![[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|
Return to untaint.C CVS log | Up to [parser3project] / parser3 / src / main |
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:
1.115.2.1! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/01/21 15:51:15 $";
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"
1.1 paf 20:
1.18 paf 21: #define escape(action) \
1.1 paf 22: { \
1.115.2.1! paf 23: const char *src=fragment->ptr; \
! 24: for(int size=fragment->size; size--; src++) \
1.18 paf 25: action \
1.1 paf 26: }
1.13 paf 27: #define _default default: *dest++=*src; break
28: #define encode(need_encode_func, prefix) \
29: if(need_encode_func(*src)) { \
1.5 paf 30: static const char *hex="0123456789ABCDEF"; \
1.9 paf 31: char chunk[3]={prefix}; \
1.13 paf 32: chunk[1]=hex[((unsigned char)*src)/0x10]; \
33: chunk[2]=hex[((unsigned char)*src)%0x10]; \
1.60 parser 34: memcpy(dest, chunk, 3); dest+=3; \
1.5 paf 35: } else \
1.13 paf 36: *dest++=*src; \
1.5 paf 37: break
1.18 paf 38: #define to_char(c) *dest++=c
39: #define to_string(b, bsize) \
1.60 parser 40: memcpy(dest, b, bsize); \
1.18 paf 41: dest+=bsize; \
1.4 paf 42:
1.9 paf 43: inline bool need_file_encode(unsigned char c){
1.108 paf 44: // russian letters and space ENABLED
45: // encoding only these...
46: return strchr(
47: "*?'\"<>|"
48: #ifndef WIN32
49: ":\\~"
1.31 paf 50: #endif
1.108 paf 51: , c)!=0;
1.9 paf 52: }
1.5 paf 53: inline bool need_uri_encode(unsigned char c){
1.13 paf 54: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 55: return false;
56:
1.5 paf 57: return !strchr("_-./", c);
58: }
1.36 paf 59: inline bool need_http_header_encode(unsigned char c){
1.18 paf 60: if(strchr(" , :", c))
1.5 paf 61: return false;
62:
63: return need_uri_encode(c);
1.4 paf 64: }
1.1 paf 65:
1.56 parser 66: //
67:
68: static const char * String_Untaint_lang_name[]={
69: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
70: "C", ///< clean
71: "T", ///< tainted, untaint language as assigned later
72: // untaint languages. assigned by ^untaint[lang]{...}
73: "P",
74: /**<
75: leave language built into string being appended.
76: just a flag, that value not stored
77: */
78: "A", ///< leave all characters intact
1.68 parser 79: "F", ///< file specification
80: "H", ///< ext in HTTP response header
1.56 parser 81: "M", ///< text in mail header
82: "URI", ///< text in uri
83: "T", ///< ^table:set body
84: "SQL", ///< ^table:sql body
85: "JS", ///< JavaScript code
1.68 parser 86: "XML", ///< ^dom:set xml
1.82 paf 87: "HTML" ///< HTML code (for editing)
1.56 parser 88: };
89:
90:
1.1 paf 91: // String
92:
1.41 paf 93: /*
94:
95: HTTP-header = field-name ":" [ field-value ] CRLF
96:
97: field-name = token
98: field-value = *( field-content | LWS )
99:
100: field-content = <the OCTETs making up the field-value
101: and consisting of either *TEXT or combinations
102: of token, tspecials, and quoted-string>
103:
104:
105:
106: word = token | quoted-string
107:
108: token = 1*<any CHAR except CTLs or tspecials>
109:
110:
111:
112: tspecials = "(" | ")" | "<" | ">" | "@"
113: | "," | ";" | ":" | "\" | <">
114: | "/" | "[" | "]" | "?" | "="
115: | "{" | "}" | SP | HT
116:
117: SP = <US-ASCII SP, space (32)>
118: HT = <US-ASCII HT, horizontal-tab (9)>
119:
120: LWS = [CRLF] 1*( SP | HT )
121: TEXT = <any OCTET except CTLs,
122: but including LWS>
123:
124: quoted-pair = "\" CHAR
125:
126: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
127: */
128: inline bool need_quote_http_header(const char *ptr, size_t size) {
129: for(; size--; ptr++)
1.42 paf 130: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 131: return true;
132: return false;
133: }
134:
1.91 paf 135: //#include "pa_sapi.h"
136: /**
1.115.2.1! paf 137: appends to other String,
1.91 paf 138: marking all tainted pieces of it with @a lang.
1.92 paf 139: or marking ALL pieces of it with a @a lang when @a forced to,
140: and propagating OPTIMIZE language bit.
1.91 paf 141: */
1.115.2.1! paf 142: String& String::append_to(String& dest, uchar lang, bool forced) const {
! 143: if(is_empty())
! 144: return dest;
! 145:
! 146: // without language-chage, small[fitting dest tail] string?
! 147: if(lang==UL_PASS_APPENDED && fused<=(dest.fallocated-dest.fused)) {
! 148: // felements[fused...]=src.felements[0..fused], fused+=src.fused
! 149: memcpy(&dest.felements[dest.fused], felements, sizeof(felements[0])*fused);
! 150: dest.fused+=fused;
! 151: return dest;
1.91 paf 152: }
153:
1.77 paf 154: // manually unrolled code to avoid do{if(const)} constructs
155: if(forced)
1.115.2.1! paf 156: STRING_FOREACH_FRAGMENT(
! 157: dest.APPEND(fragment->ptr, fragment->size,
1.77 paf 158: lang, //forcing passed lang
1.115.2.1! paf 159: fragment->origin.file, fragment->origin.line);
1.77 paf 160: )
161: else if(lang==UL_PASS_APPENDED)
1.115.2.1! paf 162: STRING_FOREACH_FRAGMENT(
! 163: dest.APPEND(fragment->ptr, fragment->size,
! 164: fragment->lang, // passing item's lang
! 165: fragment->origin.file, fragment->origin.line);
1.77 paf 166: )
167: else if(lang&UL_OPTIMIZE_BIT) // main idea here
168: // tainted piece would get OPTIMIZED bit from 'lang'
169: // clean piece would be marked OPTIMIZED manually
170: // pieces with determined languages [not tainted|clean] would retain theirs langs
1.115.2.1! paf 171: STRING_FOREACH_FRAGMENT(
! 172: dest.APPEND(fragment->ptr, fragment->size,
! 173: fragment->lang==UL_TAINTED?lang:(
! 174: fragment->lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
! 175: fragment->lang
1.77 paf 176: ),
1.115.2.1! paf 177: fragment->origin.file, fragment->origin.line);
1.77 paf 178: )
179: else
1.115.2.1! paf 180: STRING_FOREACH_FRAGMENT(
! 181: dest.APPEND(fragment->ptr, fragment->size,
! 182: fragment->lang==UL_TAINTED?lang:fragment->lang,
! 183: fragment->origin.file, fragment->origin.line);
1.77 paf 184: );
1.115.2.1! paf 185:
! 186: return dest;
1.77 paf 187: }
188:
1.75 paf 189: size_t String::cstr_bufsize(Untaint_lang lang,
190: SQL_Connection *connection,
1.115.2.1! paf 191: const char *charset_name) const {
1.77 paf 192: size_t dest=1; // for terminating 0
1.115.2.1! paf 193: STRING_FOREACH_FRAGMENT(
! 194: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 195:
196: switch(to_lang & ~UL_OPTIMIZE_BIT) {
197: case UL_CLEAN:
198: case UL_TAINTED:
199: case UL_AS_IS:
200: // clean piece
201:
202: // tainted piece, but undefined untaint language
203: // for VString.as_double of tainted values
204: // for ^process{body} evaluation
205:
206: // tainted, untaint language: as-is
1.115.2.1! paf 207: dest+=fragment->size;
1.77 paf 208: break;
209: case UL_FILE_SPEC:
210: // tainted, untaint language: file [name]
1.115.2.1! paf 211: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 212: break;
213: case UL_URI:
214: // tainted, untaint language: uri
1.115.2.1! paf 215: dest+=fragment->size*6*3/* worst utf8 x worst Z->%XX */;
1.77 paf 216: break;
217: case UL_HTTP_HEADER:
218: // tainted, untaint language: http-field-content-text
1.115.2.1! paf 219: dest+=fragment->size*3/* worst: Z->%XX */;
1.77 paf 220: break;
221: case UL_MAIL_HEADER:
222: // tainted, untaint language: mail-header
1.115.2.1! paf 223: if(charset_name) {
1.111 paf 224: int parts_count=1;
1.115.2.1! paf 225: const char *prev=fragment->ptr;
! 226: size_t size=fragment->size;
1.111 paf 227: while(const char *comma_at=(const char*)memchr(prev, ',', size)) {
228: parts_count++;
229: size-=comma_at-prev;
230: prev=comma_at+1;
231: }
1.77 paf 232: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 233: dest+=
1.115.2.1! paf 234: fragment->size*3+
! 235: parts_count*(strlen(charset_name)+MAX_STRING/* worst: =?charset?Q?=%XX?= */);
1.87 paf 236: } else
1.115.2.1! paf 237: dest+=fragment->size;
1.77 paf 238: break;
239: case UL_TABLE:
240: // tainted, untaint language: table
1.115.2.1! paf 241: dest+=fragment->size;
1.77 paf 242: break;
243: case UL_SQL:
244: // tainted, untaint language: sql
245: if(connection)
1.115.2.1! paf 246: dest+=connection->quote(0, fragment->ptr, fragment->size);
1.77 paf 247: break;
248: case UL_JS:
249: escape(switch(*src) {
250: case '"': case '\'': case '\n': case '\\': case '\xFF':
251: dest+=2; break;
252: default:
253: dest++; break;
254: });
255: break;
256: case UL_XML:
257: escape(switch(*src) {
258: case '&': case '>': case '<': case '"': case '\'':
259: dest+= 6; break;
260: default:
261: dest++; break;
262: });
263: break;
264: case UL_HTML:
265: escape(switch(*src) {
266: case '&':
267: case '>':
268: case '<':
269: case '"':
270: dest+=6; break;
271: default:
272: dest++; break;
273: });
274: break;
1.75 paf 275: }
1.77 paf 276: );
1.75 paf 277: return dest;
1.51 parser 278: }
279:
1.109 paf 280: /** http://www.ietf.org/rfc/rfc2047.txt
281: RFC
282: (3) As a replacement for a 'word' entity within a 'phrase', for example,
283: one that precedes an address in a From, To, or Cc header. The ABNF
284: definition for 'phrase' from RFC 822 thus becomes:
285:
286: phrase = 1*( encoded-word / word )
287:
288: In this case the set of characters that may be used in a "Q"-encoded
289: 'encoded-word' is restricted to: <upper and lower case ASCII
290: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
291: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
292: 'phrase' MUST be separated from any adjacent 'word', 'text' or
293: 'special' by 'linear-white-space'.
294: ...
295: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
296: represented as "_" (underscore, ASCII 95.). (This character may
297: not pass through some internetwork mail gateways, but its use
298: will greatly enhance readability of "Q" encoded data with mail
299: readers that do not support this encoding.) Note that the "_"
300: always represents hexadecimal 20, even if the SPACE character
301: occupies a different code position in the character set in use.
302:
303: paf: obviously,
304: without "=", or one could not differ "=E0" and "russian letter a"
305: and without "_", or in would mean 0x20
306: */
307: static bool mail_header_char_valid_within_Qencoded(char c) {
308: return c>='A' && c<='Z'
309: || c>='a' && c<='Z'
310: || c>='0' && c<='9'
311: || strchr("!*+-/", c);
312: }
1.43 paf 313: char *String::store_to(char *dest, Untaint_lang lang,
314: SQL_Connection *connection,
1.115.2.1! paf 315: const char *charset_name) const {
1.75 paf 316: // WARNING:
317: // before any changes check cstr_bufsize first!!!
1.44 paf 318: bool whitespace=true;
1.115.2.1! paf 319: // STRING_FOREACH_FRAGMENT(
! 320: element_type *last = felements+fused; element_type *fragment = felements; for(; fragment<last; fragment++) {
1.111 paf 321:
1.115.2.1! paf 322: uchar to_lang=lang==UL_UNSPECIFIED?fragment->lang:lang;
1.77 paf 323:
324: char *start=dest;
325:
326: switch(to_lang & ~UL_OPTIMIZE_BIT) {
327: case UL_CLEAN:
328: case UL_TAINTED:
329: case UL_AS_IS:
330: // clean piece
331:
332: // tainted piece, but undefined untaint language
333: // for VString.as_double of tainted values
334: // for ^process{body} evaluation
335:
336: // tainted, untaint language: as-is
1.115.2.1! paf 337: memcpy(dest, fragment->ptr, fragment->size);
! 338: dest+=fragment->size;
1.77 paf 339: break;
340: case UL_FILE_SPEC:
341: // tainted, untaint language: file [name]
1.83 paf 342: escape(
1.113 paf 343: // Macintosh has problems with small Russian letter 'r'
344: if( *src=='\xF0' && pool().get_source_charset().name()=="windows-1251" ) {
345: // fixing that letter for most common charset
346: to_char('p');
1.114 paf 347: } else // fallback to default
348: encode(need_file_encode, '_');
1.83 paf 349: );
1.77 paf 350: break;
351: case UL_URI:
352: // tainted, untaint language: uri
1.85 paf 353: const void *client_ptr;
354: size_t client_size;
355: Charset::transcode(pool(),
1.115.2.1! paf 356: pool().get_source_charset(), fragment->ptr, fragment->size,
1.85 paf 357: pool().get_client_charset(), client_ptr, client_size);
358: {
359: const char *src=(const char *)client_ptr;
360: for(int size=client_size; size--; src++)
1.108 paf 361: encode(need_uri_encode, '%');
1.85 paf 362: }
1.77 paf 363: break;
364: case UL_HTTP_HEADER:
365: // tainted, untaint language: http-field-content-text
1.108 paf 366: escape(
367: encode(need_uri_encode, '%');
368: );
1.77 paf 369: break;
370: case UL_MAIL_HEADER:
371: // tainted, untaint language: mail-header
1.105 paf 372: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.1! paf 373: if(charset_name) {
1.87 paf 374: const void *mail_ptr;
375: size_t mail_size;
376:
1.77 paf 377: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 378: const char *src=(const char *)mail_ptr;
1.77 paf 379: bool to_quoted_printable=false;
1.105 paf 380:
1.111 paf 381: bool email=false;
382: for(const char *end=src+mail_size; src<end; src++) {
383: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
384: if(to_quoted_printable && (*src==',' || *src=='<')) {
385: email=*src=='<';
1.105 paf 386: dest+=sprintf(dest, "?=");
1.107 paf 387: to_quoted_printable=false;
1.105 paf 388: }
1.111 paf 389: if(!email && (
1.109 paf 390: !to_quoted_printable && (*src & 0x80) // starting quote-printable-encoding on first 8bit char
391: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(*src)
1.105 paf 392: )) {
1.77 paf 393: if(!to_quoted_printable) {
1.115.2.1! paf 394: dest+=sprintf(dest, "=?%s?Q?", charset_name);
1.77 paf 395: to_quoted_printable=true;
396: }
1.105 paf 397: //RFC Upper case should be used for hexadecimal digits "A" through "F"
1.109 paf 398: if(*src == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
399: *dest++='_'; // RFC may be represented as "_"
400: else
401: dest+=sprintf(dest, "=%02X", *src & 0xFF);
1.105 paf 402: } else
1.109 paf 403: *dest++=*src;
1.111 paf 404: if(*src=='>')
405: email=false;
1.44 paf 406: }
1.112 paf 407: if(to_quoted_printable) // close
1.77 paf 408: dest+=sprintf(dest, "?=");
1.87 paf 409:
1.77 paf 410: } else {
1.115.2.1! paf 411: memcpy(dest, fragment->ptr, fragment->size);
! 412: dest+=fragment->size;
1.1 paf 413: }
1.77 paf 414: break;
415: case UL_TABLE:
416: // tainted, untaint language: table
417: escape(switch(*src) {
418: case '\t': to_char(' '); break;
419: case '\n': to_char(' '); break;
420: _default;
421: });
422: break;
423: case UL_SQL:
424: // tainted, untaint language: sql
425: if(connection)
1.115.2.1! paf 426: dest+=connection->quote(dest, fragment->ptr, fragment->size);
1.77 paf 427: else
1.99 paf 428: throw Exception(0,
1.77 paf 429: this,
430: "untaint in SQL language failed - no connection specified");
431: break;
432: case UL_JS:
433: escape(switch(*src) {
434: case '"': to_string("\\\"", 2); break;
435: case '\'': to_string("\\'", 2); break;
436: case '\n': to_string("\\n", 2); break;
437: case '\\': to_string("\\\\", 2); break;
438: case '\xFF': to_string("\\\xFF", 2); break;
439: _default;
440: });
441: break;
442: case UL_XML:
443: escape(switch(*src) {
444: case '&': to_string("&", 5); break;
445: case '>': to_string(">", 4); break;
446: case '<': to_string("<", 4); break;
447: case '"': to_string(""", 6); break;
448: case '\'': to_string("'", 6); break;
449: _default;
450: });
451: break;
452: case UL_HTML:
453: escape(switch(*src) {
454: case '&': to_string("&", 5); break;
455: case '>': to_string(">", 4); break;
456: case '<': to_string("<", 4); break;
457: case '"': to_string(""", 6); break;
458: _default;
459: });
460: break;
461: default:
1.99 paf 462: throw Exception(0,
1.77 paf 463: this,
1.81 paf 464: "unknown untaint language #%d",
1.110 paf 465: static_cast<int>(to_lang)); // should never
1.77 paf 466: break; // never
1.76 paf 467: }
1.55 parser 468:
1.77 paf 469: if(to_lang & UL_OPTIMIZE_BIT) {
470: // optimizing whitespace
471: char *stop=dest; dest=start;
472: for(char *src=start; src<stop; src++)
473: switch(*src) {
474: // of all consequent white space chars leaving only first one
1.80 paf 475: case ' ': case '\r': case '\n': case '\t':
1.77 paf 476: if(!whitespace) {
477: *dest++=*src;
478: whitespace=true;
479: }
480: break;
481: default:
482: whitespace=false;
483: *dest++=*src;
484: break;
485: };
486: } else // piece without optimization
487: whitespace=false;
1.115.2.1! paf 488: // );
! 489: }
! 490:
1.76 paf 491: return dest;
492: }
493:
494: char *String::cstr_debug_origins() const {
1.81 paf 495: //_asm int 3;
1.76 paf 496: char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2);
497: char *dest=result;
498:
1.115.2.1! paf 499: STRING_FOREACH_FRAGMENT(
1.96 paf 500: IFNDEF_NO_STRING_ORIGIN(
1.115.2.1! paf 501: if(fragment->origin.file)
1.96 paf 502: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
1.115.2.1! paf 503: fragment->origin.file,
! 504: 1+fragment->origin.line);
1.96 paf 505: else
506: dest+=sprintf(dest, "<unknown>");
507: );
1.115.2.1! paf 508: uchar show_lang=fragment->lang & ~UL_OPTIMIZE_BIT;
1.96 paf 509: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
1.99 paf 510: throw Exception(0,
1.96 paf 511: this,
512: "unknown untaint language #%d",
513: static_cast<int>(show_lang)); // sould never
514:
515: dest+=sprintf(dest, "#%s%s: ",
516: String_Untaint_lang_name[show_lang],
1.115.2.1! paf 517: fragment->lang & UL_OPTIMIZE_BIT?".O":"");
1.96 paf 518: char *dest_after_origins=dest;
1.76 paf 519:
1.115.2.1! paf 520: memcpy(dest, fragment->ptr, fragment->size);
! 521: dest+=fragment->size;
1.76 paf 522:
1.96 paf 523: remove_crlf(dest_after_origins, dest);
524: to_char('\n');
525: );
1.64 parser 526:
1.76 paf 527: *dest=0;
528: return result;
1.1 paf 529: }