Annotation of parser3/src/main/untaint.C, revision 1.115.2.9
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:
1.115.2.9! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/02/04 12:08:57 $";
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:
1.115.2.9! paf 147: // without language-chage string?
! 148: if(lang==UL_PASS_APPENDED) {
! 149: // can simply append elements
! 150: ssize_t delta=fused-(dest.fallocated-dest.fused);
! 151: if(delta>0) // they don't fit?
! 152: dest.expand(delta);
! 153:
1.115.2.1 paf 154: // felements[fused...]=src.felements[0..fused], fused+=src.fused
155: memcpy(&dest.felements[dest.fused], felements, sizeof(felements[0])*fused);
156: dest.fused+=fused;
157: return dest;
1.91 paf 158: }
159:
1.77 paf 160: // manually unrolled code to avoid do{if(const)} constructs
161: if(forced)
1.115.2.1 paf 162: STRING_FOREACH_FRAGMENT(
163: dest.APPEND(fragment->ptr, fragment->size,
1.77 paf 164: lang, //forcing passed lang
1.115.2.1 paf 165: fragment->origin.file, fragment->origin.line);
1.77 paf 166: )
167: else if(lang==UL_PASS_APPENDED)
1.115.2.1 paf 168: STRING_FOREACH_FRAGMENT(
169: dest.APPEND(fragment->ptr, fragment->size,
170: fragment->lang, // passing item's lang
171: fragment->origin.file, fragment->origin.line);
1.77 paf 172: )
173: else if(lang&UL_OPTIMIZE_BIT) // main idea here
174: // tainted piece would get OPTIMIZED bit from 'lang'
175: // clean piece would be marked OPTIMIZED manually
176: // pieces with determined languages [not tainted|clean] would retain theirs langs
1.115.2.1 paf 177: STRING_FOREACH_FRAGMENT(
178: dest.APPEND(fragment->ptr, fragment->size,
179: fragment->lang==UL_TAINTED?lang:(
180: fragment->lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
181: fragment->lang
1.77 paf 182: ),
1.115.2.1 paf 183: fragment->origin.file, fragment->origin.line);
1.77 paf 184: )
185: else
1.115.2.1 paf 186: STRING_FOREACH_FRAGMENT(
187: dest.APPEND(fragment->ptr, fragment->size,
188: fragment->lang==UL_TAINTED?lang:fragment->lang,
189: fragment->origin.file, fragment->origin.line);
1.77 paf 190: );
1.115.2.1 paf 191:
192: return dest;
1.77 paf 193: }
194:
1.75 paf 195: size_t String::cstr_bufsize(Untaint_lang lang,
196: SQL_Connection *connection,
1.115.2.3 paf 197: const Request_charsets *charsets) const {
1.77 paf 198: size_t dest=1; // for terminating 0
1.115.2.1 paf 199: STRING_FOREACH_FRAGMENT(
200: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 201:
202: switch(to_lang & ~UL_OPTIMIZE_BIT) {
203: case UL_CLEAN:
204: case UL_TAINTED:
205: case UL_AS_IS:
206: // clean piece
207:
208: // tainted piece, but undefined untaint language
209: // for VString.as_double of tainted values
210: // for ^process{body} evaluation
211:
212: // tainted, untaint language: as-is
1.115.2.1 paf 213: dest+=fragment->size;
1.77 paf 214: break;
215: case UL_FILE_SPEC:
216: // tainted, untaint language: file [name]
1.115.2.1 paf 217: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 218: break;
219: case UL_URI:
220: // tainted, untaint language: uri
1.115.2.1 paf 221: dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */;
1.77 paf 222: break;
223: case UL_HTTP_HEADER:
224: // tainted, untaint language: http-field-content-text
1.115.2.1 paf 225: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 226: break;
227: case UL_MAIL_HEADER:
228: // tainted, untaint language: mail-header
1.115.2.4 paf 229: if(charsets) {
1.111 paf 230: int parts_count=1;
1.115.2.6 paf 231: const char* prev=fragment->ptr;
1.115.2.1 paf 232: size_t size=fragment->size;
1.115.2.6 paf 233: while(const char* comma_at=(const char*)memchr(prev, ',', size)) {
1.111 paf 234: parts_count++;
235: size-=comma_at-prev;
236: prev=comma_at+1;
237: }
1.77 paf 238: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 239: dest+=
1.115.2.1 paf 240: fragment->size*3+
1.115.2.4 paf 241: parts_count*(charsets->mail().name()->size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */);
1.87 paf 242: } else
1.115.2.1 paf 243: dest+=fragment->size;
1.77 paf 244: break;
245: case UL_TABLE:
246: // tainted, untaint language: table
1.115.2.1 paf 247: dest+=fragment->size;
1.77 paf 248: break;
249: case UL_SQL:
250: // tainted, untaint language: sql
251: if(connection)
1.115.2.1 paf 252: dest+=connection->quote(0, fragment->ptr, fragment->size);
1.77 paf 253: break;
254: case UL_JS:
255: escape(switch(*src) {
256: case '"': case '\'': case '\n': case '\\': case '\xFF':
257: dest+=2; break;
258: default:
259: dest++; break;
260: });
261: break;
262: case UL_XML:
263: escape(switch(*src) {
264: case '&': case '>': case '<': case '"': case '\'':
265: dest+= 6; break;
266: default:
267: dest++; break;
268: });
269: break;
270: case UL_HTML:
271: escape(switch(*src) {
272: case '&':
273: case '>':
274: case '<':
275: case '"':
276: dest+=6; break;
277: default:
278: dest++; break;
279: });
280: break;
1.75 paf 281: }
1.77 paf 282: );
1.75 paf 283: return dest;
1.51 parser 284: }
285:
1.109 paf 286: /** http://www.ietf.org/rfc/rfc2047.txt
287: RFC
288: (3) As a replacement for a 'word' entity within a 'phrase', for example,
289: one that precedes an address in a From, To, or Cc header. The ABNF
290: definition for 'phrase' from RFC 822 thus becomes:
291:
292: phrase = 1*( encoded-word / word )
293:
294: In this case the set of characters that may be used in a "Q"-encoded
295: 'encoded-word' is restricted to: <upper and lower case ASCII
296: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
297: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
298: 'phrase' MUST be separated from any adjacent 'word', 'text' or
299: 'special' by 'linear-white-space'.
300: ...
301: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
302: represented as "_" (underscore, ASCII 95.). (This character may
303: not pass through some internetwork mail gateways, but its use
304: will greatly enhance readability of "Q" encoded data with mail
305: readers that do not support this encoding.) Note that the "_"
306: always represents hexadecimal 20, even if the SPACE character
307: occupies a different code position in the character set in use.
308:
309: paf: obviously,
310: without "=", or one could not differ "=E0" and "russian letter a"
311: and without "_", or in would mean 0x20
312: */
313: static bool mail_header_char_valid_within_Qencoded(char c) {
314: return c>='A' && c<='Z'
315: || c>='a' && c<='Z'
316: || c>='0' && c<='9'
317: || strchr("!*+-/", c);
318: }
1.43 paf 319: char *String::store_to(char *dest, Untaint_lang lang,
320: SQL_Connection *connection,
1.115.2.3 paf 321: const Request_charsets *charsets) const {
1.75 paf 322: // WARNING:
323: // before any changes check cstr_bufsize first!!!
1.44 paf 324: bool whitespace=true;
1.115.2.1 paf 325: // STRING_FOREACH_FRAGMENT(
326: element_type *last = felements+fused; element_type *fragment = felements; for(; fragment<last; fragment++) {
1.111 paf 327:
1.115.2.1 paf 328: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 329:
330: char *start=dest;
331:
332: switch(to_lang & ~UL_OPTIMIZE_BIT) {
333: case UL_CLEAN:
334: case UL_TAINTED:
335: case UL_AS_IS:
336: // clean piece
337:
338: // tainted piece, but undefined untaint language
339: // for VString.as_double of tainted values
340: // for ^process{body} evaluation
341:
342: // tainted, untaint language: as-is
1.115.2.1 paf 343: memcpy(dest, fragment->ptr, fragment->size);
344: dest+=fragment->size;
1.77 paf 345: break;
346: case UL_FILE_SPEC:
347: // tainted, untaint language: file [name]
1.83 paf 348: escape(
1.113 paf 349: // Macintosh has problems with small Russian letter 'r'
1.115.2.4 paf 350: if( *src=='\xF0' && charsets && *charsets->source().name()=="windows-1251" ) {
1.113 paf 351: // fixing that letter for most common charset
1.115.2.2 paf 352: to_char('p');
1.114 paf 353: } else // fallback to default
354: encode(need_file_encode, '_');
1.83 paf 355: );
1.77 paf 356: break;
357: case UL_URI:
358: // tainted, untaint language: uri
1.85 paf 359: const void *client_ptr;
360: size_t client_size;
1.115.2.2 paf 361: if(charsets)
1.115.2.4 paf 362: Charset::transcode(charsets->pool(),
363: charsets->source(), fragment->ptr, fragment->size,
364: charsets->client(), client_ptr, client_size);
1.115.2.2 paf 365: else {
366: client_ptr=fragment->ptr;
367: client_size=fragment->size;
368: }
369:
1.85 paf 370: {
1.115.2.6 paf 371: const char* src=(const char* )client_ptr;
1.85 paf 372: for(int size=client_size; size--; src++)
1.108 paf 373: encode(need_uri_encode, '%');
1.85 paf 374: }
1.77 paf 375: break;
376: case UL_HTTP_HEADER:
377: // tainted, untaint language: http-field-content-text
1.108 paf 378: escape(
379: encode(need_uri_encode, '%');
380: );
1.77 paf 381: break;
382: case UL_MAIL_HEADER:
383: // tainted, untaint language: mail-header
1.105 paf 384: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.4 paf 385: if(charsets) {
1.87 paf 386: const void *mail_ptr;
387: size_t mail_size;
1.115.2.4 paf 388: Charset::transcode(charsets->pool(),
389: charsets->source(), fragment->ptr, fragment->size,
390: charsets->mail(), mail_ptr, mail_size);
391: CharPtr charset_name=charsets->mail().name()->cstr();
1.87 paf 392:
1.77 paf 393: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.115.2.6 paf 394: const char* src=(const char* )mail_ptr;
1.77 paf 395: bool to_quoted_printable=false;
1.105 paf 396:
1.111 paf 397: bool email=false;
1.115.2.6 paf 398: for(const char* end=src+mail_size; src<end; src++) {
1.111 paf 399: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
400: if(to_quoted_printable && (*src==',' || *src=='<')) {
401: email=*src=='<';
1.105 paf 402: dest+=sprintf(dest, "?=");
1.107 paf 403: to_quoted_printable=false;
1.105 paf 404: }
1.111 paf 405: if(!email && (
1.109 paf 406: !to_quoted_printable && (*src & 0x80) // starting quote-printable-encoding on first 8bit char
407: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(*src)
1.105 paf 408: )) {
1.77 paf 409: if(!to_quoted_printable) {
1.115.2.7 paf 410: dest+=sprintf(dest, "=?%s?Q?", charset_name.get());
1.77 paf 411: to_quoted_printable=true;
412: }
1.105 paf 413: //RFC Upper case should be used for hexadecimal digits "A" through "F"
1.109 paf 414: if(*src == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
415: *dest++='_'; // RFC may be represented as "_"
416: else
417: dest+=sprintf(dest, "=%02X", *src & 0xFF);
1.105 paf 418: } else
1.109 paf 419: *dest++=*src;
1.111 paf 420: if(*src=='>')
421: email=false;
1.44 paf 422: }
1.112 paf 423: if(to_quoted_printable) // close
1.77 paf 424: dest+=sprintf(dest, "?=");
1.87 paf 425:
1.77 paf 426: } else {
1.115.2.1 paf 427: memcpy(dest, fragment->ptr, fragment->size);
428: dest+=fragment->size;
1.1 paf 429: }
1.77 paf 430: break;
431: case UL_TABLE:
432: // tainted, untaint language: table
433: escape(switch(*src) {
434: case '\t': to_char(' '); break;
435: case '\n': to_char(' '); break;
436: _default;
437: });
438: break;
439: case UL_SQL:
440: // tainted, untaint language: sql
441: if(connection)
1.115.2.1 paf 442: dest+=connection->quote(dest, fragment->ptr, fragment->size);
1.77 paf 443: else
1.115.2.5 paf 444: throw Exception(Exception::undefined_type,
445: Exception::undefined_source,
1.77 paf 446: "untaint in SQL language failed - no connection specified");
447: break;
448: case UL_JS:
449: escape(switch(*src) {
450: case '"': to_string("\\\"", 2); break;
451: case '\'': to_string("\\'", 2); break;
452: case '\n': to_string("\\n", 2); break;
453: case '\\': to_string("\\\\", 2); break;
454: case '\xFF': to_string("\\\xFF", 2); break;
455: _default;
456: });
457: break;
458: case UL_XML:
459: escape(switch(*src) {
460: case '&': to_string("&", 5); break;
461: case '>': to_string(">", 4); break;
462: case '<': to_string("<", 4); break;
463: case '"': to_string(""", 6); break;
464: case '\'': to_string("'", 6); break;
465: _default;
466: });
467: break;
468: case UL_HTML:
469: escape(switch(*src) {
470: case '&': to_string("&", 5); break;
471: case '>': to_string(">", 4); break;
472: case '<': to_string("<", 4); break;
473: case '"': to_string(""", 6); break;
474: _default;
475: });
476: break;
477: default:
1.115.2.5 paf 478: throw Exception(Exception::undefined_type,
479: Exception::undefined_source,
1.81 paf 480: "unknown untaint language #%d",
1.110 paf 481: static_cast<int>(to_lang)); // should never
1.77 paf 482: break; // never
1.76 paf 483: }
1.55 parser 484:
1.77 paf 485: if(to_lang & UL_OPTIMIZE_BIT) {
486: // optimizing whitespace
487: char *stop=dest; dest=start;
488: for(char *src=start; src<stop; src++)
489: switch(*src) {
490: // of all consequent white space chars leaving only first one
1.80 paf 491: case ' ': case '\r': case '\n': case '\t':
1.77 paf 492: if(!whitespace) {
493: *dest++=*src;
494: whitespace=true;
495: }
496: break;
497: default:
498: whitespace=false;
499: *dest++=*src;
500: break;
501: };
502: } else // piece without optimization
503: whitespace=false;
1.115.2.1 paf 504: // );
505: }
506:
1.76 paf 507: return dest;
508: }
509:
1.115.2.2 paf 510: char *String::cstr_debug_origins(Pool& pool) const {
1.81 paf 511: //_asm int 3;
1.115.2.2 paf 512: char *result=new(pool) char[size()+fused*MAX_STRING*2];
1.76 paf 513: char *dest=result;
514:
1.115.2.1 paf 515: STRING_FOREACH_FRAGMENT(
1.96 paf 516: IFNDEF_NO_STRING_ORIGIN(
1.115.2.1 paf 517: if(fragment->origin.file)
1.96 paf 518: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
1.115.2.1 paf 519: fragment->origin.file,
520: 1+fragment->origin.line);
1.96 paf 521: else
522: dest+=sprintf(dest, "<unknown>");
523: );
1.115.2.1 paf 524: uchar show_lang=fragment->lang & ~UL_OPTIMIZE_BIT;
1.96 paf 525: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
1.115.2.5 paf 526: throw Exception(Exception::undefined_type,
527: Exception::undefined_source,
1.96 paf 528: "unknown untaint language #%d",
529: static_cast<int>(show_lang)); // sould never
530:
531: dest+=sprintf(dest, "#%s%s: ",
532: String_Untaint_lang_name[show_lang],
1.115.2.1 paf 533: fragment->lang & UL_OPTIMIZE_BIT?".O":"");
1.96 paf 534: char *dest_after_origins=dest;
1.76 paf 535:
1.115.2.1 paf 536: memcpy(dest, fragment->ptr, fragment->size);
537: dest+=fragment->size;
1.76 paf 538:
1.96 paf 539: remove_crlf(dest_after_origins, dest);
540: to_char('\n');
541: );
1.64 parser 542:
1.76 paf 543: *dest=0;
544: return result;
1.1 paf 545: }
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