Annotation of parser3/src/main/untaint.C, revision 1.125
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:
1.125 ! paf 8: static const char * const IDENT_UNTAINT_C="$Date: 2003/11/20 17:15:12 $";
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:
101: // String
102:
1.41 paf 103: /*
104: HTTP-header = field-name ":" [ field-value ] CRLF
105:
106: field-name = token
107: field-value = *( field-content | LWS )
108:
109: field-content = <the OCTETs making up the field-value
110: and consisting of either *TEXT or combinations
111: of token, tspecials, and quoted-string>
112:
113:
114:
115: word = token | quoted-string
116:
117: token = 1*<any CHAR except CTLs or tspecials>
118:
119:
120:
121: tspecials = "(" | ")" | "<" | ">" | "@"
122: | "," | ";" | ":" | "\" | <">
123: | "/" | "[" | "]" | "?" | "="
124: | "{" | "}" | SP | HT
125:
126: SP = <US-ASCII SP, space (32)>
127: HT = <US-ASCII HT, horizontal-tab (9)>
128:
129: LWS = [CRLF] 1*( SP | HT )
130: TEXT = <any OCTET except CTLs,
131: but including LWS>
132:
133: quoted-pair = "\" CHAR
134:
135: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
136: */
1.117 paf 137: inline bool need_quote_http_header(const char* ptr, size_t size) {
1.41 paf 138: for(; size--; ptr++)
1.42 paf 139: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 140: return true;
141: return false;
142: }
143:
1.119 paf 144: #ifndef DOXYGEN
145: struct Append_fragment_info {
146: String::Language lang;
147: String::Languages* dest_languages;
148: size_t dest_body_plan_length;
149: };
150: #endif
151: int append_fragment_optimizing(char alang, size_t asize, Append_fragment_info* info) {
152: const String::Language lang=(String::Language)(unsigned char)alang;
153: // main idea here:
154: // tainted piece would get OPTIMIZED bit from 'lang'
155: // clean piece would be marked OPTIMIZED manually
156: // pieces with determined languages [not tainted|clean] would retain theirs langs
157: info->dest_languages->append(info->dest_body_plan_length,
1.120 paf 158: lang==String::L_TAINTED?
1.119 paf 159: info->lang
1.120 paf 160: :lang==String::L_CLEAN?
161: (String::Language)(String::L_CLEAN|String::L_OPTIMIZE_BIT) // ORing with OPTIMIZED flag
1.119 paf 162: :lang,
163: asize);
164: info->dest_body_plan_length+=asize;
165:
166: return 0; // 0=continue
167: }
168: int append_fragment_nonoptimizing(char alang, size_t asize, Append_fragment_info* info) {
169: const String::Language lang=(String::Language)(unsigned char)alang;
170: // The core idea: tainted pieces got marked with context's lang
171: info->dest_languages->append(info->dest_body_plan_length,
1.120 paf 172: lang==String::L_TAINTED?
1.119 paf 173: info->lang
174: :lang,
175: asize);
176: info->dest_body_plan_length+=asize;
177:
178: return 0; // 0=continue
179: }
1.117 paf 180:
1.91 paf 181: /**
1.117 paf 182: appends to other String,
1.91 paf 183: marking all tainted pieces of it with @a lang.
1.92 paf 184: or marking ALL pieces of it with a @a lang when @a forced to,
185: and propagating OPTIMIZE language bit.
1.117 paf 186: */
187: String& String::append_to(String& dest, Language lang, bool forced) const {
188: if(is_empty())
189: return dest;
1.91 paf 190:
1.119 paf 191: // first: fragment infos
1.117 paf 192:
193: if(lang==L_PASS_APPENDED) // without language-change?
1.119 paf 194: dest.langs.append(dest.body, body.length(), langs);
1.117 paf 195: else if(forced) //forcing passed lang?
1.119 paf 196: dest.langs.append(dest.body, lang, length());
197: else {
1.122 paf 198: Append_fragment_info info={lang, &dest.langs, dest.body.length()};
1.119 paf 199: langs.for_each(body, lang&L_OPTIMIZE_BIT?
200: append_fragment_optimizing
201: :append_fragment_nonoptimizing, &info);
1.91 paf 202: }
203:
1.119 paf 204: // next: letters
205: dest.body<<body;
206:
1.117 paf 207: ASSERT_STRING_INVARIANT(dest);
1.75 paf 208: return dest;
1.51 parser 209: }
210:
1.109 paf 211: /** http://www.ietf.org/rfc/rfc2047.txt
212: RFC
213: (3) As a replacement for a 'word' entity within a 'phrase', for example,
214: one that precedes an address in a From, To, or Cc header. The ABNF
215: definition for 'phrase' from RFC 822 thus becomes:
216:
217: phrase = 1*( encoded-word / word )
218:
219: In this case the set of characters that may be used in a "Q"-encoded
220: 'encoded-word' is restricted to: <upper and lower case ASCII
221: letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
222: (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
223: 'phrase' MUST be separated from any adjacent 'word', 'text' or
224: 'special' by 'linear-white-space'.
225: ...
226: (2) The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE) may be
227: represented as "_" (underscore, ASCII 95.). (This character may
228: not pass through some internetwork mail gateways, but its use
229: will greatly enhance readability of "Q" encoded data with mail
230: readers that do not support this encoding.) Note that the "_"
231: always represents hexadecimal 20, even if the SPACE character
232: occupies a different code position in the character set in use.
233:
234: paf: obviously,
235: without "=", or one could not differ "=E0" and "russian letter a"
236: and without "_", or in would mean 0x20
237: */
1.117 paf 238: inline bool mail_header_char_valid_within_Qencoded(char c) {
1.109 paf 239: return c>='A' && c<='Z'
240: || c>='a' && c<='Z'
241: || c>='0' && c<='9'
242: || strchr("!*+-/", c);
243: }
1.118 paf 244: inline bool addr_spec_soon(const char *src) {
1.124 paf 245: for(char c; (c=*src); src++)
1.118 paf 246: if(c=='<')
247: return true;
248: else if(!(c==' ' || c=='\t'))
249: return false;
250: return false;
251: }
1.117 paf 252: /**
253: RFC
254: Upper case should be used for hexadecimal digits "A" through "F"
255: The 8-bit hexadecimal value 20 (e.g., ISO-8859-1 SPACE)
256: may be represented as "_"
257: */
258: inline bool mail_header_nonspace_char(char c) {
259: return c != 0x20;
260: }
1.125 ! paf 261:
1.117 paf 262: inline void ec_append(CORD_ec& result, bool& optimize, bool& whitespace, CORD_pos pos, size_t size) {
263: while(size--) {
264: CORD_ec_append(result, CORD_pos_fetch(pos));
265: CORD_next(pos);
266: }
267: }
268: inline void pa_CORD_pos_advance(CORD_pos pos, size_t n) {
269: while(true) {
1.125 ! paf 270: long avail=CORD_pos_chars_left(pos);
! 271: if(avail<=0) {
1.117 paf 272: CORD_next(pos);
273: if(!--n)
274: break;
1.125 ! paf 275: } else if((size_t)avail<n) {
1.117 paf 276: CORD_pos_advance(pos, avail);
277: n-=avail;
278: } else { // avail>=n
279: CORD_pos_advance(pos, n);
280: break;
281: }
282: }
283: }
284:
1.119 paf 285: #ifndef DOXYGEN
286: struct Cstr_to_string_body_block_info {
287: // input
288: String::Language lang;
289: SQL_Connection* connection;
290: const Request_charsets* charsets;
291: const String::Body* body;
292:
293: // output
294: CORD_ec result;
295:
296: // private
297: CORD_pos pos;
298: size_t fragment_begin;
299: bool whitespace;
300: };
301: #endif
302: int cstr_to_string_body_block(char alang, size_t fragment_length, Cstr_to_string_body_block_info* info) {
303: const String::Language fragment_lang=(String::Language)(unsigned char)alang;
304: bool& whitespace=info->whitespace;
305: size_t fragment_end=info->fragment_begin+fragment_length;
306: //fprintf(stderr, "%d, %d\n", fragment.lang, fragment.length);
1.117 paf 307:
1.119 paf 308:
1.120 paf 309: String::Language to_lang=info->lang==String::L_UNSPECIFIED?fragment_lang:info->lang;
310: bool optimize=(to_lang & String::L_OPTIMIZE_BIT)!=0;
1.119 paf 311: if(!optimize)
312: whitespace=false;
313:
1.120 paf 314: switch(to_lang & ~String::L_OPTIMIZE_BIT) {
315: case String::L_CLEAN:
316: case String::L_TAINTED:
317: case String::L_AS_IS:
1.119 paf 318: // clean piece
319:
320: // tainted piece, but undefined untaint language
321: // for VString.as_double of tainted values
322: // for ^process{body} evaluation
323:
324: // tainted, untaint language: as-is
325: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
326: break;
1.120 paf 327: case String::L_FILE_SPEC:
1.119 paf 328: // tainted, untaint language: file [name]
329: escape(
330: // Macintosh has problems with small Russian letter 'r'
331: if( c=='\xF0' && info->charsets && info->charsets->source().NAME()=="WINDOWS-1251" ) {
332: // fixing that letter for most common charset
333: to_char('p');
334: } else // fallback to default
335: encode(need_file_encode, '_', c);
336: );
337: break;
1.120 paf 338: case String::L_URI:
1.119 paf 339: // tainted, untaint language: uri
340: {
341: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
342: // skip source [we use recoded version]
343: pa_CORD_pos_advance(info->pos, fragment_length);
344: String::C output(fragment_str, fragment_length);
345: if(info->charsets)
346: output=Charset::transcode(output,
347: info->charsets->source(),
348: info->charsets->client());
1.77 paf 349:
1.119 paf 350: char c;
1.124 paf 351: for(const char* src=output.str; (c=*src++); )
1.117 paf 352: encode(need_uri_encode, '%', c);
1.119 paf 353: }
354: break;
1.120 paf 355: case String::L_HTTP_HEADER:
1.119 paf 356: // tainted, untaint language: http-field-content-text
357: escape(
358: encode(need_uri_encode, '%', c);
359: );
360: break;
1.120 paf 361: case String::L_MAIL_HEADER:
1.119 paf 362: // tainted, untaint language: mail-header
363: // http://www.ietf.org/rfc/rfc2047.txt
364: if(info->charsets) {
365: size_t mail_size;
366: const char *mail_ptr=
367: info->body->mid(info->fragment_begin, mail_size=fragment_length).cstr();
368: // skip source [we use recoded version]
369: pa_CORD_pos_advance(info->pos, mail_size);
370:
371: const char* charset_name=info->charsets->mail().NAME().cstr();
372:
373: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
374: bool to_quoted_printable=false;
375:
376: bool email=false;
377: uchar c;
1.124 paf 378: for(const char* src=mail_ptr; (c=(uchar)*src++); ) {
1.119 paf 379: //RFC + An 'encoded-word' MUST NOT appear in any portion of an 'addr-spec'.
380: if(to_quoted_printable && (c==',' || addr_spec_soon(src) || c == '"')) {
381: email=c=='<';
382: to_string("?=");
383: to_quoted_printable=false;
384: }
385: if(!email && (
386: !to_quoted_printable && (c & 0x80) // starting quote-printable-encoding on first 8bit char
387: || to_quoted_printable && !mail_header_char_valid_within_Qencoded(c)
388: )) {
389: if(!to_quoted_printable) {
390: to_string("=?");
391: to_string(charset_name);
392: to_string("?Q?");
393: to_quoted_printable=true;
1.105 paf 394: }
1.119 paf 395: encode(mail_header_nonspace_char, '=', '_');
396: } else
397: to_char(c);
398: if(c=='>')
399: email=false;
400: }
401: if(to_quoted_printable) // close
402: to_string("?=");
403:
404: } else
405: ec_append(info->result, optimize, whitespace, info->pos, fragment_length);
406: break;
1.120 paf 407: case String::L_TABLE:
1.119 paf 408: // tainted, untaint language: table
409: escape(switch(c) {
410: case '\t': to_char(' '); break;
411: case '\n': to_char(' '); break;
412: _default;
413: });
414: break;
1.117 paf 415: #ifdef PA_SQL
1.120 paf 416: case String::L_SQL:
1.119 paf 417: // tainted, untaint language: sql
418: if(info->connection) {
419: const char *fragment_str=info->body->mid(info->fragment_begin, fragment_length).cstr();
420: // skip source [we use recoded version]
421: pa_CORD_pos_advance(info->pos, fragment_length);
422:
423: to_string(info->connection->quote(fragment_str, fragment_length));
424: } else
425: throw Exception(0,
426: 0,
427: "untaint in SQL language failed - no connection specified");
428: break;
1.117 paf 429: #endif
1.120 paf 430: case String::L_JS:
1.119 paf 431: escape(switch(c) {
432: case '"': to_string("\\\""); break;
433: case '\'': to_string("\\'"); break;
434: case '\n': to_string("\\n"); break;
435: case '\\': to_string("\\\\"); break;
436: case '\xFF': to_string("\\\xFF"); break;
437: _default;
438: });
439: break;
1.120 paf 440: case String::L_XML:
1.119 paf 441: escape(switch(c) {
442: case '&': to_string("&"); break;
443: case '>': to_string(">"); break;
444: case '<': to_string("<"); break;
445: case '"': to_string("""); break;
446: case '\'': to_string("'"); break;
447: _default;
448: });
449: break;
1.120 paf 450: case String::L_HTML:
1.119 paf 451: escape(switch(c) {
452: case '&': to_string("&"); break;
453: case '>': to_string(">"); break;
454: case '<': to_string("<"); break;
455: case '"': to_string("""); break;
456: _default;
457: });
458: break;
459: default:
460: assert(!"should never");
461: SAPI::die("unknown untaint language #%d",
462: static_cast<int>(to_lang)); // should never
463: break; // never
1.117 paf 464: }
1.55 parser 465:
1.119 paf 466: info->fragment_begin=fragment_end;
467:
468: return 0; // 0=continue
469: }
470:
471:
472: String::Body String::cstr_to_string_body(Language lang,
473: SQL_Connection* connection,
474: const Request_charsets *charsets) const {
475:
476: Cstr_to_string_body_block_info info;
477: // input
478: info.lang=lang;
479: info.connection=connection;
480: info.charsets=charsets;
481: info.body=&body;
482: // output
483: CORD_ec_init(info.result);
484: // private
485: body.set_pos(info.pos, 0);
486: info.fragment_begin=0;
487: info.whitespace=true;
488:
489: if(!is_empty())
490: langs.for_each(body, cstr_to_string_body_block, &info);
491:
492: return String::Body(CORD_ec_to_cord(info.result));
1.1 paf 493: }
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