--- parser3/src/include/pa_string.h	2001/03/07 09:29:53	1.26
+++ parser3/src/include/pa_string.h	2001/03/19 20:46:36	1.45
@@ -1,25 +1,18 @@
-/*
-  $Id: pa_string.h,v 1.26 2001/03/07 09:29:53 paf Exp $
-*/
+/** @file
+	Parser: string class decl.
 
-/*
+	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
 
-	String				Chunk0
-	======				========
-	head--------------->[ptr, size, ...]
-	append_here-------->[ptr, size, ...]
-						.
-						.
-						[ptr, size, ...]
-	link_row----------->[link to the next chunk]
+	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
 
+	$Id: pa_string.h,v 1.45 2001/03/19 20:46:36 paf Exp $
 */
 
 #ifndef PA_STRING_H
 #define PA_STRING_H
 
 #ifdef HAVE_CONFIG_H
-#include "pa_config.h"
+#	include "pa_config.h"
 #endif
 
 #include <stddef.h>
@@ -27,30 +20,89 @@
 #include "pa_pool.h"
 #include "pa_types.h"
 
+#define UNTAINT_TIMES_BIGGER 10
+
 #ifndef NO_STRING_ORIGIN
-#	define STRING_APPEND_PARAMS const char *src, size_t size, char *file, uint line
-#	define APPEND(src, size, file, line) real_append(src, size, file, line)
+#	define STRING_APPEND_PARAMS \
+		const char *src, size_t size,  \
+		bool tainted, \
+		const char *file, uint line
+/// appends clean piece to String
+#	define APPEND(src, size, file, line) real_append(src, size, false, file, line)
+/// appends tainted piece to String
+#	define APPEND_TAINTED(src, size, file, line) real_append(src, size, true, file, line)
 #else
-#	define STRING_APPEND_PARAMS const char *src, size_t size
-#	define APPEND(src, size, file, line) real_append(src, size)
+#	define STRING_APPEND_PARAMS \
+		const char *src, \
+		size_t size, \
+		bool tainted
+/// appends clean piece to String
+#	define APPEND(src, size, file, line) real_append(src, size, false)
+/// appends tainted piece to String
+#	define APPEND_TAINTED(src, size, file, line) real_append(src, size, true)
 #endif
+/// handy: appends const char* piece to String
 #define	APPEND_CONST(src) APPEND(src, 0, 0, 0)
 
+/** 
+	Pooled string.
+
+	Internal structure: @verbatim	
+
+	String				Chunk0
+	======				========
+	head--------------->[ptr, size, ...]
+	append_here-------->[ptr, size, ...]
+						.
+						.
+						[ptr, size, ...]
+	link_row----------->[link to the next chunk]
+
+	@endverbatim
+
+	All pieces remember 
+	- the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
+	- whether they are tainted or not, 
+	  and the language which should be used to detaint them
+*/
 class String : public Pooled {
 public:
 	enum {
-		CR_PREALLOCATED_COUNT=5,
-		CR_GROW_PERCENT=60
+		CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
+		CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
+	};
+
+	/// piece is tainted or not. the language to use when detaint 
+	enum Untaint_lang {
+		UNKNOWN=0, ///< when get by name fails
+		NO, ///< clean
+		YES,  ///< tainted, untaint language as assigned later 
+		// untaint languages. assigned by ^untaint[lang]{...}
+		PASS_APPENDED,
+			/**<
+				leave language built into string being appended.
+				just a flag, that value not stored
+			*/
+		AS_IS,  ///< leave all characters intact
+		FILE,   ///< filename
+		HEADER, ///< text in response header
+		URI,    ///< text in uri
+		TABLE,  ///< ^table:set body
+		SQL,    ///< ^table:sql body
+		JS,     ///< JavaScript code
+		HTML,   ///< HTML code (for editing)
+		HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing)
 	};
 
 public:
 
-	String(Pool& apool);
+	String(Pool& apool, const char *src=0, bool tainted=false);
 	String(const String& src);
 	size_t size() const { return fsize; }
-	int used_rows() const { return fused_rows; }
+	/// convert to C string
 	char *cstr() const;
 	String& real_append(STRING_APPEND_PARAMS);
+	/// @return <0 ==0 or >0 depending on comparison result
 	int cmp (const String& src) const;
 	bool operator < (const String& src) const {	return cmp(src)<0; }
 	bool operator > (const String& src) const {	return cmp(src)>0; }
@@ -63,12 +115,22 @@ public:
 	}
 	bool operator != (const String& src) const { return cmp(src)!=0; }
 
-	bool operator == (char* src) const;
-	String& operator += (const String& src);
+	bool operator == (const char* b_ptr) const;
+	/** 
+		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, Untaint_lang lang, bool forced=false);
 
+	/// simple hash code of string. used by Hash
 	uint hash_code() const;
 
-	const Origin& origin() const { return head.rows[0].item.origin; }
+#ifndef NO_STRING_ORIGIN
+	/// origin of string. calculated by first row
+	const Origin& origin() const;
+#endif
 
 private:
 
@@ -76,11 +138,14 @@ private:
 		// the number of rows in chunk
 		int count;
 		union Row {
-			// chunk item
-			struct {
-				const char *ptr;  // pointer to the start of string fragment
-				size_t size;  // length of the fragment
-				Origin origin;  // origin of this fragment
+			// fragment
+			struct { 
+				const char *ptr;  // pointer to the start
+				size_t size;  // length
+				Untaint_lang lang; // untaint flag, later untaint language
+#ifndef NO_STRING_ORIGIN
+				Origin origin;  // origin
+#endif
 			} item;
 			Chunk *link;  // link to the next chunk in chain
 		} rows[CR_PREALLOCATED_COUNT];
@@ -112,6 +177,7 @@ private:
 		return append_here == link_row;
 	}
 	void expand();
+	void set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size);
 
 private: //disabled