *src, size_t size,  \
@@ -44,6 +42,7 @@ class Table;
 /// handy: appends const char* piece to String  @see String::real_append
 #define	APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
 
+class Table;
 class Array;
 class SQL_Connection;
 class Dictionary;
@@ -74,7 +73,7 @@ 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
@@ -82,15 +81,17 @@ 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
@@ -107,14 +108,15 @@ 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.chunk.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,
-		Charset *cstr_charset=0) const {
+		Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
 
 		char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
-		char *eol=store_to(result, lang, connection, cstr_charset);
+		char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
 		*eol=0;
 		return result;
 	}
@@ -122,7 +124,7 @@ public:
 	/// puts pieces to buf
 	void serialize(size_t prolog_size,  void *& buf, size_t& buf_size) const;
 	/// appends pieces from buf to self
-	void deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
+	bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
 	/** append fragment
 		@see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
 	*/
@@ -163,12 +165,6 @@ 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); }
@@ -198,7 +194,9 @@ public:
 		const String& delim, 
 		Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
 
-	typedef void (*Row_action)(Table& table, Array *row, int start, int finish, 
+	typedef void (*Row_action)(Table& table, Array *row, 
+		int prestart, int prefinish, 
+		int poststart, int postfinish,
 		void *info);
 	/**
 		@return true if fills table.
@@ -231,45 +229,54 @@ public:
 
 private:
 
-	/// several String fragments
+	/** several String fragments
+	*/
 	struct Chunk {
-		ushort count; ///< the number of rows in chunk
+		typedef uchar count_type;
+		count_type count; ///< the number of rows in chunk
+		// here could be some padding bytes
 		/// string fragment or a link to next chunk union
-		union Row {
+		typedef 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
 			} item;
+			// we are using the fact that there's no padding before this field!
 			Chunk *link;  ///< link to the next chunk in chain
-		} rows[CR_PREALLOCATED_COUNT];
-	}
-		head;  ///< the head chunk of the chunk chain
-	Chunk *head_link; ///< next rows are here
+		} rows_type[CR_PREALLOCATED_COUNT];
+		rows_type rows;
+	};
+	/**
+		'mutable' because can write after it's end, after it was appended to somebody 
+		@see String::append
+	*/
+	mutable struct {
+		Chunk chunk;
+		Chunk *link_storage;
+	} head;  ///< the head chunk of the chunk chain
 
 	/// 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,
@@ -277,13 +284,12 @@ private:
 	/// 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,
-		Charset *store_to_charset=0) const;
+		Charset *store_to_charset=0,
+		const char *store_to_charset_name=0) const;
 
-	String& reconstruct(Pool& pool) const;
 	void join_chain(Pool& pool, 
-					   uint& ai, const Chunk*& achunk, const Chunk::Row*& arow,
+					   const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
 					   uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
-	String& replace_in_reconstructed(Pool& pool, Dictionary& dict) const;
 
 private: //disabled
 
@@ -292,18 +298,22 @@ private: //disabled
 };
 #include "pa_pragma_pack_end.h"
 
-#define STRING_PREFIX_FOREACH_ROW(src, body) { \
-	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; \
-			\
-			body \
-		} \
-		chunk=row->link; \
-	} while(chunk); \
+#define STRING_PREPARED_FOREACH_ROW(self, body) \
+	while(row!=(self).append_here) { \
+		if(countdown==0) { \
+			chunk=row->link; \
+			row=chunk->rows; \
+			countdown=chunk->count; \
+		}; \
+		{ body } \
+		row++; countdown--; \
+	} 
+
+#define STRING_PREFIX_FOREACH_ROW(self, body) { \
+	const Chunk *chunk=&(self).head.chunk;  \
+	const Chunk::Row *row=chunk->rows; \
+	uint countdown=chunk->count; \
+	STRING_PREPARED_FOREACH_ROW(self, body) \
 }
 
 #define STRING_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(*this, body)