Annotation of parser3/src/include/pa_string.h, revision 1.88
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.29 paf 4: Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
1.43 paf 5:
1.30 paf 6: Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
1.29 paf 7:
1.88 ! parser 8: $Id: pa_string.h,v 1.87 2001/05/14 13:18:07 parser Exp $
1.1 paf 9: */
10:
11: #ifndef PA_STRING_H
12: #define PA_STRING_H
13:
1.49 paf 14: #include "pa_config_includes.h"
1.9 paf 15:
1.73 paf 16: #include <string.h>
1.1 paf 17: #include <stddef.h>
18:
1.15 paf 19: #include "pa_pool.h"
1.4 paf 20: #include "pa_types.h"
21:
1.63 paf 22: class Table;
23:
1.50 paf 24: /**
25: $MAIN:html-typo table elements must enlarge string not more that that
26: that's a tradeoff - otherwise we'd have to scan string twice:
27: - first for buffer length
28: - second for replacements themselves
29: */
1.88 ! parser 30: #define UNTAINT_TIMES_BIGGER 20
1.31 paf 31:
1.9 paf 32: #ifndef NO_STRING_ORIGIN
1.33 paf 33: # define STRING_APPEND_PARAMS \
34: const char *src, size_t size, \
1.52 paf 35: String::Untaint_lang lang, \
1.33 paf 36: const char *file, uint line
1.54 paf 37: /// appends piece to String @see String::real_append
38: # define APPEND(src, size, lang, file, line) \
39: real_append(src, size, lang, file, line)
1.9 paf 40: #else
1.33 paf 41: # define STRING_APPEND_PARAMS \
42: const char *src, \
43: size_t size, \
1.52 paf 44: String::Untaint_lang lang
1.54 paf 45: /// appends piece to String @see String::real_append
46: # define APPEND(src, size, lang, file, line) \
47: real_append(src, size, lang)
48: #endif
1.83 paf 49: /// appends clean piece to String @see String::real_append
1.54 paf 50: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 51: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 52: /// appends piece to String as-is @see String::real_append
53: #define APPEND_AS_IS(src, size, file, line) \
54: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 55: /// appends tainted piece to String @see String::real_append
1.54 paf 56: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 57: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 58: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 59: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 60:
1.62 paf 61: class Array;
1.71 paf 62: class SQL_Connection;
1.62 paf 63:
1.42 paf 64: /**
1.41 paf 65: Pooled string.
66:
1.47 paf 67: Internal structure:
68: @verbatim
69: String Chunk0
70: ====== ========
71: head--------------->[ptr, size, ...]
72: append_here-------->[ptr, size, ...]
73: .
74: .
75: [ptr, size, ...]
76: link_row----------->[link to the next chunk]
1.41 paf 77: @endverbatim
78:
79: All pieces remember
80: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
81: - whether they are tainted or not,
82: and the language which should be used to detaint them
83: */
1.15 paf 84: class String : public Pooled {
1.1 paf 85: public:
1.48 paf 86:
1.1 paf 87: enum {
1.41 paf 88: CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
89: CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
1.1 paf 90: };
91:
1.41 paf 92: /// piece is tainted or not. the language to use when detaint
1.27 paf 93: enum Untaint_lang {
1.72 paf 94: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 95: UL_CLEAN, ///< clean
96: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 97: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 98: UL_PASS_APPENDED,
1.41 paf 99: /**<
100: leave language built into string being appended.
101: just a flag, that value not stored
102: */
1.48 paf 103: UL_AS_IS, ///< leave all characters intact
104: UL_FILE_NAME, ///< filename
1.74 paf 105: UL_HTTP_HEADER, ///< text in HTTP response header
106: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 107: UL_URI, ///< text in uri
108: UL_TABLE, ///< ^table:set body
109: UL_SQL, ///< ^table:sql body
110: UL_JS, ///< JavaScript code
111: UL_HTML, ///< HTML code (for editing)
1.85 paf 112: UL_USER_HTML ///< HTML code with USER chars
1.27 paf 113: };
114:
1.8 paf 115: public:
116:
1.77 paf 117: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 118: String(const String& src);
119: size_t size() const { return fsize; }
1.69 paf 120: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.82 paf 121: char *cstr(Untaint_lang lang=UL_UNSPECIFIED,
122: SQL_Connection *connection=0,
123: const char *charset=0) const {
1.88 ! parser 124:
1.50 paf 125: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
1.82 paf 126: char *eol=store_to(result, lang, connection, charset);
1.50 paf 127: *eol=0;
128: return result;
129: }
1.46 paf 130: /** append fragment
1.83 paf 131: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 132: */
1.9 paf 133: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 134: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 135: int cmp (int& partial, const String& src,
1.72 paf 136: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 137: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
138: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
139: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
140: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 141: bool operator == (const String& src) const {
142: if(size()!=src.size()) // can speed up in trivial case
143: return false;
1.56 paf 144: int p; return cmp(p, src)==0;
1.26 paf 145: }
1.56 paf 146: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 147:
1.50 paf 148: /**
149: @param partial
150: returns partial match status.
1.51 paf 151: - -1: strings too different
152: - 0: full match
153: - 1: means @c this starts @c src
154: - 2: means @src starts @this
1.50 paf 155: */
1.61 paf 156: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 157: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.50 paf 158: bool operator == (const char* src_ptr) const {
159: size_t src_size=src_ptr?strlen(src_ptr):0;
160: if(size() != src_size)
161: return false;
162: int partial; // unused
1.56 paf 163: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 164: }
1.80 paf 165: bool operator != (const char* src_ptr) const {
166: int partial; // unused
167: return cmp(partial, src_ptr, 0)!=0;
168: }
169:
1.50 paf 170:
1.42 paf 171: /**
172: appends other String.
1.41 paf 173:
1.47 paf 174: marking all tainted pieces of it with @a lang.
175: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 176: */
1.39 paf 177: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.76 paf 178: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
179: String& operator << (const char *src) { return APPEND_CONST(src); }
1.8 paf 180:
1.41 paf 181: /// simple hash code of string. used by Hash
1.14 paf 182: uint hash_code() const;
1.54 paf 183:
184: /// extracts [start, finish) piece of string
1.70 paf 185: String& mid(size_t start, size_t finish) const;
1.55 paf 186:
1.59 paf 187: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 188: int pos(const String& substr,
1.72 paf 189: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 190: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 paf 191: int pos(const char *substr, size_t substr_size,
1.72 paf 192: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 193:
194: void split(Array& result,
195: size_t *pos_after_ref,
196: const char *delim, size_t delim_size,
1.86 paf 197: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 198: void split(Array& result,
199: size_t *pos_after_ref,
200: const String& delim,
1.86 paf 201: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 202:
1.68 paf 203: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
204: void *info);
1.87 parser 205: /**
206: @return true if fills table.
207: table format is defined and fixed[can be used by others]:
208: @verbatim
209: prematch/match/postmatch/1/2/3/...
210: @endverbatim
211: */
1.81 paf 212: bool match(const unsigned char *pcre_tables,
213: const String *aorigin,
1.64 paf 214: const String& regexp,
1.65 paf 215: const String *options,
1.66 paf 216: Table **table,
217: Row_action row_action, void *info) const;
1.87 parser 218: enum Change_case_kind {
219: CC_UPPER,
220: CC_LOWER
221: };
222: String& change_case(Pool& pool, const unsigned char *pcre_tables,
223: Change_case_kind kind) const;
1.13 paf 224:
1.41 paf 225: #ifndef NO_STRING_ORIGIN
226: /// origin of string. calculated by first row
1.43 paf 227: const Origin& origin() const;
1.41 paf 228: #endif
1.8 paf 229:
1.10 paf 230: private:
231:
1.84 paf 232: /// several String fragments
1.1 paf 233: struct Chunk {
1.84 paf 234: size_t count; ///< the number of rows in chunk
235: /// string fragment or a link to next chunk union
1.1 paf 236: union Row {
1.84 paf 237: /// fragment
1.27 paf 238: struct {
1.84 paf 239: const char *ptr; ///< pointer to the start
240: size_t size; ///< length
241: Untaint_lang lang; ///< untaint flag, later untaint language
1.27 paf 242: #ifndef NO_STRING_ORIGIN
1.84 paf 243: Origin origin; ///< origin
1.27 paf 244: #endif
1.1 paf 245: } item;
1.84 paf 246: Chunk *link; ///< link to the next chunk in chain
1.2 paf 247: } rows[CR_PREALLOCATED_COUNT];
1.84 paf 248: Chunk *preallocated_link; ///< next rows are here
1.1 paf 249: }
1.84 paf 250: head; ///< the head chunk of the chunk chain
1.1 paf 251:
1.84 paf 252: /// next append would write to this record
1.1 paf 253: Chunk::Row *append_here;
254:
1.84 paf 255: /** the address of place where lies address
256: of the link to the next chunk to allocate
257: */
1.1 paf 258: Chunk::Row *link_row;
259:
1.5 paf 260: private:
1.25 paf 261: // last chunk
262: Chunk *last_chunk;
1.5 paf 263:
264: // string size
265: size_t fsize;
266:
267: // used rows in all chunks
268: int fused_rows;
269:
270: private:
1.1 paf 271:
272: bool chunk_is_full() {
273: return append_here == link_row;
274: }
275: void expand();
1.79 paf 276:
277: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 278: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
279: SQL_Connection *connection=0,
280: const char *charset=0) const;
1.7 paf 281:
282: private: //disabled
283:
1.12 paf 284: String& operator = (const String&) { return *this; }
1.7 paf 285:
1.1 paf 286: };
287:
288: #endif
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_string.h CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / include