--- parser3/src/main/untaint.C	2001/10/05 16:26:05	1.64
+++ parser3/src/main/untaint.C	2001/10/29 16:29:08	1.71
@@ -4,7 +4,7 @@
 	Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com)
 	Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
 
-	$Id: untaint.C,v 1.64 2001/10/05 16:26:05 parser Exp $
+	$Id: untaint.C,v 1.71 2001/10/29 16:29:08 paf Exp $
 */
 
 #include "pa_pool.h"
@@ -15,6 +15,7 @@
 #include "pa_globals.h"
 #include "pa_sql_connection.h"
 #include "pa_dictionary.h"
+#include "pa_common.h"
 
 #define escape(action) \
 	{ \
@@ -75,13 +76,14 @@ static const char * String_Untaint_lang_
 			just a flag, that value not stored
 		*/
 	"A",     ///< leave all characters intact
-	"F", ///< filename
-	"H",    ///< text in HTTP response header
+	"F", ///< file specification
+	"H",    ///< ext in HTTP response header
 	"M",    ///< text in mail header
 	"URI",       ///< text in uri
 	"T",     ///< ^table:set body
 	"SQL",       ///< ^table:sql body
 	"JS",        ///< JavaScript code
+	"XML",		///< ^dom:set xml
 	"HTML",      ///< HTML code (for editing)
 	"UHTML", ///< HTML code with USER chars
 };
@@ -146,7 +148,15 @@ inline bool need_quote_http_header(const
 	@todo fix potential forigins_mode buf overrun
 */
 size_t String::cstr_bufsize(Untaint_lang lang) const {
-	return (lang==UL_AS_IS?size():size()*UNTAINT_TIMES_BIGGER*(forigins_mode?10:1)) +1;
+	return  (
+			lang==UL_AS_IS?
+				size()
+			:
+				size()
+				*UNTAINT_TIMES_BIGGER
+				*(forigins_mode?10:1)
+		) 
+		+1;
 }
 
 /** @todo fix theoretical \n mem overrun in TYPO replacements
@@ -159,11 +169,10 @@ char *String::store_to(char *dest, Untai
 	Dictionary *typo_dict=user_typo_dict?user_typo_dict:default_typo_dict;
 
 	bool whitespace=true;
-	bool need_to_close_http_header_quote=false;
 	const Chunk *chunk=&head; 
 	do {
 		const Chunk::Row *row=chunk->rows;
-		for(size_t i=0; i<chunk->count; i++, row++) {
+		for(uint i=0; i<chunk->count; i++, row++) {
 			if(row==append_here)
 				goto break2;
 
@@ -194,12 +203,12 @@ char *String::store_to(char *dest, Untai
 					const char *src=row->item.ptr; 
 					for(int size=row->item.size; size--; src++)
 						switch(*src) {
-						case ' ': case '\n': case '\r': case '\t':
+						/***case ' ': case '\n': case '\t':
 							if(!whitespace) {
 								*dest++=*src;
 								whitespace=true;
 							}
-							break;
+							break;*/
 						default:
 							whitespace=false;
 							*dest++=*src;
@@ -231,18 +240,11 @@ char *String::store_to(char *dest, Untai
 				});
 				break;
 			case UL_HTTP_HEADER:
-				// tainted, untaint language: http-header
-				if(need_quote_http_header(row->item.ptr, row->item.size)) {
-					*dest++='\"';
-					escape(switch(*src) {
-						case '\"': to_string("\\\"", 2);  break;
-						_default;
-					});
-					need_to_close_http_header_quote=true;
-				} else {
-					memcpy(dest, row->item.ptr, row->item.size); 
-					dest+=row->item.size;
-				}
+				// tainted, untaint language: http-field-content-text
+				escape(switch(*src) {
+					case ' ': to_char('+');  break;
+					encode(need_uri_encode, '%');
+				});
 				break;
 			case UL_MAIL_HEADER:
 				// tainted, untaint language: mail-header
@@ -281,7 +283,7 @@ char *String::store_to(char *dest, Untai
 				if(connection)
 					dest+=connection->quote(dest, row->item.ptr, row->item.size);
 				else
-					THROW(0, 0,
+					throw Exception(0, 0,
 						this,
 						"untaint in SQL language failed - no connection specified");
 				break;
@@ -317,12 +319,12 @@ char *String::store_to(char *dest, Untai
 			case UL_USER_HTML: {
 				// tainted, untaint language: html-typo
 				if(!typo_dict) // never, always has default
-					THROW(0, 0,
+					throw Exception(0, 0,
 						this,
 						"untaint to user-html lang failed, no typo table");
 
 				char *html_for_typo=
-					(char *)malloc(row->item.size*2/* '\n' -> '\' 'n' */+1);
+					(char *)malloc(row->item.size*2/* '\n' -> '\' 'n' */+1,16);
 				// note:
 				//   there still is a possibility that user 
 				//   would not replace \n as she supposed to
@@ -334,12 +336,6 @@ char *String::store_to(char *dest, Untai
 					char *dest=html_for_typo;
 					escape(switch(*src) {
 						// convinient name for typo match "\n"
-						case '\r': 
-							to_string("\\n", 2); // \r -> "\n"
-							if(size && src[1]=='\n') { // \r\n -> remove \n
-								size--; src++;
-							}
-							break;
 						case '\n': 
 							to_string("\\n", 2);
 							break;
@@ -360,7 +356,7 @@ char *String::store_to(char *dest, Untai
 						//   b allowed to be max UNTAINT_TIMES_BIGGER then a
 						if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
 							pool().set_tag(0); // avoid recursion
-							THROW(0, 0, 
+							throw Exception(0, 0, 
 								&b, 
 								"is %g times longer then '%s', "
 								"while maximum, handled by Parser, is %d", 
@@ -381,7 +377,7 @@ char *String::store_to(char *dest, Untai
 				break;
 				}
 			default:
-				THROW(0, 0, 
+				throw Exception(0, 0, 
 					this, 
 					"unknown untaint language #%d of %d piece", 
 						static_cast<int>(row->item.lang), 
@@ -396,9 +392,7 @@ char *String::store_to(char *dest, Untai
 				if(dest==dest_after_origins) // never moved==optimized space
 					dest=dest_before_origins;
 				else {
-					for(char *p=dest_after_origins; p<dest; p++)
-						if(*p=='\n')
-							*p='|';
+					remove_crlf(dest_after_origins, dest);
 
 					to_char('\n');
 				}
@@ -406,8 +400,6 @@ char *String::store_to(char *dest, Untai
 		chunk=row->link;
 	} while(chunk);
 
-	if(need_to_close_http_header_quote)
-		*dest++='\"';
 break2:
 	return dest;
 }