--- parser3/src/include/pa_string.h	2001/07/07 17:59:02	1.97
+++ parser3/src/include/pa_string.h	2001/11/05 11:46:25	1.115
@@ -2,10 +2,9 @@
 	Parser: string class decl.
 
 	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexander Petrosyan <paf@design.ru> (http://paf.design.ru)
 
-	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
-
-	$Id: pa_string.h,v 1.97 2001/07/07 17:59:02 parser Exp $
+	$Id: pa_string.h,v 1.115 2001/11/05 11:46:25 paf Exp $
 */
 
 #ifndef PA_STRING_H
@@ -57,6 +56,7 @@ class Table;
 
 class Array;
 class SQL_Connection;
+class Dictionary;
 
 /** 
 	Pooled string.
@@ -78,6 +78,7 @@ class SQL_Connection;
 	- whether they are tainted or not, 
 	  and the language which should be used to detaint them
 */
+#include "pa_pragma_pack_begin.h"
 class String : public Pooled {
 public:
 
@@ -86,7 +87,9 @@ public:
 		CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()
 	};
 
-	/// piece is tainted or not. the language to use when detaint 
+	/** piece is tainted or not. the language to use when detaint
+		remember to change String_Untaint_lang_name @ untaint.C along
+	*/
 	enum Untaint_lang {
 		UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
 		UL_CLEAN, ///< clean
@@ -98,13 +101,14 @@ public:
 				just a flag, that value not stored
 			*/
 		UL_AS_IS,     ///< leave all characters intact
-		UL_FILE_NAME, ///< filename
+		UL_FILE_SPEC, ///< file specification
 		UL_HTTP_HEADER,    ///< text in HTTP response header
 		UL_MAIL_HEADER,    ///< text in mail header
 		UL_URI,       ///< text in uri
 		UL_TABLE,     ///< ^table:set body
 		UL_SQL,       ///< ^table:sql body
 		UL_JS,        ///< JavaScript code
+		UL_XML,		///< ^dom:set xml
 		UL_HTML,      ///< HTML code (for editing)
 		UL_USER_HTML  ///< HTML code with USER chars
 	};
@@ -114,8 +118,10 @@ public:
 	String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
 	String(const String& src);
 	size_t size() const { return fsize; }
+	/// makes store_to produce origin:text\n output
+	void set_origins_mode(bool aorigins_mode) const { forigins_mode=aorigins_mode; }
 	/// convert to C string. if 'lang' known, forcing 'lang' to it
-	char *cstr(Untaint_lang lang=UL_UNSPECIFIED, 
+	char *cstr(Untaint_lang lang=UL_AS_IS, 
 		SQL_Connection *connection=0,
 		const char *charset=0) const {
 
@@ -124,6 +130,10 @@ public:
 		*eol=0;
 		return result;
 	}
+	/// 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);
 	/** append fragment
 		@see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
 	*/
@@ -164,7 +174,6 @@ public:
 		return cmp(partial, src_ptr, 0)!=0; 
 	}
 
-
 	/** 
 		appends other String.
 
@@ -178,15 +187,18 @@ public:
 	/// simple hash code of string. used by Hash
 	uint hash_code() const;
 
+	/// extracts first char of a string
+	char first_char() const;
+
 	/// extracts [start, finish) piece of string
 	String& mid(size_t start, size_t finish) const;
 
 	/// @return position of substr in string, -1 means "not found" [String version]
 	int pos(const String& substr, 
-		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
+		int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
 	/// @return position of substr in string, -1 means "not found" [const char* version]
 	int pos(const char *substr, size_t substr_size=0, 
-		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
+		int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
 
 	void split(Array& result, 
 		size_t *pos_after_ref, 
@@ -211,13 +223,15 @@ public:
 		const String& regexp, 
 		const String *options,
 		Table **table,
-		Row_action row_action, void *info) const;
+		Row_action row_action, void *info,
+		bool *was_global=0) const;
 	enum Change_case_kind {
 		CC_UPPER,
 		CC_LOWER
 	};
 	String& change_case(Pool& pool, const unsigned char *pcre_tables, 
 		Change_case_kind kind) const;
+	String& replace(Pool& pool, Dictionary& dict) const;
 	double as_double() const;
 	int as_int() const;
 
@@ -230,14 +244,14 @@ private:
 
 	/// several String fragments
 	struct Chunk {
-		size_t count; ///< the number of rows in chunk
+		uint count; ///< the number of rows in chunk
 		/// string fragment or a link to next chunk union
 		union Row {
 			/// fragment
 			struct { 
 				const char *ptr;  ///< pointer to the start
 				size_t size;  ///< length
-				Untaint_lang lang; ///< untaint flag, later untaint language
+				unsigned char/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
 #ifndef NO_STRING_ORIGIN
 				Origin origin;  ///< origin
 #endif
@@ -257,20 +271,21 @@ private:
 	Chunk::Row *link_row;
 
 private:
-	// last chunk
+	/// last chunk
 	Chunk *last_chunk;
 
-	// string size
+	/// string size
 	size_t fsize;
 
-	// used rows in all chunks
-	int fused_rows;
+	/// origins_mode changes behaviour of store_to[cstr()]
+	mutable bool forigins_mode;
 
 private:
 
 	bool chunk_is_full() {
 		return append_here == link_row;
 	}
+	uint used_rows() const;
 	void expand();
 
 	size_t cstr_bufsize(Untaint_lang lang) const;
@@ -279,10 +294,17 @@ private:
 		SQL_Connection *connection=0,
 		const char *charset=0) const;
 
+	String& reconstruct(Pool& pool) const;
+	void 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;
+	String& replace_in_reconstructed(Pool& pool, Dictionary& dict) const;
+
 private: //disabled
 
 	String& operator = (const String&) { return *this; }
 
 };
+#include "pa_pragma_pack_end.h"
 
 #endif