Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.2
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:
1.144.2.28.2. (paf 11:): static const char* IDENT_STRING_H="$Date: 2003/03/18 15:14:17 $";
1.144.2.10 paf 12:
13: // includes
1.1 paf 14:
1.144.2.12 paf 15: #include "pa_types.h"
16: #include "pa_array.h"
1.4 paf 17:
1.144.2.28.2. (paf 18:): #include "cord.h"
19:):
1.144.2.12 paf 20: // forwards
1.144.2.10 paf 21:
1.144.2.28 paf 22: class Charset;
1.144.2.22 paf 23: class Table; DECLARE_OBJECT_PTR(Table);
1.144.2.5 paf 24: class SQL_Connection;
25: class Dictionary;
1.144.2.15 paf 26: class Request_charsets;
1.144.2.28.2. (paf 27:): typedef Array<String* > ArrayString; DECLARE_OBJECT_PTR(ArrayString);
1.144.2.5 paf 28:
29: // helpers
30:
1.9 paf 31: #ifndef NO_STRING_ORIGIN
1.33 paf 32: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 33: const char* src, size_t size, \
1.144.2.7 paf 34: String_UL lang, \
1.144.2.18 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 \
1.144.2.18 paf 41: const char* src, \
1.33 paf 42: size_t size, \
1.144.2.7 paf 43: String_UL 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
1.83 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)
1.83 paf 51: /// appends piece to String as-is @see String::real_append
52: #define APPEND_AS_IS(src, size, file, line) \
53: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 54: /// appends tainted piece to String @see String::real_append
1.54 paf 55: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 56: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 57: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 58: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 59:
1.144.2.7 paf 60: typedef uint String_UL;
61:
1.144.2.2 paf 62:
63: /**
64: String which knows the language of all it's fragments.
1.41 paf 65:
66: All pieces remember
67: - whether they are tainted or not,
68: and the language which should be used to detaint them
69: */
1.48 paf 70:
1.144.2.28.2. (paf 71:): class String: public PA_Object {
72:):
1.144.2.2 paf 73: public:
1.1 paf 74:
1.106 parser 75: /** piece is tainted or not. the language to use when detaint
76: remember to change String_Untaint_lang_name @ untaint.C along
77: */
1.27 paf 78: enum Untaint_lang {
1.72 paf 79: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 80: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 81: UL_CLEAN, ///< clean
1.122 paf 82: UL_AS_IS, ///< leave all characters intact
83:
1.48 paf 84: UL_PASS_APPENDED,
1.41 paf 85: /**<
86: leave language built into string being appended.
87: just a flag, that value not stored
88: */
1.122 paf 89: UL_TAINTED, ///< tainted, untaint language as assigned later
90: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 91: UL_FILE_SPEC, ///< file specification
1.74 paf 92: UL_HTTP_HEADER, ///< text in HTTP response header
93: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 94: UL_URI, ///< text in uri
95: UL_TABLE, ///< ^table:set body
96: UL_SQL, ///< ^table:sql body
97: UL_JS, ///< JavaScript code
1.102 parser 98: UL_XML, ///< ^dom:set xml
1.48 paf 99: UL_HTML, ///< HTML code (for editing)
1.144.2.7 paf 100: UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 101: };
102:
1.144.2.28.2. (paf 103:): struct Fragment_info {
104:): String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
105:): size_t size; ///< length
106:): Fragment_info(String_UL alang, size_t asize): lang(alang), size(asize) {}
107:): };
108:):
109:): private:
110:):
111:): CORD body; ///< all characters of string
112:): Array<Fragment_info> fragment_info; ///< fragment language+size info
113:):
1.8 paf 114: public:
115:
1.144.2.18 paf 116: explicit String(const char* src=0, size_t src_size=0, bool tainted=false);
1.14 paf 117: String(const String& src);
1.144.2.28.2. (paf 118:): bool is_empty() const { return body==CORD_EMPTY; }
119:): size_t size() const { return CORD_len(body); }
1.69 paf 120: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.28.2. (paf 121:): const char* cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 122: SQL_Connection *connection=0,
1.144.2.14 paf 123: const Request_charsets *charsets=0) const {
1.88 parser 124:
1.144.2.28.2. (paf 125:): char* result=new char[cstr_bufsize(lang, connection, charsets)];
1.144.2.12 paf 126: char *eol=store_to(result, lang, connection, charsets);
1.144.2.9 paf 127: *eol=0;
128: return result;
129: }
1.108 parser 130: /// puts pieces to buf
1.144.2.28.2. (paf 131:): void serialize(size_t prolog_size, char *& buf, size_t& buf_size) const;
1.108 parser 132: /// appends pieces from buf to self
1.144.2.18 paf 133: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.46 paf 134: /** append fragment
1.83 paf 135: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 136: */
1.9 paf 137: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 138: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 139: int cmp (int& partial, const String& src,
1.72 paf 140: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 141: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
142: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
143: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
144: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 145: bool operator == (const String& src) const {
146: if(size()!=src.size()) // can speed up in trivial case
147: return false;
1.56 paf 148: int p; return cmp(p, src)==0;
1.26 paf 149: }
1.56 paf 150: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 151:
1.50 paf 152: /**
153: @param partial
154: returns partial match status.
1.51 paf 155: - -1: strings too different
156: - 0: full match
157: - 1: means @c this starts @c src
158: - 2: means @src starts @this
1.50 paf 159: */
1.61 paf 160: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 161: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139 paf 162: /// this starts with src
163: bool starts_with(const char* src_ptr, size_t src_size=0) const {
164: int p; cmp(p, src_ptr, src_size);
165: return p==0 || p==2;
166: }
1.50 paf 167: bool operator == (const char* src_ptr) const {
168: size_t src_size=src_ptr?strlen(src_ptr):0;
169: if(size() != src_size)
170: return false;
171: int partial; // unused
1.56 paf 172: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 173: }
1.80 paf 174: bool operator != (const char* src_ptr) const {
175: int partial; // unused
176: return cmp(partial, src_ptr, 0)!=0;
177: }
178:
1.144.2.25 paf 179: String& append_to(String& dest, String_UL lang, bool forced) const;
180: String& append(const String& src, String_UL lang, bool forced=false) {
1.144.2.11 paf 181: return src.append_to(*this, lang, forced);
182: }
1.76 paf 183: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.28.2. (paf 184:): //String& operator << (const String* src) { return append(*src, UL_PASS_APPENDED); }
1.144.2.18 paf 185: String& operator << (const char* src) { return APPEND_CONST(src); }
1.8 paf 186:
1.144.2.7 paf 187: /// simple hash code of string
1.14 paf 188: uint hash_code() const;
1.100 parser 189:
1.142 paf 190: /// extracts first char of a string, if any
191: char first_char() const {
1.144.2.28.2. (paf 192:): return is_empty()?0:CORD_fetch(body, 0);
1.142 paf 193: }
1.54 paf 194:
195: /// extracts [start, finish) piece of string
1.144.2.28.2. (paf 196:): const String& mid(size_t start, size_t finish) const;
1.55 paf 197:
1.59 paf 198: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 199: int pos(const String& substr,
1.112 paf 200: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 201: /// @return position of substr in string, -1 means "not found" [const char* version]
1.144.2.18 paf 202: int pos(const char* substr, size_t substr_size=0,
1.112 paf 203: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 204:
1.144.2.22 paf 205: void split(ArrayString& result,
1.62 paf 206: size_t *pos_after_ref,
1.144.2.18 paf 207: const char* delim, size_t delim_size,
1.144.2.17 paf 208: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 209: void split(ArrayString& result,
1.62 paf 210: size_t *pos_after_ref,
211: const String& delim,
1.144.2.17 paf 212: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.62 paf 213:
1.144.2.28.2. (paf 214:): typedef void (*Row_action)(Table* table, ArrayString& row,
1.136 paf 215: int prestart, int prefinish,
216: int poststart, int postfinish,
1.144.2.9 paf 217: void *info);
1.87 parser 218: /**
1.144.2.5 paf 219: @return table of found items, if any.
1.87 parser 220: table format is defined and fixed[can be used by others]:
221: @verbatim
222: prematch/match/postmatch/1/2/3/...
223: @endverbatim
224: */
1.144.2.28.2. (paf 225:): Table* match(Charset& source_charset,
226:): const String& aorigin,
1.64 paf 227: const String& regexp,
1.144.2.28.2. (paf 228:): const String& options,
1.99 parser 229: Row_action row_action, void *info,
1.144.2.27 paf 230: bool& just_matched) const;
1.87 parser 231: enum Change_case_kind {
232: CC_UPPER,
233: CC_LOWER
234: };
1.144.2.28.2. (paf 235:): const String& change_case(Charset& source_charset,
1.144.2.17 paf 236: Change_case_kind kind);
1.144.2.28.2. (paf 237:): const String& replace(const Dictionary& dict) const;
1.144.2.17 paf 238: double as_double();
239: int as_int();
1.137 paf 240:
1.144.2.28.2. (paf 241:): const String& join_chains() const;
1.8 paf 242:
1.144.2.5 paf 243: size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
244: SQL_Connection *connection=0,
1.144.2.14 paf 245: const Request_charsets *charsets=0) const;
1.79 paf 246: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 247: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
248: SQL_Connection *connection=0,
1.144.2.14 paf 249: const Request_charsets *charsets=0) const;
1.144.2.2 paf 250: /*
1.144.2.1 paf 251: void join_chain(
1.130 paf 252: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.144.2.25 paf 253: String_UL& joined_lang, const char* & joined_ptr, size_t& joined_size) const;
1.144.2.2 paf 254: */
1.7 paf 255: private: //disabled
256:
1.12 paf 257: String& operator = (const String&) { return *this; }
1.7 paf 258:
1.1 paf 259: };
1.144.2.22 paf 260:
1.144.2.7 paf 261:
262: /// simple hash code of string. used by Hash
263: inline uint hash_code(const String& self) {
264: return self.hash_code();
265: }
1.144.2.5 paf 266:
267: // helpers
268:
269: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
270: for(; fragment<last; fragment++) { \
271: code; \
272: }
273:
274: #define STRING_FOREACH_FRAGMENT(code) \
275: { \
276: element_type *last = felements+fused; \
277: element_type *fragment = felements; \
278: STRING_PREPARED_FOREACH_FRAGMENT(code) \
279: }
1.1 paf 280:
281: #endif
E-mail: