![[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)
1.8 paf 5:
1.13 paf 6: Author: Alexander Petrosyan <paf@design.ru>(http://design.ru/paf)
1.1 paf 7:
1.41 ! paf 8: $Id: untaint.C,v 1.40 2001/04/19 15:38:00 paf Exp $
1.1 paf 9: */
10:
1.12 paf 11: #include "pa_config_includes.h"
1.1 paf 12:
13: #include "pa_pool.h"
14: #include "pa_string.h"
15: #include "pa_hash.h"
16: #include "pa_exception.h"
1.13 paf 17: #include "pa_table.h"
1.32 paf 18: #include "pa_globals.h"
1.34 paf 19: #include "pa_sql_connection.h"
1.1 paf 20:
1.18 paf 21: #define escape(action) \
1.1 paf 22: { \
1.13 paf 23: const char *src=row->item.ptr; \
24: for(int size=row->item.size; size--; src++) \
1.18 paf 25: action \
1.1 paf 26: }
1.13 paf 27: #define _default default: *dest++=*src; break
28: #define encode(need_encode_func, prefix) \
1.5 paf 29: default: \
1.13 paf 30: if(need_encode_func(*src)) { \
1.5 paf 31: static const char *hex="0123456789ABCDEF"; \
1.9 paf 32: char chunk[3]={prefix}; \
1.13 paf 33: chunk[1]=hex[((unsigned char)*src)/0x10]; \
34: chunk[2]=hex[((unsigned char)*src)%0x10]; \
35: strncpy(dest, chunk, 3); dest+=3; \
1.5 paf 36: } else \
1.13 paf 37: *dest++=*src; \
1.5 paf 38: break
1.18 paf 39: #define to_char(c) *dest++=c
40: #define to_string(b, bsize) \
41: strncpy(dest, b, bsize); \
42: dest+=bsize; \
1.4 paf 43:
1.9 paf 44: inline bool need_file_encode(unsigned char c){
1.13 paf 45: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.9 paf 46: return false;
47:
1.31 paf 48: return !strchr(
49: #ifdef WIN32
1.37 paf 50: ":\\~"
1.31 paf 51: #endif
1.39 paf 52: "./()_-", c);
1.9 paf 53: }
1.5 paf 54: inline bool need_uri_encode(unsigned char c){
1.13 paf 55: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 56: return false;
57:
1.5 paf 58: return !strchr("_-./", c);
59: }
1.36 paf 60: inline bool need_http_header_encode(unsigned char c){
1.18 paf 61: if(strchr(" , :", c))
1.5 paf 62: return false;
63:
64: return need_uri_encode(c);
1.4 paf 65: }
1.1 paf 66:
67: // String
68:
1.13 paf 69: static bool typo_present(Array::Item *value, const void *info) {
70: Array *row=static_cast<Array *>(value);
71: const char *src=static_cast<const char *>(info);
72:
73: int partial;
1.28 paf 74: row->get_string(0)->cmp(partial, src);
1.14 paf 75: return
76: partial==0 || // full match
77: partial==1; // typo left column starts 'src'
1.13 paf 78: }
79:
1.41 ! paf 80: /*
! 81:
! 82: HTTP-header = field-name ":" [ field-value ] CRLF
! 83:
! 84: field-name = token
! 85: field-value = *( field-content | LWS )
! 86:
! 87: field-content = <the OCTETs making up the field-value
! 88: and consisting of either *TEXT or combinations
! 89: of token, tspecials, and quoted-string>
! 90:
! 91:
! 92:
! 93: word = token | quoted-string
! 94:
! 95: token = 1*<any CHAR except CTLs or tspecials>
! 96:
! 97:
! 98:
! 99: tspecials = "(" | ")" | "<" | ">" | "@"
! 100: | "," | ";" | ":" | "\" | <">
! 101: | "/" | "[" | "]" | "?" | "="
! 102: | "{" | "}" | SP | HT
! 103:
! 104: SP = <US-ASCII SP, space (32)>
! 105: HT = <US-ASCII HT, horizontal-tab (9)>
! 106:
! 107: LWS = [CRLF] 1*( SP | HT )
! 108: TEXT = <any OCTET except CTLs,
! 109: but including LWS>
! 110:
! 111: quoted-pair = "\" CHAR
! 112:
! 113: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
! 114: */
! 115: inline bool need_quote_http_header(const char *ptr, size_t size) {
! 116: for(; size--; ptr++)
! 117: if(strchr("()<>@,;:\\\"[]?={} \t" /* excluded / */, *ptr))
! 118: return true;
! 119: return false;
! 120: }
! 121:
1.30 paf 122: /**
123: @test optimize whitespaces for all but 'html'
124: @todo fix theoretical \n mem overrun in TYPO replacements
1.36 paf 125: @test mail-header
1.30 paf 126: */
1.34 paf 127: char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection) const {
1.13 paf 128: // $MAIN:html-typo table
1.26 paf 129: Table *user_typo_table=static_cast<Table *>(pool().tag());
130: Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
1.1 paf 131:
132: const Chunk *chunk=&head;
133: do {
134: const Chunk::Row *row=chunk->rows;
1.28 paf 135: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 136: if(row==append_here)
137: goto break2;
138:
139: // WARNING:
140: // string can grow only UNTAINT_TIMES_BIGGER
1.35 paf 141: switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
1.29 paf 142: case UL_CLEAN:
1.1 paf 143: // clean piece
1.29 paf 144: case UL_TAINTED:
1.1 paf 145: // tainted piece, but undefined untaint language
1.23 paf 146: // for VString.as_double of tainted values
1.1 paf 147: // for ^process{body} evaluation
1.11 paf 148: case UL_AS_IS:
1.1 paf 149: // tainted, untaint language: as-is
1.13 paf 150: memcpy(dest, row->item.ptr, row->item.size);
151: dest+=row->item.size;
1.1 paf 152: break;
1.11 paf 153: case UL_FILE_NAME:
1.9 paf 154: // tainted, untaint language: file [name]
1.18 paf 155: escape(switch(*src) {
156: case ' ': to_char('_'); break;
1.39 paf 157: encode(need_file_encode, '+');
1.18 paf 158: });
1.9 paf 159: break;
1.11 paf 160: case UL_URI:
1.4 paf 161: // tainted, untaint language: uri
1.18 paf 162: escape(switch(*src) {
163: case ' ': to_char('+'); break;
1.13 paf 164: encode(need_uri_encode, '%');
1.18 paf 165: });
1.5 paf 166: break;
1.36 paf 167: case UL_HTTP_HEADER:
168: // tainted, untaint language: http-header
1.41 ! paf 169: if(need_quote_http_header(row->item.ptr, row->item.size)) {
! 170: *dest++='\"';
! 171: escape(switch(*src) {
! 172: case '\"': to_string("\\\"", 2); break;
! 173: _default;
! 174: });
! 175: *dest++='\"';
! 176: } else {
! 177: memcpy(dest, row->item.ptr, row->item.size);
! 178: dest+=row->item.size;
! 179: }
1.36 paf 180: break;
181: case UL_MAIL_HEADER:
182: // tainted, untaint language: mail-header
183: memcpy(dest, row->item.ptr, row->item.size);
184: dest+=row->item.size;
1.4 paf 185: break;
1.11 paf 186: case UL_TABLE:
1.15 paf 187: // tainted, untaint language: table
1.18 paf 188: escape(switch(*src) {
189: case '\t': to_char(' '); break;
190: case '\n': to_char(' '); break;
1.13 paf 191: _default;
1.18 paf 192: });
1.1 paf 193: break;
1.11 paf 194: case UL_SQL:
1.1 paf 195: // tainted, untaint language: sql
1.34 paf 196: if(connection)
197: dest+=connection->quote(dest, row->item.ptr, row->item.size);
198: else
199: THROW(0, 0,
200: this,
201: "untaint in SQL language failed - no connection specified");
1.1 paf 202: break;
1.11 paf 203: case UL_JS:
1.18 paf 204: escape(switch(*src) {
205: case '"': to_string("\\\"", 2); break;
206: case '\'': to_string("\\'", 2); break;
207: case '\n': to_string("\\n", 2); break;
208: case '\\': to_string("\\\\", 2); break;
209: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 210: _default;
1.18 paf 211: });
1.1 paf 212: break;
1.11 paf 213: case UL_HTML:
1.18 paf 214: escape(switch(*src) {
215: case '&': to_string("&", 5); break;
216: case '>': to_string(">", 4); break;
217: case '<': to_string("<", 4); break;
218: case '"': to_string(""", 6); break;
1.19 paf 219: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 220: _default;
1.18 paf 221: });
1.1 paf 222: break;
1.13 paf 223: case UL_HTML_TYPO: {
1.1 paf 224: // tainted, untaint language: html-typo
1.19 paf 225: char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
226: // note:
227: // there still is a possibility that user
228: // would not replace \n as she supposed to
229: // and rather replace \ and n into huge strings
230: // thus causing memory overrun
231: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 232: size_t html_for_typo_size;
1.13 paf 233: { // local dest
1.18 paf 234: char *dest=html_for_typo;
235: escape(switch(*src) {
1.16 paf 236: // convinient name for typo match "\n"
237: case '\r':
1.18 paf 238: if(typo_table) {
239: *dest++='\\'; *dest++='n'; // \r -> \n
1.24 paf 240: if(src[1]=='\n') { // \r\n -> remove \n
241: size--; src++;
242: }
1.18 paf 243: }
244: break;
245: case '\n':
246: if(typo_table)
247: to_string("\\n", 2);
1.16 paf 248: break;
1.19 paf 249: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 250: _default;
1.18 paf 251: });
1.13 paf 252: *dest=0;
1.18 paf 253: html_for_typo_size=dest-html_for_typo;
1.13 paf 254: }
255: // typo table replacements
1.21 paf 256: const char *src=html_for_typo;
257: do {
258: // there is a row where first column starts 'src'
259: if(Table::Item *item=typo_table->first_that(typo_present, src)) {
260: // get a=>b values
261: const String& a=*static_cast<Array *>(item)->get_string(0);
262: const String& b=*static_cast<Array *>(item)->get_string(1);
263: // empty 'a' | 'b' checks
264: if(a.size()==0 || b.size()==0) {
1.26 paf 265: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 266: THROW(0, 0,
267: typo_table->origin_string(),
268: "typo table column elements must not be empty");
269: }
270: // overflow check:
271: // b allowed to be max UNTAINT_TIMES_BIGGER then a
272: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.26 paf 273: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 274: THROW(0, 0,
275: &b,
276: "is %g times longer then '%s', "
277: "while maximum, handled by Parser, is %d",
278: ((double)b.size())/a.size(),
279: a.cstr(),
280: UNTAINT_TIMES_BIGGER);
281: }
282:
283: // skip 'a' in 'src'
284: src+=a.size();
285: // write 'b' to 'dest'
286: b.store_to(dest);
287: dest+=b.size();
288: } else
289: *dest++=*src++;
290: } while(*src);
1.1 paf 291: break;
1.13 paf 292: }
1.1 paf 293: default:
1.18 paf 294: THROW(0, 0,
295: this,
1.1 paf 296: "unknown untaint language #%d of %d piece",
1.18 paf 297: static_cast<int>(row->item.lang),
1.38 paf 298: i); // never
1.1 paf 299: }
300: }
301: chunk=row->link;
302: } while(chunk);
303: break2:
1.13 paf 304: return dest;
1.1 paf 305: }