Annotation of parser3/src/main/untaint.C, revision 1.119
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:
1.119 ! paf 8: static const char* IDENT_UNTAINT_C="$Date: 2003/09/24 14:32:06 $";
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) \
1.119 ! paf 62: for(; fragment_length--; CORD_next(info->pos)) { \
! 63: char c=CORD_pos_fetch(info->pos); \
1.117 paf 64: action \
1.1 paf 65: }
1.119 ! paf 66: #define _default default: CORD_ec_append(info->result, c); break
1.117 paf 67: #define encode(need_encode_func, prefix, otherwise) \
68: if(need_encode_func(c)) { \
69: static const char* hex="0123456789ABCDEF"; \
1.119 ! paf 70: CORD_ec_append(info->result, prefix); \
! 71: CORD_ec_append(info->result, hex[((unsigned char)c)/0x10]); \
! 72: CORD_ec_append(info->result, hex[((unsigned char)c)%0x10]); \
1.117 paf 73: } else \
1.119 ! paf 74: CORD_ec_append(info->result, otherwise);
! 75: #define to_char(c) CORD_ec_append(info->result, c)
! 76: #define to_string(s) CORD_ec_append_cord(info->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:
1.119 ! paf 170: #ifndef DOXYGEN
! 171: struct Append_fragment_info {
! 172: String::Language lang;
! 173: const String::Body dest_body;
! 174: String::Languages* dest_languages;
! 175: size_t dest_body_plan_length;
! 176: };
! 177: #endif
! 178: int append_fragment_optimizing(char alang, size_t asize, Append_fragment_info* info) {
! 179: const String::Language lang=(String::Language)(unsigned char)alang;
! 180: // main idea here:
! 181: // tainted piece would get OPTIMIZED bit from 'lang'
! 182: // clean piece would be marked OPTIMIZED manually
! 183: // pieces with determined languages [not tainted|clean] would retain theirs langs
! 184: info->dest_languages->append(info->dest_body_plan_length,
! 185: lang==String::Language::L_TAINTED?
! 186: info->lang
! 187: :lang==String::Language::L_CLEAN?
! 188: (String::Language)(String::Language::L_CLEAN|String::Language::L_OPTIMIZE_BIT) // ORing with OPTIMIZED flag
! 189: :lang,
! 190: asize);
! 191: info->dest_body_plan_length+=asize;
! 192:
! 193: return 0; // 0=continue
! 194: }
! 195: int append_fragment_nonoptimizing(char alang, size_t asize, Append_fragment_info* info) {
! 196: const String::Language lang=(String::Language)(unsigned char)alang;
! 197: // The core idea: tainted pieces got marked with context's lang
! 198: info->dest_languages->append(info->dest_body_plan_length,
! 199: lang==String::Language::L_TAINTED?
! 200: info->lang
! 201: :lang,
! 202: asize);
! 203: info->dest_body_plan_length+=asize;
! 204:
! 205: return 0; // 0=continue
! 206: }
1.117 paf 207:
1.91 paf 208: /**
1.117 paf 209: appends to other String,
1.91 paf 210: marking all tainted pieces of it with @a lang.
1.92 paf 211: or marking ALL pieces of it with a @a lang when @a forced to,
212: and propagating OPTIMIZE language bit.
1.117 paf 213: */
214: String& String::append_to(String& dest, Language lang, bool forced) const {
215: if(is_empty())
216: return dest;
1.91 paf 217:
1.119 ! paf 218: // first: fragment infos
1.117 paf 219:
220: if(lang==L_PASS_APPENDED) // without language-change?
1.119 ! paf 221: dest.langs.append(dest.body, body.length(), langs);
1.117 paf 222: else if(forced) //forcing passed lang?
1.119 ! paf 223: dest.langs.append(dest.body, lang, length());
! 224: else {
! 225: Append_fragment_info info={lang, dest.body, &dest.langs, dest.body.length()};
! 226: langs.for_each(body, lang&L_OPTIMIZE_BIT?
! 227: append_fragment_optimizing
! 228: :append_fragment_nonoptimizing, &info);
1.91 paf 229: }
230:
1.119 ! paf 231: // next: letters
! 232: dest.body<<body;
! 233:
1.117 paf 234: ASSERT_STRING_INVARIANT(dest);
1.75 paf 235: return dest;
1.51 parser 236: }
237:
1.109 paf 238: /** http://www.ietf.org/rfc/rfc2047.txt
239: RFC
240: (3) As a replacement for a 'word' entity within a 'phrase', for example,
241: one that precedes an address in a From, To, or Cc header. The ABNF
242: definition for 'phrase' from RFC 822 thus becomes:
243:
244: phrase = 1*( encoded-word / word )
245:
246: In this case the set of characters that may be used in a "Q"-encoded
247: 'encoded-word' is restricted to: <upper and lower case ASCII
248: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
249: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
250: 'phrase' MUST be separated from any adjacent 'word', 'text' or
251: 'special' by 'linear-white-space'.
252: ...
253: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
254: represented as "_" (underscore, ASCII 95.). (This character may
255: not pass through some internetwork mail gateways, but its use
256: will greatly enhance readability of "Q" encoded data with mail
257: readers that do not support this encoding.) Note that the "_"
258: always represents hexadecimal 20, even if the SPACE character
259: occupies a different code position in the character set in use.
260:
261: paf: obviously,
262: without "=", or one could not differ "=E0" and "russian letter a"
263: and without "_", or in would mean 0x20
264: */
1.117 paf 265: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 266: return c>='A' && c<='Z'
267: || c>='a' && c<='Z'
268: || c>='0' && c<='9'
269: || strchr("!*+-/", c);
270: }
1.118 paf 271: inline bool addr_spec_soon(const char *src) {
272: for(char c; c=*src; src++)
273: if(c=='<')
274: return true;
275: else if(!(c==' ' || c=='\t'))
276: return false;
277: return false;
278: }
1.117 paf 279: /**
280: RFC
281: Upper case should be used for hexadecimal digits "A" through "F"
282: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
283: may be represented as "_"
284: */
285: inline bool mail_header_nonspace_char(char c) {
286: return c != 0x20;
287: }
288: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
289: while(size--) {
290: CORD_ec_append(result, CORD_pos_fetch(pos));
291: CORD_next(pos);
292: }
293: }
294: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
295: while(true) {
296: size_t avail=CORD_pos_chars_left(pos);
297: if(avail==0) {
298: CORD_next(pos);
299: if(!--n)
300: break;
301: } else if(avail<n) {
302: CORD_pos_advance(pos, avail);
303: n-=avail;
304: } else { // avail>=n
305: CORD_pos_advance(pos, n);
306: break;
307: }
308: }
309: }
310:
1.119 ! paf 311: #ifndef DOXYGEN
! 312: struct Cstr_to_string_body_block_info {
! 313: // input
! 314: String::Language lang;
! 315: SQL_Connection* connection;
! 316: const Request_charsets* charsets;
! 317: const String::Body* body;
! 318:
! 319: // output
! 320: CORD_ec result;
! 321:
! 322: // private
! 323: CORD_pos pos;
! 324: size_t fragment_begin;
! 325: bool whitespace;
! 326: };
! 327: #endif
! 328: int cstr_to_string_body_block(char alang, size_t fragment_length, Cstr_to_string_body_block_info* info) {
! 329: const String::Language fragment_lang=(String::Language)(unsigned char)alang;
! 330: bool& whitespace=info->whitespace;
! 331: size_t fragment_end=info->fragment_begin+fragment_length;
! 332: //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
1.117 paf 333:
1.119 ! paf 334:
! 335: String::Language to_lang=info->lang==String::Language::L_UNSPECIFIED?fragment_lang:info->lang;
! 336: bool optimize=(to_lang & String::Language::L_OPTIMIZE_BIT)!=0;
! 337: if(!optimize)
! 338: whitespace=false;
! 339:
! 340: switch(to_lang & ~String::Language::L_OPTIMIZE_BIT) {
! 341: case String::Language::L_CLEAN:
! 342: case String::Language::L_TAINTED:
! 343: case String::Language::L_AS_IS:
! 344: // clean piece
! 345:
! 346: // tainted piece, but undefined untaint language
! 347: // for VString.as_double of tainted values
! 348: // for ^process{body} evaluation
! 349:
! 350: // tainted, untaint language: as-is
! 351: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
! 352: break;
! 353: case String::Language::L_FILE_SPEC:
! 354: // tainted, untaint language: file [name]
! 355: escape(
! 356: // Macintosh has problems with small Russian letter 'r'
! 357: if( c=='\xF0' && info->charsets && info->charsets->source().NAME()=="WINDOWS-1251" ) {
! 358: // fixing that letter for most common charset
! 359: to_char('p');
! 360: } else // fallback to default
! 361: encode(need_file_encode, '_', c);
! 362: );
! 363: break;
! 364: case String::Language::L_URI:
! 365: // tainted, untaint language: uri
! 366: {
! 367: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
! 368: // skip source [we use recoded version]
! 369: pa_CORD_pos_advance(info->pos, fragment_length);
! 370: String::C output(fragment_str, fragment_length);
! 371: if(info->charsets)
! 372: output=Charset::transcode(output,
! 373: info->charsets->source(),
! 374: info->charsets->client());
1.77 paf 375:
1.119 ! paf 376: char c;
! 377: for(const char* src=output.str; c=*src++; )
1.117 paf 378: encode(need_uri_encode, '%', c);
1.119 ! paf 379: }
! 380: break;
! 381: case String::Language::L_HTTP_HEADER:
! 382: // tainted, untaint language: http-field-content-text
! 383: escape(
! 384: encode(need_uri_encode, '%', c);
! 385: );
! 386: break;
! 387: case String::Language::L_MAIL_HEADER:
! 388: // tainted, untaint language: mail-header
! 389: // http://www.ietf.org/rfc/rfc2047.txt
! 390: if(info->charsets) {
! 391: size_t mail_size;
! 392: const char *mail_ptr=
! 393: info->body->mid(info->fragment_begin, mail_size=fragment_length).cstr();
! 394: // skip source [we use recoded version]
! 395: pa_CORD_pos_advance(info->pos, mail_size);
! 396:
! 397: const char* charset_name=info->charsets->mail().NAME().cstr();
! 398:
! 399: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
! 400: bool to_quoted_printable=false;
! 401:
! 402: bool email=false;
! 403: uchar c;
! 404: for(const char* src=mail_ptr; c=(uchar)*src++; ) {
! 405: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
! 406: if(to_quoted_printable && (c==',' || addr_spec_soon(src) || c == '"')) {
! 407: email=c=='<';
! 408: to_string("?=");
! 409: to_quoted_printable=false;
! 410: }
! 411: if(!email && (
! 412: !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
! 413: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
! 414: )) {
! 415: if(!to_quoted_printable) {
! 416: to_string("=?");
! 417: to_string(charset_name);
! 418: to_string("?Q?");
! 419: to_quoted_printable=true;
1.105 paf 420: }
1.119 ! paf 421: encode(mail_header_nonspace_char, '=', '_');
! 422: } else
! 423: to_char(c);
! 424: if(c=='>')
! 425: email=false;
! 426: }
! 427: if(to_quoted_printable) // close
! 428: to_string("?=");
! 429:
! 430: } else
! 431: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
! 432: break;
! 433: case String::Language::L_TABLE:
! 434: // tainted, untaint language: table
! 435: escape(switch(c) {
! 436: case '\t': to_char(' '); break;
! 437: case '\n': to_char(' '); break;
! 438: _default;
! 439: });
! 440: break;
1.117 paf 441: #ifdef PA_SQL
1.119 ! paf 442: case String::Language::L_SQL:
! 443: // tainted, untaint language: sql
! 444: if(info->connection) {
! 445: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
! 446: // skip source [we use recoded version]
! 447: pa_CORD_pos_advance(info->pos, fragment_length);
! 448:
! 449: to_string(info->connection->quote(fragment_str, fragment_length));
! 450: } else
! 451: throw Exception(0,
! 452: 0,
! 453: "untaint in SQL language failed - no connection specified");
! 454: break;
1.117 paf 455: #endif
1.119 ! paf 456: case String::Language::L_JS:
! 457: escape(switch(c) {
! 458: case '"': to_string("\\\""); break;
! 459: case '\'': to_string("\\'"); break;
! 460: case '\n': to_string("\\n"); break;
! 461: case '\\': to_string("\\\\"); break;
! 462: case '\xFF': to_string("\\\xFF"); break;
! 463: _default;
! 464: });
! 465: break;
! 466: case String::Language::L_XML:
! 467: escape(switch(c) {
! 468: case '&': to_string("&"); break;
! 469: case '>': to_string(">"); break;
! 470: case '<': to_string("<"); break;
! 471: case '"': to_string("""); break;
! 472: case '\'': to_string("'"); break;
! 473: _default;
! 474: });
! 475: break;
! 476: case String::Language::L_HTML:
! 477: escape(switch(c) {
! 478: case '&': to_string("&"); break;
! 479: case '>': to_string(">"); break;
! 480: case '<': to_string("<"); break;
! 481: case '"': to_string("""); break;
! 482: _default;
! 483: });
! 484: break;
! 485: default:
! 486: assert(!"should never");
! 487: SAPI::die("unknown untaint language #%d",
! 488: static_cast<int>(to_lang)); // should never
! 489: break; // never
1.117 paf 490: }
1.55 parser 491:
1.119 ! paf 492: info->fragment_begin=fragment_end;
! 493:
! 494: return 0; // 0=continue
! 495: }
! 496:
! 497:
! 498: String::Body String::cstr_to_string_body(Language lang,
! 499: SQL_Connection* connection,
! 500: const Request_charsets *charsets) const {
! 501:
! 502: Cstr_to_string_body_block_info info;
! 503: // input
! 504: info.lang=lang;
! 505: info.connection=connection;
! 506: info.charsets=charsets;
! 507: info.body=&body;
! 508: // output
! 509: CORD_ec_init(info.result);
! 510: // private
! 511: body.set_pos(info.pos, 0);
! 512: info.fragment_begin=0;
! 513: info.whitespace=true;
! 514:
! 515: if(!is_empty())
! 516: langs.for_each(body, cstr_to_string_body_block, &info);
! 517:
! 518: return String::Body(CORD_ec_to_cord(info.result));
1.1 paf 519: }
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