Annotation of parser3/src/include/pa_string.h, revision 1.105
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)
1.30 paf 5: Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
1.29 paf 6:
1.105 ! parser 7: $Id: pa_string.h,v 1.104 2001/09/21 15:35:45 parser Exp $
1.1 paf 8: */
9:
10: #ifndef PA_STRING_H
11: #define PA_STRING_H
12:
1.49 paf 13: #include "pa_config_includes.h"
1.15 paf 14: #include "pa_pool.h"
1.4 paf 15: #include "pa_types.h"
16:
1.63 paf 17: class Table;
18:
1.50 paf 19: /**
1.92 parser 20: $MAIN:html-typo table elements must enlarge string not more that that.
21:
1.50 paf 22: that's a tradeoff - otherwise we'd have to scan string twice:
23: - first for buffer length
24: - second for replacements themselves
25: */
1.92 parser 26: #define UNTAINT_TIMES_BIGGER 10
1.31 paf 27:
1.9 paf 28: #ifndef NO_STRING_ORIGIN
1.33 paf 29: # define STRING_APPEND_PARAMS \
30: const char *src, size_t size, \
1.52 paf 31: String::Untaint_lang lang, \
1.33 paf 32: const char *file, uint line
1.54 paf 33: /// appends piece to String @see String::real_append
34: # define APPEND(src, size, lang, file, line) \
35: real_append(src, size, lang, file, line)
1.9 paf 36: #else
1.33 paf 37: # define STRING_APPEND_PARAMS \
38: const char *src, \
39: size_t size, \
1.52 paf 40: String::Untaint_lang lang
1.54 paf 41: /// appends piece to String @see String::real_append
42: # define APPEND(src, size, lang, file, line) \
43: real_append(src, size, lang)
44: #endif
1.83 paf 45: /// appends clean piece to String @see String::real_append
1.54 paf 46: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 47: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 48: /// appends piece to String as-is @see String::real_append
49: #define APPEND_AS_IS(src, size, file, line) \
50: APPEND(src, size, String::UL_AS_IS, 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
1.83 paf 55: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 56:
1.62 paf 57: class Array;
1.71 paf 58: class SQL_Connection;
1.101 parser 59: class Dictionary;
1.62 paf 60:
1.42 paf 61: /**
1.41 paf 62: Pooled string.
63:
1.47 paf 64: Internal structure:
65: @verbatim
66: String Chunk0
67: ====== ========
68: head--------------->[ptr, size, ...]
69: append_here-------->[ptr, size, ...]
70: .
71: .
72: [ptr, size, ...]
73: link_row----------->[link to the next chunk]
1.41 paf 74: @endverbatim
75:
76: All pieces remember
77: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
78: - whether they are tainted or not,
79: and the language which should be used to detaint them
80: */
1.15 paf 81: class String : public Pooled {
1.1 paf 82: public:
1.48 paf 83:
1.1 paf 84: enum {
1.89 parser 85: CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
1.91 parser 86: CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()
1.1 paf 87: };
88:
1.41 paf 89: /// piece is tainted or not. the language to use when detaint
1.27 paf 90: enum Untaint_lang {
1.72 paf 91: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 92: UL_CLEAN, ///< clean
93: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 94: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 95: UL_PASS_APPENDED,
1.41 paf 96: /**<
97: leave language built into string being appended.
98: just a flag, that value not stored
99: */
1.48 paf 100: UL_AS_IS, ///< leave all characters intact
1.103 parser 101: UL_FILE_SPEC, ///< file specification
1.74 paf 102: UL_HTTP_HEADER, ///< text in HTTP response header
103: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 104: UL_URI, ///< text in uri
105: UL_TABLE, ///< ^table:set body
106: UL_SQL, ///< ^table:sql body
107: UL_JS, ///< JavaScript code
1.102 parser 108: UL_XML, ///< ^dom:set xml
1.48 paf 109: UL_HTML, ///< HTML code (for editing)
1.85 paf 110: UL_USER_HTML ///< HTML code with USER chars
1.27 paf 111: };
112:
1.8 paf 113: public:
114:
1.77 paf 115: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 116: String(const String& src);
117: size_t size() const { return fsize; }
1.98 parser 118: /// makes store_to produce origin:text\n output
119: void set_origins_mode(bool aorigins_mode) const { forigins_mode=aorigins_mode; }
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:
1.94 parser 125: char *result=(char *)malloc(cstr_bufsize(lang));
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;
1.100 parser 183:
184: /// extracts first char of a string
185: char first_char() const;
1.54 paf 186:
187: /// extracts [start, finish) piece of string
1.70 paf 188: String& mid(size_t start, size_t finish) const;
1.55 paf 189:
1.59 paf 190: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 191: int pos(const String& substr,
1.72 paf 192: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 193: /// @return position of substr in string, -1 means "not found" [const char* version]
1.97 parser 194: int pos(const char *substr, size_t substr_size=0,
1.72 paf 195: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 196:
197: void split(Array& result,
198: size_t *pos_after_ref,
199: const char *delim, size_t delim_size,
1.86 paf 200: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 201: void split(Array& result,
202: size_t *pos_after_ref,
203: const String& delim,
1.86 paf 204: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 205:
1.68 paf 206: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
207: void *info);
1.87 parser 208: /**
209: @return true if fills table.
210: table format is defined and fixed[can be used by others]:
211: @verbatim
212: prematch/match/postmatch/1/2/3/...
213: @endverbatim
214: */
1.81 paf 215: bool match(const unsigned char *pcre_tables,
216: const String *aorigin,
1.64 paf 217: const String& regexp,
1.65 paf 218: const String *options,
1.66 paf 219: Table **table,
1.99 parser 220: Row_action row_action, void *info,
221: bool *was_global=0) const;
1.87 parser 222: enum Change_case_kind {
223: CC_UPPER,
224: CC_LOWER
225: };
226: String& change_case(Pool& pool, const unsigned char *pcre_tables,
227: Change_case_kind kind) const;
1.101 parser 228: String& replace(Pool& pool, Dictionary& dict) const;
1.96 parser 229: double as_double() const;
230: int as_int() const;
1.13 paf 231:
1.41 paf 232: #ifndef NO_STRING_ORIGIN
233: /// origin of string. calculated by first row
1.43 paf 234: const Origin& origin() const;
1.41 paf 235: #endif
1.8 paf 236:
1.10 paf 237: private:
238:
1.84 paf 239: /// several String fragments
1.1 paf 240: struct Chunk {
1.84 paf 241: size_t count; ///< the number of rows in chunk
242: /// string fragment or a link to next chunk union
1.1 paf 243: union Row {
1.84 paf 244: /// fragment
1.27 paf 245: struct {
1.84 paf 246: const char *ptr; ///< pointer to the start
247: size_t size; ///< length
248: Untaint_lang lang; ///< untaint flag, later untaint language
1.27 paf 249: #ifndef NO_STRING_ORIGIN
1.84 paf 250: Origin origin; ///< origin
1.27 paf 251: #endif
1.1 paf 252: } item;
1.84 paf 253: Chunk *link; ///< link to the next chunk in chain
1.2 paf 254: } rows[CR_PREALLOCATED_COUNT];
1.84 paf 255: Chunk *preallocated_link; ///< next rows are here
1.1 paf 256: }
1.84 paf 257: head; ///< the head chunk of the chunk chain
1.1 paf 258:
1.84 paf 259: /// next append would write to this record
1.1 paf 260: Chunk::Row *append_here;
261:
1.84 paf 262: /** the address of place where lies address
263: of the link to the next chunk to allocate
264: */
1.1 paf 265: Chunk::Row *link_row;
266:
1.5 paf 267: private:
1.25 paf 268: // last chunk
269: Chunk *last_chunk;
1.5 paf 270:
271: // string size
272: size_t fsize;
273:
274: // used rows in all chunks
275: int fused_rows;
1.98 parser 276:
277: /// origins_mode changes behaviour of store_to[cstr()]
278: mutable bool forigins_mode;
1.5 paf 279:
280: private:
1.1 paf 281:
282: bool chunk_is_full() {
283: return append_here == link_row;
284: }
285: void expand();
1.79 paf 286:
1.94 parser 287: size_t cstr_bufsize(Untaint_lang lang) const;
1.79 paf 288: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 289: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
290: SQL_Connection *connection=0,
291: const char *charset=0) const;
1.7 paf 292:
293: private: //disabled
294:
1.12 paf 295: String& operator = (const String&) { return *this; }
1.7 paf 296:
1.1 paf 297: };
298:
299: #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