Annotation of parser3/src/include/pa_string.h, revision 1.104
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:
1.104 ! parser 8: $Id: pa_string.h,v 1.101 2001/08/28 09:27:42 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"
1.15 paf 15: #include "pa_pool.h"
1.4 paf 16: #include "pa_types.h"
17:
1.63 paf 18: class Table;
19:
1.50 paf 20: /**
1.92 parser 21: $MAIN:html-typo table elements must enlarge string not more that that.
22:
1.50 paf 23: that's a tradeoff - otherwise we'd have to scan string twice:
24: - first for buffer length
25: - second for replacements themselves
26: */
1.92 parser 27: #define UNTAINT_TIMES_BIGGER 10
1.31 paf 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
1.83 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)
1.83 paf 49: /// appends piece to String as-is @see String::real_append
50: #define APPEND_AS_IS(src, size, file, line) \
51: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 52: /// appends tainted piece to String @see String::real_append
1.54 paf 53: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 54: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 55: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 56: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 57:
1.62 paf 58: class Array;
1.71 paf 59: class SQL_Connection;
1.101 parser 60: class Dictionary;
1.62 paf 61:
1.42 paf 62: /**
1.41 paf 63: Pooled string.
64:
1.47 paf 65: Internal structure:
66: @verbatim
67: String Chunk0
68: ====== ========
69: head--------------->[ptr, size, ...]
70: append_here-------->[ptr, size, ...]
71: .
72: .
73: [ptr, size, ...]
74: link_row----------->[link to the next chunk]
1.41 paf 75: @endverbatim
76:
77: All pieces remember
78: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
79: - whether they are tainted or not,
80: and the language which should be used to detaint them
81: */
1.15 paf 82: class String : public Pooled {
1.1 paf 83: public:
1.48 paf 84:
1.1 paf 85: enum {
1.89 parser 86: CR_PREALLOCATED_COUNT=2, ///< default preallocated item count
1.91 parser 87: CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded()
1.1 paf 88: };
89:
1.41 paf 90: /// piece is tainted or not. the language to use when detaint
1.27 paf 91: enum Untaint_lang {
1.72 paf 92: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.62 paf 93: UL_CLEAN, ///< clean
94: UL_TAINTED, ///< tainted, untaint language as assigned later
1.27 paf 95: // untaint languages. assigned by ^untaint[lang]{...}
1.48 paf 96: UL_PASS_APPENDED,
1.41 paf 97: /**<
98: leave language built into string being appended.
99: just a flag, that value not stored
100: */
1.48 paf 101: UL_AS_IS, ///< leave all characters intact
1.103 parser 102: UL_FILE_SPEC, ///< file specification
1.74 paf 103: UL_HTTP_HEADER, ///< text in HTTP response header
104: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 105: UL_URI, ///< text in uri
106: UL_TABLE, ///< ^table:set body
107: UL_SQL, ///< ^table:sql body
108: UL_JS, ///< JavaScript code
1.102 parser 109: UL_XML, ///< ^dom:set xml
1.48 paf 110: UL_HTML, ///< HTML code (for editing)
1.85 paf 111: UL_USER_HTML ///< HTML code with USER chars
1.27 paf 112: };
113:
1.8 paf 114: public:
115:
1.77 paf 116: String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 117: String(const String& src);
118: size_t size() const { return fsize; }
1.98 parser 119: /// makes store_to produce origin:text\n output
120: void set_origins_mode(bool aorigins_mode) const { forigins_mode=aorigins_mode; }
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:
1.94 parser 126: char *result=(char *)malloc(cstr_bufsize(lang));
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;
1.100 parser 184:
185: /// extracts first char of a string
186: char first_char() const;
1.54 paf 187:
188: /// extracts [start, finish) piece of string
1.70 paf 189: String& mid(size_t start, size_t finish) const;
1.55 paf 190:
1.59 paf 191: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 192: int pos(const String& substr,
1.72 paf 193: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 194: /// @return position of substr in string, -1 means "not found" [const char* version]
1.97 parser 195: int pos(const char *substr, size_t substr_size=0,
1.72 paf 196: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 197:
198: void split(Array& result,
199: size_t *pos_after_ref,
200: const char *delim, size_t delim_size,
1.86 paf 201: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 202: void split(Array& result,
203: size_t *pos_after_ref,
204: const String& delim,
1.86 paf 205: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 206:
1.68 paf 207: typedef void (*Row_action)(Table& table, Array *row, int start, int finish,
208: void *info);
1.87 parser 209: /**
210: @return true if fills table.
211: table format is defined and fixed[can be used by others]:
212: @verbatim
213: prematch/match/postmatch/1/2/3/...
214: @endverbatim
215: */
1.81 paf 216: bool match(const unsigned char *pcre_tables,
217: const String *aorigin,
1.64 paf 218: const String& regexp,
1.65 paf 219: const String *options,
1.66 paf 220: Table **table,
1.99 parser 221: Row_action row_action, void *info,
222: bool *was_global=0) const;
1.87 parser 223: enum Change_case_kind {
224: CC_UPPER,
225: CC_LOWER
226: };
227: String& change_case(Pool& pool, const unsigned char *pcre_tables,
228: Change_case_kind kind) const;
1.101 parser 229: String& replace(Pool& pool, Dictionary& dict) const;
1.96 parser 230: double as_double() const;
231: int as_int() const;
1.13 paf 232:
1.41 paf 233: #ifndef NO_STRING_ORIGIN
234: /// origin of string. calculated by first row
1.43 paf 235: const Origin& origin() const;
1.41 paf 236: #endif
1.8 paf 237:
1.10 paf 238: private:
239:
1.84 paf 240: /// several String fragments
1.1 paf 241: struct Chunk {
1.84 paf 242: size_t count; ///< the number of rows in chunk
243: /// string fragment or a link to next chunk union
1.1 paf 244: union Row {
1.84 paf 245: /// fragment
1.27 paf 246: struct {
1.84 paf 247: const char *ptr; ///< pointer to the start
248: size_t size; ///< length
249: Untaint_lang lang; ///< untaint flag, later untaint language
1.27 paf 250: #ifndef NO_STRING_ORIGIN
1.84 paf 251: Origin origin; ///< origin
1.27 paf 252: #endif
1.1 paf 253: } item;
1.84 paf 254: Chunk *link; ///< link to the next chunk in chain
1.2 paf 255: } rows[CR_PREALLOCATED_COUNT];
1.84 paf 256: Chunk *preallocated_link; ///< next rows are here
1.1 paf 257: }
1.84 paf 258: head; ///< the head chunk of the chunk chain
1.1 paf 259:
1.84 paf 260: /// next append would write to this record
1.1 paf 261: Chunk::Row *append_here;
262:
1.84 paf 263: /** the address of place where lies address
264: of the link to the next chunk to allocate
265: */
1.1 paf 266: Chunk::Row *link_row;
267:
1.5 paf 268: private:
1.25 paf 269: // last chunk
270: Chunk *last_chunk;
1.5 paf 271:
272: // string size
273: size_t fsize;
274:
275: // used rows in all chunks
276: int fused_rows;
1.98 parser 277:
278: /// origins_mode changes behaviour of store_to[cstr()]
279: mutable bool forigins_mode;
1.5 paf 280:
281: private:
1.1 paf 282:
283: bool chunk_is_full() {
284: return append_here == link_row;
285: }
286: void expand();
1.79 paf 287:
1.94 parser 288: size_t cstr_bufsize(Untaint_lang lang) const;
1.79 paf 289: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 290: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
291: SQL_Connection *connection=0,
292: const char *charset=0) const;
1.7 paf 293:
294: private: //disabled
295:
1.12 paf 296: String& operator = (const String&) { return *this; }
1.7 paf 297:
1.1 paf 298: };
299:
300: #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