![[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:
1.35 ! paf 8: $Id: untaint.C,v 1.34 2001/04/05 13:19:43 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
50: ":\\"
51: #endif
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: }
60: inline bool need_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
83: */
1.34 paf 84: char *String::store_to(char *dest, Untaint_lang lang, SQL_Connection *connection) const {
1.13 paf 85: // $MAIN:html-typo table
1.26 paf 86: Table *user_typo_table=static_cast<Table *>(pool().tag());
87: Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
1.1 paf 88:
89: const Chunk *chunk=&head;
90: do {
91: const Chunk::Row *row=chunk->rows;
1.28 paf 92: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 93: if(row==append_here)
94: goto break2;
95:
96: // WARNING:
97: // string can grow only UNTAINT_TIMES_BIGGER
1.35 ! paf 98: switch(lang==UL_UNSPECIFIED?row->item.lang:lang) {
1.29 paf 99: case UL_CLEAN:
1.1 paf 100: // clean piece
1.29 paf 101: case UL_TAINTED:
1.1 paf 102: // tainted piece, but undefined untaint language
1.23 paf 103: // for VString.as_double of tainted values
1.1 paf 104: // for ^process{body} evaluation
1.11 paf 105: case UL_AS_IS:
1.1 paf 106: // tainted, untaint language: as-is
1.13 paf 107: memcpy(dest, row->item.ptr, row->item.size);
108: dest+=row->item.size;
1.1 paf 109: break;
1.11 paf 110: case UL_FILE_NAME:
1.9 paf 111: // tainted, untaint language: file [name]
1.18 paf 112: escape(switch(*src) {
113: case ' ': to_char('_'); break;
1.13 paf 114: encode(need_file_encode, '-');
1.18 paf 115: });
1.9 paf 116: break;
1.11 paf 117: case UL_URI:
1.4 paf 118: // tainted, untaint language: uri
1.18 paf 119: escape(switch(*src) {
120: case ' ': to_char('+'); break;
1.13 paf 121: encode(need_uri_encode, '%');
1.18 paf 122: });
1.5 paf 123: break;
1.11 paf 124: case UL_HEADER:
1.5 paf 125: // tainted, untaint language: header
1.18 paf 126: escape(switch(*src) {
1.13 paf 127: encode(need_header_encode, '%');
1.18 paf 128: });
1.4 paf 129: break;
1.11 paf 130: case UL_TABLE:
1.15 paf 131: // tainted, untaint language: table
1.18 paf 132: escape(switch(*src) {
133: case '\t': to_char(' '); break;
134: case '\n': to_char(' '); break;
1.13 paf 135: _default;
1.18 paf 136: });
1.1 paf 137: break;
1.11 paf 138: case UL_SQL:
1.1 paf 139: // tainted, untaint language: sql
1.34 paf 140: if(connection)
141: dest+=connection->quote(dest, row->item.ptr, row->item.size);
142: else
143: THROW(0, 0,
144: this,
145: "untaint in SQL language failed - no connection specified");
1.1 paf 146: break;
1.11 paf 147: case UL_JS:
1.18 paf 148: escape(switch(*src) {
149: case '"': to_string("\\\"", 2); break;
150: case '\'': to_string("\\'", 2); break;
151: case '\n': to_string("\\n", 2); break;
152: case '\\': to_string("\\\\", 2); break;
153: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 154: _default;
1.18 paf 155: });
1.1 paf 156: break;
1.11 paf 157: case UL_HTML:
1.18 paf 158: escape(switch(*src) {
159: case '&': to_string("&", 5); break;
160: case '>': to_string(">", 4); break;
161: case '<': to_string("<", 4); break;
162: case '"': to_string(""", 6); break;
1.19 paf 163: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 164: _default;
1.18 paf 165: });
1.1 paf 166: break;
1.13 paf 167: case UL_HTML_TYPO: {
1.1 paf 168: // tainted, untaint language: html-typo
1.19 paf 169: char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
170: // note:
171: // there still is a possibility that user
172: // would not replace \n as she supposed to
173: // and rather replace \ and n into huge strings
174: // thus causing memory overrun
175: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 176: size_t html_for_typo_size;
1.13 paf 177: { // local dest
1.18 paf 178: char *dest=html_for_typo;
179: escape(switch(*src) {
1.16 paf 180: // convinient name for typo match "\n"
181: case '\r':
1.18 paf 182: if(typo_table) {
183: *dest++='\\'; *dest++='n'; // \r -> \n
1.24 paf 184: if(src[1]=='\n') { // \r\n -> remove \n
185: size--; src++;
186: }
1.18 paf 187: }
188: break;
189: case '\n':
190: if(typo_table)
191: to_string("\\n", 2);
1.16 paf 192: break;
1.19 paf 193: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 194: _default;
1.18 paf 195: });
1.13 paf 196: *dest=0;
1.18 paf 197: html_for_typo_size=dest-html_for_typo;
1.13 paf 198: }
199: // typo table replacements
1.21 paf 200: const char *src=html_for_typo;
201: do {
202: // there is a row where first column starts 'src'
203: if(Table::Item *item=typo_table->first_that(typo_present, src)) {
204: // get a=>b values
205: const String& a=*static_cast<Array *>(item)->get_string(0);
206: const String& b=*static_cast<Array *>(item)->get_string(1);
207: // empty 'a' | 'b' checks
208: if(a.size()==0 || b.size()==0) {
1.26 paf 209: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 210: THROW(0, 0,
211: typo_table->origin_string(),
212: "typo table column elements must not be empty");
213: }
214: // overflow check:
215: // b allowed to be max UNTAINT_TIMES_BIGGER then a
216: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.26 paf 217: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 218: THROW(0, 0,
219: &b,
220: "is %g times longer then '%s', "
221: "while maximum, handled by Parser, is %d",
222: ((double)b.size())/a.size(),
223: a.cstr(),
224: UNTAINT_TIMES_BIGGER);
225: }
226:
227: // skip 'a' in 'src'
228: src+=a.size();
229: // write 'b' to 'dest'
230: b.store_to(dest);
231: dest+=b.size();
232: } else
233: *dest++=*src++;
234: } while(*src);
1.1 paf 235: break;
1.13 paf 236: }
1.1 paf 237: default:
1.18 paf 238: THROW(0, 0,
239: this,
1.1 paf 240: "unknown untaint language #%d of %d piece",
1.18 paf 241: static_cast<int>(row->item.lang),
1.1 paf 242: i);
243: }
244: }
245: chunk=row->link;
246: } while(chunk);
247: break2:
1.13 paf 248: return dest;
1.1 paf 249: }