--- parser3/src/main/untaint.C	2001/05/19 19:52:38	1.51
+++ parser3/src/main/untaint.C	2001/07/28 12:01:50	1.57
@@ -4,9 +4,8 @@
 	Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com)
 
 	Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
-
-	$Id: untaint.C,v 1.51 2001/05/19 19:52:38 parser Exp $
 */
+static const char *RCSId="$Id: untaint.C,v 1.57 2001/07/28 12:01:50 parser Exp $"; 
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -62,6 +61,31 @@ inline bool need_http_header_encode(unsi
 	return need_uri_encode(c);
 }
 
+//
+
+static const char * String_Untaint_lang_name[]={
+	"U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
+	"C", ///< clean
+	"T",  ///< tainted, untaint language as assigned later 
+	// untaint languages. assigned by ^untaint[lang]{...}
+	"P",
+		/**<
+			leave language built into string being appended.
+			just a flag, that value not stored
+		*/
+	"A",     ///< leave all characters intact
+	"F", ///< filename
+	"H",    ///< text in HTTP response header
+	"M",    ///< text in mail header
+	"URI",       ///< text in uri
+	"T",     ///< ^table:set body
+	"SQL",       ///< ^table:sql body
+	"JS",        ///< JavaScript code
+	"HTML",      ///< HTML code (for editing)
+	"UHTML", ///< HTML code with USER chars
+};
+
+
 // String
 
 static bool typo_present(Array::Item *value, const void *info) {
@@ -117,12 +141,16 @@ inline bool need_quote_http_header(const
 	return false;
 }
 
-/// @todo maybe additional check "are all pieces are clean?" would be profitable?
+/** @todo maybe additional check "are all pieces are clean?" would be profitable?
+	@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) +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
+/** @todo fix theoretical \n mem overrun in TYPO replacements
+@todo rename base_64 to quoted_printable [invalid name now]
+*/
 char *String::store_to(char *dest, Untaint_lang lang, 
 					   SQL_Connection *connection,
 					   const char *charset) const {
@@ -138,9 +166,27 @@ char *String::store_to(char *dest, Untai
 			if(row==append_here)
 				goto break2;
 
+			Untaint_lang to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang;
+
+			char *dest_before_origins=dest;
+
+			if(forigins_mode) {
+#ifndef NO_STRING_ORIGIN
+				if(row->item.origin.file)
+					dest+=sprintf(dest, "%s(%d)",
+						row->item.origin.file,
+						1+row->item.origin.line);
+				else
+					dest+=sprintf(dest, "unknown");
+#endif
+				dest+=sprintf(dest, "#%s: ",
+					String_Untaint_lang_name[to_lang]);
+			}
+			char *dest_after_origins=dest;
+
 			// WARNING:
 			//	string can grow only UNTAINT_TIMES_BIGGER
-			switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
+			switch(to_lang) {
 			case UL_CLEAN:
 				// clean piece
 				{ // optimizing whitespace
@@ -260,6 +306,11 @@ char *String::store_to(char *dest, Untai
 				break;
 			case UL_USER_HTML: {
 				// tainted, untaint language: html-typo
+				if(!typo_table) // never, always has default
+					THROW(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);
 				// note:
@@ -274,16 +325,13 @@ char *String::store_to(char *dest, Untai
 					escape(switch(*src) {
 						// convinient name for typo match "\n"
 						case '\r': 
-							if(typo_table) {
-								*dest++='\\';  *dest++='n'; // \r -> \n
-								if(src[1]=='\n') { // \r\n -> remove \n
-									size--; src++;
-								}
+							to_string("\\n", 2); // \r -> "\n"
+							if(size && src[1]=='\n') { // \r\n -> remove \n
+								size--; src++;
 							}
 							break;
 						case '\n': 
-							if(typo_table)
-								to_string("\\n", 2);
+							to_string("\\n", 2);
 							break;
 						//TODO: XSLT case '\'': to_string("&apos;", 6);  break;
 						_default;
@@ -340,6 +388,17 @@ char *String::store_to(char *dest, Untai
 
 			if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN)
 				whitespace=false;
+
+			if(forigins_mode)
+				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='|';
+
+					to_char('\n');
+				}
 		}
 		chunk=row->link;
 	} while(chunk);