Annotation of parser3/src/main/untaint.C, revision 1.115.2.12.2.13
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. 3(paf 8:3): static const char* IDENT_UNTAINT_C="$Date: 2003/03/24 16:08:35 $";
(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"
1.1 paf 20:
1.115.2.12.2. (paf 21:): extern "C" { // author forgot to do that
22:): #include "ec.h"
23:): }
24:):
25:): //#define PA_SQL
26:):
27:): #ifdef PA_SQL
28:): #include "pa_sql_connection.h"
29:): #endif
30:):
3(paf 31:3): // forwards
32:3):
33:3): void CORD_ec_flush_buf_optimize_whitespace(CORD_ec x);
34:3):
35:3): // defines
36:3):
37:3):
38:3): #undef CORD_ec_append
39:3): // redefining to intercept flushes and implement whitespace optimization
40:3): // of all consequent white space chars leaving only first one
41:3): #define CORD_ec_append(x, c) \
42:3): { \
43:3): bool skip=false; \
44:3): if(optimize) switch(c) { \
45:3): case ' ': case '\r': case '\n': case '\t': \
46:3): if(whitespace) \
47:3): skip=true; /*skipping subsequent*/ \
48:3): else \
49:3): whitespace=true; \
50:3): break; \
51:3): default: \
52:3): whitespace=false; \
53:3): break; \
54:3): } \
55:3): if(!skip) { \
56:3): if ((x)[0].ec_bufptr == (x)[0].ec_buf + CORD_BUFSZ) { \
57:3): CORD_ec_flush_buf(x); \
58:3): } \
59:3): *((x)[0].ec_bufptr)++ = (c); \
60:3): } \
61:3): }
62:3):
63:3):
1.18 paf 64: #define escape(action) \
1.115.2.12.2. (paf 65:): for(; fragment_length--; CORD_next(pos)) { \
66:): char c=CORD_pos_fetch(pos); \
67:): action \
1.1 paf 68: }
1.115.2.12.2. (paf 69:): #define _default default: CORD_ec_append(result, c); break
1.13 paf 70: #define encode(need_encode_func, prefix) \
1.115.2.12.2. (paf 71:): if(need_encode_func(c)) { \
72:): static const char* hex="0123456789ABCDEF"; \
73:): CORD_ec_append(result, prefix); \
74:): CORD_ec_append(result, hex[((unsigned char)c)/0x10]); \
75:): CORD_ec_append(result, hex[((unsigned char)c)%0x10]); \
76:): } else \
77:): CORD_ec_append(result, c);
78:): #define to_char(c) CORD_ec_append(result, c)
79:): #define to_string(s) CORD_ec_append_cord(result, s)
1.4 paf 80:
1.9 paf 81: inline bool need_file_encode(unsigned char c){
1.108 paf 82: // russian letters and space ENABLED
83: // encoding only these...
84: return strchr(
85: "*?'\"<>|"
86: #ifndef WIN32
87: ":\\~"
1.31 paf 88: #endif
1.108 paf 89: , c)!=0;
1.9 paf 90: }
1.5 paf 91: inline bool need_uri_encode(unsigned char c){
1.13 paf 92: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 93: return false;
94:
1.5 paf 95: return !strchr("_-./", c);
96: }
1.36 paf 97: inline bool need_http_header_encode(unsigned char c){
1.18 paf 98: if(strchr(" , :", c))
1.5 paf 99: return false;
100:
101: return need_uri_encode(c);
1.4 paf 102: }
1.1 paf 103:
1.56 parser 104: //
105:
1.115.2.6 paf 106: static const char* String_Untaint_lang_name[]={
1.56 parser 107: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
108: "C", ///< clean
109: "T", ///< tainted, untaint language as assigned later
110: // untaint languages. assigned by ^untaint[lang]{...}
111: "P",
112: /**<
113: leave language built into string being appended.
114: just a flag, that value not stored
115: */
116: "A", ///< leave all characters intact
1.68 parser 117: "F", ///< file specification
118: "H", ///< ext in HTTP response header
1.56 parser 119: "M", ///< text in mail header
120: "URI", ///< text in uri
121: "T", ///< ^table:set body
122: "SQL", ///< ^table:sql body
123: "JS", ///< JavaScript code
1.68 parser 124: "XML", ///< ^dom:set xml
1.82 paf 125: "HTML" ///< HTML code (for editing)
1.56 parser 126: };
127:
128:
1.1 paf 129: // String
130:
1.41 paf 131: /*
132:
133: HTTP-header = field-name ":" [ field-value ] CRLF
134:
135: field-name = token
136: field-value = *( field-content | LWS )
137:
138: field-content = <the OCTETs making up the field-value
139: and consisting of either *TEXT or combinations
140: of token, tspecials, and quoted-string>
141:
142:
143:
144: word = token | quoted-string
145:
146: token = 1*<any CHAR except CTLs or tspecials>
147:
148:
149:
150: tspecials = "(" | ")" | "<" | ">" | "@"
151: | "," | ";" | ":" | "\" | <">
152: | "/" | "[" | "]" | "?" | "="
153: | "{" | "}" | SP | HT
154:
155: SP = <US-ASCII SP, space (32)>
156: HT = <US-ASCII HT, horizontal-tab (9)>
157:
158: LWS = [CRLF] 1*( SP | HT )
159: TEXT = <any OCTET except CTLs,
160: but including LWS>
161:
162: quoted-pair = "\" CHAR
163:
164: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
165: */
1.115.2.6 paf 166: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 167: for(; size--; ptr++)
1.42 paf 168: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 169: return true;
170: return false;
171: }
172:
1.115.2.12.2. (paf 173:):
1.91 paf 174: /**
1.115.2.1 paf 175: appends to other String,
1.91 paf 176: marking all tainted pieces of it with @a lang.
1.92 paf 177: or marking ALL pieces of it with a @a lang when @a forced to,
178: and propagating OPTIMIZE language bit.
1.91 paf 179: */
1.115.2.12.2. (paf 180:): String& String::append_to(String& dest, Language lang, bool forced) const {
0(paf 181:3): if(is_empty())
1.115.2.1 paf 182: return dest;
183:
1.115.2.12.2. (paf 184:): // first: letters
0(paf 185:3): dest.body.append(body);
1.91 paf 186:
1.115.2.12.2. (paf 187:): // next: fragment infos
188:):
189:): if(lang==L_PASS_APPENDED) // without language-chage string?
190:): dest.fragments.append(fragments);
191:): else if(forced) //forcing passed lang?
192:): dest.fragments+=Fragment(lang, length());
193:): else if(lang&L_OPTIMIZE_BIT) {
194:): Array_iterator<ArrayFragment::element_type> i(fragments);
195:): while(i.has_next()) {
196:): const Fragment& fragment=i.next();
197:): // main idea here:
198:): // tainted piece would get OPTIMIZED bit from 'lang'
199:): // clean piece would be marked OPTIMIZED manually
200:): // pieces with determined languages [not tainted|clean] would retain theirs langs
201:): dest.fragments+=Fragment(static_cast<Language>(
202:): fragment.lang==L_TAINTED?lang:(
203:): fragment.lang==L_CLEAN?L_CLEAN|L_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
204:): fragment.lang)),
205:): fragment.length);
206:): }
207:): } else { // The core idea: tainted pieces got marked with context's lang
208:): Array_iterator<ArrayFragment::element_type> i(fragments);
209:): while(i.has_next()) {
210:): const Fragment& fragment=i.next();
211:): dest.fragments+=Fragment(
212:): fragment.lang==L_TAINTED?lang:fragment.lang,
213:): fragment.length);
214:): }
215:): }
1.115.2.1 paf 216:
217: return dest;
1.77 paf 218: }
219:
1.109 paf 220: /** http://www.ietf.org/rfc/rfc2047.txt
221: RFC
222: (3) As a replacement for a 'word' entity within a 'phrase', for example,
223: one that precedes an address in a From, To, or Cc header. The ABNF
224: definition for 'phrase' from RFC 822 thus becomes:
225:
226: phrase = 1*( encoded-word / word )
227:
228: In this case the set of characters that may be used in a "Q"-encoded
229: 'encoded-word' is restricted to: <upper and lower case ASCII
230: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
231: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
232: 'phrase' MUST be separated from any adjacent 'word', 'text' or
233: 'special' by 'linear-white-space'.
234: ...
235: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
236: represented as "_" (underscore, ASCII 95.). (This character may
237: not pass through some internetwork mail gateways, but its use
238: will greatly enhance readability of "Q" encoded data with mail
239: readers that do not support this encoding.) Note that the "_"
240: always represents hexadecimal 20, even if the SPACE character
241: occupies a different code position in the character set in use.
242:
243: paf: obviously,
244: without "=", or one could not differ "=E0" and "russian letter a"
245: and without "_", or in would mean 0x20
246: */
1.115.2.12.2. (paf 247:): inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 248: return c>='A' && c<='Z'
249: || c>='a' && c<='Z'
250: || c>='0' && c<='9'
251: || strchr("!*+-/", c);
252: }
1.115.2.12.2. 3(paf 253:3): inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos& pos, size_t size) {
(paf 254:): while(size--) {
255:): CORD_ec_append(result, CORD_pos_fetch(pos));
256:): CORD_next(pos);
257:): }
258:): }
0(paf 259:3): StringBody String::cstr_to_string_body(Language lang,
(paf 260:): SQL_Connection* connection,
261:): const Request_charsets *charsets) const {
262:): CORD_ec result; CORD_ec_init(result);
0(paf 263:3): CORD_pos pos; body.set_pos(pos, 0);
(paf 264:):
1.44 paf 265: bool whitespace=true;
1.111 paf 266:
1.115.2.12.2. (paf 267:): size_t fragment_begin=0;
268:): size_t fragment_end;
269:): for(Array_iterator<Fragment> i(fragments); i.has_next(); fragment_begin=fragment_end) {
270:): const Fragment& fragment=i.next();
271:): size_t fragment_length=fragment.length;
272:): fragment_end=fragment_begin+fragment_length;
2(paf 273:3): //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
1.77 paf 274:
1.115.2.12.2. (paf 275:): Language to_lang=lang==L_UNSPECIFIED?fragment.lang:lang;
3(paf 276:3): bool optimize=(to_lang & L_OPTIMIZE_BIT)!=0;
277:3): if(!optimize)
278:3): whitespace=false;
1.77 paf 279:
1.115.2.12.2. (paf 280:): switch(to_lang & ~L_OPTIMIZE_BIT) {
281:): case L_CLEAN:
282:): case L_TAINTED:
283:): case L_AS_IS:
1.77 paf 284: // clean piece
285:
286: // tainted piece, but undefined untaint language
287: // for VString.as_double of tainted values
288: // for ^process{body} evaluation
289:
290: // tainted, untaint language: as-is
1.115.2.12.2. 3(paf 291:3): ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 292: break;
1.115.2.12.2. (paf 293:): case L_FILE_SPEC:
1.77 paf 294: // tainted, untaint language: file [name]
1.83 paf 295: escape(
1.113 paf 296: // Macintosh has problems with small Russian letter 'r'
1.115.2.12.2. (paf 297:): if( c=='\xF0' && charsets && charsets->source().name()=="windows-1251" ) {
1.113 paf 298: // fixing that letter for most common charset
1.115.2.2 paf 299: to_char('p');
1.114 paf 300: } else // fallback to default
301: encode(need_file_encode, '_');
1.83 paf 302: );
1.77 paf 303: break;
1.115.2.12.2. (paf 304:): #ifdef LATER
305:): case L_URI:
1.77 paf 306: // tainted, untaint language: uri
1.85 paf 307: {
1.115.2.12.2. (paf 308:): const char *substr=
309:): CORD_to_const_char_star(
310:): CORD_substr(body, fragment_begin, fragment_length));
311:): const void *client_ptr;
312:): size_t client_size;
313:): if(charsets)
314:): Charset::transcode(
315:): charsets->source(), substr, fragment_length,
316:): charsets->client(), client_ptr, client_size);
317:): else {
318:): client_ptr=substr;
319:): client_size=fragment_length;
320:): }
321:):
322:): char c;
323:): for(const char* src=(const char* )client_ptr; c=*src++; )
1.108 paf 324: encode(need_uri_encode, '%');
1.85 paf 325: }
1.77 paf 326: break;
1.115.2.12.2. (paf 327:): #endif
328:): case L_HTTP_HEADER:
1.77 paf 329: // tainted, untaint language: http-field-content-text
1.108 paf 330: escape(
331: encode(need_uri_encode, '%');
332: );
1.77 paf 333: break;
1.115.2.12.2. (paf 334:): #ifdef LATER
335:): case L_MAIL_HEADER:
1.77 paf 336: // tainted, untaint language: mail-header
1.105 paf 337: // http://www.ietf.org/rfc/rfc2047.txt
1.115.2.4 paf 338: if(charsets) {
1.115.2.12.2. (paf 339:): const void *mail_ptr=fragment.ptr;
340:): size_t mail_size=fragment_length;
341:): const char* charset_name=charsets->mail().name()->cstr();
1.87 paf 342:
1.77 paf 343: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.115.2.6 paf 344: const char* src=(const char* )mail_ptr;
1.77 paf 345: bool to_quoted_printable=false;
1.105 paf 346:
1.111 paf 347: bool email=false;
1.115.2.6 paf 348: for(const char* end=src+mail_size; src<end; src++) {
1.111 paf 349: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.115.2.12.2. (paf 350:): if(to_quoted_printable && (c==',' || c=='<')) {
351:): email=c=='<';
1.105 paf 352: dest+=sprintf(dest, "?=");
1.107 paf 353: to_quoted_printable=false;
1.105 paf 354: }
1.111 paf 355: if(!email && (
1.115.2.12.2. (paf 356:): !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
357:): || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
1.105 paf 358: )) {
1.77 paf 359: if(!to_quoted_printable) {
1.115.2.7 paf 360: dest+=sprintf(dest, "=?%s?Q?", charset_name.get());
1.77 paf 361: to_quoted_printable=true;
362: }
1.105 paf 363: //RFC Upper case should be used for hexadecimal digits "A" through "F"
1.115.2.12.2. (paf 364:): if(c == 0x20) // RFC The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
1.109 paf 365: *dest++='_'; // RFC may be represented as "_"
366: else
1.115.2.12.2. (paf 367:): dest+=sprintf(dest, "=%02X", c & 0xFF);
1.105 paf 368: } else
1.115.2.12.2. (paf 369:): *dest++=c;
370:): if(c=='>')
1.111 paf 371: email=false;
1.44 paf 372: }
1.112 paf 373: if(to_quoted_printable) // close
1.77 paf 374: dest+=sprintf(dest, "?=");
1.87 paf 375:
1.77 paf 376: } else {
1.115.2.12.2. (paf 377:): memcpy(dest, fragment.ptr, fragment_length);
378:): dest+=fragment_length;
1.1 paf 379: }
1.77 paf 380: break;
1.115.2.12.2. (paf 381:): #endif
382:): case L_TABLE:
1.77 paf 383: // tainted, untaint language: table
1.115.2.12.2. (paf 384:): escape(switch(c) {
1.77 paf 385: case '\t': to_char(' '); break;
386: case '\n': to_char(' '); break;
387: _default;
388: });
389: break;
1.115.2.12.2. (paf 390:): #ifdef PA_SQL
391:): case L_SQL:
1.77 paf 392: // tainted, untaint language: sql
1.115.2.12.2. (paf 393:): if(connection) {
394:): dest+=connection->quote(0,
395:): CORD_substr(body, fragment_begin, fragment_length),
396:): fragment_length);
397:): ...^^^
398:):
399:): dest+=connection->quote(dest, fragment.ptr, fragment_length);
1.77 paf 400: else
1.115.2.12.2. (paf 401:): throw Exception(0,
402:): 0,
1.77 paf 403: "untaint in SQL language failed - no connection specified");
404: break;
1.115.2.12.2. (paf 405:): #endif
406:): case L_JS:
407:): escape(switch(c) {
408:): case '"': to_string("\\\""); break;
409:): case '\'': to_string("\\'"); break;
410:): case '\n': to_string("\\n"); break;
411:): case '\\': to_string("\\\\"); break;
412:): case '\xFF': to_string("\\\xFF"); break;
1.77 paf 413: _default;
414: });
415: break;
1.115.2.12.2. (paf 416:): case L_XML:
417:): escape(switch(c) {
418:): case '&': to_string("&"); break;
419:): case '>': to_string(">"); break;
420:): case '<': to_string("<"); break;
421:): case '"': to_string("""); break;
422:): case '\'': to_string("'"); break;
1.77 paf 423: _default;
424: });
425: break;
1.115.2.12.2. (paf 426:): case L_HTML:
427:): escape(switch(c) {
428:): case '&': to_string("&"); break;
429:): case '>': to_string(">"); break;
430:): case '<': to_string("<"); break;
431:): case '"': to_string("""); break;
1.77 paf 432: _default;
433: });
434: break;
435: default:
1.115.2.12.2. (paf 436:): throw Exception(0,
437:): 0,
1.81 paf 438: "unknown untaint language #%d",
1.110 paf 439: static_cast<int>(to_lang)); // should never
1.77 paf 440: break; // never
1.76 paf 441: }
1.115.2.1 paf 442: }
443:
1.115.2.12.2. 0(paf 444:3): return StringBody(CORD_ec_to_cord(result));
1.1 paf 445: }
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