![[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:
1.40 ! paf 8: $Id: untaint.C,v 1.39 2001/04/11 08:36:20 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:
1.30 paf 80: /**
81: @test optimize whitespaces for all but 'html'
82: @todo fix theoretical \n mem overrun in TYPO replacements
1.36 paf 83: @test mail-header
1.30 paf 84: */
1.34 paf 85: char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection) const {
1.13 paf 86: // $MAIN:html-typo table
1.26 paf 87: Table *user_typo_table=static_cast<Table *>(pool().tag());
88: Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
1.1 paf 89:
90: const Chunk *chunk=&head;
91: do {
92: const Chunk::Row *row=chunk->rows;
1.28 paf 93: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 94: if(row==append_here)
95: goto break2;
96:
97: // WARNING:
98: // string can grow only UNTAINT_TIMES_BIGGER
1.35 paf 99: switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
1.29 paf 100: case UL_CLEAN:
1.1 paf 101: // clean piece
1.29 paf 102: case UL_TAINTED:
1.1 paf 103: // tainted piece, but undefined untaint language
1.23 paf 104: // for VString.as_double of tainted values
1.1 paf 105: // for ^process{body} evaluation
1.11 paf 106: case UL_AS_IS:
1.1 paf 107: // tainted, untaint language: as-is
1.13 paf 108: memcpy(dest, row->item.ptr, row->item.size);
109: dest+=row->item.size;
1.1 paf 110: break;
1.11 paf 111: case UL_FILE_NAME:
1.9 paf 112: // tainted, untaint language: file [name]
1.18 paf 113: escape(switch(*src) {
114: case ' ': to_char('_'); break;
1.39 paf 115: encode(need_file_encode, '+');
1.18 paf 116: });
1.9 paf 117: break;
1.11 paf 118: case UL_URI:
1.4 paf 119: // tainted, untaint language: uri
1.18 paf 120: escape(switch(*src) {
121: case ' ': to_char('+'); break;
1.13 paf 122: encode(need_uri_encode, '%');
1.18 paf 123: });
1.5 paf 124: break;
1.36 paf 125: case UL_HTTP_HEADER:
126: // tainted, untaint language: http-header
1.40 ! paf 127: /*
! 128:
! 129: HTTP-header = field-name ":" [ field-value ] CRLF
! 130:
! 131: field-name = token
! 132: field-value = *( field-content | LWS )
! 133:
! 134: field-content = <the OCTETs making up the field-value
! 135: and consisting of either *TEXT or combinations
! 136: of token, tspecials, and quoted-string>
! 137:
! 138:
! 139:
! 140: word = token | quoted-string
! 141:
! 142: token = 1*<any CHAR except CTLs or tspecials>
! 143:
! 144:
! 145:
! 146: tspecials = "(" | ")" | "<" | ">" | "@"
! 147: | "," | ";" | ":" | "\" | <">
! 148: | "/" | "[" | "]" | "?" | "="
! 149: | "{" | "}" | SP | HT
! 150:
! 151: SP = <US-ASCII SP, space (32)>
! 152: HT = <US-ASCII HT, horizontal-tab (9)>
! 153:
! 154: LWS = [CRLF] 1*( SP | HT )
! 155: TEXT = <any OCTET except CTLs,
! 156: but including LWS>
! 157:
! 158: quoted-pair = "\" CHAR
! 159:
! 160:
! 161: */
! 162: *dest++='\"';
1.18 paf 163: escape(switch(*src) {
1.40 ! paf 164: case '\"': to_string("\\\"", 2); break;
! 165: _default;
1.18 paf 166: });
1.40 ! paf 167: *dest++='\"';
1.36 paf 168: break;
169: case UL_MAIL_HEADER:
170: // tainted, untaint language: mail-header
171: memcpy(dest, row->item.ptr, row->item.size);
172: dest+=row->item.size;
1.4 paf 173: break;
1.11 paf 174: case UL_TABLE:
1.15 paf 175: // tainted, untaint language: table
1.18 paf 176: escape(switch(*src) {
177: case '\t': to_char(' '); break;
178: case '\n': to_char(' '); break;
1.13 paf 179: _default;
1.18 paf 180: });
1.1 paf 181: break;
1.11 paf 182: case UL_SQL:
1.1 paf 183: // tainted, untaint language: sql
1.34 paf 184: if(connection)
185: dest+=connection->quote(dest, row->item.ptr, row->item.size);
186: else
187: THROW(0, 0,
188: this,
189: "untaint in SQL language failed - no connection specified");
1.1 paf 190: break;
1.11 paf 191: case UL_JS:
1.18 paf 192: escape(switch(*src) {
193: case '"': to_string("\\\"", 2); break;
194: case '\'': to_string("\\'", 2); break;
195: case '\n': to_string("\\n", 2); break;
196: case '\\': to_string("\\\\", 2); break;
197: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 198: _default;
1.18 paf 199: });
1.1 paf 200: break;
1.11 paf 201: case UL_HTML:
1.18 paf 202: escape(switch(*src) {
203: case '&': to_string("&", 5); break;
204: case '>': to_string(">", 4); break;
205: case '<': to_string("<", 4); break;
206: case '"': to_string(""", 6); break;
1.19 paf 207: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 208: _default;
1.18 paf 209: });
1.1 paf 210: break;
1.13 paf 211: case UL_HTML_TYPO: {
1.1 paf 212: // tainted, untaint language: html-typo
1.19 paf 213: char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
214: // note:
215: // there still is a possibility that user
216: // would not replace \n as she supposed to
217: // and rather replace \ and n into huge strings
218: // thus causing memory overrun
219: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 220: size_t html_for_typo_size;
1.13 paf 221: { // local dest
1.18 paf 222: char *dest=html_for_typo;
223: escape(switch(*src) {
1.16 paf 224: // convinient name for typo match "\n"
225: case '\r':
1.18 paf 226: if(typo_table) {
227: *dest++='\\'; *dest++='n'; // \r -> \n
1.24 paf 228: if(src[1]=='\n') { // \r\n -> remove \n
229: size--; src++;
230: }
1.18 paf 231: }
232: break;
233: case '\n':
234: if(typo_table)
235: to_string("\\n", 2);
1.16 paf 236: break;
1.19 paf 237: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 238: _default;
1.18 paf 239: });
1.13 paf 240: *dest=0;
1.18 paf 241: html_for_typo_size=dest-html_for_typo;
1.13 paf 242: }
243: // typo table replacements
1.21 paf 244: const char *src=html_for_typo;
245: do {
246: // there is a row where first column starts 'src'
247: if(Table::Item *item=typo_table->first_that(typo_present, src)) {
248: // get a=>b values
249: const String& a=*static_cast<Array *>(item)->get_string(0);
250: const String& b=*static_cast<Array *>(item)->get_string(1);
251: // empty 'a' | 'b' checks
252: if(a.size()==0 || b.size()==0) {
1.26 paf 253: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 254: THROW(0, 0,
255: typo_table->origin_string(),
256: "typo table column elements must not be empty");
257: }
258: // overflow check:
259: // b allowed to be max UNTAINT_TIMES_BIGGER then a
260: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.26 paf 261: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 262: THROW(0, 0,
263: &b,
264: "is %g times longer then '%s', "
265: "while maximum, handled by Parser, is %d",
266: ((double)b.size())/a.size(),
267: a.cstr(),
268: UNTAINT_TIMES_BIGGER);
269: }
270:
271: // skip 'a' in 'src'
272: src+=a.size();
273: // write 'b' to 'dest'
274: b.store_to(dest);
275: dest+=b.size();
276: } else
277: *dest++=*src++;
278: } while(*src);
1.1 paf 279: break;
1.13 paf 280: }
1.1 paf 281: default:
1.18 paf 282: THROW(0, 0,
283: this,
1.1 paf 284: "unknown untaint language #%d of %d piece",
1.18 paf 285: static_cast<int>(row->item.lang),
1.38 paf 286: i); // never
1.1 paf 287: }
288: }
289: chunk=row->link;
290: } while(chunk);
291: break2:
1.13 paf 292: return dest;
1.1 paf 293: }