|
|
| version 1.17, 2001/03/25 09:10:30 | version 1.115.2.12.2.11, 2003/03/24 14:53:58 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: String class part: untaint mechanizm. | Parser: String class part: untaint mechanizm. |
| Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) | Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf) | |
| $Id$ | |
| */ | */ |
| #include "pa_config_includes.h" | static const char* IDENT_UNTAINT_C="$Date$"; |
| #include "pa_pool.h" | |
| #include "pa_string.h" | #include "pa_string.h" |
| #include "pa_hash.h" | #include "pa_hash.h" |
| #include "pa_exception.h" | #include "pa_exception.h" |
| #include "pa_table.h" | #include "pa_table.h" |
| #include "pa_globals.h" | |
| #include "pa_dictionary.h" | |
| #include "pa_common.h" | |
| #include "pa_charset.h" | |
| #include "pa_request_charsets.h" | |
| #define escape(cases) \ | extern "C" { // author forgot to do that |
| { \ | #include "ec.h" |
| const char *src=row->item.ptr; \ | } |
| for(int size=row->item.size; size--; src++) \ | |
| switch(*src) { \ | //#define PA_SQL |
| cases \ | |
| } \ | #ifdef PA_SQL |
| #include "pa_sql_connection.h" | |
| #endif | |
| #define escape(action) \ | |
| for(; fragment_length--; CORD_next(pos)) { \ | |
| char c=CORD_pos_fetch(pos); \ | |
| action \ | |
| } | } |
| #define to_char(a, c) case a: *dest++=c; break | #define _default default: CORD_ec_append(result, c); break |
| #define _default default: *dest++=*src; break | |
| #define to_string(a, b, bsize) \ | |
| case a: \ | |
| strncpy(dest, b, bsize); \ | |
| dest+=bsize; \ | |
| break | |
| #define encode(need_encode_func, prefix) \ | #define encode(need_encode_func, prefix) \ |
| default: \ | if(need_encode_func(c)) { \ |
| if(need_encode_func(*src)) { \ | static const char* hex="0123456789ABCDEF"; \ |
| static const char *hex="0123456789ABCDEF"; \ | CORD_ec_append(result, prefix); \ |
| char chunk[3]={prefix}; \ | CORD_ec_append(result, hex[((unsigned char)c)/0x10]); \ |
| chunk[1]=hex[((unsigned char)*src)/0x10]; \ | CORD_ec_append(result, hex[((unsigned char)c)%0x10]); \ |
| chunk[2]=hex[((unsigned char)*src)%0x10]; \ | } else \ |
| strncpy(dest, chunk, 3); dest+=3; \ | CORD_ec_append(result, c); |
| } else \ | #define to_char(c) CORD_ec_append(result, c) |
| *dest++=*src; \ | #define to_string(s) CORD_ec_append_cord(result, s) |
| break | |
| inline bool need_file_encode(unsigned char c){ | inline bool need_file_encode(unsigned char c){ |
| if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) | // russian letters and space ENABLED |
| return false; | // encoding only these... |
| return strchr( | |
| return !strchr("./", c); | "*?'\"<>|" |
| #ifndef WIN32 | |
| ":\\~" | |
| #endif | |
| , c)!=0; | |
| } | } |
| inline bool need_uri_encode(unsigned char c){ | inline bool need_uri_encode(unsigned char c){ |
| if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) | if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) |
| Line 55 inline bool need_uri_encode(unsigned cha | Line 61 inline bool need_uri_encode(unsigned cha |
| return !strchr("_-./", c); | return !strchr("_-./", c); |
| } | } |
| inline bool need_header_encode(unsigned char c){ | inline bool need_http_header_encode(unsigned char c){ |
| if(strchr(" ,:", c)) | if(strchr(" , :", c)) |
| return false; | return false; |
| return need_uri_encode(c); | return need_uri_encode(c); |
| } | } |
| // | |
| 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 | |
| // untaint languages. assigned by ^untaint[lang]{...} | |
| "P", | |
| /**< | |
| leave language built into string being appended. | |
| just a flag, that value not stored | |
| */ | |
| "A", ///< leave all characters intact | |
| "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 | |
| "XML", ///< ^dom:set xml | |
| "HTML" ///< HTML code (for editing) | |
| }; | |
| // String | // String |
| static bool typo_present(Array::Item *value, const void *info) { | /* |
| Array *row=static_cast<Array *>(value); | |
| const char *src=static_cast<const char *>(info); | HTTP-header = field-name ":" [ field-value ] CRLF |
| int partial; | field-name = token |
| row->get_string(0)->cmp(src, partial); | field-value = *( field-content | LWS ) |
| return | |
| partial==0 || // full match | field-content = <the OCTETs making up the field-value |
| partial==1; // typo left column starts 'src' | and consisting of either *TEXT or combinations |
| } | of token, tspecials, and quoted-string> |
| /// @todo optimize whitespaces for all but 'html' | |
| char *String::store_to(char *dest) const { | |
| // $MAIN:html-typo table | word = token | quoted-string |
| Table *typo_table=static_cast<Table *>(pool().tag()); | |
| token = 1*<any CHAR except CTLs or tspecials> | |
| const Chunk *chunk=&head; | |
| do { | |
| const Chunk::Row *row=chunk->rows; | |
| for(int i=0; i<chunk->count; i++) { | tspecials = "(" | ")" | "<" | ">" | "@" |
| if(row==append_here) | | "," | ";" | ":" | "\" | <"> |
| goto break2; | | "/" | "[" | "]" | "?" | "=" |
| | "{" | "}" | SP | HT | |
| // WARNING: | |
| // string can grow only UNTAINT_TIMES_BIGGER | SP = <US-ASCII SP, space (32)> |
| switch(row->item.lang) { | HT = <US-ASCII HT, horizontal-tab (9)> |
| case UL_NO: | |
| // clean piece | LWS = [CRLF] 1*( SP | HT ) |
| case UL_YES: | TEXT = <any OCTET except CTLs, |
| // tainted piece, but undefined untaint language | but including LWS> |
| // for VString.get_double of tainted values | |
| // for ^process{body} evaluation | quoted-pair = "\" CHAR |
| case UL_AS_IS: | |
| // tainted, untaint language: as-is | if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr)) |
| memcpy(dest, row->item.ptr, row->item.size); | */ |
| dest+=row->item.size; | inline bool need_quote_http_header(const char* ptr, size_t size) { |
| break; | for(; size--; ptr++) |
| case UL_FILE_NAME: | if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr)) |
| // tainted, untaint language: file [name] | return true; |
| escape( | return false; |
| to_char(' ', '_'); | } |
| encode(need_file_encode, '-'); | |
| ); | |
| break; | /** |
| case UL_URI: | appends to other String, |
| // tainted, untaint language: uri | marking all tainted pieces of it with @a lang. |
| escape( | or marking ALL pieces of it with a @a lang when @a forced to, |
| to_char(' ', '+'); | and propagating OPTIMIZE language bit. |
| encode(need_uri_encode, '%'); | */ |
| ); | String& String::append_to(String& dest, Language lang, bool forced) const { |
| break; | if(is_empty()) |
| case UL_HEADER: | return dest; |
| // tainted, untaint language: header | |
| escape( | // first: letters |
| encode(need_header_encode, '%'); | dest.body.append(body); |
| ); | |
| break; | // next: fragment infos |
| case UL_TABLE: | |
| // tainted, untaint language: table | if(lang==L_PASS_APPENDED) // without language-chage string? |
| escape( | dest.fragments.append(fragments); |
| to_char('\t', ' '); | else if(forced) //forcing passed lang? |
| to_char('\n', ' '); | dest.fragments+=Fragment(lang, length()); |
| _default; | else if(lang&L_OPTIMIZE_BIT) { |
| ); | Array_iterator<ArrayFragment::element_type> i(fragments); |
| break; | while(i.has_next()) { |
| case UL_SQL: | const Fragment& fragment=i.next(); |
| // tainted, untaint language: sql | // main idea here: |
| // TODO: зависимость от sql сервера | // tainted piece would get OPTIMIZED bit from 'lang' |
| memset(dest, '?', row->item.size); | // clean piece would be marked OPTIMIZED manually |
| dest+=row->item.size; | // pieces with determined languages [not tainted|clean] would retain theirs langs |
| break; | dest.fragments+=Fragment(static_cast<Language>( |
| case UL_JS: | fragment.lang==L_TAINTED?lang:( |
| escape( | fragment.lang==L_CLEAN?L_CLEAN|L_OPTIMIZE_BIT: // ORing with OPTIMIZED flag |
| to_string('"', "\\\"", 2); | fragment.lang)), |
| to_string('\'', "\\'", 2); | fragment.length); |
| to_string('\n', "\\n", 2); | } |
| to_string('\\', "\\\\", 2); | } else { // The core idea: tainted pieces got marked with context's lang |
| to_string('\xFF', "\\\xFF", 2); | Array_iterator<ArrayFragment::element_type> i(fragments); |
| _default; | while(i.has_next()) { |
| ); | const Fragment& fragment=i.next(); |
| break; | dest.fragments+=Fragment( |
| case UL_HTML: | fragment.lang==L_TAINTED?lang:fragment.lang, |
| escape( | fragment.length); |
| to_string('&', "&", 5); | } |
| to_string('>', ">", 4); | } |
| to_string('<', "<",4); | |
| to_string('"', """,6); | return dest; |
| to_char('\t', ' '); | } |
| //TODO: XSLT to_string('\'', "'", 6) | |
| _default; | /** http://www.ietf.org/rfc/rfc2047.txt |
| ); | RFC |
| break; | (3) As a replacement for a 'word' entity within a 'phrase', for example, |
| case UL_HTML_TYPO: { | one that precedes an address in a From, To, or Cc header. The ABNF |
| // tainted, untaint language: html-typo | definition for 'phrase' from RFC 822 thus becomes: |
| char *html=(char *)malloc(size()*6/*""" the longest possible*/+1); | |
| size_t html_size; | phrase = 1*( encoded-word / word ) |
| { // local dest | |
| char *dest=html; | In this case the set of characters that may be used in a "Q"-encoded |
| escape( | 'encoded-word' is restricted to: <upper and lower case ASCII |
| to_string('&', "&", 5); | letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_" |
| to_string('>', ">", 4); | (underscore, ASCII 95.)>. An 'encoded-word' that appears within a |
| to_string('<', "<",4); | 'phrase' MUST be separated from any adjacent 'word', 'text' or |
| to_string('"', """,6); | 'special' by 'linear-white-space'. |
| to_char('\t', ' '); | ... |
| // convinient name for typo match "\n" | (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be |
| case '\r': | represented as "_" (underscore, ASCII 95.). (This character may |
| *dest++='\\'; *dest++='n'; // \r -> \n | not pass through some internetwork mail gateways, but its use |
| if(src[1]=='\n') // \r\n -> remove \n | will greatly enhance readability of "Q" encoded data with mail |
| src++; | readers that do not support this encoding.) Note that the "_" |
| break; | always represents hexadecimal 20, even if the SPACE character |
| to_string('\n', "\\n", 2); | occupies a different code position in the character set in use. |
| //TODO: XSLT to_string('\'', "'", 6) | |
| _default; | paf: obviously, |
| ); | without "=", or one could not differ "=E0" and "russian letter a" |
| *dest=0; | and without "_", or in would mean 0x20 |
| html_size=dest-html; | */ |
| } | inline bool mail_header_char_valid_within_Qencoded(char c) { |
| // typo table replacements | return c>='A' && c<='Z' |
| if(typo_table) { | || c>='a' && c<='Z' |
| const char *src=html; | || c>='0' && c<='9' |
| do { | || strchr("!*+-/", c); |
| // there is a row where first column starts 'src' | } |
| if(Table::Item *item=typo_table->first_that(typo_present, src)) { | inline void ec_append(CORD_ec& result, CORD_pos& pos, size_t size) { |
| // get a=>b values | while(size--) { |
| const String& a=*static_cast<Array *>(item)->get_string(0); | CORD_ec_append(result, CORD_pos_fetch(pos)); |
| const String& b=*static_cast<Array *>(item)->get_string(1); | CORD_next(pos); |
| // empty 'a' check | } |
| if(a.size()==0) { | } |
| pool().set_tag(0); // avoid recursion | StringBody String::cstr_to_string_body(Language lang, |
| THROW(0, 0, | SQL_Connection* connection, |
| typo_table->origin_string(), | const Request_charsets *charsets) const { |
| "typo table first column elements must not be empty"); | CORD_ec result; CORD_ec_init(result); |
| } | CORD_pos pos; body.set_pos(pos, 0); |
| // overflow check: | |
| // b allowed to be max UNTAINT_TIMES_BIGGER then a | bool whitespace=true; |
| if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) { | |
| pool().set_tag(0); // avoid recursion | size_t fragment_begin=0; |
| THROW(0, 0, | size_t fragment_end; |
| &b, | for(Array_iterator<Fragment> i(fragments); i.has_next(); fragment_begin=fragment_end) { |
| "is %g times longer then '%s', " | const Fragment& fragment=i.next(); |
| "while maximum, handled by Parser, is %d", | size_t fragment_length=fragment.length; |
| ((double)b.size())/a.size(), | fragment_end=fragment_begin+fragment_length; |
| a.cstr(), | |
| UNTAINT_TIMES_BIGGER); | Language to_lang=lang==L_UNSPECIFIED?fragment.lang:lang; |
| } | |
| switch(to_lang & ~L_OPTIMIZE_BIT) { | |
| // skip 'a' in 'src' | case L_CLEAN: |
| src+=a.size(); | case L_TAINTED: |
| // write 'b' to 'dest' | case L_AS_IS: |
| b.store_to(dest); | // clean piece |
| dest+=b.size(); | |
| } else | // tainted piece, but undefined untaint language |
| *dest++=*src++; | // for VString.as_double of tainted values |
| } while(*src); | // for ^process{body} evaluation |
| } else { | |
| memcpy(dest, html, html_size); | // tainted, untaint language: as-is |
| dest+=html_size; | ec_append(result, pos, fragment_length); |
| break; | |
| case L_FILE_SPEC: | |
| // tainted, untaint language: file [name] | |
| escape( | |
| // Macintosh has problems with small Russian letter 'r' | |
| if( c=='\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; | |
| #ifdef LATER | |
| case L_URI: | |
| // tainted, untaint language: uri | |
| { | |
| const char *substr= | |
| CORD_to_const_char_star( | |
| CORD_substr(body, fragment_begin, fragment_length)); | |
| const void *client_ptr; | |
| size_t client_size; | |
| if(charsets) | |
| Charset::transcode( | |
| charsets->source(), substr, fragment_length, | |
| charsets->client(), client_ptr, client_size); | |
| else { | |
| client_ptr=substr; | |
| client_size=fragment_length; | |
| } | } |
| break; | |
| char c; | |
| for(const char* src=(const char* )client_ptr; c=*src++; ) | |
| encode(need_uri_encode, '%'); | |
| } | |
| break; | |
| #endif | |
| case L_HTTP_HEADER: | |
| // tainted, untaint language: http-field-content-text | |
| escape( | |
| encode(need_uri_encode, '%'); | |
| ); | |
| break; | |
| #ifdef LATER | |
| case L_MAIL_HEADER: | |
| // tainted, untaint language: mail-header | |
| // http://www.ietf.org/rfc/rfc2047.txt | |
| if(charsets) { | |
| const void *mail_ptr=fragment.ptr; | |
| size_t mail_size=fragment_length; | |
| const char* 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<end; src++) { | |
| //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'. | |
| if(to_quoted_printable && (c==',' || c=='<')) { | |
| email=c=='<'; | |
| dest+=sprintf(dest, "?="); | |
| to_quoted_printable=false; | |
| } | |
| 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) | |
| )) { | |
| if(!to_quoted_printable) { | |
| dest+=sprintf(dest, "=?%s?Q?", charset_name.get()); | |
| to_quoted_printable=true; | |
| } | |
| //RFC Upper case should be used for hexadecimal digits "A" through "F" | |
| if(c == 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", c & 0xFF); | |
| } else | |
| *dest++=c; | |
| if(c=='>') | |
| email=false; | |
| } | } |
| default: | if(to_quoted_printable) // close |
| THROW(0,0, | dest+=sprintf(dest, "?="); |
| this, | |
| "unknown untaint language #%d of %d piece", | } else { |
| static_cast<int>(row->item.lang), | memcpy(dest, fragment.ptr, fragment_length); |
| i); | dest+=fragment_length; |
| } | } |
| row++; | break; |
| #endif | |
| case L_TABLE: | |
| // tainted, untaint language: table | |
| escape(switch(c) { | |
| case '\t': to_char(' '); break; | |
| case '\n': to_char(' '); break; | |
| _default; | |
| }); | |
| break; | |
| #ifdef PA_SQL | |
| case L_SQL: | |
| // tainted, untaint language: sql | |
| if(connection) { | |
| dest+=connection->quote(0, | |
| CORD_substr(body, fragment_begin, fragment_length), | |
| fragment_length); | |
| ...^^^ | |
| dest+=connection->quote(dest, fragment.ptr, fragment_length); | |
| else | |
| throw Exception(0, | |
| 0, | |
| "untaint in SQL language failed - no connection specified"); | |
| break; | |
| #endif | |
| case L_JS: | |
| escape(switch(c) { | |
| case '"': to_string("\\\""); break; | |
| case '\'': to_string("\\'"); break; | |
| case '\n': to_string("\\n"); break; | |
| case '\\': to_string("\\\\"); break; | |
| case '\xFF': to_string("\\\xFF"); break; | |
| _default; | |
| }); | |
| break; | |
| case L_XML: | |
| escape(switch(c) { | |
| case '&': to_string("&"); break; | |
| case '>': to_string(">"); break; | |
| case '<': to_string("<"); break; | |
| case '"': to_string("""); break; | |
| case '\'': to_string("'"); break; | |
| _default; | |
| }); | |
| break; | |
| case L_HTML: | |
| escape(switch(c) { | |
| case '&': to_string("&"); break; | |
| case '>': to_string(">"); break; | |
| case '<': to_string("<"); break; | |
| case '"': to_string("""); break; | |
| _default; | |
| }); | |
| break; | |
| default: | |
| throw Exception(0, | |
| 0, | |
| "unknown untaint language #%d", | |
| static_cast<int>(to_lang)); // should never | |
| break; // never | |
| } | } |
| chunk=row->link; | #ifdef LATER |
| } while(chunk); | if(to_lang & L_OPTIMIZE_BIT) { |
| break2: | // optimizing whitespace |
| return dest; | char *stop=dest; dest=start; |
| for(char c=start; src<stop; src++) | |
| switch(c) { | |
| // of all consequent white space chars leaving only first one | |
| case ' ': case '\r': case '\n': case '\t': | |
| if(!whitespace) { | |
| *dest++=c; | |
| whitespace=true; | |
| } | |
| break; | |
| default: | |
| whitespace=false; | |
| *dest++=c; | |
| break; | |
| }; | |
| } else // piece without optimization | |
| whitespace=false; | |
| #endif | |
| } | |
| return StringBody(CORD_ec_to_cord(result)); | |
| } | } |