Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.144.2.18 paf 4: Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
1.124 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1 paf 6: */
7:
8: #ifndef PA_STRING_H
9: #define PA_STRING_H
1.140 paf 10:
1.144.2.28! paf 11: static const char* IDENT_STRING_H="$Date: 2003/03/05 16:58:24 $";
1.144.2.10 paf 12:
13: // includes
1.1 paf 14:
1.15 paf 15: #include "pa_pool.h"
1.144.2.12 paf 16: #include "pa_types.h"
17: #include "pa_array.h"
1.4 paf 18:
1.144.2.12 paf 19: // forwards
1.144.2.10 paf 20:
1.144.2.28! paf 21: class Charset;
1.144.2.22 paf 22: class Table; DECLARE_OBJECT_PTR(Table);
1.144.2.5 paf 23: class SQL_Connection;
24: class Dictionary;
1.144.2.15 paf 25: class Request_charsets;
1.144.2.22 paf 26: typedef Array<StringPtr> ArrayString; DECLARE_OBJECT_PTR(ArrayString);
1.144.2.5 paf 27:
28: // helpers
29:
1.9 paf 30: #ifndef NO_STRING_ORIGIN
1.33 paf 31: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 32: const char* src, size_t size, \
1.144.2.7 paf 33: String_UL lang, \
1.144.2.18 paf 34: const char* file, uint line
1.54 paf 35: /// appends piece to String @see String::real_append
36: # define APPEND(src, size, lang, file, line) \
37: real_append(src, size, lang, file, line)
1.9 paf 38: #else
1.33 paf 39: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 40: const char* src, \
1.33 paf 41: size_t size, \
1.144.2.7 paf 42: String_UL lang
1.54 paf 43: /// appends piece to String @see String::real_append
44: # define APPEND(src, size, lang, file, line) \
45: real_append(src, size, lang)
46: #endif
1.83 paf 47: /// appends clean piece to String @see String::real_append
1.54 paf 48: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 49: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 50: /// appends piece to String as-is @see String::real_append
51: #define APPEND_AS_IS(src, size, file, line) \
52: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 53: /// appends tainted piece to String @see String::real_append
1.54 paf 54: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 55: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 56: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 57: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 58:
1.144.2.7 paf 59: typedef uint String_UL;
60:
1.144.2.2 paf 61: struct String_fragment {
1.144.2.7 paf 62: String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.144.2.18 paf 63: const char* ptr; ///< pointer to the start
1.144.2.2 paf 64: size_t size; ///< length
1.41 paf 65:
1.144.2.2 paf 66: #ifndef NO_STRING_ORIGIN
67: /// all String pieces hold information of where they come from
68: struct Origin {
1.144.2.18 paf 69: const char* file; ///< macros file name | load file name | sql query text
1.144.2.2 paf 70: uint line; ///< file line no | record no
71: } origin; ///< origin
72: #endif
73: };
74:
75: /**
76: String which knows the language of all it's fragments.
1.41 paf 77:
78: All pieces remember
79: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
80: - whether they are tainted or not,
81: and the language which should be used to detaint them
82: */
1.48 paf 83:
1.144.2.2 paf 84: class String: public Array<String_fragment> {
85: size_t fsize;
1.144.2.24 paf 86: int count() const { return 0; } // hiding fro accidental use instead of size()
1.144.2.2 paf 87: public:
1.1 paf 88:
1.106 parser 89: /** piece is tainted or not. the language to use when detaint
90: remember to change String_Untaint_lang_name @ untaint.C along
91: */
1.27 paf 92: enum Untaint_lang {
1.72 paf 93: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 94: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 95: UL_CLEAN, ///< clean
1.122 paf 96: UL_AS_IS, ///< leave all characters intact
97:
1.48 paf 98: UL_PASS_APPENDED,
1.41 paf 99: /**<
100: leave language built into string being appended.
101: just a flag, that value not stored
102: */
1.122 paf 103: UL_TAINTED, ///< tainted, untaint language as assigned later
104: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 105: UL_FILE_SPEC, ///< file specification
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
1.102 parser 112: UL_XML, ///< ^dom:set xml
1.48 paf 113: UL_HTML, ///< HTML code (for editing)
1.144.2.7 paf 114: UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 115: };
116:
1.8 paf 117: public:
118:
1.144.2.18 paf 119: explicit String(const char* src=0, size_t src_size=0, bool tainted=false);
1.14 paf 120: String(const String& src);
1.144.2.2 paf 121: bool is_empty() const { return !size(); }
122: size_t size() const { return fsize; }
1.69 paf 123: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.5 paf 124: CharPtr cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 125: SQL_Connection *connection=0,
1.144.2.14 paf 126: const Request_charsets *charsets=0) const {
1.88 parser 127:
1.144.2.12 paf 128: CharPtr result(new char[cstr_bufsize(lang, connection, charsets)]);
129: char *eol=store_to(result.get(), lang, connection, charsets);
1.50 paf 130: *eol=0;
131: return result;
132: }
1.144.2.9 paf 133: char *cstr(Pool& pool, Untaint_lang lang=UL_AS_IS,
134: SQL_Connection *connection=0,
1.144.2.14 paf 135: const Request_charsets *charsets=0) const {
1.144.2.12 paf 136: char *result=new(pool) char[cstr_bufsize(lang, connection, charsets)];
137: char *eol=store_to(result, lang, connection, charsets);
1.144.2.9 paf 138: *eol=0;
139: return result;
140: }
1.144.2.12 paf 141: char *cstr_debug_origins(Pool& pool) const;
1.108 parser 142: /// puts pieces to buf
1.144.2.5 paf 143: void serialize(Pool& pool, size_t prolog_size, char *& buf, size_t& buf_size) const;
1.108 parser 144: /// appends pieces from buf to self
1.144.2.18 paf 145: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.46 paf 146: /** append fragment
1.83 paf 147: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 148: */
1.9 paf 149: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 150: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 151: int cmp (int& partial, const String& src,
1.72 paf 152: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 153: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
154: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
155: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
156: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 157: bool operator == (const String& src) const {
158: if(size()!=src.size()) // can speed up in trivial case
159: return false;
1.56 paf 160: int p; return cmp(p, src)==0;
1.26 paf 161: }
1.56 paf 162: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 163:
1.50 paf 164: /**
165: @param partial
166: returns partial match status.
1.51 paf 167: - -1: strings too different
168: - 0: full match
169: - 1: means @c this starts @c src
170: - 2: means @src starts @this
1.50 paf 171: */
1.61 paf 172: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 173: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139 paf 174: /// this starts with src
175: bool starts_with(const char* src_ptr, size_t src_size=0) const {
176: int p; cmp(p, src_ptr, src_size);
177: return p==0 || p==2;
178: }
1.50 paf 179: bool operator == (const char* src_ptr) const {
180: size_t src_size=src_ptr?strlen(src_ptr):0;
181: if(size() != src_size)
182: return false;
183: int partial; // unused
1.56 paf 184: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 185: }
1.80 paf 186: bool operator != (const char* src_ptr) const {
187: int partial; // unused
188: return cmp(partial, src_ptr, 0)!=0;
189: }
190:
1.144.2.25 paf 191: String& append_to(String& dest, String_UL lang, bool forced) const;
192: String& append(const String& src, String_UL lang, bool forced=false) {
1.144.2.11 paf 193: return src.append_to(*this, lang, forced);
194: }
1.76 paf 195: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.17 paf 196: String& operator << (StringPtr src) { return append(*src, UL_PASS_APPENDED); }
1.144.2.18 paf 197: String& operator << (const char* src) { return APPEND_CONST(src); }
1.8 paf 198:
1.144.2.7 paf 199: /// simple hash code of string
1.14 paf 200: uint hash_code() const;
1.100 parser 201:
1.142 paf 202: /// extracts first char of a string, if any
203: char first_char() const {
1.144.2.2 paf 204: return is_empty()?0:*felements[0].ptr;
1.142 paf 205: }
1.54 paf 206:
207: /// extracts [start, finish) piece of string
1.144.2.4 paf 208: StringPtr mid(size_t start, size_t finish) const;
1.55 paf 209:
1.59 paf 210: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 211: int pos(const String& substr,
1.112 paf 212: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 213: /// @return position of substr in string, -1 means "not found" [const char* version]
1.144.2.18 paf 214: int pos(const char* substr, size_t substr_size=0,
1.112 paf 215: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 216:
1.144.2.22 paf 217: void split(ArrayString& result,
1.62 paf 218: size_t *pos_after_ref,
1.144.2.18 paf 219: const char* delim, size_t delim_size,
1.144.2.17 paf 220: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 221: void split(ArrayString& result,
1.62 paf 222: size_t *pos_after_ref,
223: const String& delim,
1.144.2.17 paf 224: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.62 paf 225:
1.144.2.23 paf 226: typedef void (*Row_action)(TablePtr table, ArrayStringPtr row,
1.136 paf 227: int prestart, int prefinish,
228: int poststart, int postfinish,
1.144.2.9 paf 229: void *info);
1.87 parser 230: /**
1.144.2.5 paf 231: @return table of found items, if any.
1.87 parser 232: table format is defined and fixed[can be used by others]:
233: @verbatim
234: prematch/match/postmatch/1/2/3/...
235: @endverbatim
236: */
1.144.2.9 paf 237: TablePtr match(Charset& source_charset,
1.144.2.17 paf 238: StringPtr aorigin,
1.64 paf 239: const String& regexp,
1.144.2.17 paf 240: StringPtr options,
1.99 parser 241: Row_action row_action, void *info,
1.144.2.27 paf 242: bool& just_matched) const;
1.87 parser 243: enum Change_case_kind {
244: CC_UPPER,
245: CC_LOWER
246: };
1.144.2.5 paf 247: StringPtr change_case(Pool& pool, Charset& source_charset,
1.144.2.17 paf 248: Change_case_kind kind);
1.144.2.26 paf 249: StringPtr replace(Pool& pool, const Dictionary& dict) const;
1.144.2.17 paf 250: double as_double();
251: int as_int();
1.137 paf 252:
1.144.2.5 paf 253: StringPtr join_chains(Pool& pool) const;
1.13 paf 254:
1.41 paf 255: #ifndef NO_STRING_ORIGIN
256: /// origin of string. calculated by first row
1.144.2.2 paf 257: const String_fragment::Origin& origin() const;
1.41 paf 258: #endif
1.8 paf 259:
1.144.2.5 paf 260: size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
261: SQL_Connection *connection=0,
1.144.2.14 paf 262: const Request_charsets *charsets=0) const;
1.79 paf 263: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 264: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
265: SQL_Connection *connection=0,
1.144.2.14 paf 266: const Request_charsets *charsets=0) const;
1.144.2.2 paf 267: /*
1.144.2.1 paf 268: void join_chain(
1.130 paf 269: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.144.2.25 paf 270: String_UL& joined_lang, const char* & joined_ptr, size_t& joined_size) const;
1.144.2.2 paf 271: */
1.7 paf 272: private: //disabled
273:
1.12 paf 274: String& operator = (const String&) { return *this; }
1.7 paf 275:
1.1 paf 276: };
1.144.2.22 paf 277:
1.144.2.7 paf 278:
279: /// simple hash code of string. used by Hash
280: inline uint hash_code(const String& self) {
281: return self.hash_code();
282: }
283: /// simple hash code of string. used by Hash
1.144.2.17 paf 284: inline uint hash_code(StringPtr self) {
1.144.2.7 paf 285: return self->hash_code();
286: }
1.144.2.5 paf 287:
288: // helpers
289:
290: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
291: for(; fragment<last; fragment++) { \
292: code; \
293: }
294:
295: #define STRING_FOREACH_FRAGMENT(code) \
296: { \
297: element_type *last = felements+fused; \
298: element_type *fragment = felements; \
299: STRING_PREPARED_FOREACH_FRAGMENT(code) \
300: }
1.1 paf 301:
302: #endif
E-mail: