Annotation of parser3/src/include/pa_string.h, revision 1.62
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:
1.62 ! paf 8: $Id: pa_string.h,v 1.61 2001/04/02 15:59:35 paf 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:
1.1 paf 16: #include <stddef.h>
17:
1.15 paf 18: #include "pa_pool.h"
1.4 paf 19: #include "pa_types.h"
20:
1.50 paf 21: /**
22: $MAIN:html-typo table elements must enlarge string not more that that
23: that's a tradeoff - otherwise we'd have to scan string twice:
24: - first for buffer length
25: - second for replacements themselves
26: */
1.31 paf 27: #define UNTAINT_TIMES_BIGGER 10
28:
1.9 paf 29: #ifndef NO_STRING_ORIGIN
1.33 paf 30: # define STRING_APPEND_PARAMS \
31: const char *src, size_t size, \
1.52 paf 32: String::Untaint_lang lang, \
1.33 paf 33: const char *file, uint line
1.54 paf 34: /// appends piece to String @see String::real_append
35: # define APPEND(src, size, lang, file, line) \
36: real_append(src, size, lang, file, line)
1.9 paf 37: #else
1.33 paf 38: # define STRING_APPEND_PARAMS \
39: const char *src, \
40: size_t size, \
1.52 paf 41: String::Untaint_lang lang
1.54 paf 42: /// appends piece to String @see String::real_append
43: # define APPEND(src, size, lang, file, line) \
44: real_append(src, size, lang)
45: #endif
1.48 paf 46: /// appends clean piece to String @see String::real_append
1.54 paf 47: #define APPEND_CLEAN(src, size, file, line) \
1.62 ! paf 48: APPEND(src, size, String::UL_CLEAN, file, line)
1.48 paf 49: /// appends tainted piece to String @see String::real_append
1.54 paf 50: #define APPEND_TAINTED(src, size, file, line) \
1.62 ! paf 51: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 52: /// handy: appends const char* piece to String @see String::real_append
1.54 paf 53: #define APPEND_CONST(src) APPEND_CLEAN(src, 0, 0, 0)
1.9 paf 54:
1.62 ! paf 55: class Array;
! 56:
1.42 paf 57: /**
1.41 paf 58: Pooled string.
59:
1.47 paf 60: Internal structure:
61: @verbatim
62: String Chunk0
63: ====== ========
64: head--------------->[ptr, size, ...]
65: append_here-------->[ptr, size, ...]
66: .
67: .
68: [ptr, size, ...]
69: link_row----------->[link to the next chunk]
1.41 paf 70: @endverbatim
71:
72: All pieces remember
73: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
74: - whether they are tainted or not,
75: and the language which should be used to detaint them
76: */
1.15 paf 77: class String : public Pooled {
1.1 paf 78: public:
1.48 paf 79:
1.1 paf 80: enum {
1.41 paf 81: CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
82: CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
1.1 paf 83: };
84:
1.41 paf 85: /// piece is tainted or not. the language to use when detaint
1.27 paf 86: enum Untaint_lang {
1.60 paf 87: UL_UNKNOWN=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 ! paf 88: UL_CLEAN, ///< clean
! 89: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 90: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 91: UL_PASS_APPENDED,
1.41 paf 92: /**<
93: leave language built into string being appended.
94: just a flag, that value not stored
95: */
1.48 paf 96: UL_AS_IS, ///< leave all characters intact
97: UL_FILE_NAME, ///< filename
98: UL_HEADER, ///< text in response header
99: UL_URI, ///< text in uri
100: UL_TABLE, ///< ^table:set body
101: UL_SQL, ///< ^table:sql body
102: UL_JS, ///< JavaScript code
103: UL_HTML, ///< HTML code (for editing)
104: UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing)
1.27 paf 105: };
106:
1.8 paf 107: public:
108:
1.37 paf 109: String(Pool& apool, const char *src=0, bool tainted=false);
1.14 paf 110: String(const String& src);
111: size_t size() const { return fsize; }
1.41 paf 112: /// convert to C string
1.50 paf 113: char *cstr() const {
114: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
115: char *eol=store_to(result);
116: *eol=0;
117: return result;
118: }
1.46 paf 119: /** append fragment
1.54 paf 120: @see APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 121: */
1.9 paf 122: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 123: /// @return <0 ==0 or >0 depending on comparison result
1.62 ! paf 124: int cmp (int& partial, const String& src,
! 125: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.56 paf 126: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
127: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
128: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
129: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 130: bool operator == (const String& src) const {
131: if(size()!=src.size()) // can speed up in trivial case
132: return false;
1.56 paf 133: int p; return cmp(p, src)==0;
1.26 paf 134: }
1.56 paf 135: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 136:
1.50 paf 137: /**
138: @param partial
139: returns partial match status.
1.51 paf 140: - -1: strings too different
141: - 0: full match
142: - 1: means @c this starts @c src
143: - 2: means @src starts @this
1.50 paf 144: */
1.61 paf 145: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.62 ! paf 146: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.50 paf 147: bool operator == (const char* src_ptr) const {
148: size_t src_size=src_ptr?strlen(src_ptr):0;
149: if(size() != src_size)
150: return false;
151: int partial; // unused
1.56 paf 152: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 153: }
154:
1.42 paf 155: /**
156: appends other String.
1.41 paf 157:
1.47 paf 158: marking all tainted pieces of it with @a lang.
159: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 160: */
1.39 paf 161: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.8 paf 162:
1.41 paf 163: /// simple hash code of string. used by Hash
1.14 paf 164: uint hash_code() const;
1.54 paf 165:
166: /// extracts [start, finish) piece of string
167: String& piece(size_t start, size_t finish) const;
1.55 paf 168:
1.59 paf 169: /// @return position of substr in string, -1 means "not found" [String version]
1.62 ! paf 170: int pos(const String& substr,
! 171: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.59 paf 172: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 ! paf 173: int pos(const char *substr, size_t substr_size,
! 174: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
! 175:
! 176: void split(Array& result,
! 177: size_t *pos_after_ref,
! 178: const char *delim, size_t delim_size,
! 179: Untaint_lang lang, int limit=-1) const;
! 180: void split(Array& result,
! 181: size_t *pos_after_ref,
! 182: const String& delim,
! 183: Untaint_lang lang, int limit=-1) const;
! 184:
1.13 paf 185:
1.41 paf 186: #ifndef NO_STRING_ORIGIN
187: /// origin of string. calculated by first row
1.43 paf 188: const Origin& origin() const;
1.41 paf 189: #endif
1.8 paf 190:
1.10 paf 191: private:
192:
1.1 paf 193: struct Chunk {
1.6 paf 194: // the number of rows in chunk
1.56 paf 195: size_t count;
1.1 paf 196: union Row {
1.27 paf 197: // fragment
198: struct {
199: const char *ptr; // pointer to the start
200: size_t size; // length
201: Untaint_lang lang; // untaint flag, later untaint language
202: #ifndef NO_STRING_ORIGIN
203: Origin origin; // origin
204: #endif
1.1 paf 205: } item;
206: Chunk *link; // link to the next chunk in chain
1.2 paf 207: } rows[CR_PREALLOCATED_COUNT];
1.1 paf 208: // next rows are here
209: Chunk *preallocated_link;
210: }
211: head; // the head chunk of the chunk chain
212:
213: // next append would write to this record
214: Chunk::Row *append_here;
215:
216: // the address of place where lies address
217: // of the link to the next chunk to allocate
218: Chunk::Row *link_row;
219:
1.5 paf 220: private:
1.25 paf 221: // last chunk
222: Chunk *last_chunk;
1.5 paf 223:
224: // string size
225: size_t fsize;
226:
227: // used rows in all chunks
228: int fused_rows;
229:
230: private:
1.1 paf 231:
232: bool chunk_is_full() {
233: return append_here == link_row;
234: }
235: void expand();
1.50 paf 236: char *String::store_to(char *dest) const;
1.7 paf 237:
238: private: //disabled
239:
1.12 paf 240: String& operator = (const String&) { return *this; }
1.7 paf 241:
1.1 paf 242: };
243:
244: #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