Annotation of parser3/src/include/pa_string.h, revision 1.92
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:
1.92 ! parser 8: $Id: pa_string.h,v 1.91 2001/05/15 15:51:05 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: /**
1.92 ! parser 25: $MAIN:html-typo table elements must enlarge string not more that that.
! 26:
1.50 paf 27: that's a tradeoff - otherwise we'd have to scan string twice:
28: - first for buffer length
29: - second for replacements themselves
30: */
1.92 ! parser 31: #define UNTAINT_TIMES_BIGGER 10
1.31 paf 32:
1.9 paf 33: #ifndef NO_STRING_ORIGIN
1.33 paf 34: # define STRING_APPEND_PARAMS \
35: const char *src, size_t size, \
1.52 paf 36: String::Untaint_lang lang, \
1.33 paf 37: const char *file, uint line
1.54 paf 38: /// appends piece to String @see String::real_append
39: # define APPEND(src, size, lang, file, line) \
40: real_append(src, size, lang, file, line)
1.9 paf 41: #else
1.33 paf 42: # define STRING_APPEND_PARAMS \
43: const char *src, \
44: size_t size, \
1.52 paf 45: String::Untaint_lang lang
1.54 paf 46: /// appends piece to String @see String::real_append
47: # define APPEND(src, size, lang, file, line) \
48: real_append(src, size, lang)
49: #endif
1.83 paf 50: /// appends clean piece to String @see String::real_append
1.54 paf 51: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 52: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 53: /// appends piece to String as-is @see String::real_append
54: #define APPEND_AS_IS(src, size, file, line) \
55: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 56: /// appends tainted piece to String @see String::real_append
1.54 paf 57: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 58: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 59: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 60: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 61:
1.62 paf 62: class Array;
1.71 paf 63: class SQL_Connection;
1.62 paf 64:
1.42 paf 65: /**
1.41 paf 66: Pooled string.
67:
1.47 paf 68: Internal structure:
69: @verbatim
70: String Chunk0
71: ====== ========
72: head--------------->[ptr, size, ...]
73: append_here-------->[ptr, size, ...]
74: .
75: .
76: [ptr, size, ...]
77: link_row----------->[link to the next chunk]
1.41 paf 78: @endverbatim
79:
80: All pieces remember
81: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
82: - whether they are tainted or not,
83: and the language which should be used to detaint them
84: */
1.15 paf 85: class String : public Pooled {
1.1 paf 86: public:
1.48 paf 87:
1.1 paf 88: enum {
1.89 parser 89: CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
1.91 parser 90: CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()
1.1 paf 91: };
92:
1.41 paf 93: /// piece is tainted or not. the language to use when detaint
1.27 paf 94: enum Untaint_lang {
1.72 paf 95: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 96: UL_CLEAN, ///< clean
97: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 98: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 99: UL_PASS_APPENDED,
1.41 paf 100: /**<
101: leave language built into string being appended.
102: just a flag, that value not stored
103: */
1.48 paf 104: UL_AS_IS, ///< leave all characters intact
105: UL_FILE_NAME, ///< filename
1.74 paf 106: UL_HTTP_HEADER, ///< text in HTTP response header
107: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 108: UL_URI, ///< text in uri
109: UL_TABLE, ///< ^table:set body
110: UL_SQL, ///< ^table:sql body
111: UL_JS, ///< JavaScript code
112: UL_HTML, ///< HTML code (for editing)
1.85 paf 113: UL_USER_HTML ///< HTML code with USER chars
1.27 paf 114: };
115:
1.8 paf 116: public:
117:
1.77 paf 118: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 119: String(const String& src);
120: size_t size() const { return fsize; }
1.69 paf 121: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.82 paf 122: char *cstr(Untaint_lang lang=UL_UNSPECIFIED,
123: SQL_Connection *connection=0,
124: const char *charset=0) const {
1.88 parser 125:
1.50 paf 126: char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
1.82 paf 127: char *eol=store_to(result, lang, connection, charset);
1.50 paf 128: *eol=0;
129: return result;
130: }
1.46 paf 131: /** append fragment
1.83 paf 132: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 133: */
1.9 paf 134: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 135: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 136: int cmp (int& partial, const String& src,
1.72 paf 137: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 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; }
141: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 142: bool operator == (const String& src) const {
143: if(size()!=src.size()) // can speed up in trivial case
144: return false;
1.56 paf 145: int p; return cmp(p, src)==0;
1.26 paf 146: }
1.56 paf 147: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 148:
1.50 paf 149: /**
150: @param partial
151: returns partial match status.
1.51 paf 152: - -1: strings too different
153: - 0: full match
154: - 1: means @c this starts @c src
155: - 2: means @src starts @this
1.50 paf 156: */
1.61 paf 157: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 158: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.50 paf 159: bool operator == (const char* src_ptr) const {
160: size_t src_size=src_ptr?strlen(src_ptr):0;
161: if(size() != src_size)
162: return false;
163: int partial; // unused
1.56 paf 164: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 165: }
1.80 paf 166: bool operator != (const char* src_ptr) const {
167: int partial; // unused
168: return cmp(partial, src_ptr, 0)!=0;
169: }
170:
1.50 paf 171:
1.42 paf 172: /**
173: appends other String.
1.41 paf 174:
1.47 paf 175: marking all tainted pieces of it with @a lang.
176: or marking ALL pieces of it with a @a lang when @a forced to.
1.41 paf 177: */
1.39 paf 178: String& append(const String& src, Untaint_lang lang, bool forced=false);
1.76 paf 179: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
180: String& operator << (const char *src) { return APPEND_CONST(src); }
1.8 paf 181:
1.41 paf 182: /// simple hash code of string. used by Hash
1.14 paf 183: uint hash_code() const;
1.54 paf 184:
185: /// extracts [start, finish) piece of string
1.70 paf 186: String& mid(size_t start, size_t finish) const;
1.55 paf 187:
1.59 paf 188: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 189: int pos(const String& substr,
1.72 paf 190: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 191: /// @return position of substr in string, -1 means "not found" [const char* version]
1.62 paf 192: int pos(const char *substr, size_t substr_size,
1.72 paf 193: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 194:
195: void split(Array& result,
196: size_t *pos_after_ref,
197: const char *delim, size_t delim_size,
1.86 paf 198: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 199: void split(Array& result,
200: size_t *pos_after_ref,
201: const String& delim,
1.86 paf 202: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 203:
1.68 paf 204: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
205: void *info);
1.87 parser 206: /**
207: @return true if fills table.
208: table format is defined and fixed[can be used by others]:
209: @verbatim
210: prematch/match/postmatch/1/2/3/...
211: @endverbatim
212: */
1.81 paf 213: bool match(const unsigned char *pcre_tables,
214: const String *aorigin,
1.64 paf 215: const String& regexp,
1.65 paf 216: const String *options,
1.66 paf 217: Table **table,
218: Row_action row_action, void *info) const;
1.87 parser 219: enum Change_case_kind {
220: CC_UPPER,
221: CC_LOWER
222: };
223: String& change_case(Pool& pool, const unsigned char *pcre_tables,
224: Change_case_kind kind) const;
1.13 paf 225:
1.41 paf 226: #ifndef NO_STRING_ORIGIN
227: /// origin of string. calculated by first row
1.43 paf 228: const Origin& origin() const;
1.41 paf 229: #endif
1.8 paf 230:
1.10 paf 231: private:
232:
1.84 paf 233: /// several String fragments
1.1 paf 234: struct Chunk {
1.84 paf 235: size_t count; ///< the number of rows in chunk
236: /// string fragment or a link to next chunk union
1.1 paf 237: union Row {
1.84 paf 238: /// fragment
1.27 paf 239: struct {
1.84 paf 240: const char *ptr; ///< pointer to the start
241: size_t size; ///< length
242: Untaint_lang lang; ///< untaint flag, later untaint language
1.27 paf 243: #ifndef NO_STRING_ORIGIN
1.84 paf 244: Origin origin; ///< origin
1.27 paf 245: #endif
1.1 paf 246: } item;
1.84 paf 247: Chunk *link; ///< link to the next chunk in chain
1.2 paf 248: } rows[CR_PREALLOCATED_COUNT];
1.84 paf 249: Chunk *preallocated_link; ///< next rows are here
1.1 paf 250: }
1.84 paf 251: head; ///< the head chunk of the chunk chain
1.1 paf 252:
1.84 paf 253: /// next append would write to this record
1.1 paf 254: Chunk::Row *append_here;
255:
1.84 paf 256: /** the address of place where lies address
257: of the link to the next chunk to allocate
258: */
1.1 paf 259: Chunk::Row *link_row;
260:
1.5 paf 261: private:
1.25 paf 262: // last chunk
263: Chunk *last_chunk;
1.5 paf 264:
265: // string size
266: size_t fsize;
267:
268: // used rows in all chunks
269: int fused_rows;
270:
271: private:
1.1 paf 272:
273: bool chunk_is_full() {
274: return append_here == link_row;
275: }
276: void expand();
1.79 paf 277:
278: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 279: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
280: SQL_Connection *connection=0,
281: const char *charset=0) const;
1.7 paf 282:
283: private: //disabled
284:
1.12 paf 285: String& operator = (const String&) { return *this; }
1.7 paf 286:
1.1 paf 287: };
288:
289: #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