Annotation of parser3/src/include/pa_string.h, revision 1.52
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:
1.52 ! paf 8: $Id: pa_string.h,v 1.51 2001/03/24 19:30:06 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
1.48 paf 34: /// appends clean piece to String @see String::real_append
1.52 ! paf 35: # define APPEND(src, size, file, line) \
! 36: real_append(src, size, String::UL_NO, file, line)
1.48 paf 37: /// appends tainted piece to String @see String::real_append
1.52 ! paf 38: # define APPEND_TAINTED(src, size, file, line) \
! 39: real_append(src, size, String::UL_YES, file, line)
! 40: # define APPEND_SPECIFIC_TAINTED(src, size, lang, file, line) \
! 41: real_append(src, size, lang, file, line)
1.9 paf 42: #else
1.33 paf 43: # define STRING_APPEND_PARAMS \
44: const char *src, \
45: size_t size, \
1.52 ! paf 46: String::Untaint_lang lang
1.48 paf 47: /// appends clean piece to String @see String::real_append
1.52 ! paf 48: # define APPEND(src, size, file, line) \
! 49: real_append(src, size, String::UL_NO)
1.48 paf 50: /// appends tainted piece to String @see String::real_append
1.52 ! paf 51: # define APPEND_TAINTED(src, size, file, line) \
! 52: real_append(src, size, String::UL_YES)
! 53: # define APPEND_SPECIFIC_TAINTED(src, size, lang, file, line) \
! 54: real_append(src, size, lang)
1.9 paf 55: #endif
1.48 paf 56: /// handy: appends const char* piece to String @see String::real_append
1.23 paf 57: #define APPEND_CONST(src) APPEND(src, 0, 0, 0)
1.9 paf 58:
1.42 paf 59: /**
1.41 paf 60: Pooled string.
61:
1.47 paf 62: Internal structure:
63: @verbatim
64: String Chunk0
65: ====== ========
66: head--------------->[ptr, size, ...]
67: append_here-------->[ptr, size, ...]
68: .
69: .
70: [ptr, size, ...]
71: link_row----------->[link to the next chunk]
1.41 paf 72: @endverbatim
73:
74: All pieces remember
75: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
76: - whether they are tainted or not,
77: and the language which should be used to detaint them
78: */
1.15 paf 79: class String : public Pooled {
1.1 paf 80: public:
1.48 paf 81:
1.1 paf 82: enum {
1.41 paf 83: CR_PREALLOCATED_COUNT=5, ///< default preallocated item count
84: CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded()
1.1 paf 85: };
86:
1.41 paf 87: /// piece is tainted or not. the language to use when detaint
1.27 paf 88: enum Untaint_lang {
1.48 paf 89: UL_UNKNOWN=0, ///< when get by name fails
90: UL_NO, ///< clean
91: UL_YES, ///< tainted, untaint language as assigned later
1.27 paf 92: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 93: UL_PASS_APPENDED,
1.41 paf 94: /**<
95: leave language built into string being appended.
96: just a flag, that value not stored
97: */
1.48 paf 98: UL_AS_IS, ///< leave all characters intact
99: UL_FILE_NAME, ///< filename
100: UL_HEADER, ///< text in response header
101: UL_URI, ///< text in uri
102: UL_TABLE, ///< ^table:set body
103: UL_SQL, ///< ^table:sql body
104: UL_JS, ///< JavaScript code
105: UL_HTML, ///< HTML code (for editing)
106: UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing)
1.27 paf 107: };
108:
1.8 paf 109: public:
110:
1.37 paf 111: String(Pool& apool, const char *src=0, bool tainted=false);
1.14 paf 112: String(const String& src);
113: size_t size() const { return fsize; }
1.41 paf 114: /// convert to C string
1.50 paf 115: char *cstr() const {
116: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
117: char *eol=store_to(result);
118: *eol=0;
119: return result;
120: }
1.46 paf 121: /** append fragment
122: @see APPEND, APPEND_TAINTED, APPEND_CONST
123: */
1.9 paf 124: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 125: /// @return <0 ==0 or >0 depending on comparison result
1.26 paf 126: int cmp (const String& src) const;
127: bool operator < (const String& src) const { return cmp(src)<0; }
128: bool operator > (const String& src) const { return cmp(src)>0; }
129: bool operator <= (const String& src) const { return cmp(src)<=0; }
130: bool operator >= (const String& src) const { return cmp(src)>=0; }
131: bool operator == (const String& src) const {
132: if(size()!=src.size()) // can speed up in trivial case
133: return false;
134: return cmp(src)==0;
135: }
136: bool operator != (const String& src) const { return cmp(src)!=0; }
137:
1.50 paf 138: /**
139: @param partial
140: returns partial match status.
1.51 paf 141: - -1: strings too different
142: - 0: full match
143: - 1: means @c this starts @c src
144: - 2: means @src starts @this
1.50 paf 145: */
146: int cmp(const char* src_ptr, int& partial, size_t src_size=0) const;
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
152: return cmp(src_ptr, partial, src_size)==0;
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;
1.13 paf 165:
1.41 paf 166: #ifndef NO_STRING_ORIGIN
167: /// origin of string. calculated by first row
1.43 paf 168: const Origin& origin() const;
1.41 paf 169: #endif
1.8 paf 170:
1.10 paf 171: private:
172:
1.1 paf 173: struct Chunk {
1.6 paf 174: // the number of rows in chunk
1.1 paf 175: int count;
176: union Row {
1.27 paf 177: // fragment
178: struct {
179: const char *ptr; // pointer to the start
180: size_t size; // length
181: Untaint_lang lang; // untaint flag, later untaint language
182: #ifndef NO_STRING_ORIGIN
183: Origin origin; // origin
184: #endif
1.1 paf 185: } item;
186: Chunk *link; // link to the next chunk in chain
1.2 paf 187: } rows[CR_PREALLOCATED_COUNT];
1.1 paf 188: // next rows are here
189: Chunk *preallocated_link;
190: }
191: head; // the head chunk of the chunk chain
192:
193: // next append would write to this record
194: Chunk::Row *append_here;
195:
196: // the address of place where lies address
197: // of the link to the next chunk to allocate
198: Chunk::Row *link_row;
199:
1.5 paf 200: private:
1.25 paf 201: // last chunk
202: Chunk *last_chunk;
1.5 paf 203:
204: // string size
205: size_t fsize;
206:
207: // used rows in all chunks
208: int fused_rows;
209:
210: private:
1.1 paf 211:
212: bool chunk_is_full() {
213: return append_here == link_row;
214: }
215: void expand();
1.39 paf 216: void set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size);
1.50 paf 217: char *String::store_to(char *dest) const;
1.7 paf 218:
219: private: //disabled
220:
1.12 paf 221: String& operator = (const String&) { return *this; }
1.7 paf 222:
1.1 paf 223: };
224:
225: #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