Annotation of parser3/src/main/untaint.C, revision 1.115.2.5
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.115 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:
1.115.2.5! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/01/29 11:55:22 $";
1.1 paf 9:
10: #include "pa_pool.h"
1.115.2.4 paf 11: #include "pa_value_includes.h"
1.1 paf 12: #include "pa_string.h"
13: #include "pa_hash.h"
14: #include "pa_exception.h"
1.13 paf 15: #include "pa_table.h"
1.32 paf 16: #include "pa_globals.h"
1.34 paf 17: #include "pa_sql_connection.h"
1.58 parser 18: #include "pa_dictionary.h"
1.66 parser 19: #include "pa_common.h"
1.85 paf 20: #include "pa_charset.h"
1.115.2.4 paf 21: #include "pa_request_charsets.h"
1.1 paf 22:
1.18 paf 23: #define escape(action) \
1.1 paf 24: { \
1.115.2.1 paf 25: const char *src=fragment->ptr; \
26: for(int size=fragment->size; size--; src++) \
1.18 paf 27: action \
1.1 paf 28: }
1.13 paf 29: #define _default default: *dest++=*src; break
30: #define encode(need_encode_func, prefix) \
31: if(need_encode_func(*src)) { \
1.5 paf 32: static const char *hex="0123456789ABCDEF"; \
1.9 paf 33: char chunk[3]={prefix}; \
1.13 paf 34: chunk[1]=hex[((unsigned char)*src)/0x10]; \
35: chunk[2]=hex[((unsigned char)*src)%0x10]; \
1.60 parser 36: memcpy(dest, chunk, 3); dest+=3; \
1.5 paf 37: } else \
1.13 paf 38: *dest++=*src; \
1.5 paf 39: break
1.18 paf 40: #define to_char(c) *dest++=c
41: #define to_string(b, bsize) \
1.60 parser 42: memcpy(dest, b, bsize); \
1.18 paf 43: dest+=bsize; \
1.4 paf 44:
1.9 paf 45: inline bool need_file_encode(unsigned char c){
1.108 paf 46: // russian letters and space ENABLED
47: // encoding only these...
48: return strchr(
49: "*?'\"<>|"
50: #ifndef WIN32
51: ":\\~"
1.31 paf 52: #endif
1.108 paf 53: , c)!=0;
1.9 paf 54: }
1.5 paf 55: inline bool need_uri_encode(unsigned char c){
1.13 paf 56: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 57: return false;
58:
1.5 paf 59: return !strchr("_-./", c);
60: }
1.36 paf 61: inline bool need_http_header_encode(unsigned char c){
1.18 paf 62: if(strchr(" , :", c))
1.5 paf 63: return false;
64:
65: return need_uri_encode(c);
1.4 paf 66: }
1.1 paf 67:
1.56 parser 68: //
69:
70: static const char * String_Untaint_lang_name[]={
71: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
72: "C", ///< clean
73: "T", ///< tainted, untaint language as assigned later
74: // untaint languages. assigned by ^untaint[lang]{...}
75: "P",
76: /**<
77: leave language built into string being appended.
78: just a flag, that value not stored
79: */
80: "A", ///< leave all characters intact
1.68 parser 81: "F", ///< file specification
82: "H", ///< ext in HTTP response header
1.56 parser 83: "M", ///< text in mail header
84: "URI", ///< text in uri
85: "T", ///< ^table:set body
86: "SQL", ///< ^table:sql body
87: "JS", ///< JavaScript code
1.68 parser 88: "XML", ///< ^dom:set xml
1.82 paf 89: "HTML" ///< HTML code (for editing)
1.56 parser 90: };
91:
92:
1.1 paf 93: // String
94:
1.41 paf 95: /*
96:
97: HTTP-header = field-name ":" [ field-value ] CRLF
98:
99: field-name = token
100: field-value = *( field-content | LWS )
101:
102: field-content = <the OCTETs making up the field-value
103: and consisting of either *TEXT or combinations
104: of token, tspecials, and quoted-string>
105:
106:
107:
108: word = token | quoted-string
109:
110: token = 1*<any CHAR except CTLs or tspecials>
111:
112:
113:
114: tspecials = "(" | ")" | "<" | ">" | "@"
115: | "," | ";" | ":" | "\" | <">
116: | "/" | "[" | "]" | "?" | "="
117: | "{" | "}" | SP | HT
118:
119: SP = <US-ASCII SP, space (32)>
120: HT = <US-ASCII HT, horizontal-tab (9)>
121:
122: LWS = [CRLF] 1*( SP | HT )
123: TEXT = <any OCTET except CTLs,
124: but including LWS>
125:
126: quoted-pair = "\" CHAR
127:
128: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
129: */
130: inline bool need_quote_http_header(const char *ptr, size_t size) {
131: for(; size--; ptr++)
1.42 paf 132: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 133: return true;
134: return false;
135: }
136:
1.91 paf 137: //#include "pa_sapi.h"
138: /**
1.115.2.1 paf 139: appends to other String,
1.91 paf 140: marking all tainted pieces of it with @a lang.
1.92 paf 141: or marking ALL pieces of it with a @a lang when @a forced to,
142: and propagating OPTIMIZE language bit.
1.91 paf 143: */
1.115.2.1 paf 144: String& String::append_to(String& dest, uchar lang, bool forced) const {
145: if(is_empty())
146: return dest;
147:
148: // without language-chage, small[fitting dest tail] string?
149: if(lang==UL_PASS_APPENDED && fused<=(dest.fallocated-dest.fused)) {
150: // felements[fused...]=src.felements[0..fused], fused+=src.fused
151: memcpy(&dest.felements[dest.fused], felements, sizeof(felements[0])*fused);
152: dest.fused+=fused;
153: return dest;
1.91 paf 154: }
155:
1.77 paf 156: // manually unrolled code to avoid do{if(const)} constructs
157: if(forced)
1.115.2.1 paf 158: STRING_FOREACH_FRAGMENT(
159: dest.APPEND(fragment->ptr, fragment->size,
1.77 paf 160: lang, //forcing passed lang
1.115.2.1 paf 161: fragment->origin.file, fragment->origin.line);
1.77 paf 162: )
163: else if(lang==UL_PASS_APPENDED)
1.115.2.1 paf 164: STRING_FOREACH_FRAGMENT(
165: dest.APPEND(fragment->ptr, fragment->size,
166: fragment->lang, // passing item's lang
167: fragment->origin.file, fragment->origin.line);
1.77 paf 168: )
169: else if(lang&UL_OPTIMIZE_BIT) // main idea here
170: // tainted piece would get OPTIMIZED bit from 'lang'
171: // clean piece would be marked OPTIMIZED manually
172: // pieces with determined languages [not tainted|clean] would retain theirs langs
1.115.2.1 paf 173: STRING_FOREACH_FRAGMENT(
174: dest.APPEND(fragment->ptr, fragment->size,
175: fragment->lang==UL_TAINTED?lang:(
176: fragment->lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
177: fragment->lang
1.77 paf 178: ),
1.115.2.1 paf 179: fragment->origin.file, fragment->origin.line);
1.77 paf 180: )
181: else
1.115.2.1 paf 182: STRING_FOREACH_FRAGMENT(
183: dest.APPEND(fragment->ptr, fragment->size,
184: fragment->lang==UL_TAINTED?lang:fragment->lang,
185: fragment->origin.file, fragment->origin.line);
1.77 paf 186: );
1.115.2.1 paf 187:
188: return dest;
1.77 paf 189: }
190:
1.75 paf 191: size_t String::cstr_bufsize(Untaint_lang lang,
192: SQL_Connection *connection,
1.115.2.3 paf 193: const Request_charsets *charsets) const {
1.77 paf 194: size_t dest=1; // for terminating 0
1.115.2.1 paf 195: STRING_FOREACH_FRAGMENT(
196: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 197:
198: switch(to_lang & ~UL_OPTIMIZE_BIT) {
199: case UL_CLEAN:
200: case UL_TAINTED:
201: case UL_AS_IS:
202: // clean piece
203:
204: // tainted piece, but undefined untaint language
205: // for VString.as_double of tainted values
206: // for ^process{body} evaluation
207:
208: // tainted, untaint language: as-is
1.115.2.1 paf 209: dest+=fragment->size;
1.77 paf 210: break;
211: case UL_FILE_SPEC:
212: // tainted, untaint language: file [name]
1.115.2.1 paf 213: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 214: break;
215: case UL_URI:
216: // tainted, untaint language: uri
1.115.2.1 paf 217: dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */;
1.77 paf 218: break;
219: case UL_HTTP_HEADER:
220: // tainted, untaint language: http-field-content-text
1.115.2.1 paf 221: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 222: break;
223: case UL_MAIL_HEADER:
224: // tainted, untaint language: mail-header
1.115.2.4 paf 225: if(charsets) {
1.111 paf 226: int parts_count=1;
1.115.2.1 paf 227: const char *prev=fragment->ptr;
228: size_t size=fragment->size;
1.111 paf 229: while(const char *comma_at=(const char*)memchr(prev, ',', size)) {
230: parts_count++;
231: size-=comma_at-prev;
232: prev=comma_at+1;
233: }
1.77 paf 234: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 235: dest+=
1.115.2.1 paf 236: fragment->size*3+
1.115.2.4 paf 237: parts_count*(charsets->mail().name()->size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */);
1.87 paf 238: } else
1.115.2.1 paf 239: dest+=fragment->size;
1.77 paf 240: break;
241: case UL_TABLE:
242: // tainted, untaint language: table
1.115.2.1 paf 243: dest+=fragment->size;
1.77 paf 244: break;
245: case UL_SQL:
246: // tainted, untaint language: sql
247: if(connection)
1.115.2.1 paf 248: dest+=connection->quote(0, fragment->ptr, fragment->size);
1.77 paf 249: break;
250: case UL_JS:
251: escape(switch(*src) {
252: case '"': case '\'': case '\n': case '\\': case '\xFF':
253: dest+=2; break;
254: default:
255: dest++; break;
256: });
257: break;
258: case UL_XML:
259: escape(switch(*src) {
260: case '&': case '>': case '<': case '"': case '\'':
261: dest+= 6; break;
262: default:
263: dest++; break;
264: });
265: break;
266: case UL_HTML:
267: escape(switch(*src) {
268: case '&':
269: case '>':
270: case '<':
271: case '"':
272: dest+=6; break;
273: default:
274: dest++; break;
275: });
276: break;
1.75 paf 277: }
1.77 paf 278: );
1.75 paf 279: return dest;
1.51 parser 280: }
281:
1.109 paf 282: /** http://www.ietf.org/rfc/rfc2047.txt
283: RFC
284: (3) As a replacement for a 'word' entity within a 'phrase', for example,
285: one that precedes an address in a From, To, or Cc header. The ABNF
286: definition for 'phrase' from RFC 822 thus becomes:
287:
288: phrase = 1*( encoded-word / word )
289:
290: In this case the set of characters that may be used in a "Q"-encoded
291: 'encoded-word' is restricted to: <upper and lower case ASCII
292: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
293: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
294: 'phrase' MUST be separated from any adjacent 'word', 'text' or
295: 'special' by 'linear-white-space'.
296: ...
297: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
298: represented as "_" (underscore, ASCII 95.). (This character may
299: not pass through some internetwork mail gateways, but its use
300: will greatly enhance readability of "Q" encoded data with mail
301: readers that do not support this encoding.) Note that the "_"
302: always represents hexadecimal 20, even if the SPACE character
303: occupies a different code position in the character set in use.
304:
305: paf: obviously,
306: without "=", or one could not differ "=E0" and "russian letter a"
307: and without "_", or in would mean 0x20
308: */
309: static bool mail_header_char_valid_within_Qencoded(char c) {
310: return c>='A' && c<='Z'
311: || c>='a' && c<='Z'
312: || c>='0' && c<='9'
313: || strchr("!*+-/", c);
314: }
1.43 paf 315: char *String::store_to(char *dest, Untaint_lang lang,
316: SQL_Connection *connection,
1.115.2.3 paf 317: const Request_charsets *charsets) const {
1.75 paf 318: // WARNING:
319: // before any changes check cstr_bufsize first!!!
1.44 paf 320: bool whitespace=true;
1.115.2.1 paf 321: // STRING_FOREACH_FRAGMENT(
322: element_type *last = felements+fused; element_type *fragment = felements; for(; fragment<last; fragment++) {
1.111 paf 323:
1.115.2.1 paf 324: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 325:
326: char *start=dest;
327:
328: switch(to_lang & ~UL_OPTIMIZE_BIT) {
329: case UL_CLEAN:
330: case UL_TAINTED:
331: case UL_AS_IS:
332: // clean piece
333:
334: // tainted piece, but undefined untaint language
335: // for VString.as_double of tainted values
336: // for ^process{body} evaluation
337:
338: // tainted, untaint language: as-is
1.115.2.1 paf 339: memcpy(dest, fragment->ptr, fragment->size);
340: dest+=fragment->size;
1.77 paf 341: break;
342: case UL_FILE_SPEC:
343: // tainted, untaint language: file [name]
1.83 paf 344: escape(
1.113 paf 345: // Macintosh has problems with small Russian letter 'r'
1.115.2.4 paf 346: if( *src=='\xF0' && charsets && *charsets->source().name()=="windows-1251" ) {
1.113 paf 347: // fixing that letter for most common charset
1.115.2.2 paf 348: to_char('p');
1.114 paf 349: } else // fallback to default
350: encode(need_file_encode, '_');
1.83 paf 351: );
1.77 paf 352: break;
353: case UL_URI:
354: // tainted, untaint language: uri
1.85 paf 355: const void *client_ptr;
356: size_t client_size;
1.115.2.2 paf 357: if(charsets)
1.115.2.4 paf 358: Charset::transcode(charsets->pool(),
359: charsets->source(), fragment->ptr, fragment->size,
360: charsets->client(), client_ptr, client_size);
1.115.2.2 paf 361: else {
362: client_ptr=fragment->ptr;
363: client_size=fragment->size;
364: }
365:
1.85 paf 366: {
367: const char *src=(const char *)client_ptr;
368: for(int size=client_size; size--; src++)
1.108 paf 369: encode(need_uri_encode, '%');
1.85 paf 370: }
1.77 paf 371: break;
372: case UL_HTTP_HEADER:
373: // tainted, untaint language: http-field-content-text
1.108 paf 374: escape(
375: encode(need_uri_encode, '%');
376: );
1.77 paf 377: break;
378: case UL_MAIL_HEADER:
379: // tainted, untaint language: mail-header
1.105 paf 380: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.4 paf 381: if(charsets) {
1.87 paf 382: const void *mail_ptr;
383: size_t mail_size;
1.115.2.4 paf 384: Charset::transcode(charsets->pool(),
385: charsets->source(), fragment->ptr, fragment->size,
386: charsets->mail(), mail_ptr, mail_size);
387: CharPtr charset_name=charsets->mail().name()->cstr();
1.87 paf 388:
1.77 paf 389: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 390: const char *src=(const char *)mail_ptr;
1.77 paf 391: bool to_quoted_printable=false;
1.105 paf 392:
1.111 paf 393: bool email=false;
394: for(const char *end=src+mail_size; src<end; src++) {
395: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
396: if(to_quoted_printable && (*src==',' || *src=='<')) {
397: email=*src=='<';
1.105 paf 398: dest+=sprintf(dest, "?=");
1.107 paf 399: to_quoted_printable=false;
1.105 paf 400: }
1.111 paf 401: if(!email && (
1.109 paf 402: !to_quoted_printable && (*src & 0x80) // starting quote-printable-encoding on first 8bit char
403: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(*src)
1.105 paf 404: )) {
1.77 paf 405: if(!to_quoted_printable) {
1.115.2.1 paf 406: dest+=sprintf(dest, "=?%s?Q?", charset_name);
1.77 paf 407: to_quoted_printable=true;
408: }
1.105 paf 409: //RFC Upper case should be used for hexadecimal digits "A" through "F"
1.109 paf 410: if(*src == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
411: *dest++='_'; // RFC may be represented as "_"
412: else
413: dest+=sprintf(dest, "=%02X", *src & 0xFF);
1.105 paf 414: } else
1.109 paf 415: *dest++=*src;
1.111 paf 416: if(*src=='>')
417: email=false;
1.44 paf 418: }
1.112 paf 419: if(to_quoted_printable) // close
1.77 paf 420: dest+=sprintf(dest, "?=");
1.87 paf 421:
1.77 paf 422: } else {
1.115.2.1 paf 423: memcpy(dest, fragment->ptr, fragment->size);
424: dest+=fragment->size;
1.1 paf 425: }
1.77 paf 426: break;
427: case UL_TABLE:
428: // tainted, untaint language: table
429: escape(switch(*src) {
430: case '\t': to_char(' '); break;
431: case '\n': to_char(' '); break;
432: _default;
433: });
434: break;
435: case UL_SQL:
436: // tainted, untaint language: sql
437: if(connection)
1.115.2.1 paf 438: dest+=connection->quote(dest, fragment->ptr, fragment->size);
1.77 paf 439: else
1.115.2.5! paf 440: throw Exception(Exception::undefined_type,
! 441: Exception::undefined_source,
1.77 paf 442: "untaint in SQL language failed - no connection specified");
443: break;
444: case UL_JS:
445: escape(switch(*src) {
446: case '"': to_string("\\\"", 2); break;
447: case '\'': to_string("\\'", 2); break;
448: case '\n': to_string("\\n", 2); break;
449: case '\\': to_string("\\\\", 2); break;
450: case '\xFF': to_string("\\\xFF", 2); break;
451: _default;
452: });
453: break;
454: case UL_XML:
455: escape(switch(*src) {
456: case '&': to_string("&", 5); break;
457: case '>': to_string(">", 4); break;
458: case '<': to_string("<", 4); break;
459: case '"': to_string(""", 6); break;
460: case '\'': to_string("'", 6); break;
461: _default;
462: });
463: break;
464: case UL_HTML:
465: escape(switch(*src) {
466: case '&': to_string("&", 5); break;
467: case '>': to_string(">", 4); break;
468: case '<': to_string("<", 4); break;
469: case '"': to_string(""", 6); break;
470: _default;
471: });
472: break;
473: default:
1.115.2.5! paf 474: throw Exception(Exception::undefined_type,
! 475: Exception::undefined_source,
1.81 paf 476: "unknown untaint language #%d",
1.110 paf 477: static_cast<int>(to_lang)); // should never
1.77 paf 478: break; // never
1.76 paf 479: }
1.55 parser 480:
1.77 paf 481: if(to_lang & UL_OPTIMIZE_BIT) {
482: // optimizing whitespace
483: char *stop=dest; dest=start;
484: for(char *src=start; src<stop; src++)
485: switch(*src) {
486: // of all consequent white space chars leaving only first one
1.80 paf 487: case ' ': case '\r': case '\n': case '\t':
1.77 paf 488: if(!whitespace) {
489: *dest++=*src;
490: whitespace=true;
491: }
492: break;
493: default:
494: whitespace=false;
495: *dest++=*src;
496: break;
497: };
498: } else // piece without optimization
499: whitespace=false;
1.115.2.1 paf 500: // );
501: }
502:
1.76 paf 503: return dest;
504: }
505:
1.115.2.2 paf 506: char *String::cstr_debug_origins(Pool& pool) const {
1.81 paf 507: //_asm int 3;
1.115.2.2 paf 508: char *result=new(pool) char[size()+fused*MAX_STRING*2];
1.76 paf 509: char *dest=result;
510:
1.115.2.1 paf 511: STRING_FOREACH_FRAGMENT(
1.96 paf 512: IFNDEF_NO_STRING_ORIGIN(
1.115.2.1 paf 513: if(fragment->origin.file)
1.96 paf 514: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
1.115.2.1 paf 515: fragment->origin.file,
516: 1+fragment->origin.line);
1.96 paf 517: else
518: dest+=sprintf(dest, "<unknown>");
519: );
1.115.2.1 paf 520: uchar show_lang=fragment->lang & ~UL_OPTIMIZE_BIT;
1.96 paf 521: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
1.115.2.5! paf 522: throw Exception(Exception::undefined_type,
! 523: Exception::undefined_source,
1.96 paf 524: "unknown untaint language #%d",
525: static_cast<int>(show_lang)); // sould never
526:
527: dest+=sprintf(dest, "#%s%s: ",
528: String_Untaint_lang_name[show_lang],
1.115.2.1 paf 529: fragment->lang & UL_OPTIMIZE_BIT?".O":"");
1.96 paf 530: char *dest_after_origins=dest;
1.76 paf 531:
1.115.2.1 paf 532: memcpy(dest, fragment->ptr, fragment->size);
533: dest+=fragment->size;
1.76 paf 534:
1.96 paf 535: remove_crlf(dest_after_origins, dest);
536: to_char('\n');
537: );
1.64 parser 538:
1.76 paf 539: *dest=0;
540: return result;
1.1 paf 541: }
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