--- parser3/src/main/untaint.C 2002/02/08 08:32:34 1.90 +++ parser3/src/main/untaint.C 2002/09/16 07:08:49 1.107 @@ -3,10 +3,10 @@ Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) - - $Id: untaint.C,v 1.90 2002/02/08 08:32:34 paf Exp $ */ +static const char* IDENT_UNTAINT_C="$Date: 2002/09/16 07:08:49 $"; + #include "pa_pool.h" #include "pa_string.h" #include "pa_hash.h" @@ -18,6 +18,12 @@ #include "pa_common.h" #include "pa_charset.h" +//#define DEBUG_STRING_APPENDS_VS_EXPANDS + +#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS +ulong string_string_shortcut_economy=0; +#endif + #define escape(action) \ { \ const char *src=row->item.ptr; \ @@ -138,7 +144,79 @@ inline bool need_quote_http_header(const return false; } +//#include "pa_sapi.h" +/** + appends other String, + marking all tainted pieces of it with @a lang. + or marking ALL pieces of it with a @a lang when @a forced to, + and propagating OPTIMIZE language bit. + + using architecture advantage: after string-to-string-append string never modified. + algorithm: + 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.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, + this, + "string growth stopped (append string)"); + + if(src.is_empty()) + return *this; + + // without language-chage, not-appended-before, big[not fitting our tail] string? + if(lang==UL_PASS_APPENDED + && src.last_chunk + && (uint(&last_chunk->rows[last_chunk->count]-append_here) < src.used_rows())) { +#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS + string_string_shortcut_economy+=src.used_rows()*sizeof(String::Chunk::Row); +#endif + + // using fact: + // 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.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.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.chunk; + // "your append_here is recalc-mine now" + 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.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; + } + // manually unrolled code to avoid do{if(const)} constructs if(forced) STRING_SRC_FOREACH_ROW( @@ -170,7 +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); ); -break2: +/* + for(Chunk::Row *row=last_chunk->rows; rowlink==(void*)0xcdcdcdcd) + _asm int 3;*/ return *this; } @@ -254,24 +335,17 @@ size_t String::cstr_bufsize(Untaint_lang break; } ); -break2: return dest; } 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; - // expanded STRING_FOREACH_ROW here for debugging purposes - const Chunk *chunk=&head; \ - do { \ - const Chunk::Row *row=chunk->rows; \ - for(uint i=0; icount; i++, row++) { \ - if(row==append_here) \ - goto break2; \ - \ + STRING_FOREACH_ROW( uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang; char *start=dest; @@ -321,7 +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(), @@ -331,18 +406,48 @@ 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; + while(stop>src && stop[-1]==' ') + --stop; + + 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 && to_quoted_printable) { + dest+=sprintf(dest, "?="); + closed=true; + to_quoted_printable=false; + } + if((!stop || srcname().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 { @@ -363,7 +468,7 @@ 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; @@ -397,7 +502,7 @@ char *String::store_to(char *dest, Untai }); break; default: - throw Exception(0, 0, + throw Exception(0, this, "unknown untaint language #%d", static_cast(row->item.lang)); // sould never @@ -423,12 +528,8 @@ char *String::store_to(char *dest, Untai }; } else // piece without optimization whitespace=false; + ); - } \ - chunk=row->link; \ - } while(chunk); \ - -break2: return dest; } @@ -437,43 +538,34 @@ char *String::cstr_debug_origins() const char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2); char *dest=result; - const Chunk *chunk=&head; - do { - const Chunk::Row *row=chunk->rows; - for(uint i=0; icount; i++, row++) { - if(row==append_here) - goto break2; - -#ifndef NO_STRING_ORIGIN - if(row->item.origin.file) - dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, - row->item.origin.file, - 1+row->item.origin.line); - else - dest+=sprintf(dest, ""); -#endif - 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, - this, - "unknown untaint language #%d", - static_cast(show_lang)); // sould never - - dest+=sprintf(dest, "#%s%s: ", - String_Untaint_lang_name[show_lang], - row->item.lang & UL_OPTIMIZE_BIT?".O":""); - char *dest_after_origins=dest; + STRING_FOREACH_ROW( +IFNDEF_NO_STRING_ORIGIN( + if(row->item.origin.file) + dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, + row->item.origin.file, + 1+row->item.origin.line); + else + dest+=sprintf(dest, ""); +); + 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, + this, + "unknown untaint language #%d", + static_cast(show_lang)); // sould never - memcpy(dest, row->item.ptr, row->item.size); - dest+=row->item.size; + dest+=sprintf(dest, "#%s%s: ", + String_Untaint_lang_name[show_lang], + row->item.lang & UL_OPTIMIZE_BIT?".O":""); + char *dest_after_origins=dest; - remove_crlf(dest_after_origins, dest); - to_char('\n'); - } - chunk=row->link; - } while(chunk); + memcpy(dest, row->item.ptr, row->item.size); + dest+=row->item.size; + + remove_crlf(dest_after_origins, dest); + to_char('\n'); + ); -break2: *dest=0; return result; }