--- parser3/src/main/untaint.C 2001/04/05 11:01:57 1.33 +++ parser3/src/main/untaint.C 2001/04/20 09:05:24 1.42 @@ -5,7 +5,7 @@ Author: Alexander Petrosyan (http://design.ru/paf) - $Id: untaint.C,v 1.33 2001/04/05 11:01:57 paf Exp $ + $Id: untaint.C,v 1.42 2001/04/20 09:05:24 paf Exp $ */ #include "pa_config_includes.h" @@ -16,6 +16,7 @@ #include "pa_exception.h" #include "pa_table.h" #include "pa_globals.h" +#include "pa_sql_connection.h" #define escape(action) \ { \ @@ -46,9 +47,9 @@ inline bool need_file_encode(unsigned ch return !strchr( #ifdef WIN32 - ":\\" + ":\\~" #endif - "./", c); + "./()_-", c); } inline bool need_uri_encode(unsigned char c){ if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) @@ -56,7 +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; @@ -76,11 +77,54 @@ static bool typo_present(Array::Item *va partial==1; // typo left column starts 'src' } +/* + +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; +} + /** @test optimize whitespaces for all but 'html' @todo fix theoretical \n mem overrun in TYPO replacements + @test mail-header */ -char *String::store_to(char *dest, Untaint_lang lang) const { +char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection) 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; @@ -94,7 +138,7 @@ char *String::store_to(char *dest, Untai // WARNING: // string can grow only UNTAINT_TIMES_BIGGER - switch(lang==UL_UNKNOWN?row->item.lang:lang) { + switch(lang==UL_UNSPECIFIED?row->item.lang:lang) { case UL_CLEAN: // clean piece case UL_TAINTED: @@ -110,7 +154,7 @@ char *String::store_to(char *dest, Untai // tainted, untaint language: file [name] escape(switch(*src) { case ' ': to_char('_'); break; - encode(need_file_encode, '-'); + encode(need_file_encode, '+'); }); break; case UL_URI: @@ -120,11 +164,24 @@ char *String::store_to(char *dest, Untai 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 + memcpy(dest, row->item.ptr, row->item.size); + dest+=row->item.size; break; case UL_TABLE: // tainted, untaint language: table @@ -136,10 +193,12 @@ char *String::store_to(char *dest, Untai break; case UL_SQL: // tainted, untaint language: sql - // TODO: зависимость от sql сервера - memcpy(dest, row->item.ptr, row->item.size); - //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) { @@ -236,7 +295,7 @@ char *String::store_to(char *dest, Untai this, "unknown untaint language #%d of %d piece", static_cast(row->item.lang), - i); + i); // never } } chunk=row->link;