--- parser3/src/main/untaint.C 2001/12/17 18:03:59 1.86 +++ parser3/src/main/untaint.C 2002/02/21 14:36:54 1.96 @@ -1,10 +1,10 @@ /** @file Parser: String class part: untaint mechanizm. - Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) + Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) - $Id: untaint.C,v 1.86 2001/12/17 18:03:59 paf Exp $ + $Id: untaint.C,v 1.96 2002/02/21 14:36:54 paf Exp $ */ #include "pa_pool.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 +144,86 @@ 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] + 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, + 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.count initally equeals this.head.count and shrinks-only, + // so can't be more than this.head.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)); + if(src_head_link) { + // "your body is my body" + head.rows[head.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; + // "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; rowlink==(void*)0xcdcdcdcd) + _asm int 3; + // "chopping off my tail-reserve" + last_chunk->count=append_here-last_chunk->rows; + // "you is my tail" + append_here->link=&src.head; + // "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; + } + // manually unrolled code to avoid do{if(const)} constructs if(forced) STRING_SRC_FOREACH_ROW( @@ -170,13 +255,16 @@ 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; } size_t String::cstr_bufsize(Untaint_lang lang, SQL_Connection *connection, - const char *charset) const { + Charset *buf_charset) const { size_t dest=1; // for terminating 0 STRING_FOREACH_ROW( uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang; @@ -208,12 +296,13 @@ size_t String::cstr_bufsize(Untaint_lang break; case UL_MAIL_HEADER: // tainted, untaint language: mail-header - if(charset) { + if(buf_charset) { // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?= - dest+=row->item.size*3+MAX_STRING/* worst: =?charset?Q?=%XX?= */; - } else { + dest+= + row->item.size*3+ + buf_charset->name().size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */; + } else dest+=row->item.size; - } break; case UL_TABLE: // tainted, untaint language: table @@ -253,24 +342,16 @@ size_t String::cstr_bufsize(Untaint_lang break; } ); -break2: return dest; } char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection, - const char *charset) const { + Charset *store_to_charset) 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; @@ -320,14 +401,20 @@ char *String::store_to(char *dest, Untai break; case UL_MAIL_HEADER: // tainted, untaint language: mail-header - if(charset) { + if(store_to_charset) { + const void *mail_ptr; + size_t mail_size; + Charset::transcode(pool(), + pool().get_source_charset(), row->item.ptr, row->item.size, + *store_to_charset, mail_ptr, mail_size); + // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?= - const char *src=row->item.ptr; + const char *src=(const char *)mail_ptr; bool to_quoted_printable=false; - for(int size=row->item.size; size--; src++) { + for(int size=mail_size; size--; src++) { if(*src & 0x80) { if(!to_quoted_printable) { - dest+=sprintf(dest, "=?%.15s?Q?", charset); + dest+=sprintf(dest, "=?%s?Q?", store_to_charset->name().cstr()); to_quoted_printable=true; } dest+=sprintf(dest, "=%02X", *src & 0xFF); @@ -337,6 +424,7 @@ char *String::store_to(char *dest, Untai } if(to_quoted_printable) // close dest+=sprintf(dest, "?="); + } else { memcpy(dest, row->item.ptr, row->item.size); dest+=row->item.size; @@ -415,12 +503,8 @@ char *String::store_to(char *dest, Untai }; } else // piece without optimization whitespace=false; + ); - } \ - chunk=row->link; \ - } while(chunk); \ - -break2: return dest; } @@ -429,43 +513,34 @@ 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, 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; }