--- parser3/src/main/untaint.C 2001/03/25 09:40:55 1.18 +++ parser3/src/main/untaint.C 2001/04/23 10:23:40 1.46 @@ -5,7 +5,7 @@ Author: Alexander Petrosyan (http://design.ru/paf) - $Id: untaint.C,v 1.18 2001/03/25 09:40:55 paf Exp $ + $Id: untaint.C,v 1.46 2001/04/23 10:23:40 paf Exp $ */ #include "pa_config_includes.h" @@ -15,6 +15,8 @@ #include "pa_hash.h" #include "pa_exception.h" #include "pa_table.h" +#include "pa_globals.h" +#include "pa_sql_connection.h" #define escape(action) \ { \ @@ -43,7 +45,11 @@ inline bool need_file_encode(unsigned ch if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) return false; - return !strchr("./", c); + return !strchr( +#ifdef WIN32 + ":\\~" +#endif + "./()_-", c); } inline bool need_uri_encode(unsigned char c){ if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) @@ -51,7 +57,7 @@ inline bool need_uri_encode(unsigned cha return !strchr("_-./", c); } -inline bool need_header_encode(unsigned char c){ +inline bool need_http_header_encode(unsigned char c){ if(strchr(" , :", c)) return false; @@ -65,32 +71,97 @@ static bool typo_present(Array::Item *va const char *src=static_cast(info); int partial; - row->get_string(0)->cmp(src, partial); + row->get_string(0)->cmp(partial, src); return partial==0 || // full match partial==1; // typo left column starts 'src' } -/// @todo optimize whitespaces for all but 'html' -char *String::store_to(char *dest) const { +/* + +HTTP-header = field-name ":" [ field-value ] CRLF + + field-name = token + field-value = *( field-content | LWS ) + + field-content = + + + +word = token | quoted-string + +token = 1* + + + +tspecials = "(" | ")" | "<" | ">" | "@" + | "," | ";" | ":" | "\" | <"> + | "/" | "[" | "]" | "?" | "=" + | "{" | "}" | SP | HT + +SP = +HT = + +LWS = [CRLF] 1*( SP | HT ) +TEXT = + +quoted-pair = "\" CHAR + + if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr)) +*/ +inline bool need_quote_http_header(const char *ptr, size_t size) { + for(; size--; ptr++) + if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr)) + return true; + return false; +} + +/** + @todo fix theoretical \n mem overrun in TYPO replacements +*/ +char *String::store_to(char *dest, Untaint_lang lang, + SQL_Connection *connection, + const char *charset) const { // $MAIN:html-typo table - Table *typo_table=static_cast(pool().tag()); + Table *user_typo_table=static_cast
(pool().tag()); + Table *typo_table=user_typo_table?user_typo_table:default_typo_table; + bool whitespace=true; const Chunk *chunk=&head; do { const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { + for(size_t i=0; icount; i++, row++) { if(row==append_here) goto break2; // WARNING: // string can grow only UNTAINT_TIMES_BIGGER - switch(row->item.lang) { - case UL_NO: + switch(lang==UL_UNSPECIFIED?row->item.lang:lang) { + case UL_CLEAN: // clean piece - case UL_YES: + { // 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.get_double of tainted values + // for VString.as_double of tainted values // for ^process{body} evaluation case UL_AS_IS: // tainted, untaint language: as-is @@ -101,7 +172,7 @@ char *String::store_to(char *dest) const // tainted, untaint language: file [name] escape(switch(*src) { case ' ': to_char('_'); break; - encode(need_file_encode, '-'); + encode(need_file_encode, '+'); }); break; case UL_URI: @@ -111,11 +182,43 @@ char *String::store_to(char *dest) const encode(need_uri_encode, '%'); }); break; - case UL_HEADER: - // tainted, untaint language: header - escape(switch(*src) { - encode(need_header_encode, '%'); - }); + 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 + dest+=sprintf(dest, "?="); + } else { + memcpy(dest, row->item.ptr, row->item.size); + dest+=row->item.size; + } break; case UL_TABLE: // tainted, untaint language: table @@ -127,9 +230,12 @@ char *String::store_to(char *dest) const break; case UL_SQL: // tainted, untaint language: sql - // TODO: зависимость от sql сервера - memset(dest, '?', row->item.size); - dest+=row->item.size; + if(connection) + dest+=connection->quote(dest, row->item.ptr, row->item.size); + else + THROW(0, 0, + this, + "untaint in SQL language failed - no connection specified"); break; case UL_JS: escape(switch(*src) { @@ -147,82 +253,78 @@ char *String::store_to(char *dest) const case '>': to_string(">", 4); break; case '<': to_string("<", 4); break; case '"': to_string(""", 6); break; - //TODO: XSLT to_string!'\'', "'", 6) + //TODO: XSLT case '\'': to_string("'", 6); break; _default; }); break; case UL_HTML_TYPO: { // tainted, untaint language: html-typo - char *html_for_typo= - (char *)malloc(size()*6/*""" the longest possible*/+1); + char *html_for_typo=(char *)malloc(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) { - // BEWARE: check maximum replacement length in malloc above - case '&': to_string("&", 5); break; - case '>': to_string(">", 4); break; - case '<': to_string("<", 4); break; - case '"': to_string(""", 6); break; // convinient name for typo match "\n" case '\r': if(typo_table) { *dest++='\\'; *dest++='n'; // \r -> \n - if(src[1]=='\n') // \r\n -> remove \n - src++; + if(src[1]=='\n') { // \r\n -> remove \n + size--; src++; + } } break; case '\n': if(typo_table) to_string("\\n", 2); break; - //TODO: XSLT to_string!'\'', "'", 6) + //TODO: XSLT case '\'': to_string("'", 6); break; _default; }); *dest=0; html_for_typo_size=dest-html_for_typo; } // typo table replacements - if(typo_table) { - 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(0); // 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(0); // 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); - } - - // skip 'a' in 'src' - src+=a.size(); - // write 'b' to 'dest' - b.store_to(dest); - dest+=b.size(); - } else - *dest++=*src++; - } while(*src); - } else { - memcpy(dest, html_for_typo, html_for_typo_size); - dest+=html_for_typo_size; - } + 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); + } + + // skip 'a' in 'src' + src+=a.size(); + // write 'b' to 'dest' + b.store_to(dest); + dest+=b.size(); + } else + *dest++=*src++; + } while(*src); break; } default: @@ -230,9 +332,11 @@ char *String::store_to(char *dest) const this, "unknown untaint language #%d of %d piece", static_cast(row->item.lang), - i); + i); // never } - row++; + + if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN) + whitespace=false; } chunk=row->link; } while(chunk);