Annotation of parser3/src/include/pa_string.h, revision 1.69
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:
1.69 ! paf 8: $Id: pa_string.h,v 1.68 2001/04/03 16:34:27 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:
1.63 paf 21: class Table;
22:
1.50 paf 23: /**
24: $MAIN:html-typo table elements must enlarge string not more that that
25: that's a tradeoff - otherwise we'd have to scan string twice:
26: - first for buffer length
27: - second for replacements themselves
28: */
1.31 paf 29: #define UNTAINT_TIMES_BIGGER 10
30:
1.9 paf 31: #ifndef NO_STRING_ORIGIN
1.33 paf 32: # define STRING_APPEND_PARAMS \
33: const char *src, size_t size, \
1.52 paf 34: String::Untaint_lang lang, \
1.33 paf 35: const char *file, uint line
1.54 paf 36: /// appends piece to String @see String::real_append
37: # define APPEND(src, size, lang, file, line) \
38: real_append(src, size, lang, file, line)
1.9 paf 39: #else
1.33 paf 40: # define STRING_APPEND_PARAMS \
41: const char *src, \
42: size_t size, \
1.52 paf 43: String::Untaint_lang lang
1.54 paf 44: /// appends piece to String @see String::real_append
45: # define APPEND(src, size, lang, file, line) \
46: real_append(src, size, lang)
47: #endif
1.48 paf 48: /// appends clean piece to String @see String::real_append
1.54 paf 49: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 50: APPEND(src, size, String::UL_CLEAN, file, line)
1.48 paf 51: /// appends tainted piece to String @see String::real_append
1.54 paf 52: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 53: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 54: /// handy: appends const char* piece to String @see String::real_append
1.54 paf 55: #define APPEND_CONST(src) APPEND_CLEAN(src, 0, 0, 0)
1.9 paf 56:
1.62 paf 57: class Array;
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 {
1.60 paf 89: UL_UNKNOWN=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 90: UL_CLEAN, ///< clean
91: UL_TAINTED, ///< 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; }
1.69 ! paf 114: /// convert to C string. if 'lang' known, forcing 'lang' to it
! 115: char *cstr(Untaint_lang lang=UL_UNKNOWN) const {
1.50 paf 116: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
1.69 ! paf 117: char *eol=store_to(result, lang);
1.50 paf 118: *eol=0;
119: return result;
120: }
1.46 paf 121: /** append fragment
1.54 paf 122: @see APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 123: */
1.9 paf 124: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 125: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 126: int cmp (int& partial, const String& src,
127: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.56 paf 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; }
130: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
131: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 132: bool operator == (const String& src) const {
133: if(size()!=src.size()) // can speed up in trivial case
134: return false;
1.56 paf 135: int p; return cmp(p, src)==0;
1.26 paf 136: }
1.56 paf 137: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 138:
1.50 paf 139: /**
140: @param partial
141: returns partial match status.
1.51 paf 142: - -1: strings too different
143: - 0: full match
144: - 1: means @c this starts @c src
145: - 2: means @src starts @this
1.50 paf 146: */
1.61 paf 147: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.62 paf 148: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.50 paf 149: bool operator == (const char* src_ptr) const {
150: size_t src_size=src_ptr?strlen(src_ptr):0;
151: if(size() != src_size)
152: return false;
153: int partial; // unused
1.56 paf 154: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 155: }
156:
1.42 paf 157: /**
158: appends other String.
1.41 paf 159:
1.47 paf 160: marking all tainted pieces of it with @a lang.
161: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 162: */
1.39 paf 163: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.8 paf 164:
1.41 paf 165: /// simple hash code of string. used by Hash
1.14 paf 166: uint hash_code() const;
1.54 paf 167:
168: /// extracts [start, finish) piece of string
169: String& piece(size_t start, size_t finish) const;
1.55 paf 170:
1.59 paf 171: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 172: int pos(const String& substr,
173: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
1.59 paf 174: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 paf 175: int pos(const char *substr, size_t substr_size,
176: size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
177:
178: void split(Array& result,
179: size_t *pos_after_ref,
180: const char *delim, size_t delim_size,
181: Untaint_lang lang, int limit=-1) const;
182: void split(Array& result,
183: size_t *pos_after_ref,
184: const String& delim,
185: Untaint_lang lang, int limit=-1) const;
186:
1.68 paf 187: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
188: void *info);
1.64 paf 189: bool match(const String *aorigin,
190: const String& regexp,
1.65 paf 191: const String *options,
1.66 paf 192: Table **table,
193: Row_action row_action, void *info) const;
1.13 paf 194:
1.41 paf 195: #ifndef NO_STRING_ORIGIN
196: /// origin of string. calculated by first row
1.43 paf 197: const Origin& origin() const;
1.41 paf 198: #endif
1.8 paf 199:
1.10 paf 200: private:
201:
1.1 paf 202: struct Chunk {
1.6 paf 203: // the number of rows in chunk
1.56 paf 204: size_t count;
1.1 paf 205: union Row {
1.27 paf 206: // fragment
207: struct {
208: const char *ptr; // pointer to the start
209: size_t size; // length
210: Untaint_lang lang; // untaint flag, later untaint language
211: #ifndef NO_STRING_ORIGIN
212: Origin origin; // origin
213: #endif
1.1 paf 214: } item;
215: Chunk *link; // link to the next chunk in chain
1.2 paf 216: } rows[CR_PREALLOCATED_COUNT];
1.1 paf 217: // next rows are here
218: Chunk *preallocated_link;
219: }
220: head; // the head chunk of the chunk chain
221:
222: // next append would write to this record
223: Chunk::Row *append_here;
224:
225: // the address of place where lies address
226: // of the link to the next chunk to allocate
227: Chunk::Row *link_row;
228:
1.5 paf 229: private:
1.25 paf 230: // last chunk
231: Chunk *last_chunk;
1.5 paf 232:
233: // string size
234: size_t fsize;
235:
236: // used rows in all chunks
237: int fused_rows;
238:
239: private:
1.1 paf 240:
241: bool chunk_is_full() {
242: return append_here == link_row;
243: }
244: void expand();
1.69 ! paf 245: char *String::store_to(char *dest, Untaint_lang lang=UL_UNKNOWN) const;
1.7 paf 246:
247: private: //disabled
248:
1.12 paf 249: String& operator = (const String&) { return *this; }
1.7 paf 250:
1.1 paf 251: };
252:
253: #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