![[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:
1.33 ! paf 8: $Id: untaint.C,v 1.32 2001/04/04 10:50:36 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"
1.1 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]; \
34: strncpy(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) \
40: strncpy(dest, b, bsize); \
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
49: ":\\"
50: #endif
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: }
59: inline bool need_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:
66: // String
67:
1.13 paf 68: static bool typo_present(Array::Item *value, const void *info) {
69: Array *row=static_cast<Array *>(value);
70: const char *src=static_cast<const char *>(info);
71:
72: int partial;
1.28 paf 73: row->get_string(0)->cmp(partial, src);
1.14 paf 74: return
75: partial==0 || // full match
76: partial==1; // typo left column starts 'src'
1.13 paf 77: }
78:
1.30 paf 79: /**
80: @test optimize whitespaces for all but 'html'
81: @todo fix theoretical \n mem overrun in TYPO replacements
82: */
1.31 paf 83: char *String::store_to(char *dest, Untaint_lang lang) const {
1.13 paf 84: // $MAIN:html-typo table
1.26 paf 85: Table *user_typo_table=static_cast<Table *>(pool().tag());
86: Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
1.1 paf 87:
88: const Chunk *chunk=&head;
89: do {
90: const Chunk::Row *row=chunk->rows;
1.28 paf 91: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 92: if(row==append_here)
93: goto break2;
94:
95: // WARNING:
96: // string can grow only UNTAINT_TIMES_BIGGER
1.31 paf 97: switch(lang==UL_UNKNOWN?row->item.lang:lang) {
1.29 paf 98: case UL_CLEAN:
1.1 paf 99: // clean piece
1.29 paf 100: case UL_TAINTED:
1.1 paf 101: // tainted piece, but undefined untaint language
1.23 paf 102: // for VString.as_double of tainted values
1.1 paf 103: // for ^process{body} evaluation
1.11 paf 104: case UL_AS_IS:
1.1 paf 105: // tainted, untaint language: as-is
1.13 paf 106: memcpy(dest, row->item.ptr, row->item.size);
107: dest+=row->item.size;
1.1 paf 108: break;
1.11 paf 109: case UL_FILE_NAME:
1.9 paf 110: // tainted, untaint language: file [name]
1.18 paf 111: escape(switch(*src) {
112: case ' ': to_char('_'); break;
1.13 paf 113: encode(need_file_encode, '-');
1.18 paf 114: });
1.9 paf 115: break;
1.11 paf 116: case UL_URI:
1.4 paf 117: // tainted, untaint language: uri
1.18 paf 118: escape(switch(*src) {
119: case ' ': to_char('+'); break;
1.13 paf 120: encode(need_uri_encode, '%');
1.18 paf 121: });
1.5 paf 122: break;
1.11 paf 123: case UL_HEADER:
1.5 paf 124: // tainted, untaint language: header
1.18 paf 125: escape(switch(*src) {
1.13 paf 126: encode(need_header_encode, '%');
1.18 paf 127: });
1.4 paf 128: break;
1.11 paf 129: case UL_TABLE:
1.15 paf 130: // tainted, untaint language: table
1.18 paf 131: escape(switch(*src) {
132: case '\t': to_char(' '); break;
133: case '\n': to_char(' '); break;
1.13 paf 134: _default;
1.18 paf 135: });
1.1 paf 136: break;
1.11 paf 137: case UL_SQL:
1.1 paf 138: // tainted, untaint language: sql
139: // TODO: зависимость от sql сервера
1.33 ! paf 140: memcpy(dest, row->item.ptr, row->item.size);
! 141: //memset(dest, '?', row->item.size);
1.13 paf 142: dest+=row->item.size;
1.1 paf 143: break;
1.11 paf 144: case UL_JS:
1.18 paf 145: escape(switch(*src) {
146: case '"': to_string("\\\"", 2); break;
147: case '\'': to_string("\\'", 2); break;
148: case '\n': to_string("\\n", 2); break;
149: case '\\': to_string("\\\\", 2); break;
150: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 151: _default;
1.18 paf 152: });
1.1 paf 153: break;
1.11 paf 154: case UL_HTML:
1.18 paf 155: escape(switch(*src) {
156: case '&': to_string("&", 5); break;
157: case '>': to_string(">", 4); break;
158: case '<': to_string("<", 4); break;
159: case '"': to_string(""", 6); break;
1.19 paf 160: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 161: _default;
1.18 paf 162: });
1.1 paf 163: break;
1.13 paf 164: case UL_HTML_TYPO: {
1.1 paf 165: // tainted, untaint language: html-typo
1.19 paf 166: char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
167: // note:
168: // there still is a possibility that user
169: // would not replace \n as she supposed to
170: // and rather replace \ and n into huge strings
171: // thus causing memory overrun
172: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 173: size_t html_for_typo_size;
1.13 paf 174: { // local dest
1.18 paf 175: char *dest=html_for_typo;
176: escape(switch(*src) {
1.16 paf 177: // convinient name for typo match "\n"
178: case '\r':
1.18 paf 179: if(typo_table) {
180: *dest++='\\'; *dest++='n'; // \r -> \n
1.24 paf 181: if(src[1]=='\n') { // \r\n -> remove \n
182: size--; src++;
183: }
1.18 paf 184: }
185: break;
186: case '\n':
187: if(typo_table)
188: to_string("\\n", 2);
1.16 paf 189: break;
1.19 paf 190: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 191: _default;
1.18 paf 192: });
1.13 paf 193: *dest=0;
1.18 paf 194: html_for_typo_size=dest-html_for_typo;
1.13 paf 195: }
196: // typo table replacements
1.21 paf 197: const char *src=html_for_typo;
198: do {
199: // there is a row where first column starts 'src'
200: if(Table::Item *item=typo_table->first_that(typo_present, src)) {
201: // get a=>b values
202: const String& a=*static_cast<Array *>(item)->get_string(0);
203: const String& b=*static_cast<Array *>(item)->get_string(1);
204: // empty 'a' | 'b' checks
205: if(a.size()==0 || b.size()==0) {
1.26 paf 206: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 207: THROW(0, 0,
208: typo_table->origin_string(),
209: "typo table column elements must not be empty");
210: }
211: // overflow check:
212: // b allowed to be max UNTAINT_TIMES_BIGGER then a
213: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.26 paf 214: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 215: THROW(0, 0,
216: &b,
217: "is %g times longer then '%s', "
218: "while maximum, handled by Parser, is %d",
219: ((double)b.size())/a.size(),
220: a.cstr(),
221: UNTAINT_TIMES_BIGGER);
222: }
223:
224: // skip 'a' in 'src'
225: src+=a.size();
226: // write 'b' to 'dest'
227: b.store_to(dest);
228: dest+=b.size();
229: } else
230: *dest++=*src++;
231: } while(*src);
1.1 paf 232: break;
1.13 paf 233: }
1.1 paf 234: default:
1.18 paf 235: THROW(0, 0,
236: this,
1.1 paf 237: "unknown untaint language #%d of %d piece",
1.18 paf 238: static_cast<int>(row->item.lang),
1.1 paf 239: i);
240: }
241: }
242: chunk=row->link;
243: } while(chunk);
244: break2:
1.13 paf 245: return dest;
1.1 paf 246: }