Annotation of parser3/src/main/untaint.C, revision 1.118.2.1
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.117 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:
1.118.2.1! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/08/18 08:27:41 $";
1.117 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"
1.117 paf 19: #include "pa_request_charsets.h"
20: #include "pa_sapi.h"
1.1 paf 21:
1.117 paf 22: extern "C" { // author forgot to do that
23: #include "ec.h"
24: }
1.91 paf 25:
1.117 paf 26: #define PA_SQL
27:
28: #ifdef PA_SQL
29: #include "pa_sql_connection.h"
1.91 paf 30: #endif
31:
1.117 paf 32: // defines
33:
34:
35: #undef CORD_ec_append
36: // redefining to intercept flushes and implement whitespace optimization
37: // of all consequent white space chars leaving only first one
38: #define CORD_ec_append(x, c) \
39: { \
40: bool skip=false; \
41: if(optimize) switch(c) { \
42: case ' ': case '\r': case '\n': case '\t': \
43: if(whitespace) \
44: skip=true; /*skipping subsequent*/ \
45: else \
46: whitespace=true; \
47: break; \
48: default: \
49: whitespace=false; \
50: break; \
51: } \
52: if(!skip) { \
53: if ((x)[0].ec_bufptr == (x)[0].ec_buf + CORD_BUFSZ) { \
54: CORD_ec_flush_buf(x); \
55: } \
56: *((x)[0].ec_bufptr)++ = (c); \
57: } \
58: }
59:
60:
1.18 paf 61: #define escape(action) \
1.117 paf 62: for(; fragment_length--; CORD_next(pos)) { \
63: char c=CORD_pos_fetch(pos); \
64: action \
1.1 paf 65: }
1.117 paf 66: #define _default default: CORD_ec_append(result, c); break
67: #define encode(need_encode_func, prefix, otherwise) \
68: if(need_encode_func(c)) { \
69: static const char* hex="0123456789ABCDEF"; \
70: CORD_ec_append(result, prefix); \
71: CORD_ec_append(result, hex[((unsigned char)c)/0x10]); \
72: CORD_ec_append(result, hex[((unsigned char)c)%0x10]); \
73: } else \
74: CORD_ec_append(result, otherwise);
75: #define to_char(c) CORD_ec_append(result, c)
76: #define to_string(s) CORD_ec_append_cord(result, s)
1.4 paf 77:
1.9 paf 78: inline bool need_file_encode(unsigned char c){
1.108 paf 79: // russian letters and space ENABLED
80: // encoding only these...
81: return strchr(
82: "*?'\"<>|"
83: #ifndef WIN32
1.116 paf 84: ":\\"
1.31 paf 85: #endif
1.108 paf 86: , c)!=0;
1.9 paf 87: }
1.5 paf 88: inline bool need_uri_encode(unsigned char c){
1.13 paf 89: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 90: return false;
91:
1.5 paf 92: return !strchr("_-./", c);
93: }
1.36 paf 94: inline bool need_http_header_encode(unsigned char c){
1.18 paf 95: if(strchr(" , :", c))
1.5 paf 96: return false;
97:
98: return need_uri_encode(c);
1.4 paf 99: }
1.1 paf 100:
1.56 parser 101: //
102:
1.117 paf 103: static const char* String_Untaint_lang_name[]={
1.56 parser 104: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
105: "C", ///< clean
106: "T", ///< tainted, untaint language as assigned later
107: // untaint languages. assigned by ^untaint[lang]{...}
108: "P",
109: /**<
110: leave language built into string being appended.
111: just a flag, that value not stored
112: */
113: "A", ///< leave all characters intact
1.68 parser 114: "F", ///< file specification
115: "H", ///< ext in HTTP response header
1.56 parser 116: "M", ///< text in mail header
117: "URI", ///< text in uri
118: "T", ///< ^table:set body
119: "SQL", ///< ^table:sql body
120: "JS", ///< JavaScript code
1.68 parser 121: "XML", ///< ^dom:set xml
1.82 paf 122: "HTML" ///< HTML code (for editing)
1.56 parser 123: };
124:
125:
1.1 paf 126: // String
127:
1.41 paf 128: /*
129:
130: HTTP-header = field-name ":" [ field-value ] CRLF
131:
132: field-name = token
133: field-value = *( field-content | LWS )
134:
135: field-content = <the OCTETs making up the field-value
136: and consisting of either *TEXT or combinations
137: of token, tspecials, and quoted-string>
138:
139:
140:
141: word = token | quoted-string
142:
143: token = 1*<any CHAR except CTLs or tspecials>
144:
145:
146:
147: tspecials = "(" | ")" | "<" | ">" | "@"
148: | "," | ";" | ":" | "\" | <">
149: | "/" | "[" | "]" | "?" | "="
150: | "{" | "}" | SP | HT
151:
152: SP = <US-ASCII SP, space (32)>
153: HT = <US-ASCII HT, horizontal-tab (9)>
154:
155: LWS = [CRLF] 1*( SP | HT )
156: TEXT = <any OCTET except CTLs,
157: but including LWS>
158:
159: quoted-pair = "\" CHAR
160:
161: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
162: */
1.117 paf 163: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 164: for(; size--; ptr++)
1.42 paf 165: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 166: return true;
167: return false;
168: }
169:
1.118.2.1! paf 170: #ifndef DOXYGEN
! 171: struct Append_fragment_info {
! 172: String::Language lang;
! 173: const StringBody dest_body;
! 174: String::Languages* dest_languages;
! 175: };
! 176: #endif
! 177: int append_fragment_optimizing(char alang, size_t asize, Append_fragment_info& info) {
! 178: String::Language lang=(String::Language)alang;
! 179: // main idea here:
! 180: // tainted piece would get OPTIMIZED bit from 'lang'
! 181: // clean piece would be marked OPTIMIZED manually
! 182: // pieces with determined languages [not tainted|clean] would retain theirs langs
! 183: info.dest_languages->append(info.dest_body,
! 184: lang==String::Language::L_TAINTED?
! 185: info.lang
! 186: :lang==String::Language::L_CLEAN?
! 187: (String::Language)(String::Language::L_CLEAN|String::Language::L_OPTIMIZE_BIT) // ORing with OPTIMIZED flag
! 188: :lang,
! 189: asize);
! 190:
! 191: return 0; // 0=continue
! 192: }
! 193: int append_fragment_nonoptimizing(char alang, size_t asize, Append_fragment_info& info) {
! 194: String::Language lang=(String::Language)alang;
! 195: // The core idea: tainted pieces got marked with context's lang
! 196: info.dest_languages->append(info.dest_body,
! 197: lang==String::Language::L_TAINTED?
! 198: info.lang
! 199: :lang,
! 200: asize);
! 201:
! 202: return 0; // 0=continue
! 203: }
1.117 paf 204:
1.91 paf 205: /**
1.117 paf 206: appends to other String,
1.91 paf 207: marking all tainted pieces of it with @a lang.
1.92 paf 208: or marking ALL pieces of it with a @a lang when @a forced to,
209: and propagating OPTIMIZE language bit.
1.117 paf 210: */
211: String& String::append_to(String& dest, Language lang, bool forced) const {
212: if(is_empty())
213: return dest;
1.91 paf 214:
1.118.2.1! paf 215: // first: fragment infos
1.117 paf 216:
217: if(lang==L_PASS_APPENDED) // without language-change?
1.118.2.1! paf 218: dest.langs.append(dest.body, body, langs);
1.117 paf 219: else if(forced) //forcing passed lang?
1.118.2.1! paf 220: dest.langs.append(dest.body, lang, length());
! 221: else {
! 222: Append_fragment_info info={lang, dest.body, &dest.langs};
! 223: langs.for_each(body, lang&L_OPTIMIZE_BIT?
! 224: append_fragment_optimizing
! 225: :append_fragment_nonoptimizing, info);
1.91 paf 226: }
227:
1.118.2.1! paf 228: // next: letters
! 229: dest.body<<body;
! 230:
1.75 paf 231: return dest;
1.51 parser 232: }
233:
1.109 paf 234: /** http://www.ietf.org/rfc/rfc2047.txt
235: RFC
236: (3) As a replacement for a 'word' entity within a 'phrase', for example,
237: one that precedes an address in a From, To, or Cc header. The ABNF
238: definition for 'phrase' from RFC 822 thus becomes:
239:
240: phrase = 1*( encoded-word / word )
241:
242: In this case the set of characters that may be used in a "Q"-encoded
243: 'encoded-word' is restricted to: <upper and lower case ASCII
244: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
245: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
246: 'phrase' MUST be separated from any adjacent 'word', 'text' or
247: 'special' by 'linear-white-space'.
248: ...
249: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
250: represented as "_" (underscore, ASCII 95.). (This character may
251: not pass through some internetwork mail gateways, but its use
252: will greatly enhance readability of "Q" encoded data with mail
253: readers that do not support this encoding.) Note that the "_"
254: always represents hexadecimal 20, even if the SPACE character
255: occupies a different code position in the character set in use.
256:
257: paf: obviously,
258: without "=", or one could not differ "=E0" and "russian letter a"
259: and without "_", or in would mean 0x20
260: */
1.117 paf 261: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 262: return c>='A' && c<='Z'
263: || c>='a' && c<='Z'
264: || c>='0' && c<='9'
265: || strchr("!*+-/", c);
266: }
1.118 paf 267: inline bool addr_spec_soon(const char *src) {
268: for(char c; c=*src; src++)
269: if(c=='<')
270: return true;
271: else if(!(c==' ' || c=='\t'))
272: return false;
273: return false;
274: }
1.117 paf 275: /**
276: RFC
277: Upper case should be used for hexadecimal digits "A" through "F"
278: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
279: may be represented as "_"
280: */
281: inline bool mail_header_nonspace_char(char c) {
282: return c != 0x20;
283: }
284: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
285: while(size--) {
286: CORD_ec_append(result, CORD_pos_fetch(pos));
287: CORD_next(pos);
288: }
289: }
290: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
291: while(true) {
292: size_t avail=CORD_pos_chars_left(pos);
293: if(avail==0) {
294: CORD_next(pos);
295: if(!--n)
296: break;
297: } else if(avail<n) {
298: CORD_pos_advance(pos, avail);
299: n-=avail;
300: } else { // avail>=n
301: CORD_pos_advance(pos, n);
302: break;
303: }
304: }
305: }
306: StringBody String::cstr_to_string_body(Language lang,
307: SQL_Connection* connection,
308: const Request_charsets *charsets) const {
309: CORD_ec result; CORD_ec_init(result);
310: CORD_pos pos; body.set_pos(pos, 0);
311:
1.44 paf 312: bool whitespace=true;
1.117 paf 313:
314: size_t fragment_begin=0;
315: size_t fragment_end;
316: for(Array_iterator<Fragment> i(fragments); i.has_next(); fragment_begin=fragment_end) {
317: const Fragment fragment=i.next();
318: size_t fragment_length=fragment.length;
319: fragment_end=fragment_begin+fragment_length;
320: //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
321:
322: Language to_lang=lang==L_UNSPECIFIED?fragment.lang:lang;
323: bool optimize=(to_lang & L_OPTIMIZE_BIT)!=0;
324: if(!optimize)
325: whitespace=false;
326:
327: switch(to_lang & ~L_OPTIMIZE_BIT) {
328: case L_CLEAN:
329: case L_TAINTED:
330: case L_AS_IS:
1.77 paf 331: // clean piece
332:
333: // tainted piece, but undefined untaint language
334: // for VString.as_double of tainted values
335: // for ^process{body} evaluation
336:
337: // tainted, untaint language: as-is
1.117 paf 338: ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 339: break;
1.117 paf 340: case L_FILE_SPEC:
1.77 paf 341: // tainted, untaint language: file [name]
1.83 paf 342: escape(
1.113 paf 343: // Macintosh has problems with small Russian letter 'r'
1.117 paf 344: if( c=='\xF0' && charsets && charsets->source().NAME()=="WINDOWS-1251" ) {
1.113 paf 345: // fixing that letter for most common charset
1.117 paf 346: to_char('p');
1.114 paf 347: } else // fallback to default
1.117 paf 348: encode(need_file_encode, '_', c);
1.83 paf 349: );
1.77 paf 350: break;
1.117 paf 351: case L_URI:
1.77 paf 352: // tainted, untaint language: uri
1.85 paf 353: {
1.117 paf 354: const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
355: // skip source [we use recoded version]
356: pa_CORD_pos_advance(pos, fragment_length);
357: String::C output(fragment_str, fragment_length);
358: if(charsets)
359: output=Charset::transcode(output,
360: charsets->source(),
361: charsets->client());
362:
363: char c;
364: for(const char* src=output.str; c=*src++; )
365: encode(need_uri_encode, '%', c);
1.85 paf 366: }
1.77 paf 367: break;
1.117 paf 368: case L_HTTP_HEADER:
1.77 paf 369: // tainted, untaint language: http-field-content-text
1.108 paf 370: escape(
1.117 paf 371: encode(need_uri_encode, '%', c);
1.108 paf 372: );
1.77 paf 373: break;
1.117 paf 374: case L_MAIL_HEADER:
1.77 paf 375: // tainted, untaint language: mail-header
1.105 paf 376: // http://www.ietf.org/rfc/rfc2047.txt
1.117 paf 377: if(charsets) {
1.87 paf 378: size_t mail_size;
1.117 paf 379: const char *mail_ptr=
380: body.mid(fragment_begin, mail_size=fragment_length).cstr();
381: // skip source [we use recoded version]
382: pa_CORD_pos_advance(pos, mail_size);
383:
384: const char* charset_name=charsets->mail().NAME().cstr();
1.87 paf 385:
1.77 paf 386: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
387: bool to_quoted_printable=false;
1.105 paf 388:
1.111 paf 389: bool email=false;
1.117 paf 390: uchar c;
391: for(const char* src=mail_ptr; c=(uchar)*src++; ) {
1.111 paf 392: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.118 paf 393: if(to_quoted_printable && (c==',' || addr_spec_soon(src) || c == '"')) {
1.117 paf 394: email=c=='<';
395: to_string("?=");
1.107 paf 396: to_quoted_printable=false;
1.105 paf 397: }
1.111 paf 398: if(!email && (
1.117 paf 399: !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
400: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
1.105 paf 401: )) {
1.77 paf 402: if(!to_quoted_printable) {
1.117 paf 403: to_string("=?");
404: to_string(charset_name);
405: to_string("?Q?");
1.77 paf 406: to_quoted_printable=true;
407: }
1.117 paf 408: encode(mail_header_nonspace_char, '=', '_');
1.105 paf 409: } else
1.117 paf 410: to_char(c);
411: if(c=='>')
1.111 paf 412: email=false;
1.44 paf 413: }
1.112 paf 414: if(to_quoted_printable) // close
1.117 paf 415: to_string("?=");
1.87 paf 416:
1.117 paf 417: } else
418: ec_append(result, optimize, whitespace, pos, fragment_length);
1.77 paf 419: break;
1.117 paf 420: case L_TABLE:
1.77 paf 421: // tainted, untaint language: table
1.117 paf 422: escape(switch(c) {
1.77 paf 423: case '\t': to_char(' '); break;
424: case '\n': to_char(' '); break;
425: _default;
426: });
427: break;
1.117 paf 428: #ifdef PA_SQL
429: case L_SQL:
1.77 paf 430: // tainted, untaint language: sql
1.117 paf 431: if(connection) {
432: const char *fragment_str=body.mid(fragment_begin, fragment_length).cstr();
433: // skip source [we use recoded version]
434: pa_CORD_pos_advance(pos, fragment_length);
435:
436: to_string(connection->quote(fragment_str, fragment_length));
437: } else
1.99 paf 438: throw Exception(0,
1.117 paf 439: 0,
1.77 paf 440: "untaint in SQL language failed - no connection specified");
441: break;
1.117 paf 442: #endif
443: case L_JS:
444: escape(switch(c) {
445: case '"': to_string("\\\""); break;
446: case '\'': to_string("\\'"); break;
447: case '\n': to_string("\\n"); break;
448: case '\\': to_string("\\\\"); break;
449: case '\xFF': to_string("\\\xFF"); break;
1.77 paf 450: _default;
451: });
452: break;
1.117 paf 453: case L_XML:
454: escape(switch(c) {
455: case '&': to_string("&"); break;
456: case '>': to_string(">"); break;
457: case '<': to_string("<"); break;
458: case '"': to_string("""); break;
459: case '\'': to_string("'"); break;
1.77 paf 460: _default;
461: });
462: break;
1.117 paf 463: case L_HTML:
464: escape(switch(c) {
465: case '&': to_string("&"); break;
466: case '>': to_string(">"); break;
467: case '<': to_string("<"); break;
468: case '"': to_string("""); break;
1.77 paf 469: _default;
470: });
471: break;
472: default:
1.117 paf 473: SAPI::die("unknown untaint language #%d",
474: static_cast<int>(to_lang)); // should never
1.77 paf 475: break; // never
1.76 paf 476: }
1.117 paf 477: }
1.55 parser 478:
1.117 paf 479: return StringBody(CORD_ec_to_cord(result));
1.1 paf 480: }
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