--- parser3/src/main/untaint.C	2001/04/11 08:36:20	1.39
+++ parser3/src/main/untaint.C	2001/06/28 07:41:59	1.52
@@ -5,10 +5,9 @@
 
 	Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
 
-	$Id: untaint.C,v 1.39 2001/04/11 08:36:20 paf Exp $
+	$Id: untaint.C,v 1.52 2001/06/28 07:41:59 parser Exp $
 */
-
-#include "pa_config_includes.h"
+static char *RCSId="$Id: untaint.C,v 1.52 2001/06/28 07:41:59 parser Exp $"; 
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -77,16 +76,62 @@ static bool typo_present(Array::Item *va
 		partial==1; // typo left column starts 'src'
 }
 
-/**
-	@test optimize whitespaces for all but 'html'
-	@todo fix theoretical \n mem overrun in TYPO replacements
-	@test mail-header
+/*
+
+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))
 */
-char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection) const {
+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 maybe additional check "are all pieces are clean?" would be profitable?
+size_t String::cstr_bufsize(Untaint_lang lang) const {
+	return (lang==UL_AS_IS?size():size()*UNTAINT_TIMES_BIGGER) +1;
+}
+
+///	@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;
@@ -99,6 +144,23 @@ char *String::store_to(char *dest, Untai
 			switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
 			case UL_CLEAN:
 				// clean piece
+				{ // 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
@@ -124,14 +186,41 @@ char *String::store_to(char *dest, Untai
 				break;
 			case UL_HTTP_HEADER:
 				// tainted, untaint language: http-header
-				escape(switch(*src) {
-					encode(need_http_header_encode, '%');
-				});
+				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
-				memcpy(dest, row->item.ptr, row->item.size); 
-				dest+=row->item.size;
+				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
@@ -170,9 +259,10 @@ char *String::store_to(char *dest, Untai
 					_default;
 				});
 				break;
-			case UL_HTML_TYPO: {
+			case UL_USER_HTML: {
 				// tainted, untaint language: html-typo
-				char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
+				char *html_for_typo=
+					(char *)malloc(row->item.size*2/* '\n' -> '\' 'n' */+1);
 				// note:
 				//   there still is a possibility that user 
 				//   would not replace \n as she supposed to
@@ -225,9 +315,9 @@ char *String::store_to(char *dest, Untai
 								&b, 
 								"is %g times longer then '%s', "
 								"while maximum, handled by Parser, is %d", 
-								((double)b.size())/a.size(), 
-								a.cstr(), 
-								UNTAINT_TIMES_BIGGER);
+									((double)b.size())/a.size(), 
+									a.cstr(), 
+									UNTAINT_TIMES_BIGGER);
 						}
 						
 						// skip 'a' in 'src'
@@ -246,7 +336,11 @@ char *String::store_to(char *dest, Untai
 					"unknown untaint language #%d of %d piece", 
 						static_cast<int>(row->item.lang), 
 						i); // never
+				break; // never
 			}
+
+			if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN)
+				whitespace=false;
 		}
 		chunk=row->link;
 	} while(chunk);