--- parser3/src/main/untaint.C	2002/02/21 14:36:54	1.96
+++ parser3/src/main/untaint.C	2002/09/12 08:37:32	1.105
@@ -3,10 +3,10 @@
 
 	Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
-
-	$Id: untaint.C,v 1.96 2002/02/21 14:36:54 paf Exp $
 */
 
+static const char* IDENT_UNTAINT_C="$Date: 2002/09/12 08:37:32 $";
+
 #include "pa_pool.h"
 #include "pa_string.h"
 #include "pa_hash.h"
@@ -156,13 +156,13 @@ inline bool need_quote_http_header(const
 	if no language-change specified and src not yet appended to some other string[last_chunk!=0]
 		shrinking dest last_chunk[preparing it for linking],
 		///shrinking src last_chunk[preparing it to be linked, consequent dest.appends would go there],
-		linking[dest.last_chunk = src.head]
+		linking[dest.last_chunk = src.head.chunk]
 	if some language-change specified or src already appended to some other string[last_chunk==0]
 		cloning pieces.
 */
 String& String::append(const String& src, uchar lang, bool forced) {
 	if(!last_chunk) // growth stopped [we're appended as string to somebody]
-		throw Exception(0, 0,
+		throw Exception(0,
 			this,
 			"string growth stopped (append string)");
 
@@ -178,50 +178,43 @@ String& String::append(const String& src
 #endif
 
 		// using fact: 
-		// src.head.count initally equeals this.head.count and shrinks-only, 
-		// so can't be more than this.head.count, 
+		// src.head.chunk.count initally equeals this.head.chunk.count and shrinks-only, 
+		// so can't be more than this.head.chunk.count, 
 		// which means that we know that 
-		// src.head would fit into this.head
-		if(is_empty()) { // our head is empty
-/*			// they have more than head? we need all head : we need only filled-part of head
-			Chunk *src_head_link=src.head.rows[src.head.count].link;
-			size_t head_count=src_head_link?src.head.count:(src.append_here-src.head.rows);
-			// "your head is my head"
-			memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*(head_count));
+		// src.head.chunk would fit into this.head.chunk
+		if(is_empty()) { // our head.chunk is empty
+			// they have more than head.chunk? we need all head.chunk : we need only filled-part of head.chunk
+			Chunk *src_head_link=src.head.chunk.rows[src.head.chunk.count].link;
+			size_t head_count=src_head_link?src.head.chunk.count:(src.append_here-src.head.chunk.rows);
+			// "your head.chunk is my head.chunk"
+			memcpy(head.chunk.rows, src.head.chunk.rows, sizeof(Chunk::Row)*(head_count));
 			if(src_head_link) {
 				// "your body is my body"
-				head.rows[head.count=head_count].link=src_head_link;
+				head.chunk.rows[head.chunk.count=head_count].link=src_head_link;
 				// "your last_chunk is mine now"
 				last_chunk=src.last_chunk;
 				// "your append_here is mine now"
 				append_here=src.append_here;
 			} else {
 				// "your last_chunk is mine now"
-				last_chunk=&head;
+				last_chunk=&head.chunk;
 				// "your append_here is recalc-mine now"
-				append_here=head.rows+head_count;
-			}*/
-		} else { // our head contains something
-			/*
-			for(Chunk::Row *row=src.last_chunk->rows; row<src.append_here; row++)
-				if(row->link==(void*)0xcdcdcdcd)
-					_asm int 3;
+				append_here=head.chunk.rows+head_count;
+			}
+		} else { // our head.chunk contains something
 			// "chopping off my tail-reserve"
 			last_chunk->count=append_here-last_chunk->rows;
 			// "you is my tail"
-			append_here->link=&src.head;
+			append_here->link=&src.head.chunk;
 			// "your last_chunk is mine now"
 			last_chunk=src.last_chunk;
 			// "your append_here is mine now"
 			append_here=src.append_here;
-		// stop-growing mark
-		src.last_chunk=0;
-		return *this;*/
 		}
 
 		// stop-growing mark
-//		src.last_chunk=0;
-//		return *this;
+		src.last_chunk=0;
+		return *this;
 	}
 
 	// manually unrolled code to avoid do{if(const)} constructs
@@ -255,10 +248,10 @@ String& String::append(const String& src
 				row->item.lang==UL_TAINTED?lang:row->item.lang,
 				row->item.origin.file, row->item.origin.line);
 		);
-
+/*
 			for(Chunk::Row *row=last_chunk->rows; row<append_here; row++)
 				if(row->link==(void*)0xcdcdcdcd)
-					_asm int 3;
+					_asm int 3;*/
 	return *this;
 }
 
@@ -347,7 +340,8 @@ size_t String::cstr_bufsize(Untaint_lang
 
 char *String::store_to(char *dest, Untaint_lang lang, 
 					   SQL_Connection *connection,
-					   Charset *store_to_charset) const {
+					   Charset *store_to_charset,
+					   const char *store_to_charset_name) const {
 	// WARNING:
 	//	 before any changes check cstr_bufsize first!!!
 	bool whitespace=true;
@@ -401,7 +395,8 @@ char *String::store_to(char *dest, Untai
 			break;
 		case UL_MAIL_HEADER:
 			// tainted, untaint language: mail-header
-			if(store_to_charset) {
+			// http://www.ietf.org/rfc/rfc2047.txt
+			if(store_to_charset && store_to_charset_name) {
 				const void *mail_ptr;
 				size_t mail_size;
 				Charset::transcode(pool(), 
@@ -411,18 +406,45 @@ char *String::store_to(char *dest, Untai
 				// Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
 				const char *src=(const char *)mail_ptr;
 				bool to_quoted_printable=false;
+
+				//RFC   + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
+				const char *tail=src+mail_size;
+				if(*--tail=='>') {
+					for(int size=mail_size-1; size--; tail--)
+						if(*tail=='<')
+							break;
+				}
+				const char *stop=*tail--=='<'?tail:0;
+
+				bool closed=false;
 				for(int size=mail_size; size--; src++) {
-					if(*src & 0x80) {
+/*RFC
+(3) 8-bit values which correspond to printable ASCII characters other
+   than "=", "?", and "_" (underscore), MAY be represented as those
+   characters.  (But see section 5 for restrictions.)  In
+   particular, SPACE and TAB MUST NOT be represented as themselves
+   within encoded words.
+*/
+					if(src==stop) {
+						dest+=sprintf(dest, "?=");
+						closed=true;
+					}
+					if((!stop || src<stop) && (
+						(*src & 0x80)  // starting quote-printable-encoding on first 8bit char
+						|| (to_quoted_printable 
+							&& (*src==' ' || *src=='=' || *src=='?') || *src=='_')
+						)) {
 						if(!to_quoted_printable) {
-							dest+=sprintf(dest, "=?%s?Q?", store_to_charset->name().cstr());
+							dest+=sprintf(dest, "=?%s?Q?", store_to_charset_name);
 							to_quoted_printable=true;
+
 						}
+						//RFC Upper case should be used for hexadecimal digits "A" through "F"
 						dest+=sprintf(dest, "=%02X", *src & 0xFF);
-					} else {
+					} else
 						*dest++=*src;						
-					}
 				}
-				if(to_quoted_printable) // close
+				if(to_quoted_printable && !closed) // close
 					dest+=sprintf(dest, "?=");
 			
 			} else {
@@ -443,7 +465,7 @@ char *String::store_to(char *dest, Untai
 			if(connection)
 				dest+=connection->quote(dest, row->item.ptr, row->item.size);
 			else
-				throw Exception(0, 0,
+				throw Exception(0,
 					this,
 					"untaint in SQL language failed - no connection specified");
 			break;
@@ -477,7 +499,7 @@ char *String::store_to(char *dest, Untai
 			});
 			break;
 		default:
-			throw Exception(0, 0, 
+			throw Exception(0, 
 				this, 
 				"unknown untaint language #%d", 
 					static_cast<int>(row->item.lang)); // sould never
@@ -524,7 +546,7 @@ IFNDEF_NO_STRING_ORIGIN(
 );
 		uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT;
 		if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
-			throw Exception(0, 0, 
+			throw Exception(0, 
 				this, 
 				"unknown untaint language #%d", 
 					static_cast<int>(show_lang)); // sould never