![[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:
1.88 paf 4: Copyright(c) 2001, 2002 ArtLebedev Group(http://www.artlebedev.com)
1.89 ! paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.8 paf 6:
1.89 ! paf 7: $Id: untaint.C,v 1.88 2002/02/08 07:27:49 paf Exp $
1.1 paf 8: */
9:
10: #include "pa_pool.h"
11: #include "pa_string.h"
12: #include "pa_hash.h"
13: #include "pa_exception.h"
1.13 paf 14: #include "pa_table.h"
1.32 paf 15: #include "pa_globals.h"
1.34 paf 16: #include "pa_sql_connection.h"
1.58 parser 17: #include "pa_dictionary.h"
1.66 parser 18: #include "pa_common.h"
1.85 paf 19: #include "pa_charset.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) \
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]; \
1.60 parser 34: memcpy(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) \
1.60 parser 40: memcpy(dest, b, bsize); \
1.18 paf 41: dest+=bsize; \
1.4 paf 42:
1.9 paf 43: inline bool need_file_encode(unsigned char c){
1.83 paf 44: // theoretical problem with, for instance, "_2B" and "." fragments,
45: // they would yield the same
46: // because need_file_encode('_')=false
47: // but we need to delete such files somehow, getting names from ^index
48:
1.13 paf 49: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.9 paf 50: return false;
51:
1.31 paf 52: return !strchr(
1.86 paf 53: " _./()-"
1.31 paf 54: #ifdef WIN32
1.37 paf 55: ":\\~"
1.31 paf 56: #endif
1.83 paf 57: , c);
1.9 paf 58: }
1.5 paf 59: inline bool need_uri_encode(unsigned char c){
1.13 paf 60: if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z'))
1.4 paf 61: return false;
62:
1.5 paf 63: return !strchr("_-./", c);
64: }
1.36 paf 65: inline bool need_http_header_encode(unsigned char c){
1.18 paf 66: if(strchr(" , :", c))
1.5 paf 67: return false;
68:
69: return need_uri_encode(c);
1.4 paf 70: }
1.1 paf 71:
1.56 parser 72: //
73:
74: static const char * String_Untaint_lang_name[]={
75: "U", ///< zero value handy for hash lookup @see untaint_lang_name2enum
76: "C", ///< clean
77: "T", ///< tainted, untaint language as assigned later
78: // untaint languages. assigned by ^untaint[lang]{...}
79: "P",
80: /**<
81: leave language built into string being appended.
82: just a flag, that value not stored
83: */
84: "A", ///< leave all characters intact
1.68 parser 85: "F", ///< file specification
86: "H", ///< ext in HTTP response header
1.56 parser 87: "M", ///< text in mail header
88: "URI", ///< text in uri
89: "T", ///< ^table:set body
90: "SQL", ///< ^table:sql body
91: "JS", ///< JavaScript code
1.68 parser 92: "XML", ///< ^dom:set xml
1.82 paf 93: "HTML" ///< HTML code (for editing)
1.56 parser 94: };
95:
96:
1.1 paf 97: // String
98:
1.41 paf 99: /*
100:
101: HTTP-header = field-name ":" [ field-value ] CRLF
102:
103: field-name = token
104: field-value = *( field-content | LWS )
105:
106: field-content = <the OCTETs making up the field-value
107: and consisting of either *TEXT or combinations
108: of token, tspecials, and quoted-string>
109:
110:
111:
112: word = token | quoted-string
113:
114: token = 1*<any CHAR except CTLs or tspecials>
115:
116:
117:
118: tspecials = "(" | ")" | "<" | ">" | "@"
119: | "," | ";" | ":" | "\" | <">
120: | "/" | "[" | "]" | "?" | "="
121: | "{" | "}" | SP | HT
122:
123: SP = <US-ASCII SP, space (32)>
124: HT = <US-ASCII HT, horizontal-tab (9)>
125:
126: LWS = [CRLF] 1*( SP | HT )
127: TEXT = <any OCTET except CTLs,
128: but including LWS>
129:
130: quoted-pair = "\" CHAR
131:
132: if(strchr("()<>@,;:\\\"/[]?={} \t", *ptr))
133: */
134: inline bool need_quote_http_header(const char *ptr, size_t size) {
135: for(; size--; ptr++)
1.42 paf 136: if(strchr(";\\\"= \t" /* excluded ()<>@, :/ ? []{} */, *ptr))
1.41 paf 137: return true;
138: return false;
139: }
140:
1.77 paf 141: String& String::append(const String& src, uchar lang, bool forced) {
142: // manually unrolled code to avoid do{if(const)} constructs
143: if(forced)
144: STRING_SRC_FOREACH_ROW(
145: APPEND(row->item.ptr, row->item.size,
146: lang, //forcing passed lang
147: row->item.origin.file, row->item.origin.line);
148: )
149: else if(lang==UL_PASS_APPENDED)
150: STRING_SRC_FOREACH_ROW(
151: APPEND(row->item.ptr, row->item.size,
152: row->item.lang, // passing item's lang
153: row->item.origin.file, row->item.origin.line);
154: )
155: else if(lang&UL_OPTIMIZE_BIT) // main idea here
156: // tainted piece would get OPTIMIZED bit from 'lang'
157: // clean piece would be marked OPTIMIZED manually
158: // pieces with determined languages [not tainted|clean] would retain theirs langs
159: STRING_SRC_FOREACH_ROW(
160: APPEND(row->item.ptr, row->item.size,
161: row->item.lang==UL_TAINTED?lang:(
162: row->item.lang==UL_CLEAN?UL_CLEAN|UL_OPTIMIZE_BIT: // ORing with OPTIMIZED flag
163: row->item.lang
164: ),
165: row->item.origin.file, row->item.origin.line);
166: )
167: else
168: STRING_SRC_FOREACH_ROW(
169: APPEND(row->item.ptr, row->item.size,
170: row->item.lang==UL_TAINTED?lang:row->item.lang,
171: row->item.origin.file, row->item.origin.line);
172: );
173: break2:
174: return *this;
175: }
176:
1.75 paf 177: size_t String::cstr_bufsize(Untaint_lang lang,
178: SQL_Connection *connection,
1.87 paf 179: Charset *buf_charset) const {
1.77 paf 180: size_t dest=1; // for terminating 0
181: STRING_FOREACH_ROW(
182: uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang;
183:
184: switch(to_lang & ~UL_OPTIMIZE_BIT) {
185: case UL_CLEAN:
186: case UL_TAINTED:
187: case UL_AS_IS:
188: // clean piece
189:
190: // tainted piece, but undefined untaint language
191: // for VString.as_double of tainted values
192: // for ^process{body} evaluation
193:
194: // tainted, untaint language: as-is
195: dest+=row->item.size;
196: break;
197: case UL_FILE_SPEC:
198: // tainted, untaint language: file [name]
199: dest+=row->item.size*3/* worst: Z->%XX */;
200: break;
201: case UL_URI:
202: // tainted, untaint language: uri
1.84 paf 203: dest+=row->item.size*6*3/* worst utf8 x worst Z->%XX */;
1.77 paf 204: break;
205: case UL_HTTP_HEADER:
206: // tainted, untaint language: http-field-content-text
207: dest+=row->item.size*3/* worst: Z->%XX */;
208: break;
209: case UL_MAIL_HEADER:
210: // tainted, untaint language: mail-header
1.87 paf 211: if(buf_charset) {
1.77 paf 212: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 213: dest+=
214: row->item.size*3+
215: buf_charset->name().size()+MAX_STRING/* worst: =?charset?Q?=%XX?= */;
216: } else
1.75 paf 217: dest+=row->item.size;
1.77 paf 218: break;
219: case UL_TABLE:
220: // tainted, untaint language: table
221: dest+=row->item.size;
222: break;
223: case UL_SQL:
224: // tainted, untaint language: sql
225: if(connection)
226: dest+=connection->quote(0, row->item.ptr, row->item.size);
227: break;
228: case UL_JS:
229: escape(switch(*src) {
230: case '"': case '\'': case '\n': case '\\': case '\xFF':
231: dest+=2; break;
232: default:
233: dest++; break;
234: });
235: break;
236: case UL_XML:
237: escape(switch(*src) {
238: case '&': case '>': case '<': case '"': case '\'':
239: dest+= 6; break;
240: default:
241: dest++; break;
242: });
243: break;
244: case UL_HTML:
245: escape(switch(*src) {
246: case '&':
247: case '>':
248: case '<':
249: case '"':
250: dest+=6; break;
251: default:
252: dest++; break;
253: });
254: break;
1.75 paf 255: }
1.77 paf 256: );
1.75 paf 257: break2:
258: return dest;
1.51 parser 259: }
260:
1.43 paf 261: char *String::store_to(char *dest, Untaint_lang lang,
262: SQL_Connection *connection,
1.87 paf 263: Charset *store_to_charset) const {
1.75 paf 264: // WARNING:
265: // before any changes check cstr_bufsize first!!!
1.44 paf 266: bool whitespace=true;
1.79 paf 267: // expanded STRING_FOREACH_ROW here for debugging purposes
1.78 paf 268: const Chunk *chunk=&head; \
269: do { \
270: const Chunk::Row *row=chunk->rows; \
271: for(uint i=0; i<chunk->count; i++, row++) { \
272: if(row==append_here) \
273: goto break2; \
274: \
1.77 paf 275: uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang;
276:
277: char *start=dest;
278:
279: switch(to_lang & ~UL_OPTIMIZE_BIT) {
280: case UL_CLEAN:
281: case UL_TAINTED:
282: case UL_AS_IS:
283: // clean piece
284:
285: // tainted piece, but undefined untaint language
286: // for VString.as_double of tainted values
287: // for ^process{body} evaluation
288:
289: // tainted, untaint language: as-is
290: memcpy(dest, row->item.ptr, row->item.size);
291: dest+=row->item.size;
292: break;
293: case UL_FILE_SPEC:
294: // tainted, untaint language: file [name]
1.83 paf 295: escape(
296: encode(need_file_encode, '_');
297: );
1.77 paf 298: break;
299: case UL_URI:
300: // tainted, untaint language: uri
1.85 paf 301: const void *client_ptr;
302: size_t client_size;
303: Charset::transcode(pool(),
304: pool().get_source_charset(), row->item.ptr, row->item.size,
305: pool().get_client_charset(), client_ptr, client_size);
306: {
307: const char *src=(const char *)client_ptr;
308: for(int size=client_size; size--; src++)
309: switch(*src) {
310: case ' ': to_char('+'); break;
311: default: encode(need_uri_encode, '%');
312: };
313: }
1.77 paf 314: break;
315: case UL_HTTP_HEADER:
316: // tainted, untaint language: http-field-content-text
317: escape(switch(*src) {
318: case ' ': to_char('+'); break;
1.83 paf 319: default: encode(need_uri_encode, '%');
1.77 paf 320: });
321: break;
322: case UL_MAIL_HEADER:
323: // tainted, untaint language: mail-header
1.87 paf 324: if(store_to_charset) {
325: const void *mail_ptr;
326: size_t mail_size;
327: Charset::transcode(pool(),
328: pool().get_source_charset(), row->item.ptr, row->item.size,
329: *store_to_charset, mail_ptr, mail_size);
330:
1.77 paf 331: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
1.87 paf 332: const char *src=(const char *)mail_ptr;
1.77 paf 333: bool to_quoted_printable=false;
1.87 paf 334: for(int size=mail_size; size--; src++) {
1.77 paf 335: if(*src & 0x80) {
336: if(!to_quoted_printable) {
1.87 paf 337: dest+=sprintf(dest, "=?%s?Q?", store_to_charset->name().cstr());
1.77 paf 338: to_quoted_printable=true;
339: }
340: dest+=sprintf(dest, "=%02X", *src & 0xFF);
341: } else {
342: *dest++=*src;
343: }
1.44 paf 344: }
1.77 paf 345: if(to_quoted_printable) // close
346: dest+=sprintf(dest, "?=");
1.87 paf 347:
1.77 paf 348: } else {
1.13 paf 349: memcpy(dest, row->item.ptr, row->item.size);
350: dest+=row->item.size;
1.1 paf 351: }
1.77 paf 352: break;
353: case UL_TABLE:
354: // tainted, untaint language: table
355: escape(switch(*src) {
356: case '\t': to_char(' '); break;
357: case '\n': to_char(' '); break;
358: _default;
359: });
360: break;
361: case UL_SQL:
362: // tainted, untaint language: sql
363: if(connection)
364: dest+=connection->quote(dest, row->item.ptr, row->item.size);
365: else
366: throw Exception(0, 0,
367: this,
368: "untaint in SQL language failed - no connection specified");
369: break;
370: case UL_JS:
371: escape(switch(*src) {
372: case '"': to_string("\\\"", 2); break;
373: case '\'': to_string("\\'", 2); break;
374: case '\n': to_string("\\n", 2); break;
375: case '\\': to_string("\\\\", 2); break;
376: case '\xFF': to_string("\\\xFF", 2); break;
377: _default;
378: });
379: break;
380: case UL_XML:
381: escape(switch(*src) {
382: case '&': to_string("&", 5); break;
383: case '>': to_string(">", 4); break;
384: case '<': to_string("<", 4); break;
385: case '"': to_string(""", 6); break;
386: case '\'': to_string("'", 6); break;
387: _default;
388: });
389: break;
390: case UL_HTML:
391: escape(switch(*src) {
392: case '&': to_string("&", 5); break;
393: case '>': to_string(">", 4); break;
394: case '<': to_string("<", 4); break;
395: case '"': to_string(""", 6); break;
396: _default;
397: });
398: break;
399: default:
400: throw Exception(0, 0,
401: this,
1.81 paf 402: "unknown untaint language #%d",
403: static_cast<int>(row->item.lang)); // sould never
1.77 paf 404: break; // never
1.76 paf 405: }
1.55 parser 406:
1.77 paf 407: if(to_lang & UL_OPTIMIZE_BIT) {
408: // optimizing whitespace
409: char *stop=dest; dest=start;
410: for(char *src=start; src<stop; src++)
411: switch(*src) {
412: // of all consequent white space chars leaving only first one
1.80 paf 413: case ' ': case '\r': case '\n': case '\t':
1.77 paf 414: if(!whitespace) {
415: *dest++=*src;
416: whitespace=true;
417: }
418: break;
419: default:
420: whitespace=false;
421: *dest++=*src;
422: break;
423: };
424: } else // piece without optimization
425: whitespace=false;
1.78 paf 426:
427: } \
428: chunk=row->link; \
429: } while(chunk); \
430:
1.76 paf 431: break2:
432: return dest;
433: }
434:
435: char *String::cstr_debug_origins() const {
1.81 paf 436: //_asm int 3;
1.76 paf 437: char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2);
438: char *dest=result;
439:
440: const Chunk *chunk=&head;
441: do {
442: const Chunk::Row *row=chunk->rows;
443: for(uint i=0; i<chunk->count; i++, row++) {
444: if(row==append_here)
445: goto break2;
1.55 parser 446:
1.76 paf 447: #ifndef NO_STRING_ORIGIN
448: if(row->item.origin.file)
449: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
450: row->item.origin.file,
451: 1+row->item.origin.line);
452: else
453: dest+=sprintf(dest, "<unknown>");
454: #endif
1.81 paf 455: uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT;
456: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
457: throw Exception(0, 0,
458: this,
459: "unknown untaint language #%d",
460: static_cast<int>(show_lang)); // sould never
461:
462: dest+=sprintf(dest, "#%s%s: ",
463: String_Untaint_lang_name[show_lang],
464: row->item.lang & UL_OPTIMIZE_BIT?".O":"");
1.76 paf 465: char *dest_after_origins=dest;
466:
467: memcpy(dest, row->item.ptr, row->item.size);
468: dest+=row->item.size;
469:
470: remove_crlf(dest_after_origins, dest);
471: to_char('\n');
1.1 paf 472: }
473: chunk=row->link;
474: } while(chunk);
1.64 parser 475:
1.1 paf 476: break2:
1.76 paf 477: *dest=0;
478: return result;
1.1 paf 479: }