![[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:
1.31 ! paf 8: $Id: untaint.C,v 1.30 2001/04/03 07:32:46 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"
1.1 paf 18:
1.18 paf 19: #define escape(action) \
1.1 paf 20: { \
1.13 paf 21: const char *src=row->item.ptr; \
22: for(int size=row->item.size; size--; src++) \
1.18 paf 23: action \
1.1 paf 24: }
1.13 paf 25: #define _default default: *dest++=*src; break
26: #define encode(need_encode_func, prefix) \
1.5 paf 27: default: \
1.13 paf 28: if(need_encode_func(*src)) { \
1.5 paf 29: static const char *hex="0123456789ABCDEF"; \
1.9 paf 30: char chunk[3]={prefix}; \
1.13 paf 31: chunk[1]=hex[((unsigned char)*src)/0x10]; \
32: chunk[2]=hex[((unsigned char)*src)%0x10]; \
33: strncpy(dest, chunk, 3); dest+=3; \
1.5 paf 34: } else \
1.13 paf 35: *dest++=*src; \
1.5 paf 36: break
1.18 paf 37: #define to_char(c) *dest++=c
38: #define to_string(b, bsize) \
39: strncpy(dest, b, bsize); \
40: dest+=bsize; \
1.4 paf 41:
1.9 paf 42: inline bool need_file_encode(unsigned char c){
1.13 paf 43: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.9 paf 44: return false;
45:
1.31 ! paf 46: return !strchr(
! 47: #ifdef WIN32
! 48: ":\\"
! 49: #endif
! 50: "./", c);
1.9 paf 51: }
1.5 paf 52: inline bool need_uri_encode(unsigned char c){
1.13 paf 53: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 54: return false;
55:
1.5 paf 56: return !strchr("_-./", c);
57: }
58: inline bool need_header_encode(unsigned char c){
1.18 paf 59: if(strchr(" , :", c))
1.5 paf 60: return false;
61:
62: return need_uri_encode(c);
1.4 paf 63: }
1.1 paf 64:
65: // String
66:
1.13 paf 67: static bool typo_present(Array::Item *value, const void *info) {
68: Array *row=static_cast<Array *>(value);
69: const char *src=static_cast<const char *>(info);
70:
71: int partial;
1.28 paf 72: row->get_string(0)->cmp(partial, src);
1.14 paf 73: return
74: partial==0 || // full match
75: partial==1; // typo left column starts 'src'
1.13 paf 76: }
77:
1.30 paf 78: /**
79: @test optimize whitespaces for all but 'html'
80: @todo fix theoretical \n mem overrun in TYPO replacements
81: */
1.31 ! paf 82: char *String::store_to(char *dest, Untaint_lang lang) const {
1.13 paf 83: // $MAIN:html-typo table
1.26 paf 84: Table *user_typo_table=static_cast<Table *>(pool().tag());
85: Table *typo_table=user_typo_table?user_typo_table:default_typo_table;
1.1 paf 86:
87: const Chunk *chunk=&head;
88: do {
89: const Chunk::Row *row=chunk->rows;
1.28 paf 90: for(size_t i=0; i<chunk->count; i++, row++) {
1.1 paf 91: if(row==append_here)
92: goto break2;
93:
94: // WARNING:
95: // string can grow only UNTAINT_TIMES_BIGGER
1.31 ! paf 96: switch(lang==UL_UNKNOWN?row->item.lang:lang) {
1.29 paf 97: case UL_CLEAN:
1.1 paf 98: // clean piece
1.29 paf 99: case UL_TAINTED:
1.1 paf 100: // tainted piece, but undefined untaint language
1.23 paf 101: // for VString.as_double of tainted values
1.1 paf 102: // for ^process{body} evaluation
1.11 paf 103: case UL_AS_IS:
1.1 paf 104: // tainted, untaint language: as-is
1.13 paf 105: memcpy(dest, row->item.ptr, row->item.size);
106: dest+=row->item.size;
1.1 paf 107: break;
1.11 paf 108: case UL_FILE_NAME:
1.9 paf 109: // tainted, untaint language: file [name]
1.18 paf 110: escape(switch(*src) {
111: case ' ': to_char('_'); break;
1.13 paf 112: encode(need_file_encode, '-');
1.18 paf 113: });
1.9 paf 114: break;
1.11 paf 115: case UL_URI:
1.4 paf 116: // tainted, untaint language: uri
1.18 paf 117: escape(switch(*src) {
118: case ' ': to_char('+'); break;
1.13 paf 119: encode(need_uri_encode, '%');
1.18 paf 120: });
1.5 paf 121: break;
1.11 paf 122: case UL_HEADER:
1.5 paf 123: // tainted, untaint language: header
1.18 paf 124: escape(switch(*src) {
1.13 paf 125: encode(need_header_encode, '%');
1.18 paf 126: });
1.4 paf 127: break;
1.11 paf 128: case UL_TABLE:
1.15 paf 129: // tainted, untaint language: table
1.18 paf 130: escape(switch(*src) {
131: case '\t': to_char(' '); break;
132: case '\n': to_char(' '); break;
1.13 paf 133: _default;
1.18 paf 134: });
1.1 paf 135: break;
1.11 paf 136: case UL_SQL:
1.1 paf 137: // tainted, untaint language: sql
138: // TODO: зависимость от sql сервера
1.13 paf 139: memset(dest, '?', row->item.size);
140: dest+=row->item.size;
1.1 paf 141: break;
1.11 paf 142: case UL_JS:
1.18 paf 143: escape(switch(*src) {
144: case '"': to_string("\\\"", 2); break;
145: case '\'': to_string("\\'", 2); break;
146: case '\n': to_string("\\n", 2); break;
147: case '\\': to_string("\\\\", 2); break;
148: case '\xFF': to_string("\\\xFF", 2); break;
1.13 paf 149: _default;
1.18 paf 150: });
1.1 paf 151: break;
1.11 paf 152: case UL_HTML:
1.18 paf 153: escape(switch(*src) {
154: case '&': to_string("&", 5); break;
155: case '>': to_string(">", 4); break;
156: case '<': to_string("<", 4); break;
157: case '"': to_string(""", 6); break;
1.19 paf 158: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 159: _default;
1.18 paf 160: });
1.1 paf 161: break;
1.13 paf 162: case UL_HTML_TYPO: {
1.1 paf 163: // tainted, untaint language: html-typo
1.19 paf 164: char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1);
165: // note:
166: // there still is a possibility that user
167: // would not replace \n as she supposed to
168: // and rather replace \ and n into huge strings
169: // thus causing memory overrun
170: // this can be dealed by allocating *2 memory, but that's too expensive
1.18 paf 171: size_t html_for_typo_size;
1.13 paf 172: { // local dest
1.18 paf 173: char *dest=html_for_typo;
174: escape(switch(*src) {
1.16 paf 175: // convinient name for typo match "\n"
176: case '\r':
1.18 paf 177: if(typo_table) {
178: *dest++='\\'; *dest++='n'; // \r -> \n
1.24 paf 179: if(src[1]=='\n') { // \r\n -> remove \n
180: size--; src++;
181: }
1.18 paf 182: }
183: break;
184: case '\n':
185: if(typo_table)
186: to_string("\\n", 2);
1.16 paf 187: break;
1.19 paf 188: //TODO: XSLT case '\'': to_string("'", 6); break;
1.13 paf 189: _default;
1.18 paf 190: });
1.13 paf 191: *dest=0;
1.18 paf 192: html_for_typo_size=dest-html_for_typo;
1.13 paf 193: }
194: // typo table replacements
1.21 paf 195: const char *src=html_for_typo;
196: do {
197: // there is a row where first column starts 'src'
198: if(Table::Item *item=typo_table->first_that(typo_present, src)) {
199: // get a=>b values
200: const String& a=*static_cast<Array *>(item)->get_string(0);
201: const String& b=*static_cast<Array *>(item)->get_string(1);
202: // empty 'a' | 'b' checks
203: if(a.size()==0 || b.size()==0) {
1.26 paf 204: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 205: THROW(0, 0,
206: typo_table->origin_string(),
207: "typo table column elements must not be empty");
208: }
209: // overflow check:
210: // b allowed to be max UNTAINT_TIMES_BIGGER then a
211: if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) {
1.26 paf 212: pool().set_tag(default_typo_table); // avoid recursion
1.21 paf 213: THROW(0, 0,
214: &b,
215: "is %g times longer then '%s', "
216: "while maximum, handled by Parser, is %d",
217: ((double)b.size())/a.size(),
218: a.cstr(),
219: UNTAINT_TIMES_BIGGER);
220: }
221:
222: // skip 'a' in 'src'
223: src+=a.size();
224: // write 'b' to 'dest'
225: b.store_to(dest);
226: dest+=b.size();
227: } else
228: *dest++=*src++;
229: } while(*src);
1.1 paf 230: break;
1.13 paf 231: }
1.1 paf 232: default:
1.18 paf 233: THROW(0, 0,
234: this,
1.1 paf 235: "unknown untaint language #%d of %d piece",
1.18 paf 236: static_cast<int>(row->item.lang),
1.1 paf 237: i);
238: }
239: }
240: chunk=row->link;
241: } while(chunk);
242: break2:
1.13 paf 243: return dest;
1.1 paf 244: }