Annotation of parser3/src/main/untaint.C, revision 1.117
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.117 ! paf 4: Copyright(c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
1.89 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.103 paf 6: */
1.8 paf 7:
1.117 ! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/04/18 10:36:03 $";
! 9:
1.1 paf 10:
11: #include "pa_string.h"
12: #include "pa_hash.h"
13: #include "pa_exception.h"
1.13 paf 14: #include "pa_table.h"
1.32 paf 15: #include "pa_globals.h"
1.58 parser 16: #include "pa_dictionary.h"
1.66 parser 17: #include "pa_common.h"
1.85 paf 18: #include "pa_charset.h"
1.117 ! paf 19: #include "pa_request_charsets.h"
! 20: #include "pa_sapi.h"
1.1 paf 21:
1.117 ! paf 22: extern "C" { // author forgot to do that
! 23: #include "ec.h"
! 24: }
1.91 paf 25:
1.117 ! paf 26: #define PA_SQL
! 27:
! 28: #ifdef PA_SQL
! 29: #include "pa_sql_connection.h"
1.91 paf 30: #endif
31:
1.117 ! paf 32: // defines
! 33:
! 34:
! 35: #undef CORD_ec_append
! 36: // redefining to intercept flushes and implement whitespace optimization
! 37: // of all consequent white space chars leaving only first one
! 38: #define CORD_ec_append(x, c) \
! 39: { \
! 40: bool skip=false; \
! 41: if(optimize) switch(c) { \
! 42: case ' ': case '\r': case '\n': case '\t': \
! 43: if(whitespace) \
! 44: skip=true; /*skipping subsequent*/ \
! 45: else \
! 46: whitespace=true; \
! 47: break; \
! 48: default: \
! 49: whitespace=false; \
! 50: break; \
! 51: } \
! 52: if(!skip) { \
! 53: if ((x)[0].ec_bufptr == (x)[0].ec_buf + CORD_BUFSZ) { \
! 54: CORD_ec_flush_buf(x); \
! 55: } \
! 56: *((x)[0].ec_bufptr)++ = (c); \
! 57: } \
! 58: }
! 59:
! 60:
1.18 paf 61: #define escape(action) \
1.117 ! paf 62: for(; fragment_length--; CORD_next(pos)) { \
! 63: char c=CORD_pos_fetch(pos); \
! 64: action \
1.1 paf 65: }
1.117 ! paf 66: #define _default default: CORD_ec_append(result, c); break
! 67: #define encode(need_encode_func, prefix, otherwise) \
! 68: if(need_encode_func(c)) { \
! 69: static const char* hex="0123456789ABCDEF"; \
! 70: CORD_ec_append(result, prefix); \
! 71: CORD_ec_append(result, hex[((unsigned char)c)/0x10]); \
! 72: CORD_ec_append(result, hex[((unsigned char)c)%0x10]); \
! 73: } else \
! 74: CORD_ec_append(result, otherwise);
! 75: #define to_char(c) CORD_ec_append(result, c)
! 76: #define to_string(s) CORD_ec_append_cord(result, s)
1.4 paf 77:
1.9 paf 78: inline bool need_file_encode(unsigned char c){
1.108 paf 79: // russian letters and space ENABLED
80: // encoding only these...
81: return strchr(
82: "*?'\"<>|"
83: #ifndef WIN32
1.116 paf 84: ":\\"
1.31 paf 85: #endif
1.108 paf 86: , c)!=0;
1.9 paf 87: }
1.5 paf 88: inline bool need_uri_encode(unsigned char c){
1.13 paf 89: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 90: return false;
91:
1.5 paf 92: return !strchr("_-./", c);
93: }
1.36 paf 94: inline bool need_http_header_encode(unsigned char c){
1.18 paf 95: if(strchr(" , :", c))
1.5 paf 96: return false;
97:
98: return need_uri_encode(c);
1.4 paf 99: }
1.1 paf 100:
1.56 parser 101: //
102:
1.117 ! paf 103: static const char* String_Untaint_lang_name[]={
1.56 parser 104: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
105: "C", ///< clean
106: "T", ///< tainted, untaint language as assigned later
107: // untaint languages. assigned by ^untaint[lang]{...}
108: "P",
109: /**<
110: leave language built into string being appended.
111: just a flag, that value not stored
112: */
113: "A", ///< leave all characters intact
1.68 parser 114: "F", ///< file specification
115: "H", ///< ext in HTTP response header
1.56 parser 116: "M", ///< text in mail header
117: "URI", ///< text in uri
118: "T", ///< ^table:set body
119: "SQL", ///< ^table:sql body
120: "JS", ///< JavaScript code
1.68 parser 121: "XML", ///< ^dom:set xml
1.82 paf 122: "HTML" ///< HTML code (for editing)
1.56 parser 123: };
124:
125:
1.1 paf 126: // String
127:
1.41 paf 128: /*
129:
130: HTTP-header = field-name ":" [ field-value ] CRLF
131:
132: field-name = token
133: field-value = *( field-content | LWS )
134:
135: field-content = <the OCTETs making up the field-value
136: and consisting of either *TEXT or combinations
137: of token, tspecials, and quoted-string>
138:
139:
140:
141: word = token | quoted-string
142:
143: token = 1*<any CHAR except CTLs or tspecials>
144:
145:
146:
147: tspecials = "(" | ")" | "<" | ">" | "@"
148: | "," | ";" | ":" | "\" | <">
149: | "/" | "[" | "]" | "?" | "="
150: | "{" | "}" | SP | HT
151:
152: SP = <US-ASCII SP, space (32)>
153: HT = <US-ASCII HT, horizontal-tab (9)>
154:
155: LWS = [CRLF] 1*( SP | HT )
156: TEXT = <any OCTET except CTLs,
157: but including LWS>
158:
159: quoted-pair = "\" CHAR
160:
161: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
162: */
1.117 ! paf 163: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 164: for(; size--; ptr++)
1.42 paf 165: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 166: return true;
167: return false;
168: }
169:
1.117 ! paf 170:
1.91 paf 171: /**
1.117 ! paf 172: appends to other String,
1.91 paf 173: marking all tainted pieces of it with @a lang.
1.92 paf 174: or marking ALL pieces of it with a @a lang when @a forced to,
175: and propagating OPTIMIZE language bit.
1.117 ! paf 176: */
! 177: String& String::append_to(String& dest, Language lang, bool forced) const {
! 178: if(is_empty())
! 179: return dest;
1.91 paf 180:
1.117 ! paf 181: // first: letters
! 182: dest.body<<body;
1.94 paf 183:
1.117 ! paf 184: // next: fragment infos
! 185:
! 186: if(lang==L_PASS_APPENDED) // without language-change?
! 187: dest.fragments.append(fragments);
! 188: else if(forced) //forcing passed lang?
! 189: dest.fragments+=Fragment(lang, length());
! 190: else if(lang&L_OPTIMIZE_BIT) {
! 191: for(Array_iterator<ArrayFragment::element_type> i(fragments); i.has_next(); ) {
! 192: const Fragment fragment=i.next();
! 193: // main idea here:
! 194: // tainted piece would get OPTIMIZED bit from 'lang'
! 195: // clean piece would be marked OPTIMIZED manually
! 196: // pieces with determined languages [not tainted|clean] would retain theirs langs
! 197: dest.fragments+=Fragment(static_cast<Language>(
! 198: fragment.lang==L_TAINTED?lang:(
! 199: fragment.lang==L_CLEAN?L_CLEAN|L_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
! 200: fragment.lang)),
! 201: fragment.length);
! 202: }
! 203: } else { // The core idea: tainted pieces got marked with context's lang
! 204: for(Array_iterator<ArrayFragment::element_type> i(fragments); i.has_next(); ) {
! 205: const Fragment fragment=i.next();
! 206: dest.fragments+=Fragment(
! 207: fragment.lang==L_TAINTED?lang:fragment.lang,
! 208: fragment.length);
1.93 paf 209: }
1.91 paf 210: }
211:
1.117 ! paf 212: ASSERT_STRING_INVARIANT(dest);
1.75 paf 213: return dest;
1.51 parser 214: }
215:
1.109 paf 216: /** http://www.ietf.org/rfc/rfc2047.txt
217: RFC
218: (3) As a replacement for a 'word' entity within a 'phrase', for example,
219: one that precedes an address in a From, To, or Cc header. The ABNF
220: definition for 'phrase' from RFC 822 thus becomes:
221:
222: phrase = 1*( encoded-word / word )
223:
224: In this case the set of characters that may be used in a "Q"-encoded
225: 'encoded-word' is restricted to: <upper and lower case ASCII
226: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
227: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
228: 'phrase' MUST be separated from any adjacent 'word', 'text' or
229: 'special' by 'linear-white-space'.
230: ...
231: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
232: represented as "_" (underscore, ASCII 95.). (This character may
233: not pass through some internetwork mail gateways, but its use
234: will greatly enhance readability of "Q" encoded data with mail
235: readers that do not support this encoding.) Note that the "_"
236: always represents hexadecimal 20, even if the SPACE character
237: occupies a different code position in the character set in use.
238:
239: paf: obviously,
240: without "=", or one could not differ "=E0" and "russian letter a"
241: and without "_", or in would mean 0x20
242: */
1.117 ! paf 243: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 244: return c>='A' && c<='Z'
245: || c>='a' && c<='Z'
246: || c>='0' && c<='9'
247: || strchr("!*+-/", c);
248: }
1.117 ! paf 249: /**
! 250: RFC
! 251: Upper case should be used for hexadecimal digits "A" through "F"
! 252: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
! 253: may be represented as "_"
! 254: */
! 255: inline bool mail_header_nonspace_char(char c) {
! 256: return c != 0x20;
! 257: }
! 258: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
! 259: while(size--) {
! 260: CORD_ec_append(result, CORD_pos_fetch(pos));
! 261: CORD_next(pos);
! 262: }
! 263: }
! 264: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
! 265: while(true) {
! 266: size_t avail=CORD_pos_chars_left(pos);
! 267: if(avail==0) {
! 268: CORD_next(pos);
! 269: if(!--n)
! 270: break;
! 271: } else if(avail<n) {
! 272: CORD_pos_advance(pos, avail);
! 273: n-=avail;
! 274: } else { // avail>=n
! 275: CORD_pos_advance(pos, n);
! 276: break;
! 277: }
! 278: }
! 279: }
! 280: StringBody String::cstr_to_string_body(Language lang,
! 281: SQL_Connection* connection,
! 282: const Request_charsets *charsets) const {
! 283: CORD_ec result; CORD_ec_init(result);
! 284: CORD_pos pos; body.set_pos(pos, 0);
! 285:
1.44 paf 286: bool whitespace=true;
1.117 ! paf 287:
! 288: size_t fragment_begin=0;
! 289: size_t fragment_end;
! 290: for(Array_iterator<Fragment> i(fragments); i.has_next(); fragment_begin=fragment_end) {
! 291: const Fragment fragment=i.next();
! 292: size_t fragment_length=fragment.length;
! 293: fragment_end=fragment_begin+fragment_length;
! 294: //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
! 295:
! 296: Language to_lang=lang==L_UNSPECIFIED?fragment.lang:lang;
! 297: bool optimize=(to_lang & L_OPTIMIZE_BIT)!=0;
! 298: if(!optimize)
! 299: whitespace=false;
! 300:
! 301: switch(to_lang & ~L_OPTIMIZE_BIT) {
! 302: case L_CLEAN:
! 303: case L_TAINTED:
! 304: case L_AS_IS:
1.77 paf 305: // clean piece
306:
307: // tainted piece, but undefined untaint language
308: // for VString.as_double of tainted values
309: // for ^process{body} evaluation
310:
311: // tainted, untaint language: as-is
1.117 ! paf 312: ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 313: break;
1.117 ! paf 314: case L_FILE_SPEC:
1.77 paf 315: // tainted, untaint language: file [name]
1.83 paf 316: escape(
1.113 paf 317: // Macintosh has problems with small Russian letter 'r'
1.117 ! paf 318: if( c=='\xF0' && charsets && charsets->source().NAME()=="WINDOWS-1251" ) {
1.113 paf 319: // fixing that letter for most common charset
1.117 ! paf 320: to_char('p');
1.114 paf 321: } else // fallback to default
1.117 ! paf 322: encode(need_file_encode, '_', c);
1.83 paf 323: );
1.77 paf 324: break;
1.117 ! paf 325: case L_URI:
1.77 paf 326: // tainted, untaint language: uri
1.85 paf 327: {
1.117 ! paf 328: const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
! 329: // skip source [we use recoded version]
! 330: pa_CORD_pos_advance(pos, fragment_length);
! 331: String::C output(fragment_str, fragment_length);
! 332: if(charsets)
! 333: output=Charset::transcode(output,
! 334: charsets->source(),
! 335: charsets->client());
! 336:
! 337: char c;
! 338: for(const char* src=output.str; c=*src++; )
! 339: encode(need_uri_encode, '%', c);
1.85 paf 340: }
1.77 paf 341: break;
1.117 ! paf 342: case L_HTTP_HEADER:
1.77 paf 343: // tainted, untaint language: http-field-content-text
1.108 paf 344: escape(
1.117 ! paf 345: encode(need_uri_encode, '%', c);
1.108 paf 346: );
1.77 paf 347: break;
1.117 ! paf 348: case L_MAIL_HEADER:
1.77 paf 349: // tainted, untaint language: mail-header
1.105 paf 350: // http://www.ietf.org/rfc/rfc2047.txt
1.117 ! paf 351: if(charsets) {
1.87 paf 352: size_t mail_size;
1.117 ! paf 353: const char *mail_ptr=
! 354: body.mid(fragment_begin, mail_size=fragment_length).cstr();
! 355: // skip source [we use recoded version]
! 356: pa_CORD_pos_advance(pos, mail_size);
! 357:
! 358: const char* charset_name=charsets->mail().NAME().cstr();
1.87 paf 359:
1.77 paf 360: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
361: bool to_quoted_printable=false;
1.105 paf 362:
1.111 paf 363: bool email=false;
1.117 ! paf 364: uchar c;
! 365: for(const char* src=mail_ptr; c=(uchar)*src++; ) {
1.111 paf 366: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.117 ! paf 367: if(to_quoted_printable && (c==',' || c=='<')) {
! 368: email=c=='<';
! 369: to_string("?=");
1.107 paf 370: to_quoted_printable=false;
1.105 paf 371: }
1.111 paf 372: if(!email && (
1.117 ! paf 373: !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
! 374: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
1.105 paf 375: )) {
1.77 paf 376: if(!to_quoted_printable) {
1.117 ! paf 377: to_string("=?");
! 378: to_string(charset_name);
! 379: to_string("?Q?");
1.77 paf 380: to_quoted_printable=true;
381: }
1.117 ! paf 382: encode(mail_header_nonspace_char, '=', '_');
1.105 paf 383: } else
1.117 ! paf 384: to_char(c);
! 385: if(c=='>')
1.111 paf 386: email=false;
1.44 paf 387: }
1.112 paf 388: if(to_quoted_printable) // close
1.117 ! paf 389: to_string("?=");
1.87 paf 390:
1.117 ! paf 391: } else
! 392: ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 393: break;
1.117 ! paf 394: case L_TABLE:
1.77 paf 395: // tainted, untaint language: table
1.117 ! paf 396: escape(switch(c) {
1.77 paf 397: case '\t': to_char(' '); break;
398: case '\n': to_char(' '); break;
399: _default;
400: });
401: break;
1.117 ! paf 402: #ifdef PA_SQL
! 403: case L_SQL:
1.77 paf 404: // tainted, untaint language: sql
1.117 ! paf 405: if(connection) {
! 406: const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
! 407: // skip source [we use recoded version]
! 408: pa_CORD_pos_advance(pos, fragment_length);
! 409:
! 410: to_string(connection->quote(fragment_str, fragment_length));
! 411: } else
1.99 paf 412: throw Exception(0,
1.117 ! paf 413: 0,
1.77 paf 414: "untaint in SQL language failed - no connection specified");
415: break;
1.117 ! paf 416: #endif
! 417: case L_JS:
! 418: escape(switch(c) {
! 419: case '"': to_string("\\\""); break;
! 420: case '\'': to_string("\\'"); break;
! 421: case '\n': to_string("\\n"); break;
! 422: case '\\': to_string("\\\\"); break;
! 423: case '\xFF': to_string("\\\xFF"); break;
1.77 paf 424: _default;
425: });
426: break;
1.117 ! paf 427: case L_XML:
! 428: escape(switch(c) {
! 429: case '&': to_string("&"); break;
! 430: case '>': to_string(">"); break;
! 431: case '<': to_string("<"); break;
! 432: case '"': to_string("""); break;
! 433: case '\'': to_string("'"); break;
1.77 paf 434: _default;
435: });
436: break;
1.117 ! paf 437: case L_HTML:
! 438: escape(switch(c) {
! 439: case '&': to_string("&"); break;
! 440: case '>': to_string(">"); break;
! 441: case '<': to_string("<"); break;
! 442: case '"': to_string("""); break;
1.77 paf 443: _default;
444: });
445: break;
446: default:
1.117 ! paf 447: SAPI::die("unknown untaint language #%d",
! 448: static_cast<int>(to_lang)); // should never
1.77 paf 449: break; // never
1.76 paf 450: }
1.117 ! paf 451: }
1.55 parser 452:
1.117 ! paf 453: return StringBody(CORD_ec_to_cord(result));
1.1 paf 454: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to untaint.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main