--- parser3/src/include/pa_string.h	2001/02/21 11:10:02	1.23
+++ parser3/src/include/pa_string.h	2001/03/18 13:22:05	1.36
@@ -1,5 +1,9 @@
 /*
-  $Id: pa_string.h,v 1.23 2001/02/21 11:10:02 paf Exp $
+	Parser
+	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
+
+	$Id: pa_string.h,v 1.36 2001/03/18 13:22:05 paf Exp $
 */
 
 /*
@@ -19,7 +23,7 @@
 #define PA_STRING_H
 
 #ifdef HAVE_CONFIG_H
-#include "pa_config.h"
+#	include "pa_config.h"
 #endif
 
 #include <stddef.h>
@@ -27,52 +31,91 @@
 #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
+#	define APPEND(src, size, file, line) real_append(src, size, false, file, line)
+#	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
+#	define APPEND(src, size, file, line) real_append(src, size, false)
+#	define APPEND_TAINTED(src, size, file, line) real_append(src, size, true)
 #endif
 #define	APPEND_CONST(src) APPEND(src, 0, 0, 0)
 
-//class String_iterator;
-
 class String : public Pooled {
-//	friend String_iterator;
 public:
 	enum {
 		CR_PREALLOCATED_COUNT=5,
 		CR_GROW_PERCENT=60
 	};
 
+	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_APPEND,
+			// leave language built into string being appended
+			// just a flag, that value not stored
+		AS_IS,
+		TABLE,
+		SQL,
+		JS,
+		HTML,
+		HTML_TYPO
+	};
+
 public:
 
 	String(Pool& apool);
 	String(const String& src);
-	//String(const String_iterator& begin, const String_iterator& end);
 	size_t size() const { return fsize; }
 	int used_rows() const { return fused_rows; }
 	char *cstr() const;
 	String& real_append(STRING_APPEND_PARAMS);
-	bool operator == (const String& src) const;
-//	String& append(const String_iterator& begin, const String_iterator& end);
+	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; }
+	bool operator <= (const String& src) const { return cmp(src)<=0; }
+	bool operator >= (const String& src) const { return cmp(src)>=0; }
+	bool operator == (const String& src) const {
+		if(size()!=src.size()) // can speed up in trivial case
+			return false;
+		return cmp(src)==0;
+	}
+	bool operator != (const String& src) const { return cmp(src)!=0; }
+
+	bool operator == (const char* b_ptr) const;
+	String& append(const String& src, Untaint_lang lang);
 
 	uint hash_code() const;
 
 	const Origin& origin() const { return head.rows[0].item.origin; }
 
+	void change_lang(Untaint_lang lang);
+
 private:
 
 	struct Chunk {
 		// 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];
@@ -89,8 +132,8 @@ private:
 	Chunk::Row *link_row;
 
 private:
-	// last chank allocated count
-	int curr_chunk_rows;
+	// last chunk
+	Chunk *last_chunk;
 
 	// string size
 	size_t fsize;
@@ -104,6 +147,7 @@ private:
 		return append_here == link_row;
 	}
 	void expand();
+	void set_lang(Chunk::Row *row, Untaint_lang lang, size_t size);
 
 private: //disabled
 
@@ -111,52 +155,4 @@ private: //disabled
 
 };
 
-/*
-class Char_types {
-public:
-	Char_types();
-	void set(char from, char to, int type);
-	void set(char c, int type) { 
-		types[static_cast<unsigned int>(c)]=static_cast<char>(type); 
-	}
-	int get(char c) { 
-		return static_cast<int>(types[static_cast<unsigned int>(c)]); 
-	}
-private:	
-	char types[0x100];
-};
-
-class String_iterator {
-public:
-	String_iterator(String& astring);
-	String_iterator(String_iterator& asi);
-
-	void operator ++() { skip(); }
-	void operator ++(int) { skip(); }
-
-	int skip_to(Char_types& types);
-	bool skip_to(char c);
-
-	bool eof() { return position==0; }
-
-	// current char
-	char operator() () const;
-
-protected:
-	// home string
-	String& string;
-	// the row in which we are
-	String::Chunk::Row *read_here;
-	// position in text, eof when 0 
-	const char *position;
-	// when read_here reaches this row, move to the next chunk
-	String::Chunk::Row *link_row;
-
-protected:
-
-	// advances position by one char
-	void skip();
-};
-*/
-
 #endif