--- parser3/src/main/untaint.C 2009/08/21 08:38:55 1.151 +++ parser3/src/main/untaint.C 2015/10/03 00:03:49 1.167 @@ -1,11 +1,11 @@ /** @file Parser: String class part: untaint mechanizm. - Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_UNTAINT_C="$Date: 2009/08/21 08:38:55 $"; +volatile const char * IDENT_UNTAINT_C="$Id: untaint.C,v 1.167 2015/10/03 00:03:49 moko Exp $"; #include "pa_string.h" @@ -52,29 +52,34 @@ extern "C" { // author forgot to do that } \ } - #define escape_fragment(action) \ for(; fragment_length--; CORD_next(info->pos)) { \ char c=CORD_pos_fetch(info->pos); \ action \ } -#define _default CORD_ec_append(info->result, c) + #define encode(need_encode_func, prefix, otherwise) \ if(need_encode_func(c)) { \ - static const char* hex="0123456789ABCDEF"; \ CORD_ec_append(info->result, prefix); \ - CORD_ec_append(info->result, hex[((unsigned char)c)/0x10]); \ - CORD_ec_append(info->result, hex[((unsigned char)c)%0x10]); \ + to_hex(c); \ } else \ CORD_ec_append(info->result, otherwise); + +#define to_hex(c) \ + { \ + CORD_ec_append(info->result, hex_digits[((unsigned char)c) >> 4]); \ + CORD_ec_append(info->result, hex_digits[((unsigned char)c) & 0x0F]); \ + } + #define to_char(c) { CORD_ec_append(info->result, c); whitespace=false; } #define to_string(s) { CORD_ec_append_cord(info->result, s); whitespace=false; } +#define _default CORD_ec_append(info->result, c) inline bool need_file_encode(unsigned char c){ // russian letters and space ENABLED // encoding only these... return strchr( - "*?'\"<>|" + "*?\"<>|" #ifndef WIN32 ":\\" #endif @@ -82,17 +87,7 @@ inline bool need_file_encode(unsigned ch } inline bool need_uri_encode(unsigned char c){ - if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) - return false; - - return !strchr("_-./", c); -} - -inline bool need_http_header_encode(unsigned char c){ - if(strchr(" , :", c)) - return false; - - return need_uri_encode(c); + return !(pa_isalnum(c) || strchr("_-./*", c)); } inline bool need_regex_escape(unsigned char c){ @@ -108,34 +103,35 @@ inline bool need_parser_code_escape(unsi /* HTTP-header = field-name ":" [ field-value ] CRLF - field-name = token - field-value = *( field-content | LWS ) +field-name = token +field-value = *( field-content | LWS ) - field-content = - -word = token | quoted-string - token = 1* +word = token | quoted-string +quoted-string = ( <"> *(qdtext | quoted-pair ) <"> ) +qdtext = > +quoted-pair = "\" CHAR - +OCTET = +CHAR = tspecials = "(" | ")" | "<" | ">" | "@" - | "," | ";" | ":" | "\" | <"> - | "/" | "[" | "]" | "?" | "=" - | "{" | "}" | SP | HT + | "," | ";" | ":" | "\" | <"> + | "/" | "[" | "]" | "?" | "=" + | "{" | "}" | SP | HT SP = HT = LWS = [CRLF] 1*( SP | HT ) -TEXT = +TEXT = +CTL = -quoted-pair = "\" CHAR if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr)) */ @@ -257,10 +253,7 @@ RFC and without "_", or in would mean 0x20 */ inline bool mail_header_char_valid_within_Qencoded(char c) { - return c>='A' && c<='Z' - || c>='a' && c<='Z' - || c>='0' && c<='9' - || strchr("!*+-/", c); + return (pa_isalnum((unsigned char)c) || strchr("!*+-/", c)); } inline bool addr_spec_soon(const char *src) { for(char c; (c=*src); src++) @@ -348,32 +341,35 @@ int cstr_to_string_body_block(String::La case String::L_FILE_SPEC: // tainted, untaint language: file [name] { - bool is1251=(info->charsets && info->charsets->source().NAME()=="WINDOWS-1251"); escape_fragment( - // Macintosh has problems with small Russian letter 'r' - if( is1251 && c=='\xF0' ) { - // fixing that letter for most common charset - to_char('p'); - } else // fallback to default - encode(need_file_encode, '_', c); + encode(need_file_encode, '_', c); ); } break; - case String::L_FILE_POST: - { - escape_fragment(switch(c) { - case '\0': to_string("\\0"); break; - case '\\': to_string("\\\\"); break; - default: _default; break; - }); + case String::L_URI: + // tainted, untaint language: uri + { + const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr(); + // skip source [we use recoded version] + pa_CORD_pos_advance(info->pos, fragment_length); + String::C output(fragment_str, fragment_length); + if(info->charsets) + output=Charset::transcode(output, + info->charsets->source(), + info->charsets->client()); + + char c; + for(const char* src=output.str; (c=*src++); ) + encode(need_uri_encode, '%', c); } break; - case String::L_URI: case String::L_HTTP_HEADER: // tainted, untaint language: http-field-content-text - escape_fragment( - encode(need_uri_encode, '%', c); - ); + escape_fragment(switch(c) { + case '\n': + case '\r': to_string(" "); break; + default: _default; break; + }); break; case String::L_MAIL_HEADER: // tainted, untaint language: mail-header @@ -402,8 +398,8 @@ int cstr_to_string_body_block(String::La } //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'. if(!email && ( - !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char - || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c) + ( !to_quoted_printable && (c & 0x80) ) // starting quote-printable-encoding on first 8bit char + || ( to_quoted_printable && !mail_header_char_valid_within_Qencoded(c) ) )) { if(!to_quoted_printable) { to_string("=?"); @@ -498,19 +494,72 @@ int cstr_to_string_body_block(String::La _default; ); break; + case String::L_JSON: + // tainted, untaint language: json + // escape '"' '\' '/' '\n' '\t' '\r' '\b' '\f' chars and escape chars as \uXXXX if output charset != UTF-8 + { + if(info->charsets==NULL || info->charsets->client().isUTF8()){ + // escaping to \uXXXX is not needed + escape_fragment(switch((unsigned char)c) { + case '\n': to_string("\\n"); break; + case '"' : to_string("\\\""); break; + case '\\': to_string("\\\\"); break; + case '/' : to_string("\\/"); break; + case '\t': to_string("\\t"); break; + case '\r': to_string("\\r"); break; + case '\b': to_string("\\b"); break; + case '\f': to_string("\\f"); break; + case 0xE2: // \u2028 and \u2029 (line/paragraph separators), check bug #1023 + if(info->charsets && info->charsets->source().isUTF8() && fragment_length>=2){ + CORD_next(info->pos); + char c1=CORD_pos_fetch(info->pos); + CORD_next(info->pos); + char c2=CORD_pos_fetch(info->pos); + if((unsigned char)c1 == 0x80 && ((unsigned char)c2 >= 0xA8 && (unsigned char)c2 <= 0xAF)){ + to_string("\\u20"); + to_hex(((unsigned char)c2-0x80)); + } else { + CORD_ec_append(info->result, c); + CORD_ec_append(info->result, c1); + CORD_ec_append(info->result, c2); + } + fragment_length-=2; + } else { + _default; + } + break; + default: + if((unsigned char)c < 0x20){ + to_string("\\u00"); + to_hex(c); + } else { + _default; + } + break; + }); + } else { + const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr(); + // skip source [we use recoded version] + pa_CORD_pos_advance(info->pos, fragment_length); + String::C output(fragment_str, fragment_length); + + output=Charset::escape_JSON(output, info->charsets->source()); + to_string(output); + } + } + break; case String::L_HTTP_COOKIE: // tainted, untaint language: cookie (3.3.0 and higher: %uXXXX in UTF-8) - { + if(info->charsets) { const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr(); // skip source [we use recoded version] pa_CORD_pos_advance(info->pos, fragment_length); String::C output(fragment_str, fragment_length); output=Charset::escape(output, info->charsets->source()); - //throw Exception(0, 0, output); to_string(output); - - } + } else + ec_append(info->result, optimize, whitespace, info->pos, fragment_length); break; case String::L_PARSER_CODE: // for auto-untaint in process @@ -603,61 +652,10 @@ String::Body String::cstr_to_string_body return String::Body(CORD_ec_to_cord(info.result)); } -const char* String::transcode_and_untaint_cstr(Language lang, const Request_charsets *charsets) const { +const char* String::untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const { if(charsets && &charsets->source() != &charsets->client()){ - return cstr_to_string_body_transcode_and_untaint(lang, charsets).cstr(); - } else { + // Note: L_URI is allready transcoded during untaint, but transcode does not affect %XX + return Charset::transcode(cstr_to_string_body_untaint(lang, 0, charsets), charsets->source(), charsets->client()).cstr(); + } else return cstr_to_string_body_untaint(lang, 0, charsets).cstr(); - } -} - -int cstr_to_string_body_block_transcode_and_untaint(char alang, size_t fragment_length, Cstr_to_string_body_block_info* info){ - String::C output_c=Charset::transcode( - String::C(info->body->mid(info->fragment_begin, fragment_length).cstr(), fragment_length), - info->charsets->source(), - info->charsets->client() - ); - String::Body output_body=String::Body(output_c); - - size_t fragment_end=info->fragment_begin+fragment_length; - const String::Body* info_body=info->body; - - info->fragment_begin=0; - info->body=&output_body; - info->body->set_pos(info->pos, 0); - - int result=cstr_to_string_body_block_untaint(alang, output_c.length, info); - - info->fragment_begin=fragment_end; - info->body=info_body; - - return result; -} - -String::Body String::cstr_to_string_body_transcode_and_untaint(Language lang, const Request_charsets *charsets) const { - if(is_empty()) - return String::Body(); - - Cstr_to_string_body_block_info info; - // input - info.lang=lang; - info.connection=0; - info.charsets=charsets; - info.body=&body; - // output - CORD_ec_init(info.result); - // private - body.set_pos(info.pos, 0); - info.fragment_begin=0; - info.exception=0; - info.whitespace=true; - - langs.for_each(body, cstr_to_string_body_block_transcode_and_untaint, &info); - - if(info.exception) - throw Exception(0, - 0, - info.exception); - - return String::Body(CORD_ec_to_cord(info.result)); }