Annotation of parser3/src/main/untaint.C, revision 1.169
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.163 moko 4: Copyright (c) 2001-2012 Art. Lebedev Studio (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:
1.169 ! moko 8: volatile const char * IDENT_UNTAINT_C="$Id: untaint.C,v 1.168 2015/10/08 18:29:16 moko Exp $";
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: #include "pa_sql_connection.h"
1.91 paf 27:
1.117 paf 28: // defines
29:
30: #undef CORD_ec_append
31: // redefining to intercept flushes and implement whitespace optimization
32: // of all consequent white space chars leaving only first one
33: #define CORD_ec_append(x, c) \
34: { \
35: bool skip=false; \
36: if(optimize) switch(c) { \
1.131 paf 37: case ' ': case '\n': case '\t': \
1.117 paf 38: if(whitespace) \
39: skip=true; /*skipping subsequent*/ \
40: else \
41: whitespace=true; \
42: break; \
43: default: \
44: whitespace=false; \
45: break; \
46: } \
47: if(!skip) { \
48: if ((x)[0].ec_bufptr == (x)[0].ec_buf + CORD_BUFSZ) { \
49: CORD_ec_flush_buf(x); \
50: } \
51: *((x)[0].ec_bufptr)++ = (c); \
52: } \
53: }
54:
1.140 misha 55: #define escape_fragment(action) \
1.119 paf 56: for(; fragment_length--; CORD_next(info->pos)) { \
57: char c=CORD_pos_fetch(info->pos); \
1.117 paf 58: action \
1.1 paf 59: }
1.167 moko 60:
1.117 paf 61: #define encode(need_encode_func, prefix, otherwise) \
62: if(need_encode_func(c)) { \
1.119 paf 63: CORD_ec_append(info->result, prefix); \
1.167 moko 64: to_hex(c); \
65: } else \
66: CORD_ec_append(info->result, otherwise);
67:
68: #define to_hex(c) \
69: { \
1.165 moko 70: CORD_ec_append(info->result, hex_digits[((unsigned char)c) >> 4]); \
71: CORD_ec_append(info->result, hex_digits[((unsigned char)c) & 0x0F]); \
1.167 moko 72: }
73:
1.126 paf 74: #define to_char(c) { CORD_ec_append(info->result, c); whitespace=false; }
1.168 moko 75: #define to_string(s) { CORD_ec_append_cord(info->result, (CORD)(s)); whitespace=false; }
1.167 moko 76: #define _default CORD_ec_append(info->result, c)
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(
1.164 misha 82: "*?\"<>|"
1.108 paf 83: #ifndef WIN32
1.143 misha 84: ":\\"
1.31 paf 85: #endif
1.143 misha 86: , c)!=0;
1.9 paf 87: }
1.143 misha 88:
1.5 paf 89: inline bool need_uri_encode(unsigned char c){
1.162 misha 90: return !(pa_isalnum(c) || strchr("_-./*", c));
1.4 paf 91: }
1.143 misha 92:
1.136 paf 93: inline bool need_regex_escape(unsigned char c){
1.142 misha 94: return strchr("\\^$.[]|()?*+{}-", c)!=0;
1.136 paf 95: }
1.1 paf 96:
1.150 misha 97: inline bool need_parser_code_escape(unsigned char c){
98: return strchr("^$;@()[]{}:#\"", c)!=0;
99: }
100:
1.1 paf 101: // String
102:
1.41 paf 103: /*
104: HTTP-header = field-name ":" [ field-value ] CRLF
105:
1.152 misha 106: field-name = token
107: field-value = *( field-content | LWS )
1.41 paf 108:
1.152 misha 109: field-content = <the OCTETs making up the field-value
1.41 paf 110: and consisting of either *TEXT or combinations
111: of token, tspecials, and quoted-string>
112:
113:
1.152 misha 114: token = 1*<any CHAR except CTLs or tspecials>
1.41 paf 115: word = token | quoted-string
1.152 misha 116: quoted-string = ( <"> *(qdtext | quoted-pair ) <"> )
117: qdtext = <any TEXT except <">>
118: quoted-pair = "\" CHAR
1.41 paf 119:
1.152 misha 120: OCTET = <any 8-bit sequence of data>
121: CHAR = <any US-ASCII character (octets 0 - 127)>
1.41 paf 122:
123: tspecials = "(" | ")" | "<" | ">" | "@"
1.152 misha 124: | "," | ";" | ":" | "\" | <">
125: | "/" | "[" | "]" | "?" | "="
126: | "{" | "}" | SP | HT
1.41 paf 127:
128: SP = <US-ASCII SP, space (32)>
129: HT = <US-ASCII HT, horizontal-tab (9)>
130:
131: LWS = [CRLF] 1*( SP | HT )
1.152 misha 132: TEXT = <any OCTET except CTLs, but including LWS>
133: CTL = <any US-ASCII control character (octets 0 - 31) and DEL (127)>
1.41 paf 134:
135:
136: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
137: */
1.117 paf 138: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 139: for(; size--; ptr++)
1.42 paf 140: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 141: return true;
142: return false;
143: }
144:
1.119 paf 145: #ifndef DOXYGEN
146: struct Append_fragment_info {
147: String::Language lang;
148: String::Languages* dest_languages;
149: size_t dest_body_plan_length;
150: };
151: #endif
152: int append_fragment_optimizing(char alang, size_t asize, Append_fragment_info* info) {
153: const String::Language lang=(String::Language)(unsigned char)alang;
154: // main idea here:
155: // tainted piece would get OPTIMIZED bit from 'lang'
156: // clean piece would be marked OPTIMIZED manually
157: // pieces with determined languages [not tainted|clean] would retain theirs langs
158: info->dest_languages->append(info->dest_body_plan_length,
1.120 paf 159: lang==String::L_TAINTED?
1.119 paf 160: info->lang
1.120 paf 161: :lang==String::L_CLEAN?
162: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT) // ORing with OPTIMIZED flag
1.119 paf 163: :lang,
164: asize);
165: info->dest_body_plan_length+=asize;
166:
167: return 0; // 0=continue
168: }
169: int append_fragment_nonoptimizing(char alang, size_t asize, Append_fragment_info* info) {
170: const String::Language lang=(String::Language)(unsigned char)alang;
171: // The core idea: tainted pieces got marked with context's lang
172: info->dest_languages->append(info->dest_body_plan_length,
1.120 paf 173: lang==String::L_TAINTED?
1.119 paf 174: info->lang
175: :lang,
176: asize);
177: info->dest_body_plan_length+=asize;
178:
179: return 0; // 0=continue
180: }
1.117 paf 181:
1.91 paf 182: /**
1.117 paf 183: appends to other String,
1.91 paf 184: marking all tainted pieces of it with @a lang.
1.92 paf 185: or marking ALL pieces of it with a @a lang when @a forced to,
186: and propagating OPTIMIZE language bit.
1.117 paf 187: */
1.147 misha 188: String& String::append_to(String& dest, Language ilang, bool forced) const {
1.117 paf 189: if(is_empty())
190: return dest;
1.91 paf 191:
1.119 paf 192: // first: fragment infos
1.117 paf 193:
1.147 misha 194: if(ilang==L_PASS_APPENDED) // without language-change?
195: dest.langs.appendHelper(dest.body, langs, body);
1.117 paf 196: else if(forced) //forcing passed lang?
1.147 misha 197: dest.langs.appendHelper(dest.body, ilang, body);
198: else {
199: if(langs.opt.is_not_just_lang){
200: Append_fragment_info info={ilang, &dest.langs, dest.body.length()};
201: langs.for_each(body, ilang&L_OPTIMIZE_BIT?
202: append_fragment_optimizing
203: :append_fragment_nonoptimizing, &info);
204: } else {
205: Language lang=langs.opt.lang;
206: // see append_fragment_* for explanation
207: if(ilang&L_OPTIMIZE_BIT){
208: dest.langs.appendHelper(dest.body,
209: lang==String::L_TAINTED?
210: ilang
211: :lang==String::L_CLEAN?
212: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT)
213: :lang,
214: body);
215: } else {
216: dest.langs.appendHelper(dest.body, lang==String::L_TAINTED ? ilang:lang, body);
217: }
218: }
1.91 paf 219: }
220:
1.119 paf 221: // next: letters
222: dest.body<<body;
223:
1.117 paf 224: ASSERT_STRING_INVARIANT(dest);
1.75 paf 225: return dest;
1.51 parser 226: }
227:
1.109 paf 228: /** http://www.ietf.org/rfc/rfc2047.txt
229: RFC
230: (3) As a replacement for a 'word' entity within a 'phrase', for example,
231: one that precedes an address in a From, To, or Cc header. The ABNF
232: definition for 'phrase' from RFC 822 thus becomes:
233:
234: phrase = 1*( encoded-word / word )
235:
236: In this case the set of characters that may be used in a "Q"-encoded
237: 'encoded-word' is restricted to: <upper and lower case ASCII
238: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
239: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
240: 'phrase' MUST be separated from any adjacent 'word', 'text' or
241: 'special' by 'linear-white-space'.
242: ...
243: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
244: represented as "_" (underscore, ASCII 95.). (This character may
245: not pass through some internetwork mail gateways, but its use
246: will greatly enhance readability of "Q" encoded data with mail
247: readers that do not support this encoding.) Note that the "_"
248: always represents hexadecimal 20, even if the SPACE character
249: occupies a different code position in the character set in use.
250:
251: paf: obviously,
252: without "=", or one could not differ "=E0" and "russian letter a"
253: and without "_", or in would mean 0x20
254: */
1.117 paf 255: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.162 misha 256: return (pa_isalnum((unsigned char)c) || strchr("!*+-/", c));
1.109 paf 257: }
1.118 paf 258: inline bool addr_spec_soon(const char *src) {
1.124 paf 259: for(char c; (c=*src); src++)
1.118 paf 260: if(c=='<')
261: return true;
262: else if(!(c==' ' || c=='\t'))
263: return false;
264: return false;
265: }
1.117 paf 266: /**
267: RFC
268: Upper case should be used for hexadecimal digits "A" through "F"
269: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
270: may be represented as "_"
271: */
272: inline bool mail_header_nonspace_char(char c) {
273: return c != 0x20;
274: }
1.125 paf 275:
1.117 paf 276: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
277: while(size--) {
278: CORD_ec_append(result, CORD_pos_fetch(pos));
279: CORD_next(pos);
280: }
281: }
282: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
283: while(true) {
1.125 paf 284: long avail=CORD_pos_chars_left(pos);
285: if(avail<=0) {
1.117 paf 286: CORD_next(pos);
287: if(!--n)
288: break;
1.125 paf 289: } else if((size_t)avail<n) {
1.117 paf 290: CORD_pos_advance(pos, avail);
291: n-=avail;
292: } else { // avail>=n
293: CORD_pos_advance(pos, n);
294: break;
295: }
296: }
297: }
298:
1.119 paf 299: #ifndef DOXYGEN
300: struct Cstr_to_string_body_block_info {
301: // input
302: String::Language lang;
303: SQL_Connection* connection;
304: const Request_charsets* charsets;
305: const String::Body* body;
306:
307: // output
308: CORD_ec result;
309:
310: // private
311: CORD_pos pos;
312: size_t fragment_begin;
313: bool whitespace;
1.138 misha 314: const char* exception;
1.119 paf 315: };
316: #endif
1.151 misha 317:
318: // @todo: replace info->body->mid with something that uses info->pos
1.148 misha 319: int cstr_to_string_body_block(String::Language to_lang, size_t fragment_length, Cstr_to_string_body_block_info* info) {
1.119 paf 320: bool& whitespace=info->whitespace;
321: size_t fragment_end=info->fragment_begin+fragment_length;
1.148 misha 322: //fprintf(stderr, "%d, %d =%s=\n", to_lang, fragment_length, info->body->cstr());
1.119 paf 323:
1.120 paf 324: bool optimize=(to_lang & String::L_OPTIMIZE_BIT)!=0;
1.119 paf 325: if(!optimize)
326: whitespace=false;
327:
1.120 paf 328: switch(to_lang & ~String::L_OPTIMIZE_BIT) {
329: case String::L_CLEAN:
330: case String::L_TAINTED:
331: case String::L_AS_IS:
1.119 paf 332: // clean piece
333:
334: // tainted piece, but undefined untaint language
335: // for VString.as_double of tainted values
336: // for ^process{body} evaluation
337:
338: // tainted, untaint language: as-is
339: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
340: break;
1.120 paf 341: case String::L_FILE_SPEC:
1.119 paf 342: // tainted, untaint language: file [name]
1.140 misha 343: {
344: escape_fragment(
1.154 misha 345: encode(need_file_encode, '_', c);
1.140 misha 346: );
347: }
1.119 paf 348: break;
1.120 paf 349: case String::L_URI:
1.169 ! moko 350: // tainted, untaint language: uri
! 351: {
! 352: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
! 353: // skip source [we use recoded version]
! 354: pa_CORD_pos_advance(info->pos, fragment_length);
! 355: String::C output(fragment_str, fragment_length);
! 356: if(info->charsets)
! 357: output=Charset::transcode(output, info->charsets->source(), info->charsets->client());
! 358:
! 359: char c;
! 360: for(const char* src=output.str; (c=*src++); )
! 361: encode(need_uri_encode, '%', c);
1.155 misha 362: }
363: break;
1.120 paf 364: case String::L_HTTP_HEADER:
1.119 paf 365: // tainted, untaint language: http-field-content-text
1.161 moko 366: escape_fragment(switch(c) {
367: case '\n':
368: case '\r': to_string(" "); break;
369: default: _default; break;
370: });
1.119 paf 371: break;
1.120 paf 372: case String::L_MAIL_HEADER:
1.119 paf 373: // tainted, untaint language: mail-header
374: // http://www.ietf.org/rfc/rfc2047.txt
375: if(info->charsets) {
376: size_t mail_size;
377: const char *mail_ptr=
378: info->body->mid(info->fragment_begin, mail_size=fragment_length).cstr();
379: // skip source [we use recoded version]
380: pa_CORD_pos_advance(info->pos, mail_size);
381:
382: const char* charset_name=info->charsets->mail().NAME().cstr();
383:
384: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
385: bool to_quoted_printable=false;
386:
387: bool email=false;
388: uchar c;
1.124 paf 389: for(const char* src=mail_ptr; (c=(uchar)*src++); ) {
1.137 paf 390: if(c=='\r' || c=='\n')
391: c=' ';
1.130 paf 392: if(to_quoted_printable && (c==',' || c == '"' || addr_spec_soon(src-1/*position to 'c'*/))) {
1.119 paf 393: email=c=='<';
394: to_string("?=");
395: to_quoted_printable=false;
396: }
1.130 paf 397: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.119 paf 398: if(!email && (
1.166 moko 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) )
1.119 paf 401: )) {
402: if(!to_quoted_printable) {
403: to_string("=?");
404: to_string(charset_name);
405: to_string("?Q?");
406: to_quoted_printable=true;
1.105 paf 407: }
1.119 paf 408: encode(mail_header_nonspace_char, '=', '_');
409: } else
410: to_char(c);
411: if(c=='>')
412: email=false;
413: }
414: if(to_quoted_printable) // close
415: to_string("?=");
416:
417: } else
418: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
419: break;
1.120 paf 420: case String::L_SQL:
1.119 paf 421: // tainted, untaint language: sql
422: if(info->connection) {
423: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
424: // skip source [we use recoded version]
425: pa_CORD_pos_advance(info->pos, fragment_length);
426:
427: to_string(info->connection->quote(fragment_str, fragment_length));
1.138 misha 428: } else {
429: info->exception="untaint in SQL language failed - no connection specified";
430: info->fragment_begin=fragment_end;
431: return 1; // stop processing. can't throw exception here
432: }
1.119 paf 433: break;
1.120 paf 434: case String::L_JS:
1.140 misha 435: escape_fragment(switch(c) {
436: case '\n': to_string("\\n"); break;
1.119 paf 437: case '"': to_string("\\\""); break;
438: case '\'': to_string("\\'"); break;
439: case '\\': to_string("\\\\"); break;
440: case '\xFF': to_string("\\\xFF"); break;
1.146 misha 441: case '\r': to_string("\\r"); break;
1.133 paf 442: default: _default; break;
1.119 paf 443: });
444: break;
1.120 paf 445: case String::L_XML:
1.133 paf 446: // [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
1.140 misha 447: escape_fragment(switch(c) {
448: case '\x20':
449: case '\x9':
450: case '\xA':
1.133 paf 451: case '\xD': // this is usually removed on input
452: _default;
453: break;
1.119 paf 454: case '&': to_string("&"); break;
455: case '>': to_string(">"); break;
456: case '<': to_string("<"); break;
457: case '"': to_string("""); break;
458: case '\'': to_string("'"); break;
1.133 paf 459: default:
460: if(((unsigned char)c)<0x20) {
461: // fixing it, so that libxml would not result
462: // in fatal error parsing text
463: // though it really violates standard.
464: // to indicate there were an error
465: // replace bad char not to it's code,
466: // which we can do,
467: // but rather to '!' to show that input were actually
468: // invalid.
469: // life: shows that MSIE can somehow garble form values
470: // so that they contain these chars.
471: to_char('!');
472: } else {
473: _default;
474: }
475: break;
1.119 paf 476: });
477: break;
1.120 paf 478: case String::L_HTML:
1.140 misha 479: escape_fragment(switch(c) {
1.119 paf 480: case '&': to_string("&"); break;
481: case '>': to_string(">"); break;
482: case '<': to_string("<"); break;
483: case '"': to_string("""); break;
1.133 paf 484: default: _default; break;
1.119 paf 485: });
486: break;
1.136 paf 487: case String::L_REGEX:
488: // tainted, untaint language: regex
1.140 misha 489: escape_fragment(
1.136 paf 490: if(need_regex_escape(c))
491: to_char('\\')
492: _default;
493: );
494: break;
1.156 misha 495: case String::L_JSON:
496: // tainted, untaint language: json <http://json.org/>
497: // escape '"' '\' '/' '\n' '\t' '\r' '\b' '\f' chars and escape chars as \uXXXX if output charset != UTF-8
498: {
1.159 moko 499: if(info->charsets==NULL || info->charsets->client().isUTF8()){
1.156 misha 500: // escaping to \uXXXX is not needed
1.167 moko 501: escape_fragment(switch((unsigned char)c) {
1.156 misha 502: case '\n': to_string("\\n"); break;
503: case '"' : to_string("\\\""); break;
504: case '\\': to_string("\\\\"); break;
505: case '/' : to_string("\\/"); break;
506: case '\t': to_string("\\t"); break;
507: case '\r': to_string("\\r"); break;
508: case '\b': to_string("\\b"); break;
509: case '\f': to_string("\\f"); break;
1.167 moko 510: case 0xE2: // \u2028 and \u2029 (line/paragraph separators), check bug #1023
511: if(info->charsets && info->charsets->source().isUTF8() && fragment_length>=2){
512: CORD_next(info->pos);
513: char c1=CORD_pos_fetch(info->pos);
514: CORD_next(info->pos);
515: char c2=CORD_pos_fetch(info->pos);
516: if((unsigned char)c1 == 0x80 && ((unsigned char)c2 >= 0xA8 && (unsigned char)c2 <= 0xAF)){
517: to_string("\\u20");
518: to_hex(((unsigned char)c2-0x80));
519: } else {
520: CORD_ec_append(info->result, c);
521: CORD_ec_append(info->result, c1);
522: CORD_ec_append(info->result, c2);
523: }
524: fragment_length-=2;
525: } else {
526: _default;
527: }
528: break;
1.165 moko 529: default:
530: if((unsigned char)c < 0x20){
531: to_string("\\u00");
1.167 moko 532: to_hex(c);
1.165 moko 533: } else {
534: _default;
535: }
536: break;
1.156 misha 537: });
538: } else {
539: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
540: // skip source [we use recoded version]
541: pa_CORD_pos_advance(info->pos, fragment_length);
1.169 ! moko 542: to_string(Charset::escape_JSON(String::C(fragment_str, fragment_length), info->charsets->source()).str);
1.156 misha 543: }
544: }
545: break;
1.140 misha 546: case String::L_HTTP_COOKIE:
547: // tainted, untaint language: cookie (3.3.0 and higher: %uXXXX in UTF-8)
1.159 moko 548: if(info->charsets) {
1.140 misha 549: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
550: // skip source [we use recoded version]
551: pa_CORD_pos_advance(info->pos, fragment_length);
1.169 ! moko 552: to_string(Charset::escape(String::C(fragment_str, fragment_length), info->charsets->source()).str);
1.159 moko 553: } else
554: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
1.140 misha 555: break;
1.150 misha 556: case String::L_PARSER_CODE:
557: // for auto-untaint in process
558: escape_fragment(
559: if(need_parser_code_escape(c))
560: to_char('^');
561: _default;
562: );
563: break;
1.119 paf 564: default:
1.127 paf 565: SAPI::abort("unknown untaint language #%d",
1.119 paf 566: static_cast<int>(to_lang)); // should never
567: break; // never
1.117 paf 568: }
1.55 parser 569:
1.119 paf 570: info->fragment_begin=fragment_end;
571:
572: return 0; // 0=continue
573: }
574:
575:
1.148 misha 576: String::Body String::cstr_to_string_body_taint(Language lang, SQL_Connection* connection, const Request_charsets *charsets) const {
1.134 paf 577: if(is_empty())
578: return String::Body();
1.119 paf 579:
580: Cstr_to_string_body_block_info info;
581: // input
582: info.lang=lang;
583: info.connection=connection;
584: info.charsets=charsets;
585: info.body=&body;
586: // output
587: CORD_ec_init(info.result);
588: // private
589: body.set_pos(info.pos, 0);
590: info.fragment_begin=0;
1.138 misha 591: info.exception=0;
1.119 paf 592: info.whitespace=true;
593:
1.148 misha 594: cstr_to_string_body_block(lang, length(), &info);
595:
596: if(info.exception)
597: throw Exception(0,
598: 0,
599: info.exception);
600:
1.160 moko 601: return String::Body(CORD_ec_to_cord(info.result));
1.148 misha 602: }
603:
604: int cstr_to_string_body_block_untaint(char alang, size_t fragment_length, Cstr_to_string_body_block_info* info){
605: const String::Language lang=(String::Language)(unsigned char)alang;
606: // see append_fragment_* for explanation
607: if(info->lang&String::L_OPTIMIZE_BIT)
608: return cstr_to_string_body_block(
609: lang==String::L_TAINTED?
610: info->lang
611: :lang==String::L_CLEAN?
612: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT)
613: :lang,
614: fragment_length, info);
615: else
616: return cstr_to_string_body_block(lang==String::L_TAINTED ? info->lang:lang, fragment_length, info);
617: }
618:
619: String::Body String::cstr_to_string_body_untaint(Language lang, SQL_Connection* connection, const Request_charsets *charsets) const {
620: if(is_empty())
621: return String::Body();
622:
623: Cstr_to_string_body_block_info info;
624: // input
625: info.lang=lang;
626: info.connection=connection;
627: info.charsets=charsets;
628: info.body=&body;
629: // output
630: CORD_ec_init(info.result);
631: // private
632: body.set_pos(info.pos, 0);
633: info.fragment_begin=0;
634: info.exception=0;
635: info.whitespace=true;
636:
637: langs.for_each(body, cstr_to_string_body_block_untaint, &info);
1.151 misha 638:
639: if(info.exception)
640: throw Exception(0,
641: 0,
642: info.exception);
643:
1.160 moko 644: return String::Body(CORD_ec_to_cord(info.result));
1.151 misha 645: }
646:
1.155 misha 647: const char* String::untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const {
1.151 misha 648: if(charsets && &charsets->source() != &charsets->client()){
1.155 misha 649: // Note: L_URI is allready transcoded during untaint, but transcode does not affect %XX
650: return Charset::transcode(cstr_to_string_body_untaint(lang, 0, charsets), charsets->source(), charsets->client()).cstr();
651: } else
1.151 misha 652: return cstr_to_string_body_untaint(lang, 0, charsets).cstr();
1.1 paf 653: }
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