![[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|
Return to untaint.C CVS log | Up to [parser3project] / parser3 / src / main |
1.7 paf 1: /** @file
1.8 paf 2: Parser: String class part: untaint mechanizm.
3:
1.13 paf 4: Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com)
1.63 parser 5: Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
1.8 paf 6:
1.68 ! parser 7: $Id: untaint.C,v 1.67 2001/10/08 14:09:18 parser Exp $
1.1 paf 8: */
9:
10: #include "pa_pool.h"
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"
1.34 paf 16: #include "pa_sql_connection.h"
1.58 parser 17: #include "pa_dictionary.h"
1.66 parser 18: #include "pa_common.h"
1.1 paf 19:
1.18 paf 20: #define escape(action) \
1.1 paf 21: { \
1.13 paf 22: const char *src=row->item.ptr; \
23: for(int size=row->item.size; size--; src++) \
1.18 paf 24: action \
1.1 paf 25: }
1.13 paf 26: #define _default default: *dest++=*src; break
27: #define encode(need_encode_func, prefix) \
1.5 paf 28: default: \
1.13 paf 29: if(need_encode_func(*src)) { \
1.5 paf 30: static const char *hex="0123456789ABCDEF"; \
1.9 paf 31: char chunk[3]={prefix}; \
1.13 paf 32: chunk[1]=hex[((unsigned char)*src)/0x10]; \
33: chunk[2]=hex[((unsigned char)*src)%0x10]; \
1.60 parser 34: memcpy(dest, chunk, 3); dest+=3; \
1.5 paf 35: } else \
1.13 paf 36: *dest++=*src; \
1.5 paf 37: break
1.18 paf 38: #define to_char(c) *dest++=c
39: #define to_string(b, bsize) \
1.60 parser 40: memcpy(dest, b, bsize); \
1.18 paf 41: dest+=bsize; \
1.4 paf 42:
1.9 paf 43: inline bool need_file_encode(unsigned char c){
1.13 paf 44: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.9 paf 45: return false;
46:
1.31 paf 47: return !strchr(
48: #ifdef WIN32
1.37 paf 49: ":\\~"
1.31 paf 50: #endif
1.39 paf 51: "./()_-", c);
1.9 paf 52: }
1.5 paf 53: inline bool need_uri_encode(unsigned char c){
1.13 paf 54: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 55: return false;
56:
1.5 paf 57: return !strchr("_-./", c);
58: }
1.36 paf 59: inline bool need_http_header_encode(unsigned char c){
1.18 paf 60: if(strchr(" , :", c))
1.5 paf 61: return false;
62:
63: return need_uri_encode(c);
1.4 paf 64: }
1.1 paf 65:
1.56 parser 66: //
67:
68: static const char * String_Untaint_lang_name[]={
69: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
70: "C", ///< clean
71: "T", ///< tainted, untaint language as assigned later
72: // untaint languages. assigned by ^untaint[lang]{...}
73: "P",
74: /**<
75: leave language built into string being appended.
76: just a flag, that value not stored
77: */
78: "A", ///< leave all characters intact
1.68 ! parser 79: "F", ///< file specification
! 80: "H", ///< ext in HTTP response header
1.56 parser 81: "M", ///< text in mail header
82: "URI", ///< text in uri
83: "T", ///< ^table:set body
84: "SQL", ///< ^table:sql body
85: "JS", ///< JavaScript code
1.68 ! parser 86: "XML", ///< ^dom:set xml
1.56 parser 87: "HTML", ///< HTML code (for editing)
88: "UHTML", ///< HTML code with USER chars
89: };
90:
91:
1.1 paf 92: // String
93:
1.13 paf 94: static bool typo_present(Array::Item *value, const void *info) {
95: Array *row=static_cast<Array *>(value);
96: const char *src=static_cast<const char *>(info);
97:
98: int partial;
1.28 paf 99: row->get_string(0)->cmp(partial, src);
1.14 paf 100: return
101: partial==0 || // full match
102: partial==1; // typo left column starts 'src'
1.13 paf 103: }
104:
1.41 paf 105: /*
106:
107: HTTP-header = field-name ":" [ field-value ] CRLF
108:
109: field-name = token
110: field-value = *( field-content | LWS )
111:
112: field-content = <the OCTETs making up the field-value
113: and consisting of either *TEXT or combinations
114: of token, tspecials, and quoted-string>
115:
116:
117:
118: word = token | quoted-string
119:
120: token = 1*<any CHAR except CTLs or tspecials>
121:
122:
123:
124: tspecials = "(" | ")" | "<" | ">" | "@"
125: | "," | ";" | ":" | "\" | <">
126: | "/" | "[" | "]" | "?" | "="
127: | "{" | "}" | SP | HT
128:
129: SP = <US-ASCII SP, space (32)>
130: HT = <US-ASCII HT, horizontal-tab (9)>
131:
132: LWS = [CRLF] 1*( SP | HT )
133: TEXT = <any OCTET except CTLs,
134: but including LWS>
135:
136: quoted-pair = "\" CHAR
137:
138: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
139: */
140: inline bool need_quote_http_header(const char *ptr, size_t size) {
141: for(; size--; ptr++)
1.42 paf 142: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 143: return true;
144: return false;
145: }
146:
1.55 parser 147: /** @todo maybe additional check "are all pieces are clean?" would be profitable?
148: @todo fix potential forigins_mode buf overrun
149: */
1.51 parser 150: size_t String::cstr_bufsize(Untaint_lang lang) const {
1.55 parser 151: return (lang==UL_AS_IS?size():size()*UNTAINT_TIMES_BIGGER*(forigins_mode?10:1)) +1;
1.51 parser 152: }
153:
1.54 parser 154: /** @todo fix theoretical \n mem overrun in TYPO replacements
155: */
1.43 paf 156: char *String::store_to(char *dest, Untaint_lang lang,
157: SQL_Connection *connection,
158: const char *charset) const {
1.13 paf 159: // $MAIN:html-typo table
1.58 parser 160: Dictionary *user_typo_dict=static_cast<Dictionary *>(pool().tag());
161: Dictionary *typo_dict=user_typo_dict?user_typo_dict:default_typo_dict;
1.1 paf 162:
1.44 paf 163: bool whitespace=true;
1.1 paf 164: const Chunk *chunk=&head;
165: do {
166: const Chunk::Row *row=chunk->rows;
1.28 paf 167: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 168: if(row==append_here)
169: goto break2;
170:
1.55 parser 171: Untaint_lang to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang;
172:
173: char *dest_before_origins=dest;
174:
175: if(forigins_mode) {
176: #ifndef NO_STRING_ORIGIN
177: if(row->item.origin.file)
178: dest+=sprintf(dest, "%s(%d)",
179: row->item.origin.file,
180: 1+row->item.origin.line);
181: else
182: dest+=sprintf(dest, "unknown");
183: #endif
1.56 parser 184: dest+=sprintf(dest, "#%s: ",
185: String_Untaint_lang_name[to_lang]);
1.55 parser 186: }
187: char *dest_after_origins=dest;
188:
1.1 paf 189: // WARNING:
190: // string can grow only UNTAINT_TIMES_BIGGER
1.55 parser 191: switch(to_lang) {
1.29 paf 192: case UL_CLEAN:
1.1 paf 193: // clean piece
1.44 paf 194: { // optimizing whitespace
195: const char *src=row->item.ptr;
196: for(int size=row->item.size; size--; src++)
197: switch(*src) {
1.67 parser 198: case ' ': case '\n': case '\t':
1.44 paf 199: if(!whitespace) {
200: *dest++=*src;
201: whitespace=true;
202: }
203: break;
204: default:
205: whitespace=false;
206: *dest++=*src;
207: break;
208: }
209: }
210: break;
1.29 paf 211: case UL_TAINTED:
1.1 paf 212: // tainted piece, but undefined untaint language
1.23 paf 213: // for VString.as_double of tainted values
1.1 paf 214: // for ^process{body} evaluation
1.11 paf 215: case UL_AS_IS:
1.1 paf 216: // tainted, untaint language: as-is
1.13 paf 217: memcpy(dest, row->item.ptr, row->item.size);
218: dest+=row->item.size;
1.1 paf 219: break;
1.62 parser 220: case UL_FILE_SPEC:
1.9 paf 221: // tainted, untaint language: file [name]
1.18 paf 222: escape(switch(*src) {
223: case ' ': to_char('_'); break;
1.39 paf 224: encode(need_file_encode, '+');
1.18 paf 225: });
1.9 paf 226: break;
1.11 paf 227: case UL_URI:
1.4 paf 228: // tainted, untaint language: uri
1.18 paf 229: escape(switch(*src) {
230: case ' ': to_char('+'); break;
1.13 paf 231: encode(need_uri_encode, '%');
1.18 paf 232: });
1.5 paf 233: break;
1.36 paf 234: case UL_HTTP_HEADER:
1.67 parser 235: // tainted, untaint language: http-field-content-text
236: escape(switch(*src) {
237: case ' ': to_char('+'); break;
238: encode(need_uri_encode, '%');
239: });
1.36 paf 240: break;
241: case UL_MAIL_HEADER:
242: // tainted, untaint language: mail-header
1.46 paf 243: if(charset) {
1.43 paf 244: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
245: const char *src=row->item.ptr;
1.59 parser 246: bool to_quoted_printable=false;
1.43 paf 247: for(int size=row->item.size; size--; src++) {
248: if(*src & 0x80) {
1.59 parser 249: if(!to_quoted_printable) {
1.43 paf 250: dest+=sprintf(dest, "=?%.15s?Q?", charset);
1.59 parser 251: to_quoted_printable=true;
1.43 paf 252: }
1.45 paf 253: dest+=sprintf(dest, "=%02X", *src & 0xFF);
1.43 paf 254: } else {
255: *dest++=*src;
256: }
257: }
1.59 parser 258: if(to_quoted_printable) // close
1.43 paf 259: dest+=sprintf(dest, "?=");
1.46 paf 260: } else {
261: memcpy(dest, row->item.ptr, row->item.size);
262: dest+=row->item.size;
1.43 paf 263: }
1.4 paf 264: break;
1.11 paf 265: case UL_TABLE:
1.15 paf 266: // tainted, untaint language: table
1.18 paf 267: escape(switch(*src) {
268: case '\t': to_char(' '); break;
269: case '\n': to_char(' '); break;
1.13 paf 270: _default;
1.18 paf 271: });
1.1 paf 272: break;
1.11 paf 273: case UL_SQL:
1.1 paf 274: // tainted, untaint language: sql
1.34 paf 275: if(connection)
276: dest+=connection->quote(dest, row->item.ptr, row->item.size);
277: else
278: THROW(0, 0,
279: this,
280: "untaint in SQL language failed - no connection specified");
1.1 paf 281: break;
1.11 paf 282: case UL_JS:
1.18 paf 283: escape(switch(*src) {
284: case '"': to_string("\\\"", 2); break;
285: case '\'': to_string("\\'", 2); break;
286: case '\n': to_string("\\n", 2); break;
287: case '\\': to_string("\\\\", 2); break;
288: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 289: _default;
1.18 paf 290: });
1.1 paf 291: break;
1.61 parser 292: case UL_XML:
293: escape(switch(*src) {
294: case '&': to_string("&", 5); break;
295: case '>': to_string(">", 4); break;
296: case '<': to_string("<", 4); break;
297: case '"': to_string(""", 6); break;
298: case '\'': to_string("'", 6); break;
299: _default;
300: });
301: break;
1.11 paf 302: case UL_HTML:
1.18 paf 303: escape(switch(*src) {
304: case '&': to_string("&", 5); break;
305: case '>': to_string(">", 4); break;
306: case '<': to_string("<", 4); break;
307: case '"': to_string(""", 6); break;
1.13 paf 308: _default;
1.18 paf 309: });
1.1 paf 310: break;
1.47 paf 311: case UL_USER_HTML: {
1.1 paf 312: // tainted, untaint language: html-typo
1.58 parser 313: if(!typo_dict) // never, always has default
1.57 parser 314: THROW(0, 0,
315: this,
316: "untaint to user-html lang failed, no typo table");
317:
1.50 parser 318: char *html_for_typo=
319: (char *)malloc(row->item.size*2/* '\n' -> '\' 'n' */+1);
1.19 paf 320: // note:
321: // there still is a possibility that user
322: // would not replace \n as she supposed to
323: // and rather replace \ and n into huge strings
324: // thus causing memory overrun
325: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 326: size_t html_for_typo_size;
1.13 paf 327: { // local dest
1.18 paf 328: char *dest=html_for_typo;
329: escape(switch(*src) {
1.16 paf 330: // convinient name for typo match "\n"
1.18 paf 331: case '\n':
1.57 parser 332: to_string("\\n", 2);
1.16 paf 333: break;
1.13 paf 334: _default;
1.18 paf 335: });
1.13 paf 336: *dest=0;
1.18 paf 337: html_for_typo_size=dest-html_for_typo;
1.13 paf 338: }
339: // typo table replacements
1.21 paf 340: const char *src=html_for_typo;
341: do {
342: // there is a row where first column starts 'src'
1.58 parser 343: if(Table::Item *item=typo_dict->first_that_starts(src)) {
1.21 paf 344: // get a=>b values
345: const String& a=*static_cast<Array *>(item)->get_string(0);
346: const String& b=*static_cast<Array *>(item)->get_string(1);
347: // overflow check:
348: // b allowed to be max UNTAINT_TIMES_BIGGER then a
349: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.58 parser 350: pool().set_tag(0); // avoid recursion
1.21 paf 351: THROW(0, 0,
352: &b,
353: "is %g times longer then '%s', "
354: "while maximum, handled by Parser, is %d",
1.50 parser 355: ((double)b.size())/a.size(),
356: a.cstr(),
357: UNTAINT_TIMES_BIGGER);
1.21 paf 358: }
359:
360: // skip 'a' in 'src'
361: src+=a.size();
362: // write 'b' to 'dest'
363: b.store_to(dest);
1.59 parser 364: // skip 'b' in 'dest'
1.21 paf 365: dest+=b.size();
366: } else
367: *dest++=*src++;
368: } while(*src);
1.1 paf 369: break;
1.13 paf 370: }
1.1 paf 371: default:
1.18 paf 372: THROW(0, 0,
373: this,
1.1 paf 374: "unknown untaint language #%d of %d piece",
1.18 paf 375: static_cast<int>(row->item.lang),
1.38 paf 376: i); // never
1.48 parser 377: break; // never
1.1 paf 378: }
1.44 paf 379:
380: if((lang==UL_UNSPECIFIED?row->item.lang:lang)!=UL_CLEAN)
381: whitespace=false;
1.55 parser 382:
383: if(forigins_mode)
384: if(dest==dest_after_origins) // never moved==optimized space
385: dest=dest_before_origins;
386: else {
1.66 parser 387: remove_crlf(dest_after_origins, dest);
1.55 parser 388:
389: to_char('\n');
390: }
1.1 paf 391: }
392: chunk=row->link;
393: } while(chunk);
1.64 parser 394:
1.1 paf 395: break2:
1.13 paf 396: return dest;
1.1 paf 397: }