--- parser3/src/main/untaint.C 2001/07/18 13:18:00 1.56 +++ parser3/src/main/untaint.C 2003/01/29 11:55:22 1.115.2.4 @@ -1,52 +1,56 @@ /** @file Parser: String class part: untaint mechanizm. - Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) + Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char *RCSId="$Id: untaint.C,v 1.56 2001/07/18 13:18:00 parser Exp $"; + +static const char* IDENT_UNTAINT_C="$Date: 2003/01/29 11:55:22 $"; #include "pa_pool.h" +#include "pa_value_includes.h" #include "pa_string.h" #include "pa_hash.h" #include "pa_exception.h" #include "pa_table.h" #include "pa_globals.h" #include "pa_sql_connection.h" +#include "pa_dictionary.h" +#include "pa_common.h" +#include "pa_charset.h" +#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) \ - default: \ if(need_encode_func(*src)) { \ static const char *hex="0123456789ABCDEF"; \ char chunk[3]={prefix}; \ chunk[1]=hex[((unsigned char)*src)/0x10]; \ chunk[2]=hex[((unsigned char)*src)%0x10]; \ - strncpy(dest, chunk, 3); dest+=3; \ + memcpy(dest, chunk, 3); dest+=3; \ } else \ *dest++=*src; \ break #define to_char(c) *dest++=c #define to_string(b, bsize) \ - strncpy(dest, b, bsize); \ + memcpy(dest, b, bsize); \ dest+=bsize; \ inline bool need_file_encode(unsigned char c){ - 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')) @@ -74,31 +78,20 @@ static const char * String_Untaint_lang_ just a flag, that value not stored */ "A", ///< leave all characters intact - "F", ///< filename - "H", ///< text in HTTP response header + "F", ///< file specification + "H", ///< ext in HTTP response header "M", ///< text in mail header "URI", ///< text in uri "T", ///< ^table:set body "SQL", ///< ^table:sql body "JS", ///< JavaScript code - "HTML", ///< HTML code (for editing) - "UHTML", ///< HTML code with USER chars + "XML", ///< ^dom:set xml + "HTML" ///< HTML code (for editing) }; // String -static bool typo_present(Array::Item *value, const void *info) { - Array *row=static_cast(value); - const char *src=static_cast(info); - - int partial; - row->get_string(0)->cmp(partial, src); - return - partial==0 || // full match - partial==1; // typo left column starts 'src' -} - /* HTTP-header = field-name ":" [ field-value ] CRLF @@ -141,265 +134,408 @@ inline bool need_quote_http_header(const return false; } -/** @todo maybe additional check "are all pieces are clean?" would be profitable? - @todo fix potential forigins_mode buf overrun +//#include "pa_sapi.h" +/** + 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. */ -size_t String::cstr_bufsize(Untaint_lang lang) const { - return (lang==UL_AS_IS?size():size()*UNTAINT_TIMES_BIGGER*(forigins_mode?10:1)) +1; +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_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, + lang, //forcing passed lang + fragment->origin.file, fragment->origin.line); + ) + else if(lang==UL_PASS_APPENDED) + 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_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 + ), + fragment->origin.file, fragment->origin.line); + ) + else + STRING_FOREACH_FRAGMENT( + dest.APPEND(fragment->ptr, fragment->size, + fragment->lang==UL_TAINTED?lang:fragment->lang, + fragment->origin.file, fragment->origin.line); + ); + + return dest; } -/** @todo fix theoretical \n mem overrun in TYPO replacements -@todo rename base_64 to quoted_printable [invalid name now] +size_t String::cstr_bufsize(Untaint_lang lang, + SQL_Connection *connection, + const Request_charsets *charsets) const { + size_t dest=1; // for terminating 0 + STRING_FOREACH_FRAGMENT( + uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang; + + switch(to_lang & ~UL_OPTIMIZE_BIT) { + case UL_CLEAN: + case UL_TAINTED: + case UL_AS_IS: + // clean piece + + // tainted piece, but undefined untaint language + // for VString.as_double of tainted values + // for ^process{body} evaluation + + // tainted, untaint language: as-is + dest+=fragment->size; + break; + case UL_FILE_SPEC: + // tainted, untaint language: file [name] + dest+=fragment->size*3/* worst: Z->%XX */; + break; + case UL_URI: + // tainted, untaint language: uri + dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */; + break; + case UL_HTTP_HEADER: + // tainted, untaint language: http-field-content-text + dest+=fragment->size*3/* worst: Z->%XX */; + break; + case UL_MAIL_HEADER: + // tainted, untaint language: mail-header + 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+= + fragment->size*3+ + parts_count*(charsets->mail().name()->size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */); + } else + dest+=fragment->size; + break; + case UL_TABLE: + // tainted, untaint language: table + dest+=fragment->size; + break; + case UL_SQL: + // tainted, untaint language: sql + if(connection) + dest+=connection->quote(0, fragment->ptr, fragment->size); + break; + case UL_JS: + escape(switch(*src) { + case '"': case '\'': case '\n': case '\\': case '\xFF': + dest+=2; break; + default: + dest++; break; + }); + break; + case UL_XML: + escape(switch(*src) { + case '&': case '>': case '<': case '"': case '\'': + dest+= 6; break; + default: + dest++; break; + }); + break; + case UL_HTML: + escape(switch(*src) { + case '&': + case '>': + case '<': + case '"': + dest+=6; break; + default: + dest++; break; + }); + break; + } + ); + 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, - const char *charset) const { - // $MAIN:html-typo table - Table *user_typo_table=static_cast(pool().tag()); - Table *typo_table=user_typo_table?user_typo_table:default_typo_table; - + const Request_charsets *charsets) const { + // WARNING: + // before any changes check cstr_bufsize first!!! bool whitespace=true; - const Chunk *chunk=&head; - do { - const Chunk::Row *row=chunk->rows; - for(size_t i=0; icount; i++, row++) { - if(row==append_here) - goto break2; - - Untaint_lang to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang; - - char *dest_before_origins=dest; - - if(forigins_mode) { -#ifndef NO_STRING_ORIGIN - if(row->item.origin.file) - dest+=sprintf(dest, "%s(%d)", - row->item.origin.file, - 1+row->item.origin.line); - else - dest+=sprintf(dest, "unknown"); -#endif - dest+=sprintf(dest, "#%s: ", - String_Untaint_lang_name[to_lang]); +// STRING_FOREACH_FRAGMENT( + element_type *last = felements+fused; element_type *fragment = felements; for(; fragmentlang:lang; + + char *start=dest; + + switch(to_lang & ~UL_OPTIMIZE_BIT) { + case UL_CLEAN: + case UL_TAINTED: + case UL_AS_IS: + // clean piece + + // tainted piece, but undefined untaint language + // for VString.as_double of tainted values + // for ^process{body} evaluation + + // tainted, untaint language: as-is + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; + break; + case UL_FILE_SPEC: + // tainted, untaint language: file [name] + escape( + // 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; + 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; } - char *dest_after_origins=dest; - // WARNING: - // string can grow only UNTAINT_TIMES_BIGGER - switch(to_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 - // for ^process{body} evaluation - case UL_AS_IS: - // tainted, untaint language: as-is - memcpy(dest, row->item.ptr, row->item.size); - dest+=row->item.size; - break; - case UL_FILE_NAME: - // tainted, untaint language: file [name] - escape(switch(*src) { - case ' ': to_char('_'); break; - encode(need_file_encode, '+'); - }); - break; - case UL_URI: - // tainted, untaint language: uri - escape(switch(*src) { - case ' ': to_char('+'); break; + { + const char *src=(const char *)client_ptr; + for(int size=client_size; size--; src++) encode(need_uri_encode, '%'); - }); - break; - case UL_HTTP_HEADER: - // tainted, untaint language: http-header - 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 - 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 + } + break; + case UL_HTTP_HEADER: + // tainted, untaint language: http-field-content-text + escape( + encode(need_uri_encode, '%'); + ); + break; + case UL_MAIL_HEADER: + // tainted, untaint language: mail-header + // http://www.ietf.org/rfc/rfc2047.txt + if(charsets) { + const void *mail_ptr; + size_t 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; + bool to_quoted_printable=false; + + bool email=false; + for(const char *end=src+mail_size; src': to_string(">", 4); break; - case '<': to_string("<", 4); break; - case '"': to_string(""", 6); break; - //TODO: XSLT case '\'': to_string("'", 6); break; - _default; - }); - break; - case UL_USER_HTML: { - // tainted, untaint language: html-typo - 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 - // and rather replace \ and n into huge strings - // thus causing memory overrun - // this can be dealed by allocating *2 memory, but that's too expensive - size_t html_for_typo_size; - { // local dest - char *dest=html_for_typo; - escape(switch(*src) { - // convinient name for typo match "\n" - case '\r': - if(typo_table) { - *dest++='\\'; *dest++='n'; // \r -> \n - if(src[1]=='\n') { // \r\n -> remove \n - size--; src++; - } - } - break; - case '\n': - if(typo_table) - to_string("\\n", 2); - break; - //TODO: XSLT case '\'': to_string("'", 6); break; - _default; - }); - *dest=0; - html_for_typo_size=dest-html_for_typo; - } - // typo table replacements - const char *src=html_for_typo; - do { - // there is a row where first column starts 'src' - if(Table::Item *item=typo_table->first_that(typo_present, src)) { - // get a=>b values - const String& a=*static_cast(item)->get_string(0); - const String& b=*static_cast(item)->get_string(1); - // empty 'a' | 'b' checks - if(a.size()==0 || b.size()==0) { - pool().set_tag(default_typo_table); // avoid recursion - THROW(0, 0, - typo_table->origin_string(), - "typo table column elements must not be empty"); - } - // overflow check: - // b allowed to be max UNTAINT_TIMES_BIGGER then a - if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) { - pool().set_tag(default_typo_table); // avoid recursion - THROW(0, 0, - &b, - "is %g times longer then '%s', " - "while maximum, handled by Parser, is %d", - ((double)b.size())/a.size(), - a.cstr(), - UNTAINT_TIMES_BIGGER); + to_quoted_printable=false; + } + if(!email && ( + !to_quoted_printable && (*src & 0x80) // starting quote-printable-encoding on first 8bit char + || to_quoted_printable && !mail_header_char_valid_within_Qencoded(*src) + )) { + if(!to_quoted_printable) { + dest+=sprintf(dest, "=?%s?Q?", charset_name); + to_quoted_printable=true; } - - // skip 'a' in 'src' - src+=a.size(); - // write 'b' to 'dest' - b.store_to(dest); - dest+=b.size(); + //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++; - } while(*src); - break; + *dest++=*src; + if(*src=='>') + email=false; } - default: - THROW(0, 0, - this, - "unknown untaint language #%d of %d piece", - static_cast(row->item.lang), - i); // never - break; // never + if(to_quoted_printable) // close + dest+=sprintf(dest, "?="); + + } else { + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; } + break; + case UL_TABLE: + // tainted, untaint language: table + escape(switch(*src) { + case '\t': to_char(' '); break; + case '\n': to_char(' '); break; + _default; + }); + break; + case UL_SQL: + // tainted, untaint language: sql + if(connection) + dest+=connection->quote(dest, fragment->ptr, fragment->size); + else + throw Exception(0, + ConstStringPtr(this), + "untaint in SQL language failed - no connection specified"); + break; + case UL_JS: + escape(switch(*src) { + case '"': to_string("\\\"", 2); break; + case '\'': to_string("\\'", 2); break; + case '\n': to_string("\\n", 2); break; + case '\\': to_string("\\\\", 2); break; + case '\xFF': to_string("\\\xFF", 2); break; + _default; + }); + break; + case UL_XML: + escape(switch(*src) { + case '&': to_string("&", 5); break; + case '>': to_string(">", 4); break; + case '<': to_string("<", 4); break; + case '"': to_string(""", 6); break; + case '\'': to_string("'", 6); break; + _default; + }); + break; + case UL_HTML: + escape(switch(*src) { + case '&': to_string("&", 5); break; + case '>': to_string(">", 4); break; + case '<': to_string("<", 4); break; + case '"': to_string(""", 6); break; + _default; + }); + break; + default: + throw Exception(0, + ConstStringPtr(this), + "unknown untaint language #%d", + static_cast(to_lang)); // should never + break; // never + } - if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN) - whitespace=false; - - if(forigins_mode) - if(dest==dest_after_origins) // never moved==optimized space - dest=dest_before_origins; - else { - for(char *p=dest_after_origins; plink; - } while(chunk); -break2: return dest; } + +char *String::cstr_debug_origins(Pool& pool) const { + //_asm int 3; + char *result=new(pool) char[size()+fused*MAX_STRING*2]; + char *dest=result; + + STRING_FOREACH_FRAGMENT( +IFNDEF_NO_STRING_ORIGIN( + if(fragment->origin.file) + dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, + fragment->origin.file, + 1+fragment->origin.line); + else + dest+=sprintf(dest, ""); +); + 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, + ConstStringPtr(this), + "unknown untaint language #%d", + static_cast(show_lang)); // sould never + + dest+=sprintf(dest, "#%s%s: ", + String_Untaint_lang_name[show_lang], + fragment->lang & UL_OPTIMIZE_BIT?".O":""); + char *dest_after_origins=dest; + + memcpy(dest, fragment->ptr, fragment->size); + dest+=fragment->size; + + remove_crlf(dest_after_origins, dest); + to_char('\n'); + ); + + *dest=0; + return result; +}