Annotation of parser3/src/main/untaint.C, revision 1.174
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.172 moko 4: Copyright (c) 2001-2017 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:
1.174 ! moko 8: volatile const char * IDENT_UNTAINT_C="$Id: untaint.C,v 1.173 2017/05/29 11:00:22 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
1.174 ! moko 23: #include "../lib/cord/include/ec.h"
1.117 paf 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:
1.171 moko 182:
183: /**
184: appends to other String without language change
185: */
186:
187: String& String::append_to(String& dest) const {
188: if(is_empty())
189: return dest;
190:
191: // first: fragment infos
192: dest.langs.appendHelper(dest.body, langs, body);
193:
194: // next: letters
195: dest.body<<body;
196:
197: ASSERT_STRING_INVARIANT(dest);
198: return dest;
199: }
200:
1.91 paf 201: /**
1.117 paf 202: appends to other String,
1.91 paf 203: marking all tainted pieces of it with @a lang.
1.92 paf 204: or marking ALL pieces of it with a @a lang when @a forced to,
205: and propagating OPTIMIZE language bit.
1.117 paf 206: */
1.147 misha 207: String& String::append_to(String& dest, Language ilang, bool forced) const {
1.117 paf 208: if(is_empty())
209: return dest;
1.91 paf 210:
1.119 paf 211: // first: fragment infos
1.117 paf 212:
1.171 moko 213: if(forced) //forcing passed lang?
1.147 misha 214: dest.langs.appendHelper(dest.body, ilang, body);
215: else {
216: if(langs.opt.is_not_just_lang){
217: Append_fragment_info info={ilang, &dest.langs, dest.body.length()};
218: langs.for_each(body, ilang&L_OPTIMIZE_BIT?
219: append_fragment_optimizing
220: :append_fragment_nonoptimizing, &info);
221: } else {
222: Language lang=langs.opt.lang;
223: // see append_fragment_* for explanation
224: if(ilang&L_OPTIMIZE_BIT){
225: dest.langs.appendHelper(dest.body,
226: lang==String::L_TAINTED?
227: ilang
228: :lang==String::L_CLEAN?
229: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT)
230: :lang,
231: body);
232: } else {
233: dest.langs.appendHelper(dest.body, lang==String::L_TAINTED ? ilang:lang, body);
234: }
235: }
1.91 paf 236: }
237:
1.119 paf 238: // next: letters
239: dest.body<<body;
240:
1.117 paf 241: ASSERT_STRING_INVARIANT(dest);
1.75 paf 242: return dest;
1.51 parser 243: }
244:
1.109 paf 245: /** http://www.ietf.org/rfc/rfc2047.txt
246: RFC
247: (3) As a replacement for a 'word' entity within a 'phrase', for example,
248: one that precedes an address in a From, To, or Cc header. The ABNF
249: definition for 'phrase' from RFC 822 thus becomes:
250:
251: phrase = 1*( encoded-word / word )
252:
253: In this case the set of characters that may be used in a "Q"-encoded
254: 'encoded-word' is restricted to: <upper and lower case ASCII
255: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
256: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
257: 'phrase' MUST be separated from any adjacent 'word', 'text' or
258: 'special' by 'linear-white-space'.
259: ...
260: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
261: represented as "_" (underscore, ASCII 95.). (This character may
262: not pass through some internetwork mail gateways, but its use
263: will greatly enhance readability of "Q" encoded data with mail
264: readers that do not support this encoding.) Note that the "_"
265: always represents hexadecimal 20, even if the SPACE character
266: occupies a different code position in the character set in use.
267:
268: paf: obviously,
269: without "=", or one could not differ "=E0" and "russian letter a"
270: and without "_", or in would mean 0x20
271: */
1.117 paf 272: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.162 misha 273: return (pa_isalnum((unsigned char)c) || strchr("!*+-/", c));
1.109 paf 274: }
1.118 paf 275: inline bool addr_spec_soon(const char *src) {
1.124 paf 276: for(char c; (c=*src); src++)
1.118 paf 277: if(c=='<')
278: return true;
279: else if(!(c==' ' || c=='\t'))
280: return false;
281: return false;
282: }
1.117 paf 283: /**
284: RFC
285: Upper case should be used for hexadecimal digits "A" through "F"
286: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
287: may be represented as "_"
288: */
289: inline bool mail_header_nonspace_char(char c) {
290: return c != 0x20;
291: }
1.125 paf 292:
1.117 paf 293: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
294: while(size--) {
295: CORD_ec_append(result, CORD_pos_fetch(pos));
296: CORD_next(pos);
297: }
298: }
299: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
300: while(true) {
1.125 paf 301: long avail=CORD_pos_chars_left(pos);
302: if(avail<=0) {
1.117 paf 303: CORD_next(pos);
304: if(!--n)
305: break;
1.125 paf 306: } else if((size_t)avail<n) {
1.117 paf 307: CORD_pos_advance(pos, avail);
308: n-=avail;
309: } else { // avail>=n
310: CORD_pos_advance(pos, n);
311: break;
312: }
313: }
314: }
315:
1.119 paf 316: #ifndef DOXYGEN
317: struct Cstr_to_string_body_block_info {
318: // input
319: String::Language lang;
320: SQL_Connection* connection;
321: const Request_charsets* charsets;
322: const String::Body* body;
323:
324: // output
325: CORD_ec result;
326:
327: // private
328: CORD_pos pos;
329: size_t fragment_begin;
330: bool whitespace;
1.138 misha 331: const char* exception;
1.119 paf 332: };
333: #endif
1.151 misha 334:
335: // @todo: replace info->body->mid with something that uses info->pos
1.148 misha 336: int cstr_to_string_body_block(String::Language to_lang, size_t fragment_length, Cstr_to_string_body_block_info* info) {
1.119 paf 337: bool& whitespace=info->whitespace;
338: size_t fragment_end=info->fragment_begin+fragment_length;
1.148 misha 339: //fprintf(stderr, "%d, %d =%s=\n", to_lang, fragment_length, info->body->cstr());
1.119 paf 340:
1.120 paf 341: bool optimize=(to_lang & String::L_OPTIMIZE_BIT)!=0;
1.119 paf 342: if(!optimize)
343: whitespace=false;
344:
1.120 paf 345: switch(to_lang & ~String::L_OPTIMIZE_BIT) {
346: case String::L_CLEAN:
347: case String::L_TAINTED:
348: case String::L_AS_IS:
1.119 paf 349: // clean piece
350:
351: // tainted piece, but undefined untaint language
352: // for VString.as_double of tainted values
353: // for ^process{body} evaluation
354:
355: // tainted, untaint language: as-is
356: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
357: break;
1.120 paf 358: case String::L_FILE_SPEC:
1.119 paf 359: // tainted, untaint language: file [name]
1.140 misha 360: {
361: escape_fragment(
1.154 misha 362: encode(need_file_encode, '_', c);
1.140 misha 363: );
364: }
1.119 paf 365: break;
1.120 paf 366: case String::L_URI:
1.169 moko 367: // tainted, untaint language: uri
368: {
369: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
370: // skip source [we use recoded version]
371: pa_CORD_pos_advance(info->pos, fragment_length);
372: String::C output(fragment_str, fragment_length);
373: if(info->charsets)
374: output=Charset::transcode(output, info->charsets->source(), info->charsets->client());
375:
376: char c;
377: for(const char* src=output.str; (c=*src++); )
378: encode(need_uri_encode, '%', c);
1.155 misha 379: }
380: break;
1.120 paf 381: case String::L_HTTP_HEADER:
1.119 paf 382: // tainted, untaint language: http-field-content-text
1.161 moko 383: escape_fragment(switch(c) {
384: case '\n':
385: case '\r': to_string(" "); break;
386: default: _default; break;
387: });
1.119 paf 388: break;
1.120 paf 389: case String::L_MAIL_HEADER:
1.119 paf 390: // tainted, untaint language: mail-header
391: // http://www.ietf.org/rfc/rfc2047.txt
392: if(info->charsets) {
393: size_t mail_size;
394: const char *mail_ptr=
395: info->body->mid(info->fragment_begin, mail_size=fragment_length).cstr();
396: // skip source [we use recoded version]
397: pa_CORD_pos_advance(info->pos, mail_size);
398:
399: const char* charset_name=info->charsets->mail().NAME().cstr();
400:
401: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
402: bool to_quoted_printable=false;
403:
404: bool email=false;
405: uchar c;
1.173 moko 406: for(const char* src=mail_ptr; c=(uchar)*src; src++) {
1.137 paf 407: if(c=='\r' || c=='\n')
408: c=' ';
1.173 moko 409: if(to_quoted_printable && (c==',' || c == '"' || addr_spec_soon(src))) {
1.119 paf 410: email=c=='<';
411: to_string("?=");
412: to_quoted_printable=false;
413: }
1.130 paf 414: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
1.119 paf 415: if(!email && (
1.173 moko 416: ( !to_quoted_printable && (c & 0x80 || (c == ' ' && src == mail_ptr) ) ) // starting quote-printable-encoding on first 8bit char or leading space (issue #123)
1.166 moko 417: || ( to_quoted_printable && !mail_header_char_valid_within_Qencoded(c) )
1.119 paf 418: )) {
419: if(!to_quoted_printable) {
420: to_string("=?");
421: to_string(charset_name);
422: to_string("?Q?");
423: to_quoted_printable=true;
1.105 paf 424: }
1.119 paf 425: encode(mail_header_nonspace_char, '=', '_');
426: } else
427: to_char(c);
428: if(c=='>')
429: email=false;
430: }
431: if(to_quoted_printable) // close
432: to_string("?=");
433:
434: } else
435: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
436: break;
1.120 paf 437: case String::L_SQL:
1.119 paf 438: // tainted, untaint language: sql
439: if(info->connection) {
440: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
441: // skip source [we use recoded version]
442: pa_CORD_pos_advance(info->pos, fragment_length);
443:
444: to_string(info->connection->quote(fragment_str, fragment_length));
1.138 misha 445: } else {
446: info->exception="untaint in SQL language failed - no connection specified";
447: info->fragment_begin=fragment_end;
448: return 1; // stop processing. can't throw exception here
449: }
1.119 paf 450: break;
1.120 paf 451: case String::L_JS:
1.140 misha 452: escape_fragment(switch(c) {
453: case '\n': to_string("\\n"); break;
1.119 paf 454: case '"': to_string("\\\""); break;
455: case '\'': to_string("\\'"); break;
456: case '\\': to_string("\\\\"); break;
457: case '\xFF': to_string("\\\xFF"); break;
1.146 misha 458: case '\r': to_string("\\r"); break;
1.133 paf 459: default: _default; break;
1.119 paf 460: });
461: break;
1.120 paf 462: case String::L_XML:
1.133 paf 463: // [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
1.140 misha 464: escape_fragment(switch(c) {
465: case '\x20':
466: case '\x9':
467: case '\xA':
1.133 paf 468: case '\xD': // this is usually removed on input
469: _default;
470: break;
1.119 paf 471: case '&': to_string("&"); break;
472: case '>': to_string(">"); break;
473: case '<': to_string("<"); break;
474: case '"': to_string("""); break;
475: case '\'': to_string("'"); break;
1.133 paf 476: default:
477: if(((unsigned char)c)<0x20) {
478: // fixing it, so that libxml would not result
479: // in fatal error parsing text
480: // though it really violates standard.
481: // to indicate there were an error
482: // replace bad char not to it's code,
483: // which we can do,
484: // but rather to '!' to show that input were actually
485: // invalid.
486: // life: shows that MSIE can somehow garble form values
487: // so that they contain these chars.
488: to_char('!');
489: } else {
490: _default;
491: }
492: break;
1.119 paf 493: });
494: break;
1.120 paf 495: case String::L_HTML:
1.140 misha 496: escape_fragment(switch(c) {
1.119 paf 497: case '&': to_string("&"); break;
498: case '>': to_string(">"); break;
499: case '<': to_string("<"); break;
500: case '"': to_string("""); break;
1.133 paf 501: default: _default; break;
1.119 paf 502: });
503: break;
1.136 paf 504: case String::L_REGEX:
505: // tainted, untaint language: regex
1.140 misha 506: escape_fragment(
1.136 paf 507: if(need_regex_escape(c))
508: to_char('\\')
509: _default;
510: );
511: break;
1.156 misha 512: case String::L_JSON:
513: // tainted, untaint language: json <http://json.org/>
514: // escape '"' '\' '/' '\n' '\t' '\r' '\b' '\f' chars and escape chars as \uXXXX if output charset != UTF-8
515: {
1.159 moko 516: if(info->charsets==NULL || info->charsets->client().isUTF8()){
1.156 misha 517: // escaping to \uXXXX is not needed
1.167 moko 518: escape_fragment(switch((unsigned char)c) {
1.156 misha 519: case '\n': to_string("\\n"); break;
520: case '"' : to_string("\\\""); break;
521: case '\\': to_string("\\\\"); break;
522: case '/' : to_string("\\/"); break;
523: case '\t': to_string("\\t"); break;
524: case '\r': to_string("\\r"); break;
525: case '\b': to_string("\\b"); break;
526: case '\f': to_string("\\f"); break;
1.167 moko 527: case 0xE2: // \u2028 and \u2029 (line/paragraph separators), check bug #1023
528: if(info->charsets && info->charsets->source().isUTF8() && fragment_length>=2){
529: CORD_next(info->pos);
530: char c1=CORD_pos_fetch(info->pos);
531: CORD_next(info->pos);
532: char c2=CORD_pos_fetch(info->pos);
533: if((unsigned char)c1 == 0x80 && ((unsigned char)c2 >= 0xA8 && (unsigned char)c2 <= 0xAF)){
534: to_string("\\u20");
535: to_hex(((unsigned char)c2-0x80));
536: } else {
537: CORD_ec_append(info->result, c);
538: CORD_ec_append(info->result, c1);
539: CORD_ec_append(info->result, c2);
540: }
541: fragment_length-=2;
542: } else {
543: _default;
544: }
545: break;
1.165 moko 546: default:
547: if((unsigned char)c < 0x20){
548: to_string("\\u00");
1.167 moko 549: to_hex(c);
1.165 moko 550: } else {
551: _default;
552: }
553: break;
1.156 misha 554: });
555: } else {
556: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
557: // skip source [we use recoded version]
558: pa_CORD_pos_advance(info->pos, fragment_length);
1.169 moko 559: to_string(Charset::escape_JSON(String::C(fragment_str, fragment_length), info->charsets->source()).str);
1.156 misha 560: }
561: }
562: break;
1.140 misha 563: case String::L_HTTP_COOKIE:
564: // tainted, untaint language: cookie (3.3.0 and higher: %uXXXX in UTF-8)
1.159 moko 565: if(info->charsets) {
1.140 misha 566: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
567: // skip source [we use recoded version]
568: pa_CORD_pos_advance(info->pos, fragment_length);
1.169 moko 569: to_string(Charset::escape(String::C(fragment_str, fragment_length), info->charsets->source()).str);
1.159 moko 570: } else
571: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
1.140 misha 572: break;
1.150 misha 573: case String::L_PARSER_CODE:
574: // for auto-untaint in process
575: escape_fragment(
576: if(need_parser_code_escape(c))
577: to_char('^');
578: _default;
579: );
580: break;
1.119 paf 581: default:
1.127 paf 582: SAPI::abort("unknown untaint language #%d",
1.119 paf 583: static_cast<int>(to_lang)); // should never
584: break; // never
1.117 paf 585: }
1.55 parser 586:
1.119 paf 587: info->fragment_begin=fragment_end;
588:
589: return 0; // 0=continue
590: }
591:
592:
1.148 misha 593: String::Body String::cstr_to_string_body_taint(Language lang, SQL_Connection* connection, const Request_charsets *charsets) const {
1.134 paf 594: if(is_empty())
595: return String::Body();
1.119 paf 596:
597: Cstr_to_string_body_block_info info;
598: // input
599: info.lang=lang;
600: info.connection=connection;
601: info.charsets=charsets;
602: info.body=&body;
603: // output
604: CORD_ec_init(info.result);
605: // private
606: body.set_pos(info.pos, 0);
607: info.fragment_begin=0;
1.138 misha 608: info.exception=0;
1.119 paf 609: info.whitespace=true;
610:
1.148 misha 611: cstr_to_string_body_block(lang, length(), &info);
612:
613: if(info.exception)
614: throw Exception(0,
615: 0,
616: info.exception);
617:
1.160 moko 618: return String::Body(CORD_ec_to_cord(info.result));
1.148 misha 619: }
620:
621: int cstr_to_string_body_block_untaint(char alang, size_t fragment_length, Cstr_to_string_body_block_info* info){
622: const String::Language lang=(String::Language)(unsigned char)alang;
623: // see append_fragment_* for explanation
624: if(info->lang&String::L_OPTIMIZE_BIT)
625: return cstr_to_string_body_block(
626: lang==String::L_TAINTED?
627: info->lang
628: :lang==String::L_CLEAN?
629: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT)
630: :lang,
631: fragment_length, info);
632: else
633: return cstr_to_string_body_block(lang==String::L_TAINTED ? info->lang:lang, fragment_length, info);
634: }
635:
636: String::Body String::cstr_to_string_body_untaint(Language lang, SQL_Connection* connection, const Request_charsets *charsets) const {
637: if(is_empty())
638: return String::Body();
639:
640: Cstr_to_string_body_block_info info;
641: // input
642: info.lang=lang;
643: info.connection=connection;
644: info.charsets=charsets;
645: info.body=&body;
646: // output
647: CORD_ec_init(info.result);
648: // private
649: body.set_pos(info.pos, 0);
650: info.fragment_begin=0;
651: info.exception=0;
652: info.whitespace=true;
653:
654: langs.for_each(body, cstr_to_string_body_block_untaint, &info);
1.151 misha 655:
656: if(info.exception)
657: throw Exception(0,
658: 0,
659: info.exception);
660:
1.160 moko 661: return String::Body(CORD_ec_to_cord(info.result));
1.151 misha 662: }
663:
1.155 misha 664: const char* String::untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const {
1.151 misha 665: if(charsets && &charsets->source() != &charsets->client()){
1.155 misha 666: // Note: L_URI is allready transcoded during untaint, but transcode does not affect %XX
667: return Charset::transcode(cstr_to_string_body_untaint(lang, 0, charsets), charsets->source(), charsets->client()).cstr();
668: } else
1.151 misha 669: return cstr_to_string_body_untaint(lang, 0, charsets).cstr();
1.1 paf 670: }
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