--- parser3/src/main/untaint.C 2001/04/04 10:50:36 1.32 +++ parser3/src/main/untaint.C 2002/12/05 12:53:48 1.111 @@ -1,14 +1,11 @@ /** @file Parser: String class part: untaint mechanizm. - Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) - - $Id: untaint.C,v 1.32 2001/04/04 10:50:36 paf Exp $ + Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -#include "pa_config_includes.h" +static const char* IDENT_UNTAINT_C="$Date: 2002/12/05 12:53:48 $"; #include "pa_pool.h" #include "pa_string.h" @@ -16,6 +13,16 @@ #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" + +//#define DEBUG_STRING_APPENDS_VS_EXPANDS + +#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS +ulong string_string_shortcut_economy=0; +#endif #define escape(action) \ { \ @@ -25,30 +32,29 @@ } #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')) @@ -56,190 +62,545 @@ 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; 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 -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 + + 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 +//#include "pa_sapi.h" +/** + appends 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. + + using architecture advantage: after string-to-string-append string never modified. + algorithm: + if no language-change specified and src not yet appended to some other string[last_chunk!=0] + shrinking dest last_chunk[preparing it for linking], + ///shrinking src last_chunk[preparing it to be linked, consequent dest.appends would go there], + linking[dest.last_chunk = src.head.chunk] + if some language-change specified or src already appended to some other string[last_chunk==0] + cloning pieces. */ -char *String::store_to(char *dest, Untaint_lang lang) 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 Chunk *chunk=&head; - do { - const Chunk::Row *row=chunk->rows; - for(size_t i=0; icount; i++, row++) { - if(row==append_here) - goto break2; - - // WARNING: - // string can grow only UNTAINT_TIMES_BIGGER - switch(lang==UL_UNKNOWN?row->item.lang:lang) { - case UL_CLEAN: - // clean piece - 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); +String& String::append(const String& src, uchar lang, bool forced) { + if(!last_chunk) // growth stopped [we're appended as string to somebody] + throw Exception(0, + this, + "string growth stopped (append string)"); + + if(src.is_empty()) + return *this; + + // without language-chage, not-appended-before, big[not fitting our tail] string? + if(lang==UL_PASS_APPENDED + && src.last_chunk + && (uint(&last_chunk->rows[last_chunk->count]-append_here) < src.used_rows())) { +#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS + string_string_shortcut_economy+=src.used_rows()*sizeof(String::Chunk::Row); +#endif + + // using fact: + // src.head.chunk.count initally equeals this.head.chunk.count and shrinks-only, + // so can't be more than this.head.chunk.count, + // which means that we know that + // src.head.chunk would fit into this.head.chunk + if(is_empty()) { // our head.chunk is empty + // they have more than head.chunk? we need all head.chunk : we need only filled-part of head.chunk + Chunk *src_head_link=src.head.chunk.rows[src.head.chunk.count].link; + size_t head_count=src_head_link?src.head.chunk.count:(src.append_here-src.head.chunk.rows); + // "your head.chunk is my head.chunk" + memcpy(head.chunk.rows, src.head.chunk.rows, sizeof(Chunk::Row)*(head_count)); + if(src_head_link) { + // "your body is my body" + head.chunk.rows[head.chunk.count=head_count].link=src_head_link; + // "your last_chunk is mine now" + last_chunk=src.last_chunk; + // "your append_here is mine now" + append_here=src.append_here; + } else { + // "your last_chunk is mine now" + last_chunk=&head.chunk; + // "your append_here is recalc-mine now" + append_here=head.chunk.rows+head_count; + } + } else { // our head.chunk contains something + // "chopping off my tail-reserve" + last_chunk->count=append_here-last_chunk->rows; + // "you is my tail" + append_here->link=&src.head.chunk; + // "your last_chunk is mine now" + last_chunk=src.last_chunk; + // "your append_here is mine now" + append_here=src.append_here; + } + + // stop-growing mark + src.last_chunk=0; + return *this; + } + + // manually unrolled code to avoid do{if(const)} constructs + if(forced) + STRING_SRC_FOREACH_ROW( + APPEND(row->item.ptr, row->item.size, + lang, //forcing passed lang + row->item.origin.file, row->item.origin.line); + ) + else if(lang==UL_PASS_APPENDED) + STRING_SRC_FOREACH_ROW( + APPEND(row->item.ptr, row->item.size, + row->item.lang, // passing item's lang + row->item.origin.file, row->item.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_SRC_FOREACH_ROW( + APPEND(row->item.ptr, row->item.size, + row->item.lang==UL_TAINTED?lang:( + row->item.lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag + row->item.lang + ), + row->item.origin.file, row->item.origin.line); + ) + else + STRING_SRC_FOREACH_ROW( + APPEND(row->item.ptr, row->item.size, + row->item.lang==UL_TAINTED?lang:row->item.lang, + row->item.origin.file, row->item.origin.line); + ); +/* + for(Chunk::Row *row=last_chunk->rows; rowlink==(void*)0xcdcdcdcd) + _asm int 3;*/ + return *this; +} + +size_t String::cstr_bufsize(Untaint_lang lang, + SQL_Connection *connection, + Charset *buf_charset) const { + size_t dest=1; // for terminating 0 + STRING_FOREACH_ROW( + uchar to_lang=lang==UL_UNSPECIFIED?row->item.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+=row->item.size; + break; + case UL_FILE_SPEC: + // tainted, untaint language: file [name] + dest+=row->item.size*3/* worst: Z->%XX */; + break; + case UL_URI: + // tainted, untaint language: uri + dest+=row->item.size*6*3/* worst utf8 x worst Z->%XX */; + break; + case UL_HTTP_HEADER: + // tainted, untaint language: http-field-content-text + dest+=row->item.size*3/* worst: Z->%XX */; + break; + case UL_MAIL_HEADER: + // tainted, untaint language: mail-header + if(buf_charset) { + int parts_count=1; + const char *prev=row->item.ptr; + size_t size=row->item.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+= + row->item.size*3+ + parts_count*(buf_charset->name().size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */); + } else 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; + break; + case UL_TABLE: + // tainted, untaint language: table + dest+=row->item.size; + break; + case UL_SQL: + // tainted, untaint language: sql + if(connection) + dest+=connection->quote(0, row->item.ptr, row->item.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, + Charset *store_to_charset, + const char *store_to_charset_name) const { + // WARNING: + // before any changes check cstr_bufsize first!!! + bool whitespace=true; +//debug + const Chunk *chunk=&(*this).head.chunk; \ + const Chunk::Row *row=chunk->rows; \ + uint countdown=chunk->count; \ + while(row!=(*this).append_here) { \ + if(countdown==0) { \ + chunk=row->link; \ + row=chunk->rows; \ + countdown=chunk->count; \ + }; +//*/ +// STRING_FOREACH_ROW( + + uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang: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, row->item.ptr, row->item.size); + dest+=row->item.size; + break; + case UL_FILE_SPEC: + // tainted, untaint language: file [name] + escape( + encode(need_file_encode, '_'); + ); + break; + case UL_URI: + // tainted, untaint language: uri + const void *client_ptr; + size_t client_size; + Charset::transcode(pool(), + pool().get_source_charset(), row->item.ptr, row->item.size, + pool().get_client_charset(), client_ptr, client_size); + { + const char *src=(const char *)client_ptr; + for(int size=client_size; size--; src++) encode(need_uri_encode, '%'); - }); - break; - case UL_HEADER: - // tainted, untaint language: header - escape(switch(*src) { - encode(need_header_encode, '%'); - }); - 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 - // TODO: зависимость от sql сервера - memset(dest, '?', row->item.size); - dest+=row->item.size; - 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_HTML: - escape(switch(*src) { - case '&': to_string("&", 5); break; - case '>': 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_HTML_TYPO: { - // tainted, untaint language: html-typo - 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) { - // 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); + } + 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(store_to_charset && store_to_charset_name) { + const void *mail_ptr; + size_t mail_size; + Charset::transcode(pool(), + pool().get_source_charset(), row->item.ptr, row->item.size, + *store_to_charset, mail_ptr, mail_size); + + // 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 closed=false; + bool email=false; + for(const char *end=src+mail_size; src') + email=false; } - default: - THROW(0, 0, - this, - "unknown untaint language #%d of %d piece", - static_cast(row->item.lang), - i); + if(to_quoted_printable && !closed) // 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 + 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, row->item.ptr, row->item.size); + else + throw Exception(0, + 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, + this, + "unknown untaint language #%d", + static_cast(to_lang)); // should never + break; // never } - chunk=row->link; - } while(chunk); -break2: + + if(to_lang & UL_OPTIMIZE_BIT) { + // optimizing whitespace + char *stop=dest; dest=start; + for(char *src=start; srcitem.origin.file) + dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, + row->item.origin.file, + 1+row->item.origin.line); + else + dest+=sprintf(dest, ""); +); + uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT; + if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0])) + throw Exception(0, + this, + "unknown untaint language #%d", + static_cast(show_lang)); // sould never + + dest+=sprintf(dest, "#%s%s: ", + String_Untaint_lang_name[show_lang], + row->item.lang & UL_OPTIMIZE_BIT?".O":""); + char *dest_after_origins=dest; + + memcpy(dest, row->item.ptr, row->item.size); + dest+=row->item.size; + + remove_crlf(dest_after_origins, dest); + to_char('\n'); + ); + + *dest=0; + return result; +}