--- parser3/src/main/untaint.C 2002/02/21 14:44:37 1.97 +++ parser3/src/main/untaint.C 2003/01/31 12:34:38 1.115.2.6 @@ -1,13 +1,14 @@ /** @file Parser: String class part: untaint mechanizm. - Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) - - $Id: untaint.C,v 1.97 2002/02/21 14:44:37 paf Exp $ */ +static const char* IDENT_UNTAINT_C="$Date: 2003/01/31 12:34:38 $"; + #include "pa_pool.h" +#include "pa_value_includes.h" #include "pa_string.h" #include "pa_hash.h" #include "pa_exception.h" @@ -17,23 +18,18 @@ #include "pa_dictionary.h" #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 +#include "pa_request_charsets.h" #define escape(action) \ { \ - const char *src=row->item.ptr; \ - for(int size=row->item.size; size--; src++) \ + const char* src=fragment->ptr; \ + for(int size=fragment->size; size--; src++) \ action \ } #define _default default: *dest++=*src; break #define encode(need_encode_func, prefix) \ if(need_encode_func(*src)) { \ - static const char *hex="0123456789ABCDEF"; \ + static const char* hex="0123456789ABCDEF"; \ char chunk[3]={prefix}; \ chunk[1]=hex[((unsigned char)*src)/0x10]; \ chunk[2]=hex[((unsigned char)*src)%0x10]; \ @@ -47,20 +43,14 @@ ulong string_string_shortcut_economy=0; dest+=bsize; \ inline bool need_file_encode(unsigned char c){ - // theoretical problem with, for instance, "_2B" and "." fragments, - // they would yield the same - // because need_file_encode('_')=false - // but we need to delete such files somehow, getting names from ^index - - if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) - return false; - - return !strchr( - " _./()-" -#ifdef WIN32 - ":\\~" + // russian letters and space ENABLED + // encoding only these... + return strchr( + "*?'\"<>|" +#ifndef WIN32 + ":\\~" #endif - , c); + , c)!=0; } inline bool need_uri_encode(unsigned char c){ if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) @@ -77,7 +67,7 @@ inline bool need_http_header_encode(unsi // -static const char * String_Untaint_lang_name[]={ +static const char* String_Untaint_lang_name[]={ "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum "C", ///< clean "T", ///< tainted, untaint language as assigned later @@ -137,7 +127,7 @@ quoted-pair = "\" CHAR if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr)) */ -inline bool need_quote_http_header(const char *ptr, size_t size) { +inline bool need_quote_http_header(const char* ptr, size_t size) { for(; size--; ptr++) if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr)) return true; @@ -146,124 +136,64 @@ inline bool need_quote_http_header(const //#include "pa_sapi.h" /** - appends other String, + appends to 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; +String& String::append_to(String& dest, uchar lang, bool forced) const { + if(is_empty()) + return dest; + + // without language-chage, small[fitting dest tail] string? + if(lang==UL_PASS_APPENDED && fused<=(dest.fallocated-dest.fused)) { + // felements[fused...]=src.felements[0..fused], fused+=src.fused + memcpy(&dest.felements[dest.fused], felements, sizeof(felements[0])*fused); + dest.fused+=fused; + return dest; } // manually unrolled code to avoid do{if(const)} constructs if(forced) - STRING_SRC_FOREACH_ROW( - APPEND(row->item.ptr, row->item.size, + STRING_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, lang, //forcing passed lang - row->item.origin.file, row->item.origin.line); + fragment->origin.file, fragment->origin.line); ) else if(lang==UL_PASS_APPENDED) - STRING_SRC_FOREACH_ROW( - APPEND(row->item.ptr, row->item.size, - row->item.lang, // passing item's lang - row->item.origin.file, row->item.origin.line); + STRING_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, + fragment->lang, // passing item's lang + fragment->origin.file, fragment->origin.line); ) else if(lang&UL_OPTIMIZE_BIT) // main idea here // tainted piece would get OPTIMIZED bit from 'lang' // clean piece would be marked OPTIMIZED manually // pieces with determined languages [not tainted|clean] would retain theirs langs - STRING_SRC_FOREACH_ROW( - APPEND(row->item.ptr, row->item.size, - row->item.lang==UL_TAINTED?lang:( - row->item.lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag - row->item.lang + STRING_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, + fragment->lang==UL_TAINTED?lang:( + fragment->lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag + fragment->lang ), - row->item.origin.file, row->item.origin.line); + fragment->origin.file, fragment->origin.line); ) else - STRING_SRC_FOREACH_ROW( - APPEND(row->item.ptr, row->item.size, - row->item.lang==UL_TAINTED?lang:row->item.lang, - row->item.origin.file, row->item.origin.line); + STRING_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, + fragment->lang==UL_TAINTED?lang:fragment->lang, + fragment->origin.file, fragment->origin.line); ); -/* - for(Chunk::Row *row=last_chunk->rows; rowlink==(void*)0xcdcdcdcd) - _asm int 3;*/ - return *this; + + return dest; } size_t String::cstr_bufsize(Untaint_lang lang, SQL_Connection *connection, - Charset *buf_charset) const { + const Request_charsets *charsets) const { size_t dest=1; // for terminating 0 - STRING_FOREACH_ROW( - uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang; + STRING_FOREACH_FRAGMENT( + uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang; switch(to_lang & ~UL_OPTIMIZE_BIT) { case UL_CLEAN: @@ -276,38 +206,46 @@ size_t String::cstr_bufsize(Untaint_lang // for ^process{body} evaluation // tainted, untaint language: as-is - dest+=row->item.size; + dest+=fragment->size; break; case UL_FILE_SPEC: // tainted, untaint language: file [name] - dest+=row->item.size*3/* worst: Z->%XX */; + dest+=fragment->size*3/* worst: Z->%XX */; break; case UL_URI: // tainted, untaint language: uri - dest+=row->item.size*6*3/* worst utf8 x worst Z->%XX */; + dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */; break; case UL_HTTP_HEADER: // tainted, untaint language: http-field-content-text - dest+=row->item.size*3/* worst: Z->%XX */; + dest+=fragment->size*3/* worst: Z->%XX */; break; case UL_MAIL_HEADER: // tainted, untaint language: mail-header - if(buf_charset) { + if(charsets) { + int parts_count=1; + const char* prev=fragment->ptr; + size_t size=fragment->size; + while(const char* comma_at=(const char*)memchr(prev, ',', size)) { + parts_count++; + size-=comma_at-prev; + prev=comma_at+1; + } // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?= dest+= - row->item.size*3+ - buf_charset->name().size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */; + fragment->size*3+ + parts_count*(charsets->mail().name()->size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */); } else - dest+=row->item.size; + dest+=fragment->size; break; case UL_TABLE: // tainted, untaint language: table - dest+=row->item.size; + dest+=fragment->size; break; case UL_SQL: // tainted, untaint language: sql if(connection) - dest+=connection->quote(0, row->item.ptr, row->item.size); + dest+=connection->quote(0, fragment->ptr, fragment->size); break; case UL_JS: escape(switch(*src) { @@ -341,14 +279,49 @@ size_t String::cstr_bufsize(Untaint_lang return dest; } +/** http://www.ietf.org/rfc/rfc2047.txt +RFC +(3) As a replacement for a 'word' entity within a 'phrase', for example, + one that precedes an address in a From, To, or Cc header. The ABNF + definition for 'phrase' from RFC 822 thus becomes: + + phrase = 1*( encoded-word / word ) + + In this case the set of characters that may be used in a "Q"-encoded + 'encoded-word' is restricted to: . An 'encoded-word' that appears within a + 'phrase' MUST be separated from any adjacent 'word', 'text' or + 'special' by 'linear-white-space'. +... + (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be + represented as "_" (underscore, ASCII 95.). (This character may + not pass through some internetwork mail gateways, but its use + will greatly enhance readability of "Q" encoded data with mail + readers that do not support this encoding.) Note that the "_" + always represents hexadecimal 20, even if the SPACE character + occupies a different code position in the character set in use. + + paf: obviously, + without "=", or one could not differ "=E0" and "russian letter a" + and without "_", or in would mean 0x20 +*/ +static bool mail_header_char_valid_within_Qencoded(char c) { + return c>='A' && c<='Z' + || c>='a' && c<='Z' + || c>='0' && c<='9' + || strchr("!*+-/", c); +} char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection, - Charset *store_to_charset) const { + const Request_charsets *charsets) const { // WARNING: // before any changes check cstr_bufsize first!!! bool whitespace=true; - STRING_FOREACH_ROW( - uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang; +// STRING_FOREACH_FRAGMENT( + element_type *last = felements+fused; element_type *fragment = felements; for(; fragmentlang:lang; char *start=dest; @@ -363,67 +336,92 @@ char *String::store_to(char *dest, Untai // for ^process{body} evaluation // tainted, untaint language: as-is - memcpy(dest, row->item.ptr, row->item.size); - dest+=row->item.size; + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; break; case UL_FILE_SPEC: // tainted, untaint language: file [name] escape( - encode(need_file_encode, '_'); + // Macintosh has problems with small Russian letter 'r' + if( *src=='\xF0' && charsets && *charsets->source().name()=="windows-1251" ) { + // fixing that letter for most common charset + to_char('p'); + } else // fallback to default + encode(need_file_encode, '_'); ); break; case UL_URI: // tainted, untaint language: uri const void *client_ptr; size_t client_size; - Charset::transcode(pool(), - pool().get_source_charset(), row->item.ptr, row->item.size, - pool().get_client_charset(), client_ptr, client_size); + if(charsets) + Charset::transcode(charsets->pool(), + charsets->source(), fragment->ptr, fragment->size, + charsets->client(), client_ptr, client_size); + else { + client_ptr=fragment->ptr; + client_size=fragment->size; + } + { - const char *src=(const char *)client_ptr; + const char* src=(const char* )client_ptr; for(int size=client_size; size--; src++) - switch(*src) { - case ' ': to_char('+'); break; - default: encode(need_uri_encode, '%'); - }; + encode(need_uri_encode, '%'); } break; case UL_HTTP_HEADER: // tainted, untaint language: http-field-content-text - escape(switch(*src) { - case ' ': to_char('+'); break; - default: encode(need_uri_encode, '%'); - }); + escape( + encode(need_uri_encode, '%'); + ); break; case UL_MAIL_HEADER: // tainted, untaint language: mail-header - if(store_to_charset) { + // http://www.ietf.org/rfc/rfc2047.txt + if(charsets) { 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); + Charset::transcode(charsets->pool(), + charsets->source(), fragment->ptr, fragment->size, + charsets->mail(), mail_ptr, mail_size); + CharPtr charset_name=charsets->mail().name()->cstr(); // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?= - const char *src=(const char *)mail_ptr; + const char* src=(const char* )mail_ptr; bool to_quoted_printable=false; - for(int size=mail_size; size--; src++) { - if(*src & 0x80) { + + bool email=false; + for(const char* end=src+mail_size; srcname().cstr()); + dest+=sprintf(dest, "=?%s?Q?", charset_name); to_quoted_printable=true; } - dest+=sprintf(dest, "=%02X", *src & 0xFF); - } else { - *dest++=*src; - } + //RFC Upper case should be used for hexadecimal digits "A" through "F" + if(*src == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) + *dest++='_'; // RFC may be represented as "_" + else + dest+=sprintf(dest, "=%02X", *src & 0xFF); + } else + *dest++=*src; + if(*src=='>') + email=false; } if(to_quoted_printable) // close dest+=sprintf(dest, "?="); } else { - memcpy(dest, row->item.ptr, row->item.size); - dest+=row->item.size; + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; } break; case UL_TABLE: @@ -437,10 +435,10 @@ char *String::store_to(char *dest, Untai case UL_SQL: // tainted, untaint language: sql if(connection) - dest+=connection->quote(dest, row->item.ptr, row->item.size); + dest+=connection->quote(dest, fragment->ptr, fragment->size); else - throw Exception(0, 0, - this, + throw Exception(Exception::undefined_type, + Exception::undefined_source, "untaint in SQL language failed - no connection specified"); break; case UL_JS: @@ -473,10 +471,10 @@ char *String::store_to(char *dest, Untai }); break; default: - throw Exception(0, 0, - this, + throw Exception(Exception::undefined_type, + Exception::undefined_source, "unknown untaint language #%d", - static_cast(row->item.lang)); // sould never + static_cast(to_lang)); // should never break; // never } @@ -499,39 +497,40 @@ char *String::store_to(char *dest, Untai }; } else // piece without optimization whitespace=false; - ); +// ); + } return dest; } -char *String::cstr_debug_origins() const { +char *String::cstr_debug_origins(Pool& pool) const { //_asm int 3; - char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2); + char *result=new(pool) char[size()+fused*MAX_STRING*2]; char *dest=result; - STRING_FOREACH_ROW( + STRING_FOREACH_FRAGMENT( IFNDEF_NO_STRING_ORIGIN( - if(row->item.origin.file) + if(fragment->origin.file) dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, - row->item.origin.file, - 1+row->item.origin.line); + fragment->origin.file, + 1+fragment->origin.line); else dest+=sprintf(dest, ""); ); - uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT; + uchar show_lang=fragment->lang & ~UL_OPTIMIZE_BIT; if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0])) - throw Exception(0, 0, - this, + throw Exception(Exception::undefined_type, + Exception::undefined_source, "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":""); + fragment->lang & UL_OPTIMIZE_BIT?".O":""); char *dest_after_origins=dest; - memcpy(dest, row->item.ptr, row->item.size); - dest+=row->item.size; + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; remove_crlf(dest_after_origins, dest); to_char('\n');