--- parser3/src/main/untaint.C	2001/03/29 16:12:46	1.28
+++ parser3/src/main/untaint.C	2001/04/23 10:23:40	1.46
@@ -5,7 +5,7 @@
 
 	Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
 
-	$Id: untaint.C,v 1.28 2001/03/29 16:12:46 paf Exp $
+	$Id: untaint.C,v 1.46 2001/04/23 10:23:40 paf Exp $
 */
 
 #include "pa_config_includes.h"
@@ -15,6 +15,8 @@
 #include "pa_hash.h"
 #include "pa_exception.h"
 #include "pa_table.h"
+#include "pa_globals.h"
+#include "pa_sql_connection.h"
 
 #define escape(action) \
 	{ \
@@ -43,7 +45,11 @@ inline bool need_file_encode(unsigned ch
     if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) 
 		return false;
 
-    return !strchr("./", c);
+    return !strchr(
+#ifdef WIN32
+		":\\~"
+#endif
+		"./()_-", c);
 }
 inline bool need_uri_encode(unsigned char c){
     if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) 
@@ -51,7 +57,7 @@ inline bool need_uri_encode(unsigned cha
 
     return !strchr("_-./", c);
 }
-inline bool need_header_encode(unsigned char c){
+inline bool need_http_header_encode(unsigned char c){
     if(strchr(" , :", c))
 		return false;
 
@@ -71,12 +77,59 @@ static bool typo_present(Array::Item *va
 		partial==1; // typo left column starts 'src'
 }
 
-/// @todo @b now: optimize whitespaces for all but 'html'
-char *String::store_to(char *dest) const {
+/*
+
+HTTP-header    = field-name ":" [ field-value ] CRLF
+
+       field-name     = token
+       field-value    = *( field-content | LWS )
+
+       field-content  = <the OCTETs making up the field-value
+                        and consisting of either *TEXT or combinations
+                        of token, tspecials, and quoted-string>
+
+
+
+word           = token | quoted-string
+
+token          = 1*<any CHAR except CTLs or tspecials>
+
+
+
+tspecials      = "(" | ")" | "<" | ">" | "@"
+                      | "," | ";" | ":" | "\" | <">
+                      | "/" | "[" | "]" | "?" | "="
+                      | "{" | "}" | SP | HT
+
+SP             = <US-ASCII SP, space (32)>
+HT             = <US-ASCII HT, horizontal-tab (9)>
+
+LWS            = [CRLF] 1*( SP | HT )
+TEXT           = <any OCTET except CTLs,
+                        but including LWS>
+
+quoted-pair    = "\" CHAR
+
+  if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
+*/
+inline bool need_quote_http_header(const char *ptr, size_t size) {
+	for(; size--; ptr++)
+		if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
+			return true;
+	return false;
+}
+
+/**
+	@todo fix theoretical \n mem overrun in TYPO replacements
+*/
+char *String::store_to(char *dest, Untaint_lang lang, 
+					   SQL_Connection *connection,
+					   const char *charset) const {
 	// $MAIN:html-typo table
 	Table *user_typo_table=static_cast<Table *>(pool().tag());
 	Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
 
+	bool whitespace=true;
 	const Chunk *chunk=&head; 
 	do {
 		const Chunk::Row *row=chunk->rows;
@@ -86,10 +139,27 @@ char *String::store_to(char *dest) const
 
 			// WARNING:
 			//	string can grow only UNTAINT_TIMES_BIGGER
-			switch(row->item.lang) {
-			case UL_NO:
+			switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
+			case UL_CLEAN:
 				// clean piece
-			case UL_YES:
+				{ // optimizing whitespace
+					const char *src=row->item.ptr; 
+					for(int size=row->item.size; size--; src++)
+						switch(*src) {
+						case ' ': case '\n': case '\r': case '\t':
+							if(!whitespace) {
+								*dest++=*src;
+								whitespace=true;
+							}
+							break;
+						default:
+							whitespace=false;
+							*dest++=*src;
+							break;
+						}
+				}
+				break;
+			case UL_TAINTED:
 				// tainted piece, but undefined untaint language
 				// for VString.as_double of tainted values
 				// for ^process{body} evaluation
@@ -102,7 +172,7 @@ char *String::store_to(char *dest) const
 				// tainted, untaint language: file [name]
 				escape(switch(*src) {
 					case ' ': to_char('_');  break;
-					encode(need_file_encode, '-');
+					encode(need_file_encode, '+');
 				});
 				break;
 			case UL_URI:
@@ -112,11 +182,43 @@ char *String::store_to(char *dest) const
 					encode(need_uri_encode, '%');
 				});
 				break;
-			case UL_HEADER:
-				// tainted, untaint language: header
-				escape(switch(*src) {
-					encode(need_header_encode, '%');
-				});
+			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;
+					});
+					*dest++='\"';
+				} else {
+					memcpy(dest, row->item.ptr, row->item.size); 
+					dest+=row->item.size;
+				}
+				break;
+			case UL_MAIL_HEADER:
+				// tainted, untaint language: mail-header
+				if(charset) {
+					// Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
+					const char *src=row->item.ptr; 
+					bool to_base_64=false;
+					for(int size=row->item.size; size--; src++) {
+						if(*src & 0x80) {
+							if(!to_base_64) {
+								dest+=sprintf(dest, "=?%.15s?Q?", charset);
+								to_base_64=true;
+							}
+							dest+=sprintf(dest, "=%02X", *src & 0xFF);
+						} else {
+							*dest++=*src;						
+						}
+					}
+					if(to_base_64) // close
+						dest+=sprintf(dest, "?=");
+				} else {
+					memcpy(dest, row->item.ptr, row->item.size); 
+					dest+=row->item.size;
+				}
 				break;
 			case UL_TABLE: 
 				// tainted, untaint language: table
@@ -128,9 +230,12 @@ char *String::store_to(char *dest) const
 				break;
 			case UL_SQL:
 				// tainted, untaint language: sql
-				// TODO: зависимость от sql сервера
-				memset(dest, '?', row->item.size); 
-				dest+=row->item.size;
+				if(connection)
+					dest+=connection->quote(dest, row->item.ptr, row->item.size);
+				else
+					THROW(0, 0,
+						this,
+						"untaint in SQL language failed - no connection specified");
 				break;
 			case UL_JS:
 				escape(switch(*src) {
@@ -161,7 +266,6 @@ char *String::store_to(char *dest) const
 				//   and rather replace \ and n into huge strings
 				//   thus causing memory overrun
 				//   this can be dealed by allocating *2 memory, but that's too expensive
-				// todo parser4: fix that
 				size_t html_for_typo_size;
 				{ // local dest
 					char *dest=html_for_typo;
@@ -228,8 +332,11 @@ char *String::store_to(char *dest) const
 					this, 
 					"unknown untaint language #%d of %d piece", 
 						static_cast<int>(row->item.lang), 
-						i);
+						i); // never
 			}
+
+			if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN)
+				whitespace=false;
 		}
 		chunk=row->link;
 	} while(chunk);