![[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)
1.74 paf 5: Author: Alexander Petrosyan <paf@design.ru>(http://paf.design.ru)
1.8 paf 6:
1.85 ! paf 7: $Id: untaint.C,v 1.84 2001/12/15 21:28:22 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(
1.83 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,
179: const char *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
211: if(charset) {
212: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
213: dest+=row->item.size*3+MAX_STRING/* worst: =?charset?Q?=%XX?= */;
214: } else {
1.75 paf 215: dest+=row->item.size;
216: }
1.77 paf 217: break;
218: case UL_TABLE:
219: // tainted, untaint language: table
220: dest+=row->item.size;
221: break;
222: case UL_SQL:
223: // tainted, untaint language: sql
224: if(connection)
225: dest+=connection->quote(0, row->item.ptr, row->item.size);
226: break;
227: case UL_JS:
228: escape(switch(*src) {
229: case '"': case '\'': case '\n': case '\\': case '\xFF':
230: dest+=2; break;
231: default:
232: dest++; break;
233: });
234: break;
235: case UL_XML:
236: escape(switch(*src) {
237: case '&': case '>': case '<': case '"': case '\'':
238: dest+= 6; break;
239: default:
240: dest++; break;
241: });
242: break;
243: case UL_HTML:
244: escape(switch(*src) {
245: case '&':
246: case '>':
247: case '<':
248: case '"':
249: dest+=6; break;
250: default:
251: dest++; break;
252: });
253: break;
1.75 paf 254: }
1.77 paf 255: );
1.75 paf 256: break2:
257: return dest;
1.51 parser 258: }
259:
1.43 paf 260: char *String::store_to(char *dest, Untaint_lang lang,
261: SQL_Connection *connection,
262: const char *charset) const {
1.75 paf 263: // WARNING:
264: // before any changes check cstr_bufsize first!!!
1.44 paf 265: bool whitespace=true;
1.79 paf 266: // expanded STRING_FOREACH_ROW here for debugging purposes
1.78 paf 267: const Chunk *chunk=&head; \
268: do { \
269: const Chunk::Row *row=chunk->rows; \
270: for(uint i=0; i<chunk->count; i++, row++) { \
271: if(row==append_here) \
272: goto break2; \
273: \
1.77 paf 274: uchar to_lang=lang==UL_UNSPECIFIED?row->item.lang:lang;
275:
276: char *start=dest;
277:
278: switch(to_lang & ~UL_OPTIMIZE_BIT) {
279: case UL_CLEAN:
280: case UL_TAINTED:
281: case UL_AS_IS:
282: // clean piece
283:
284: // tainted piece, but undefined untaint language
285: // for VString.as_double of tainted values
286: // for ^process{body} evaluation
287:
288: // tainted, untaint language: as-is
289: memcpy(dest, row->item.ptr, row->item.size);
290: dest+=row->item.size;
291: break;
292: case UL_FILE_SPEC:
293: // tainted, untaint language: file [name]
1.83 paf 294: escape(
295: encode(need_file_encode, '_');
296: );
1.77 paf 297: break;
298: case UL_URI:
299: // tainted, untaint language: uri
1.85 ! paf 300: const void *client_ptr;
! 301: size_t client_size;
! 302: Charset::transcode(pool(),
! 303: pool().get_source_charset(), row->item.ptr, row->item.size,
! 304: pool().get_client_charset(), client_ptr, client_size);
! 305: {
! 306: const char *src=(const char *)client_ptr;
! 307: for(int size=client_size; size--; src++)
! 308: switch(*src) {
! 309: case ' ': to_char('+'); break;
! 310: default: encode(need_uri_encode, '%');
! 311: };
! 312: }
1.77 paf 313: break;
314: case UL_HTTP_HEADER:
315: // tainted, untaint language: http-field-content-text
316: escape(switch(*src) {
317: case ' ': to_char('+'); break;
1.83 paf 318: default: encode(need_uri_encode, '%');
1.77 paf 319: });
320: break;
321: case UL_MAIL_HEADER:
322: // tainted, untaint language: mail-header
323: if(charset) {
324: // Subject: Re: parser3: =?koi8-r?Q?=D3=C5=CD=C9=CE=C1=D2?=
325: const char *src=row->item.ptr;
326: bool to_quoted_printable=false;
327: for(int size=row->item.size; size--; src++) {
328: if(*src & 0x80) {
329: if(!to_quoted_printable) {
330: dest+=sprintf(dest, "=?%.15s?Q?", charset);
331: to_quoted_printable=true;
332: }
333: dest+=sprintf(dest, "=%02X", *src & 0xFF);
334: } else {
335: *dest++=*src;
336: }
1.44 paf 337: }
1.77 paf 338: if(to_quoted_printable) // close
339: dest+=sprintf(dest, "?=");
340: } else {
1.13 paf 341: memcpy(dest, row->item.ptr, row->item.size);
342: dest+=row->item.size;
1.1 paf 343: }
1.77 paf 344: break;
345: case UL_TABLE:
346: // tainted, untaint language: table
347: escape(switch(*src) {
348: case '\t': to_char(' '); break;
349: case '\n': to_char(' '); break;
350: _default;
351: });
352: break;
353: case UL_SQL:
354: // tainted, untaint language: sql
355: if(connection)
356: dest+=connection->quote(dest, row->item.ptr, row->item.size);
357: else
358: throw Exception(0, 0,
359: this,
360: "untaint in SQL language failed - no connection specified");
361: break;
362: case UL_JS:
363: escape(switch(*src) {
364: case '"': to_string("\\\"", 2); break;
365: case '\'': to_string("\\'", 2); break;
366: case '\n': to_string("\\n", 2); break;
367: case '\\': to_string("\\\\", 2); break;
368: case '\xFF': to_string("\\\xFF", 2); break;
369: _default;
370: });
371: break;
372: case UL_XML:
373: escape(switch(*src) {
374: case '&': to_string("&", 5); break;
375: case '>': to_string(">", 4); break;
376: case '<': to_string("<", 4); break;
377: case '"': to_string(""", 6); break;
378: case '\'': to_string("'", 6); break;
379: _default;
380: });
381: break;
382: case UL_HTML:
383: escape(switch(*src) {
384: case '&': to_string("&", 5); break;
385: case '>': to_string(">", 4); break;
386: case '<': to_string("<", 4); break;
387: case '"': to_string(""", 6); break;
388: _default;
389: });
390: break;
391: default:
392: throw Exception(0, 0,
393: this,
1.81 paf 394: "unknown untaint language #%d",
395: static_cast<int>(row->item.lang)); // sould never
1.77 paf 396: break; // never
1.76 paf 397: }
1.55 parser 398:
1.77 paf 399: if(to_lang & UL_OPTIMIZE_BIT) {
400: // optimizing whitespace
401: char *stop=dest; dest=start;
402: for(char *src=start; src<stop; src++)
403: switch(*src) {
404: // of all consequent white space chars leaving only first one
1.80 paf 405: case ' ': case '\r': case '\n': case '\t':
1.77 paf 406: if(!whitespace) {
407: *dest++=*src;
408: whitespace=true;
409: }
410: break;
411: default:
412: whitespace=false;
413: *dest++=*src;
414: break;
415: };
416: } else // piece without optimization
417: whitespace=false;
1.78 paf 418:
419: } \
420: chunk=row->link; \
421: } while(chunk); \
422:
1.76 paf 423: break2:
424: return dest;
425: }
426:
427: char *String::cstr_debug_origins() const {
1.81 paf 428: //_asm int 3;
1.76 paf 429: char *result=(char *)malloc(size()+used_rows()*MAX_STRING*2);
430: char *dest=result;
431:
432: const Chunk *chunk=&head;
433: do {
434: const Chunk::Row *row=chunk->rows;
435: for(uint i=0; i<chunk->count; i++, row++) {
436: if(row==append_here)
437: goto break2;
1.55 parser 438:
1.76 paf 439: #ifndef NO_STRING_ORIGIN
440: if(row->item.origin.file)
441: dest+=sprintf(dest, ORIGIN_FILE_LINE_FORMAT,
442: row->item.origin.file,
443: 1+row->item.origin.line);
444: else
445: dest+=sprintf(dest, "<unknown>");
446: #endif
1.81 paf 447: uchar show_lang=row->item.lang & ~UL_OPTIMIZE_BIT;
448: if(show_lang>=sizeof(String_Untaint_lang_name)/sizeof(String_Untaint_lang_name[0]))
449: throw Exception(0, 0,
450: this,
451: "unknown untaint language #%d",
452: static_cast<int>(show_lang)); // sould never
453:
454: dest+=sprintf(dest, "#%s%s: ",
455: String_Untaint_lang_name[show_lang],
456: row->item.lang & UL_OPTIMIZE_BIT?".O":"");
1.76 paf 457: char *dest_after_origins=dest;
458:
459: memcpy(dest, row->item.ptr, row->item.size);
460: dest+=row->item.size;
461:
462: remove_crlf(dest_after_origins, dest);
463: to_char('\n');
1.1 paf 464: }
465: chunk=row->link;
466: } while(chunk);
1.64 parser 467:
1.1 paf 468: break2:
1.76 paf 469: *dest=0;
470: return result;
1.1 paf 471: }