Annotation of parser3/src/main/untaint.C, revision 1.115.2.12.2.19
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:
1.115.2.12.2. 9(paf 8:3): static const char* IDENT_UNTAINT_C="$Date: 2003/03/27 10:53:01 $";
(paf 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"
1.115.2.4 paf 19: #include "pa_request_charsets.h"
1.115.2.12.2. 9(paf 20:3): #include "pa_sapi.h"
1.1 paf 21:
1.115.2.12.2. (paf 22:): extern "C" { // author forgot to do that
23:): #include "ec.h"
24:): }
25:):
6(paf 26:3): #define PA_SQL
(paf 27:):
28:): #ifdef PA_SQL
29:): #include "pa_sql_connection.h"
30:): #endif
31:):
3(paf 32:3): // forwards
33:3):
34:3): void CORD_ec_flush_buf_optimize_whitespace(CORD_ec x);
35:3):
36:3): // defines
37:3):
38:3):
39:3): #undef CORD_ec_append
40:3): // redefining to intercept flushes and implement whitespace optimization
41:3): // of all consequent white space chars leaving only first one
42:3): #define CORD_ec_append(x, c) \
43:3): { \
44:3): bool skip=false; \
45:3): if(optimize) switch(c) { \
46:3): case ' ': case '\r': case '\n': case '\t': \
47:3): if(whitespace) \
48:3): skip=true; /*skipping subsequent*/ \
49:3): else \
50:3): whitespace=true; \
51:3): break; \
52:3): default: \
53:3): whitespace=false; \
54:3): break; \
55:3): } \
56:3): if(!skip) { \
57:3): if ((x)[0].ec_bufptr == (x)[0].ec_buf + CORD_BUFSZ) { \
58:3): CORD_ec_flush_buf(x); \
59:3): } \
60:3): *((x)[0].ec_bufptr)++ = (c); \
61:3): } \
62:3): }
63:3):
64:3):
1.18 paf 65: #define escape(action) \
1.115.2.12.2. (paf 66:): for(; fragment_length--; CORD_next(pos)) { \
67:): char c=CORD_pos_fetch(pos); \
68:): action \
1.1 paf 69: }
1.115.2.12.2. (paf 70:): #define _default default: CORD_ec_append(result, c); break
5(paf 71:3): #define encode(need_encode_func, prefix, otherwise) \
(paf 72:): if(need_encode_func(c)) { \
73:): static const char* hex="0123456789ABCDEF"; \
74:): CORD_ec_append(result, prefix); \
75:): CORD_ec_append(result, hex[((unsigned char)c)/0x10]); \
76:): CORD_ec_append(result, hex[((unsigned char)c)%0x10]); \
77:): } else \
5(paf 78:3): CORD_ec_append(result, otherwise);
(paf 79:): #define to_char(c) CORD_ec_append(result, c)
80:): #define to_string(s) CORD_ec_append_cord(result, s)
1.4 paf 81:
1.9 paf 82: inline bool need_file_encode(unsigned char c){
1.108 paf 83: // russian letters and space ENABLED
84: // encoding only these...
85: return strchr(
86: "*?'\"<>|"
87: #ifndef WIN32
88: ":\\~"
1.31 paf 89: #endif
1.108 paf 90: , c)!=0;
1.9 paf 91: }
1.5 paf 92: inline bool need_uri_encode(unsigned char c){
1.13 paf 93: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 94: return false;
95:
1.5 paf 96: return !strchr("_-./", c);
97: }
1.36 paf 98: inline bool need_http_header_encode(unsigned char c){
1.18 paf 99: if(strchr(" , :", c))
1.5 paf 100: return false;
101:
102: return need_uri_encode(c);
1.4 paf 103: }
1.1 paf 104:
1.56 parser 105: //
106:
1.115.2.6 paf 107: static const char* String_Untaint_lang_name[]={
1.56 parser 108: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
109: "C", ///< clean
110: "T", ///< tainted, untaint language as assigned later
111: // untaint languages. assigned by ^untaint[lang]{...}
112: "P",
113: /**<
114: leave language built into string being appended.
115: just a flag, that value not stored
116: */
117: "A", ///< leave all characters intact
1.68 parser 118: "F", ///< file specification
119: "H", ///< ext in HTTP response header
1.56 parser 120: "M", ///< text in mail header
121: "URI", ///< text in uri
122: "T", ///< ^table:set body
123: "SQL", ///< ^table:sql body
124: "JS", ///< JavaScript code
1.68 parser 125: "XML", ///< ^dom:set xml
1.82 paf 126: "HTML" ///< HTML code (for editing)
1.56 parser 127: };
128:
129:
1.1 paf 130: // String
131:
1.41 paf 132: /*
133:
134: HTTP-header = field-name ":" [ field-value ] CRLF
135:
136: field-name = token
137: field-value = *( field-content | LWS )
138:
139: field-content = <the OCTETs making up the field-value
140: and consisting of either *TEXT or combinations
141: of token, tspecials, and quoted-string>
142:
143:
144:
145: word = token | quoted-string
146:
147: token = 1*<any CHAR except CTLs or tspecials>
148:
149:
150:
151: tspecials = "(" | ")" | "<" | ">" | "@"
152: | "," | ";" | ":" | "\" | <">
153: | "/" | "[" | "]" | "?" | "="
154: | "{" | "}" | SP | HT
155:
156: SP = <US-ASCII SP, space (32)>
157: HT = <US-ASCII HT, horizontal-tab (9)>
158:
159: LWS = [CRLF] 1*( SP | HT )
160: TEXT = <any OCTET except CTLs,
161: but including LWS>
162:
163: quoted-pair = "\" CHAR
164:
165: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
166: */
1.115.2.6 paf 167: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 168: for(; size--; ptr++)
1.42 paf 169: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 170: return true;
171: return false;
172: }
173:
1.115.2.12.2. (paf 174:):
1.91 paf 175: /**
1.115.2.1 paf 176: appends to other String,
1.91 paf 177: marking all tainted pieces of it with @a lang.
1.92 paf 178: or marking ALL pieces of it with a @a lang when @a forced to,
179: and propagating OPTIMIZE language bit.
1.91 paf 180: */
1.115.2.12.2. (paf 181:): String& String::append_to(String& dest, Language lang, bool forced) const {
0(paf 182:3): if(is_empty())
1.115.2.1 paf 183: return dest;
184:
1.115.2.12.2. (paf 185:): // first: letters
7(paf 186:3): dest.body<<body;
1.91 paf 187:
1.115.2.12.2. (paf 188:): // next: fragment infos
189:):
190:): if(lang==L_PASS_APPENDED) // without language-chage string?
191:): dest.fragments.append(fragments);
192:): else if(forced) //forcing passed lang?
193:): dest.fragments+=Fragment(lang, length());
194:): else if(lang&L_OPTIMIZE_BIT) {
195:): Array_iterator<ArrayFragment::element_type> i(fragments);
196:): while(i.has_next()) {
8(paf 197:3): const Fragment fragment=i.next();
(paf 198:): // main idea here:
199:): // tainted piece would get OPTIMIZED bit from 'lang'
200:): // clean piece would be marked OPTIMIZED manually
201:): // pieces with determined languages [not tainted|clean] would retain theirs langs
202:): dest.fragments+=Fragment(static_cast<Language>(
203:): fragment.lang==L_TAINTED?lang:(
204:): fragment.lang==L_CLEAN?L_CLEAN|L_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
205:): fragment.lang)),
206:): fragment.length);
207:): }
208:): } else { // The core idea: tainted pieces got marked with context's lang
209:): Array_iterator<ArrayFragment::element_type> i(fragments);
210:): while(i.has_next()) {
8(paf 211:3): const Fragment fragment=i.next();
(paf 212:): dest.fragments+=Fragment(
213:): fragment.lang==L_TAINTED?lang:fragment.lang,
214:): fragment.length);
215:): }
216:): }
1.115.2.1 paf 217:
218: return dest;
1.77 paf 219: }
220:
1.109 paf 221: /** http://www.ietf.org/rfc/rfc2047.txt
222: RFC
223: (3) As a replacement for a 'word' entity within a 'phrase', for example,
224: one that precedes an address in a From, To, or Cc header. The ABNF
225: definition for 'phrase' from RFC 822 thus becomes:
226:
227: phrase = 1*( encoded-word / word )
228:
229: In this case the set of characters that may be used in a "Q"-encoded
230: 'encoded-word' is restricted to: <upper and lower case ASCII
231: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
232: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
233: 'phrase' MUST be separated from any adjacent 'word', 'text' or
234: 'special' by 'linear-white-space'.
235: ...
236: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
237: represented as "_" (underscore, ASCII 95.). (This character may
238: not pass through some internetwork mail gateways, but its use
239: will greatly enhance readability of "Q" encoded data with mail
240: readers that do not support this encoding.) Note that the "_"
241: always represents hexadecimal 20, even if the SPACE character
242: occupies a different code position in the character set in use.
243:
244: paf: obviously,
245: without "=", or one could not differ "=E0" and "russian letter a"
246: and without "_", or in would mean 0x20
247: */
1.115.2.12.2. (paf 248:): inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 249: return c>='A' && c<='Z'
250: || c>='a' && c<='Z'
251: || c>='0' && c<='9'
252: || strchr("!*+-/", c);
253: }
1.115.2.12.2. 5(paf 254:3): /**
255:3): RFC
256:3): Upper case should be used for hexadecimal digits "A" through "F"
257:3): The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
258:3): may be represented as "_"
259:3): */
260:3): inline bool mail_header_nonspace_char(char c) {
261:3): return c != 0x20;
262:3): }
3(paf 263:3): inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos& pos, size_t size) {
(paf 264:): while(size--) {
265:): CORD_ec_append(result, CORD_pos_fetch(pos));
266:): CORD_next(pos);
267:): }
268:): }
0(paf 269:3): StringBody String::cstr_to_string_body(Language lang,
(paf 270:): SQL_Connection* connection,
271:): const Request_charsets *charsets) const {
272:): CORD_ec result; CORD_ec_init(result);
0(paf 273:3): CORD_pos pos; body.set_pos(pos, 0);
(paf 274:):
1.44 paf 275: bool whitespace=true;
1.111 paf 276:
1.115.2.12.2. (paf 277:): size_t fragment_begin=0;
278:): size_t fragment_end;
279:): for(Array_iterator<Fragment> i(fragments); i.has_next(); fragment_begin=fragment_end) {
8(paf 280:3): const Fragment fragment=i.next();
(paf 281:): size_t fragment_length=fragment.length;
282:): fragment_end=fragment_begin+fragment_length;
2(paf 283:3): //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
1.77 paf 284:
1.115.2.12.2. (paf 285:): Language to_lang=lang==L_UNSPECIFIED?fragment.lang:lang;
3(paf 286:3): bool optimize=(to_lang & L_OPTIMIZE_BIT)!=0;
287:3): if(!optimize)
288:3): whitespace=false;
1.77 paf 289:
1.115.2.12.2. (paf 290:): switch(to_lang & ~L_OPTIMIZE_BIT) {
291:): case L_CLEAN:
292:): case L_TAINTED:
293:): case L_AS_IS:
1.77 paf 294: // clean piece
295:
296: // tainted piece, but undefined untaint language
297: // for VString.as_double of tainted values
298: // for ^process{body} evaluation
299:
300: // tainted, untaint language: as-is
1.115.2.12.2. 3(paf 301:3): ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 302: break;
1.115.2.12.2. (paf 303:): case L_FILE_SPEC:
1.77 paf 304: // tainted, untaint language: file [name]
1.83 paf 305: escape(
1.113 paf 306: // Macintosh has problems with small Russian letter 'r'
1.115.2.12.2. (paf 307:): if( c=='\xF0' && charsets && charsets->source().name()=="windows-1251" ) {
1.113 paf 308: // fixing that letter for most common charset
1.115.2.2 paf 309: to_char('p');
1.114 paf 310: } else // fallback to default
1.115.2.12.2. 5(paf 311:3): encode(need_file_encode, '_', c);
1.83 paf 312: );
1.77 paf 313: break;
1.115.2.12.2. (paf 314:): case L_URI:
1.77 paf 315: // tainted, untaint language: uri
1.85 paf 316: {
1.115.2.12.2. 6(paf 317:3): const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
4(paf 318:3): // skip source [we use recoded version]
6(paf 319:3): CORD_pos_advance(pos, fragment_length);
(paf 320:): const void *client_ptr;
321:): size_t client_size;
322:): if(charsets)
323:): Charset::transcode(
6(paf 324:3): charsets->source(), fragment_str, fragment_length,
(paf 325:): charsets->client(), client_ptr, client_size);
326:): else {
6(paf 327:3): client_ptr=fragment_str;
328:3): client_size=fragment_length;
(paf 329:): }
330:):
331:): char c;
332:): for(const char* src=(const char* )client_ptr; c=*src++; )
5(paf 333:3): encode(need_uri_encode, '%', c);
1.85 paf 334: }
1.77 paf 335: break;
1.115.2.12.2. (paf 336:): case L_HTTP_HEADER:
1.77 paf 337: // tainted, untaint language: http-field-content-text
1.108 paf 338: escape(
1.115.2.12.2. 5(paf 339:3): encode(need_uri_encode, '%', c);
1.108 paf 340: );
1.77 paf 341: break;
1.115.2.12.2. (paf 342:): case L_MAIL_HEADER:
1.77 paf 343: // tainted, untaint language: mail-header
1.105 paf 344: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.4 paf 345: if(charsets) {
1.115.2.12.2. 5(paf 346:3): size_t mail_size;
347:3): const char *mail_ptr=
348:3): body.mid(fragment_begin, mail_size=fragment_length).cstr();
349:3): // skip source [we use recoded version]
350:3): CORD_pos_advance(pos, mail_size);
351:3):
352:3): const char* charset_name=charsets->mail().name().cstr();
1.87 paf 353:
1.77 paf 354: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
355: bool to_quoted_printable=false;
1.105 paf 356:
1.111 paf 357: bool email=false;
1.115.2.12.2. 5(paf 358:3): uchar c;
359:3): for(const char* src=mail_ptr; c=(uchar)*src++; ) {
1.111 paf 360: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.115.2.12.2. (paf 361:): if(to_quoted_printable && (c==',' || c=='<')) {
362:): email=c=='<';
5(paf 363:3): to_string("?=");
1.107 paf 364: to_quoted_printable=false;
1.105 paf 365: }
1.111 paf 366: if(!email && (
1.115.2.12.2. (paf 367:): !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
368:): || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
1.105 paf 369: )) {
1.77 paf 370: if(!to_quoted_printable) {
1.115.2.12.2. 5(paf 371:3): to_string("=?");
372:3): to_string(charset_name);
373:3): to_string("?Q?");
1.77 paf 374: to_quoted_printable=true;
375: }
1.115.2.12.2. 5(paf 376:3): encode(mail_header_nonspace_char, '=', '_');
1.105 paf 377: } else
1.115.2.12.2. 5(paf 378:3): to_char(c);
(paf 379:): if(c=='>')
1.111 paf 380: email=false;
1.44 paf 381: }
1.112 paf 382: if(to_quoted_printable) // close
1.115.2.12.2. 5(paf 383:3): to_string("?=");
1.87 paf 384:
1.115.2.12.2. 5(paf 385:3): } else
386:3): ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 387: break;
1.115.2.12.2. (paf 388:): case L_TABLE:
1.77 paf 389: // tainted, untaint language: table
1.115.2.12.2. (paf 390:): escape(switch(c) {
1.77 paf 391: case '\t': to_char(' '); break;
392: case '\n': to_char(' '); break;
393: _default;
394: });
395: break;
1.115.2.12.2. (paf 396:): #ifdef PA_SQL
397:): case L_SQL:
1.77 paf 398: // tainted, untaint language: sql
1.115.2.12.2. (paf 399:): if(connection) {
6(paf 400:3): const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
401:3): // skip source [we use recoded version]
402:3): CORD_pos_advance(pos, fragment_length);
(paf 403:):
6(paf 404:3): to_string(connection->quote(fragment_str, fragment_length));
405:3): } else
(paf 406:): throw Exception(0,
407:): 0,
1.77 paf 408: "untaint in SQL language failed - no connection specified");
409: break;
1.115.2.12.2. (paf 410:): #endif
411:): case L_JS:
412:): escape(switch(c) {
413:): case '"': to_string("\\\""); break;
414:): case '\'': to_string("\\'"); break;
415:): case '\n': to_string("\\n"); break;
416:): case '\\': to_string("\\\\"); break;
417:): case '\xFF': to_string("\\\xFF"); break;
1.77 paf 418: _default;
419: });
420: break;
1.115.2.12.2. (paf 421:): case L_XML:
422:): escape(switch(c) {
423:): case '&': to_string("&"); break;
424:): case '>': to_string(">"); break;
425:): case '<': to_string("<"); break;
426:): case '"': to_string("""); break;
427:): case '\'': to_string("'"); break;
1.77 paf 428: _default;
429: });
430: break;
1.115.2.12.2. (paf 431:): case L_HTML:
432:): escape(switch(c) {
433:): case '&': to_string("&"); break;
434:): case '>': to_string(">"); break;
435:): case '<': to_string("<"); break;
436:): case '"': to_string("""); break;
1.77 paf 437: _default;
438: });
439: break;
440: default:
1.115.2.12.2. 9(paf 441:3): SAPI::die("unknown untaint language #%d",
442:3): static_cast<int>(to_lang)); // should never
1.77 paf 443: break; // never
1.76 paf 444: }
1.115.2.1 paf 445: }
446:
1.115.2.12.2. 0(paf 447:3): return StringBody(CORD_ec_to_cord(result));
1.1 paf 448: }
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