Annotation of parser3/src/main/untaint.C, revision 1.115.2.8
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.115.2.6 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:
1.115.2.8! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/02/04 10:17:06 $";
1.1 paf 9:
10: #include "pa_pool.h"
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"
1.34 paf 16: #include "pa_sql_connection.h"
1.58 parser 17: #include "pa_dictionary.h"
1.66 parser 18: #include "pa_common.h"
1.85 paf 19: #include "pa_charset.h"
1.115.2.4 paf 20: #include "pa_request_charsets.h"
1.1 paf 21:
1.18 paf 22: #define escape(action) \
1.1 paf 23: { \
1.115.2.6 paf 24: const char* src=fragment->ptr; \
1.115.2.1 paf 25: for(int size=fragment->size; size--; src++) \
1.18 paf 26: action \
1.1 paf 27: }
1.13 paf 28: #define _default default: *dest++=*src; break
29: #define encode(need_encode_func, prefix) \
30: if(need_encode_func(*src)) { \
1.115.2.6 paf 31: static const char* hex="0123456789ABCDEF"; \
1.9 paf 32: char chunk[3]={prefix}; \
1.13 paf 33: chunk[1]=hex[((unsigned char)*src)/0x10]; \
34: chunk[2]=hex[((unsigned char)*src)%0x10]; \
1.60 parser 35: memcpy(dest, chunk, 3); dest+=3; \
1.5 paf 36: } else \
1.13 paf 37: *dest++=*src; \
1.5 paf 38: break
1.18 paf 39: #define to_char(c) *dest++=c
40: #define to_string(b, bsize) \
1.60 parser 41: memcpy(dest, b, bsize); \
1.18 paf 42: dest+=bsize; \
1.4 paf 43:
1.9 paf 44: inline bool need_file_encode(unsigned char c){
1.108 paf 45: // russian letters and space ENABLED
46: // encoding only these...
47: return strchr(
48: "*?'\"<>|"
49: #ifndef WIN32
50: ":\\~"
1.31 paf 51: #endif
1.108 paf 52: , c)!=0;
1.9 paf 53: }
1.5 paf 54: inline bool need_uri_encode(unsigned char c){
1.13 paf 55: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 56: return false;
57:
1.5 paf 58: return !strchr("_-./", c);
59: }
1.36 paf 60: inline bool need_http_header_encode(unsigned char c){
1.18 paf 61: if(strchr(" , :", c))
1.5 paf 62: return false;
63:
64: return need_uri_encode(c);
1.4 paf 65: }
1.1 paf 66:
1.56 parser 67: //
68:
1.115.2.6 paf 69: static const char* String_Untaint_lang_name[]={
1.56 parser 70: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
71: "C", ///< clean
72: "T", ///< tainted, untaint language as assigned later
73: // untaint languages. assigned by ^untaint[lang]{...}
74: "P",
75: /**<
76: leave language built into string being appended.
77: just a flag, that value not stored
78: */
79: "A", ///< leave all characters intact
1.68 parser 80: "F", ///< file specification
81: "H", ///< ext in HTTP response header
1.56 parser 82: "M", ///< text in mail header
83: "URI", ///< text in uri
84: "T", ///< ^table:set body
85: "SQL", ///< ^table:sql body
86: "JS", ///< JavaScript code
1.68 parser 87: "XML", ///< ^dom:set xml
1.82 paf 88: "HTML" ///< HTML code (for editing)
1.56 parser 89: };
90:
91:
1.1 paf 92: // String
93:
1.41 paf 94: /*
95:
96: HTTP-header = field-name ":" [ field-value ] CRLF
97:
98: field-name = token
99: field-value = *( field-content | LWS )
100:
101: field-content = <the OCTETs making up the field-value
102: and consisting of either *TEXT or combinations
103: of token, tspecials, and quoted-string>
104:
105:
106:
107: word = token | quoted-string
108:
109: token = 1*<any CHAR except CTLs or tspecials>
110:
111:
112:
113: tspecials = "(" | ")" | "<" | ">" | "@"
114: | "," | ";" | ":" | "\" | <">
115: | "/" | "[" | "]" | "?" | "="
116: | "{" | "}" | SP | HT
117:
118: SP = <US-ASCII SP, space (32)>
119: HT = <US-ASCII HT, horizontal-tab (9)>
120:
121: LWS = [CRLF] 1*( SP | HT )
122: TEXT = <any OCTET except CTLs,
123: but including LWS>
124:
125: quoted-pair = "\" CHAR
126:
127: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
128: */
1.115.2.6 paf 129: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 130: for(; size--; ptr++)
1.42 paf 131: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 132: return true;
133: return false;
134: }
135:
1.91 paf 136: //#include "pa_sapi.h"
137: /**
1.115.2.1 paf 138: appends to other String,
1.91 paf 139: marking all tainted pieces of it with @a lang.
1.92 paf 140: or marking ALL pieces of it with a @a lang when @a forced to,
141: and propagating OPTIMIZE language bit.
1.91 paf 142: */
1.115.2.1 paf 143: String& String::append_to(String& dest, uchar lang, bool forced) const {
144: if(is_empty())
145: return dest;
146:
147: // without language-chage, small[fitting dest tail] string?
148: if(lang==UL_PASS_APPENDED && fused<=(dest.fallocated-dest.fused)) {
149: // felements[fused...]=src.felements[0..fused], fused+=src.fused
150: memcpy(&dest.felements[dest.fused], felements, sizeof(felements[0])*fused);
151: dest.fused+=fused;
152: return dest;
1.91 paf 153: }
154:
1.77 paf 155: // manually unrolled code to avoid do{if(const)} constructs
156: if(forced)
1.115.2.1 paf 157: STRING_FOREACH_FRAGMENT(
158: dest.APPEND(fragment->ptr, fragment->size,
1.77 paf 159: lang, //forcing passed lang
1.115.2.1 paf 160: fragment->origin.file, fragment->origin.line);
1.77 paf 161: )
162: else if(lang==UL_PASS_APPENDED)
1.115.2.1 paf 163: STRING_FOREACH_FRAGMENT(
164: dest.APPEND(fragment->ptr, fragment->size,
165: fragment->lang, // passing item's lang
166: fragment->origin.file, fragment->origin.line);
1.77 paf 167: )
168: else if(lang&UL_OPTIMIZE_BIT) // main idea here
169: // tainted piece would get OPTIMIZED bit from 'lang'
170: // clean piece would be marked OPTIMIZED manually
171: // pieces with determined languages [not tainted|clean] would retain theirs langs
1.115.2.1 paf 172: STRING_FOREACH_FRAGMENT(
173: dest.APPEND(fragment->ptr, fragment->size,
174: fragment->lang==UL_TAINTED?lang:(
175: fragment->lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
176: fragment->lang
1.77 paf 177: ),
1.115.2.1 paf 178: fragment->origin.file, fragment->origin.line);
1.77 paf 179: )
180: else
1.115.2.1 paf 181: STRING_FOREACH_FRAGMENT(
182: dest.APPEND(fragment->ptr, fragment->size,
183: fragment->lang==UL_TAINTED?lang:fragment->lang,
184: fragment->origin.file, fragment->origin.line);
1.77 paf 185: );
1.115.2.1 paf 186:
187: return dest;
1.77 paf 188: }
189:
1.75 paf 190: size_t String::cstr_bufsize(Untaint_lang lang,
191: SQL_Connection *connection,
1.115.2.3 paf 192: const Request_charsets *charsets) const {
1.77 paf 193: size_t dest=1; // for terminating 0
1.115.2.1 paf 194: STRING_FOREACH_FRAGMENT(
195: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 196:
197: switch(to_lang & ~UL_OPTIMIZE_BIT) {
198: case UL_CLEAN:
199: case UL_TAINTED:
200: case UL_AS_IS:
201: // clean piece
202:
203: // tainted piece, but undefined untaint language
204: // for VString.as_double of tainted values
205: // for ^process{body} evaluation
206:
207: // tainted, untaint language: as-is
1.115.2.1 paf 208: dest+=fragment->size;
1.77 paf 209: break;
210: case UL_FILE_SPEC:
211: // tainted, untaint language: file [name]
1.115.2.1 paf 212: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 213: break;
214: case UL_URI:
215: // tainted, untaint language: uri
1.115.2.1 paf 216: dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */;
1.77 paf 217: break;
218: case UL_HTTP_HEADER:
219: // tainted, untaint language: http-field-content-text
1.115.2.1 paf 220: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 221: break;
222: case UL_MAIL_HEADER:
223: // tainted, untaint language: mail-header
1.115.2.4 paf 224: if(charsets) {
1.111 paf 225: int parts_count=1;
1.115.2.6 paf 226: const char* prev=fragment->ptr;
1.115.2.1 paf 227: size_t size=fragment->size;
1.115.2.6 paf 228: while(const char* comma_at=(const char*)memchr(prev, ',', size)) {
1.111 paf 229: parts_count++;
230: size-=comma_at-prev;
231: prev=comma_at+1;
232: }
1.77 paf 233: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 234: dest+=
1.115.2.1 paf 235: fragment->size*3+
1.115.2.4 paf 236: parts_count*(charsets->mail().name()->size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */);
1.87 paf 237: } else
1.115.2.1 paf 238: dest+=fragment->size;
1.77 paf 239: break;
240: case UL_TABLE:
241: // tainted, untaint language: table
1.115.2.1 paf 242: dest+=fragment->size;
1.77 paf 243: break;
244: case UL_SQL:
245: // tainted, untaint language: sql
246: if(connection)
1.115.2.1 paf 247: dest+=connection->quote(0, fragment->ptr, fragment->size);
1.77 paf 248: break;
249: case UL_JS:
250: escape(switch(*src) {
251: case '"': case '\'': case '\n': case '\\': case '\xFF':
252: dest+=2; break;
253: default:
254: dest++; break;
255: });
256: break;
257: case UL_XML:
258: escape(switch(*src) {
259: case '&': case '>': case '<': case '"': case '\'':
260: dest+= 6; break;
261: default:
262: dest++; break;
263: });
264: break;
265: case UL_HTML:
266: escape(switch(*src) {
267: case '&':
268: case '>':
269: case '<':
270: case '"':
271: dest+=6; break;
272: default:
273: dest++; break;
274: });
275: break;
1.75 paf 276: }
1.77 paf 277: );
1.75 paf 278: return dest;
1.51 parser 279: }
280:
1.109 paf 281: /** http://www.ietf.org/rfc/rfc2047.txt
282: RFC
283: (3) As a replacement for a 'word' entity within a 'phrase', for example,
284: one that precedes an address in a From, To, or Cc header. The ABNF
285: definition for 'phrase' from RFC 822 thus becomes:
286:
287: phrase = 1*( encoded-word / word )
288:
289: In this case the set of characters that may be used in a "Q"-encoded
290: 'encoded-word' is restricted to: <upper and lower case ASCII
291: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
292: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
293: 'phrase' MUST be separated from any adjacent 'word', 'text' or
294: 'special' by 'linear-white-space'.
295: ...
296: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
297: represented as "_" (underscore, ASCII 95.). (This character may
298: not pass through some internetwork mail gateways, but its use
299: will greatly enhance readability of "Q" encoded data with mail
300: readers that do not support this encoding.) Note that the "_"
301: always represents hexadecimal 20, even if the SPACE character
302: occupies a different code position in the character set in use.
303:
304: paf: obviously,
305: without "=", or one could not differ "=E0" and "russian letter a"
306: and without "_", or in would mean 0x20
307: */
308: static bool mail_header_char_valid_within_Qencoded(char c) {
309: return c>='A' && c<='Z'
310: || c>='a' && c<='Z'
311: || c>='0' && c<='9'
312: || strchr("!*+-/", c);
313: }
1.43 paf 314: char *String::store_to(char *dest, Untaint_lang lang,
315: SQL_Connection *connection,
1.115.2.3 paf 316: const Request_charsets *charsets) const {
1.75 paf 317: // WARNING:
318: // before any changes check cstr_bufsize first!!!
1.44 paf 319: bool whitespace=true;
1.115.2.1 paf 320: // STRING_FOREACH_FRAGMENT(
321: element_type *last = felements+fused; element_type *fragment = felements; for(; fragment<last; fragment++) {
1.111 paf 322:
1.115.2.1 paf 323: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 324:
325: char *start=dest;
326:
327: switch(to_lang & ~UL_OPTIMIZE_BIT) {
328: case UL_CLEAN:
329: case UL_TAINTED:
330: case UL_AS_IS:
331: // clean piece
332:
333: // tainted piece, but undefined untaint language
334: // for VString.as_double of tainted values
335: // for ^process{body} evaluation
336:
337: // tainted, untaint language: as-is
1.115.2.1 paf 338: memcpy(dest, fragment->ptr, fragment->size);
339: dest+=fragment->size;
1.77 paf 340: break;
341: case UL_FILE_SPEC:
342: // tainted, untaint language: file [name]
1.83 paf 343: escape(
1.113 paf 344: // Macintosh has problems with small Russian letter 'r'
1.115.2.4 paf 345: if( *src=='\xF0' && charsets && *charsets->source().name()=="windows-1251" ) {
1.113 paf 346: // fixing that letter for most common charset
1.115.2.2 paf 347: to_char('p');
1.114 paf 348: } else // fallback to default
349: encode(need_file_encode, '_');
1.83 paf 350: );
1.77 paf 351: break;
352: case UL_URI:
353: // tainted, untaint language: uri
1.85 paf 354: const void *client_ptr;
355: size_t client_size;
1.115.2.2 paf 356: if(charsets)
1.115.2.4 paf 357: Charset::transcode(charsets->pool(),
358: charsets->source(), fragment->ptr, fragment->size,
359: charsets->client(), client_ptr, client_size);
1.115.2.2 paf 360: else {
361: client_ptr=fragment->ptr;
362: client_size=fragment->size;
363: }
364:
1.85 paf 365: {
1.115.2.6 paf 366: const char* src=(const char* )client_ptr;
1.85 paf 367: for(int size=client_size; size--; src++)
1.108 paf 368: encode(need_uri_encode, '%');
1.85 paf 369: }
1.77 paf 370: break;
371: case UL_HTTP_HEADER:
372: // tainted, untaint language: http-field-content-text
1.108 paf 373: escape(
374: encode(need_uri_encode, '%');
375: );
1.77 paf 376: break;
377: case UL_MAIL_HEADER:
378: // tainted, untaint language: mail-header
1.105 paf 379: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.4 paf 380: if(charsets) {
1.87 paf 381: const void *mail_ptr;
382: size_t mail_size;
1.115.2.4 paf 383: Charset::transcode(charsets->pool(),
384: charsets->source(), fragment->ptr, fragment->size,
385: charsets->mail(), mail_ptr, mail_size);
386: CharPtr charset_name=charsets->mail().name()->cstr();
1.87 paf 387:
1.77 paf 388: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.115.2.6 paf 389: const char* src=(const char* )mail_ptr;
1.77 paf 390: bool to_quoted_printable=false;
1.105 paf 391:
1.111 paf 392: bool email=false;
1.115.2.6 paf 393: for(const char* end=src+mail_size; src<end; src++) {
1.111 paf 394: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
395: if(to_quoted_printable && (*src==',' || *src=='<')) {
396: email=*src=='<';
1.105 paf 397: dest+=sprintf(dest, "?=");
1.107 paf 398: to_quoted_printable=false;
1.105 paf 399: }
1.111 paf 400: if(!email && (
1.109 paf 401: !to_quoted_printable && (*src & 0x80) // starting quote-printable-encoding on first 8bit char
402: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(*src)
1.105 paf 403: )) {
1.77 paf 404: if(!to_quoted_printable) {
1.115.2.7 paf 405: dest+=sprintf(dest, "=?%s?Q?", charset_name.get());
1.77 paf 406: to_quoted_printable=true;
407: }
1.105 paf 408: //RFC Upper case should be used for hexadecimal digits "A" through "F"
1.109 paf 409: if(*src == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
410: *dest++='_'; // RFC may be represented as "_"
411: else
412: dest+=sprintf(dest, "=%02X", *src & 0xFF);
1.105 paf 413: } else
1.109 paf 414: *dest++=*src;
1.111 paf 415: if(*src=='>')
416: email=false;
1.44 paf 417: }
1.112 paf 418: if(to_quoted_printable) // close
1.77 paf 419: dest+=sprintf(dest, "?=");
1.87 paf 420:
1.77 paf 421: } else {
1.115.2.1 paf 422: memcpy(dest, fragment->ptr, fragment->size);
423: dest+=fragment->size;
1.1 paf 424: }
1.77 paf 425: break;
426: case UL_TABLE:
427: // tainted, untaint language: table
428: escape(switch(*src) {
429: case '\t': to_char(' '); break;
430: case '\n': to_char(' '); break;
431: _default;
432: });
433: break;
434: case UL_SQL:
435: // tainted, untaint language: sql
436: if(connection)
1.115.2.1 paf 437: dest+=connection->quote(dest, fragment->ptr, fragment->size);
1.77 paf 438: else
1.115.2.5 paf 439: throw Exception(Exception::undefined_type,
440: Exception::undefined_source,
1.77 paf 441: "untaint in SQL language failed - no connection specified");
442: break;
443: case UL_JS:
444: escape(switch(*src) {
445: case '"': to_string("\\\"", 2); break;
446: case '\'': to_string("\\'", 2); break;
447: case '\n': to_string("\\n", 2); break;
448: case '\\': to_string("\\\\", 2); break;
449: case '\xFF': to_string("\\\xFF", 2); break;
450: _default;
451: });
452: break;
453: case UL_XML:
454: escape(switch(*src) {
455: case '&': to_string("&", 5); break;
456: case '>': to_string(">", 4); break;
457: case '<': to_string("<", 4); break;
458: case '"': to_string(""", 6); break;
459: case '\'': to_string("'", 6); break;
460: _default;
461: });
462: break;
463: case UL_HTML:
464: escape(switch(*src) {
465: case '&': to_string("&", 5); break;
466: case '>': to_string(">", 4); break;
467: case '<': to_string("<", 4); break;
468: case '"': to_string(""", 6); break;
469: _default;
470: });
471: break;
472: default:
1.115.2.5 paf 473: throw Exception(Exception::undefined_type,
474: Exception::undefined_source,
1.81 paf 475: "unknown untaint language #%d",
1.110 paf 476: static_cast<int>(to_lang)); // should never
1.77 paf 477: break; // never
1.76 paf 478: }
1.55 parser 479:
1.77 paf 480: if(to_lang & UL_OPTIMIZE_BIT) {
481: // optimizing whitespace
482: char *stop=dest; dest=start;
483: for(char *src=start; src<stop; src++)
484: switch(*src) {
485: // of all consequent white space chars leaving only first one
1.80 paf 486: case ' ': case '\r': case '\n': case '\t':
1.77 paf 487: if(!whitespace) {
488: *dest++=*src;
489: whitespace=true;
490: }
491: break;
492: default:
493: whitespace=false;
494: *dest++=*src;
495: break;
496: };
497: } else // piece without optimization
498: whitespace=false;
1.115.2.1 paf 499: // );
500: }
501:
1.76 paf 502: return dest;
503: }
504:
1.115.2.2 paf 505: char *String::cstr_debug_origins(Pool& pool) const {
1.81 paf 506: //_asm int 3;
1.115.2.2 paf 507: char *result=new(pool) char[size()+fused*MAX_STRING*2];
1.76 paf 508: char *dest=result;
509:
1.115.2.1 paf 510: STRING_FOREACH_FRAGMENT(
1.96 paf 511: IFNDEF_NO_STRING_ORIGIN(
1.115.2.1 paf 512: if(fragment->origin.file)
1.96 paf 513: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
1.115.2.1 paf 514: fragment->origin.file,
515: 1+fragment->origin.line);
1.96 paf 516: else
517: dest+=sprintf(dest, "<unknown>");
518: );
1.115.2.1 paf 519: uchar show_lang=fragment->lang & ~UL_OPTIMIZE_BIT;
1.96 paf 520: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
1.115.2.5 paf 521: throw Exception(Exception::undefined_type,
522: Exception::undefined_source,
1.96 paf 523: "unknown untaint language #%d",
524: static_cast<int>(show_lang)); // sould never
525:
526: dest+=sprintf(dest, "#%s%s: ",
527: String_Untaint_lang_name[show_lang],
1.115.2.1 paf 528: fragment->lang & UL_OPTIMIZE_BIT?".O":"");
1.96 paf 529: char *dest_after_origins=dest;
1.76 paf 530:
1.115.2.1 paf 531: memcpy(dest, fragment->ptr, fragment->size);
532: dest+=fragment->size;
1.76 paf 533:
1.96 paf 534: remove_crlf(dest_after_origins, dest);
535: to_char('\n');
536: );
1.64 parser 537:
1.76 paf 538: *dest=0;
539: return result;
1.1 paf 540: }
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