|
|
| version 1.17, 2001/03/25 09:10:30 | version 1.90, 2002/02/08 08:32:34 |
|---|---|
| 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, 2002 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) | |
| $Id$ | $Id$ |
| */ | */ |
| #include "pa_config_includes.h" | |
| #include "pa_pool.h" | #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_sql_connection.h" | |
| #include "pa_dictionary.h" | |
| #include "pa_common.h" | |
| #include "pa_charset.h" | |
| #define escape(cases) \ | #define escape(action) \ |
| { \ | { \ |
| const char *src=row->item.ptr; \ | const char *src=row->item.ptr; \ |
| for(int size=row->item.size; size--; src++) \ | for(int size=row->item.size; size--; src++) \ |
| switch(*src) { \ | action \ |
| cases \ | |
| } \ | |
| } | } |
| #define to_char(a, c) case a: *dest++=c; break | |
| #define _default default: *dest++=*src; 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(*src)) { \ | if(need_encode_func(*src)) { \ |
| static const char *hex="0123456789ABCDEF"; \ | static const char *hex="0123456789ABCDEF"; \ |
| char chunk[3]={prefix}; \ | char chunk[3]={prefix}; \ |
| chunk[1]=hex[((unsigned char)*src)/0x10]; \ | chunk[1]=hex[((unsigned char)*src)/0x10]; \ |
| chunk[2]=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 \ | } else \ |
| *dest++=*src; \ | *dest++=*src; \ |
| break | break |
| #define to_char(c) *dest++=c | |
| #define to_string(b, bsize) \ | |
| memcpy(dest, b, bsize); \ | |
| dest+=bsize; \ | |
| inline bool need_file_encode(unsigned char c){ | inline bool need_file_encode(unsigned char c){ |
| // theoretical problem with, for instance, "_2B" and "." fragments, | |
| // they would yield the same | |
| // because need_file_encode('_')=false | |
| // but we need to delete such files somehow, getting names from ^index | |
| 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')) |
| return false; | return false; |
| return !strchr("./", c); | return !strchr( |
| " _./()-" | |
| #ifdef WIN32 | |
| ":\\~" | |
| #endif | |
| , c); | |
| } | } |
| 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 62 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> | |
| word = token | quoted-string | |
| token = 1*<any CHAR except CTLs or tspecials> | |
| tspecials = "(" | ")" | "<" | ">" | "@" | |
| | "," | ";" | ":" | "\" | <"> | |
| | "/" | "[" | "]" | "?" | "=" | |
| | "{" | "}" | SP | HT | |
| SP = <US-ASCII SP, space (32)> | |
| HT = <US-ASCII HT, horizontal-tab (9)> | |
| LWS = [CRLF] 1*( SP | HT ) | |
| TEXT = <any OCTET except CTLs, | |
| but including LWS> | |
| 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; | |
| } | |
| String& String::append(const String& src, uchar lang, bool forced) { | |
| // 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); | |
| ); | |
| break2: | |
| 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) { | |
| // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?= | |
| dest+= | |
| row->item.size*3+ | |
| buf_charset->name().size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */; | |
| } else | |
| dest+=row->item.size; | |
| 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; | |
| } | |
| ); | |
| break2: | |
| return dest; | |
| } | } |
| /// @todo optimize whitespaces for all but 'html' | char *String::store_to(char *dest, Untaint_lang lang, |
| char *String::store_to(char *dest) const { | SQL_Connection *connection, |
| // $MAIN:html-typo table | Charset *store_to_charset) const { |
| Table *typo_table=static_cast<Table *>(pool().tag()); | // WARNING: |
| // before any changes check cstr_bufsize first!!! | |
| bool whitespace=true; | |
| // expanded STRING_FOREACH_ROW here for debugging purposes | |
| const Chunk *chunk=&head; \ | |
| do { \ | |
| const Chunk::Row *row=chunk->rows; \ | |
| for(uint i=0; i<chunk->count; i++, row++) { \ | |
| if(row==append_here) \ | |
| goto break2; \ | |
| \ | |
| 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++) | |
| switch(*src) { | |
| case ' ': to_char('+'); break; | |
| default: encode(need_uri_encode, '%'); | |
| }; | |
| } | |
| break; | |
| case UL_HTTP_HEADER: | |
| // tainted, untaint language: http-field-content-text | |
| escape(switch(*src) { | |
| case ' ': to_char('+'); break; | |
| default: encode(need_uri_encode, '%'); | |
| }); | |
| break; | |
| case UL_MAIL_HEADER: | |
| // tainted, untaint language: mail-header | |
| if(store_to_charset) { | |
| 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; | |
| for(int size=mail_size; size--; src++) { | |
| if(*src & 0x80) { | |
| if(!to_quoted_printable) { | |
| dest+=sprintf(dest, "=?%s?Q?", store_to_charset->name().cstr()); | |
| to_quoted_printable=true; | |
| } | |
| dest+=sprintf(dest, "=%02X", *src & 0xFF); | |
| } else { | |
| *dest++=*src; | |
| } | |
| } | |
| if(to_quoted_printable) // 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, 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, 0, | |
| this, | |
| "unknown untaint language #%d", | |
| static_cast<int>(row->item.lang)); // sould never | |
| break; // never | |
| } | |
| if(to_lang & UL_OPTIMIZE_BIT) { | |
| // optimizing whitespace | |
| char *stop=dest; dest=start; | |
| for(char *src=start; src<stop; src++) | |
| switch(*src) { | |
| // of all consequent white space chars leaving only first one | |
| case ' ': case '\r': case '\n': case '\t': | |
| if(!whitespace) { | |
| *dest++=*src; | |
| whitespace=true; | |
| } | |
| break; | |
| default: | |
| whitespace=false; | |
| *dest++=*src; | |
| break; | |
| }; | |
| } else // piece without optimization | |
| whitespace=false; | |
| } \ | |
| chunk=row->link; \ | |
| } while(chunk); \ | |
| break2: | |
| return dest; | |
| } | |
| char *String::cstr_debug_origins() const { | |
| //_asm int 3; | |
| char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2); | |
| char *dest=result; | |
| const Chunk *chunk=&head; | const Chunk *chunk=&head; |
| do { | do { |
| const Chunk::Row *row=chunk->rows; | const Chunk::Row *row=chunk->rows; |
| for(int i=0; i<chunk->count; i++) { | for(uint i=0; i<chunk->count; i++, row++) { |
| if(row==append_here) | if(row==append_here) |
| goto break2; | goto break2; |
| // WARNING: | #ifndef NO_STRING_ORIGIN |
| // string can grow only UNTAINT_TIMES_BIGGER | if(row->item.origin.file) |
| switch(row->item.lang) { | dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT, |
| case UL_NO: | row->item.origin.file, |
| // clean piece | 1+row->item.origin.line); |
| case UL_YES: | else |
| // tainted piece, but undefined untaint language | dest+=sprintf(dest, "<unknown>"); |
| // for VString.get_double of tainted values | #endif |
| // for ^process{body} evaluation | uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT; |
| case UL_AS_IS: | if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0])) |
| // tainted, untaint language: as-is | throw Exception(0, 0, |
| memcpy(dest, row->item.ptr, row->item.size); | this, |
| dest+=row->item.size; | "unknown untaint language #%d", |
| break; | static_cast<int>(show_lang)); // sould never |
| case UL_FILE_NAME: | |
| // tainted, untaint language: file [name] | dest+=sprintf(dest, "#%s%s: ", |
| escape( | String_Untaint_lang_name[show_lang], |
| to_char(' ', '_'); | row->item.lang & UL_OPTIMIZE_BIT?".O":""); |
| encode(need_file_encode, '-'); | char *dest_after_origins=dest; |
| ); | |
| break; | memcpy(dest, row->item.ptr, row->item.size); |
| case UL_URI: | dest+=row->item.size; |
| // tainted, untaint language: uri | |
| escape( | remove_crlf(dest_after_origins, dest); |
| to_char(' ', '+'); | to_char('\n'); |
| encode(need_uri_encode, '%'); | |
| ); | |
| break; | |
| case UL_HEADER: | |
| // tainted, untaint language: header | |
| escape( | |
| encode(need_header_encode, '%'); | |
| ); | |
| break; | |
| case UL_TABLE: | |
| // tainted, untaint language: table | |
| escape( | |
| to_char('\t', ' '); | |
| to_char('\n', ' '); | |
| _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( | |
| to_string('"', "\\\"", 2); | |
| to_string('\'', "\\'", 2); | |
| to_string('\n', "\\n", 2); | |
| to_string('\\', "\\\\", 2); | |
| to_string('\xFF', "\\\xFF", 2); | |
| _default; | |
| ); | |
| break; | |
| case UL_HTML: | |
| escape( | |
| to_string('&', "&", 5); | |
| to_string('>', ">", 4); | |
| to_string('<', "<",4); | |
| to_string('"', """,6); | |
| to_char('\t', ' '); | |
| //TODO: XSLT to_string('\'', "'", 6) | |
| _default; | |
| ); | |
| break; | |
| case UL_HTML_TYPO: { | |
| // tainted, untaint language: html-typo | |
| char *html=(char *)malloc(size()*6/*""" the longest possible*/+1); | |
| size_t html_size; | |
| { // local dest | |
| char *dest=html; | |
| escape( | |
| to_string('&', "&", 5); | |
| to_string('>', ">", 4); | |
| to_string('<', "<",4); | |
| to_string('"', """,6); | |
| to_char('\t', ' '); | |
| // convinient name for typo match "\n" | |
| case '\r': | |
| *dest++='\\'; *dest++='n'; // \r -> \n | |
| if(src[1]=='\n') // \r\n -> remove \n | |
| src++; | |
| break; | |
| to_string('\n', "\\n", 2); | |
| //TODO: XSLT to_string('\'', "'", 6) | |
| _default; | |
| ); | |
| *dest=0; | |
| html_size=dest-html; | |
| } | |
| // typo table replacements | |
| if(typo_table) { | |
| const char *src=html; | |
| 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<Array *>(item)->get_string(0); | |
| const String& b=*static_cast<Array *>(item)->get_string(1); | |
| // empty 'a' check | |
| if(a.size()==0) { | |
| pool().set_tag(0); // avoid recursion | |
| THROW(0, 0, | |
| typo_table->origin_string(), | |
| "typo table first 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, html_size); | |
| dest+=html_size; | |
| } | |
| break; | |
| } | |
| default: | |
| THROW(0,0, | |
| this, | |
| "unknown untaint language #%d of %d piece", | |
| static_cast<int>(row->item.lang), | |
| i); | |
| } | |
| row++; | |
| } | } |
| chunk=row->link; | chunk=row->link; |
| } while(chunk); | } while(chunk); |
| break2: | break2: |
| return dest; | *dest=0; |
| return result; | |
| } | } |