Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.3
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 16:00:07 $";
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:): class String;
28:): typedef Array<const String* > ArrayString;
1.144.2.5 paf 29:
30: // helpers
31:
1.9 paf 32: #ifndef NO_STRING_ORIGIN
1.33 paf 33: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 34: const char* src, size_t size, \
1.144.2.28.2. (paf 35:): Untaint_lang lang, \
1.144.2.18 paf 36: const char* file, uint line
1.54 paf 37: /// appends piece to String @see String::real_append
38: # define APPEND(src, size, lang, file, line) \
39: real_append(src, size, lang, file, line)
1.9 paf 40: #else
1.33 paf 41: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 42: const char* src, \
1.33 paf 43: size_t size, \
1.144.2.28.2. (paf 44:): Untaint_lang lang
1.54 paf 45: /// appends piece to String @see String::real_append
46: # define APPEND(src, size, lang, file, line) \
47: real_append(src, size, lang)
48: #endif
1.83 paf 49: /// appends clean piece to String @see String::real_append
1.54 paf 50: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 51: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 52: /// appends piece to String as-is @see String::real_append
53: #define APPEND_AS_IS(src, size, file, line) \
54: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 55: /// appends tainted piece to String @see String::real_append
1.54 paf 56: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 57: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 58: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 59: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 60:
1.144.2.2 paf 61: /**
62: String which knows the language of all it's fragments.
1.41 paf 63:
64: All pieces remember
65: - whether they are tainted or not,
66: and the language which should be used to detaint them
67: */
1.48 paf 68:
1.144.2.28.2. (paf 69:): class String: public PA_Object {
70:):
1.144.2.2 paf 71: public:
1.1 paf 72:
1.106 parser 73: /** piece is tainted or not. the language to use when detaint
74: remember to change String_Untaint_lang_name @ untaint.C along
75: */
1.27 paf 76: enum Untaint_lang {
1.72 paf 77: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 78: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 79: UL_CLEAN, ///< clean
1.122 paf 80: UL_AS_IS, ///< leave all characters intact
81:
1.48 paf 82: UL_PASS_APPENDED,
1.41 paf 83: /**<
84: leave language built into string being appended.
85: just a flag, that value not stored
86: */
1.122 paf 87: UL_TAINTED, ///< tainted, untaint language as assigned later
88: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 89: UL_FILE_SPEC, ///< file specification
1.74 paf 90: UL_HTTP_HEADER, ///< text in HTTP response header
91: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 92: UL_URI, ///< text in uri
93: UL_TABLE, ///< ^table:set body
94: UL_SQL, ///< ^table:sql body
95: UL_JS, ///< JavaScript code
1.102 parser 96: UL_XML, ///< ^dom:set xml
1.48 paf 97: UL_HTML, ///< HTML code (for editing)
1.144.2.7 paf 98: UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 99: };
100:
1.144.2.28.2. (paf 101:): struct Fragment {
102:): Untaint_lang lang; ///< untaint flag, later untaint language
103:): size_t size; ///< size
104:): Fragment(Untaint_lang alang, size_t asize): lang(alang), size(asize) {}
105:): };
106:):
107:): class ArrayFragment: public Array<Fragment> {
108:): void simple_append(element_type src) {
109:): *static_cast<Array<Fragment> *>(this)+=src;
110:): }
111:): public:
112:): ArrayFragment& operator += (element_type src) {
113:): if(count()) { // not empty?
114:): // try to join with last
115:): Fragment& last=get(count()-1);
116:): if(last.lang==src.lang) {
117:): last.size+=src.size;
118:): return *this;
119:): }
120:): }
121:): simple_append(src);
122:): return *this;
123:): }
124:): ArrayFragment& append(const ArrayFragment& src, size_t substr_start, size_t substr_end);
125:): };
126:):
127:): private:
128:):
129:): CORD body; ///< all characters of string
130:): ArrayFragment fragments; ///< fragment language+size info
131:):
1.8 paf 132: public:
133:
1.144.2.18 paf 134: explicit String(const char* src=0, size_t src_size=0, bool tainted=false);
1.14 paf 135: String(const String& src);
1.144.2.28.2. (paf 136:): operator bool() const { return body!=CORD_EMPTY; }
137:): size_t size() const { return CORD_len(body); }
138:):
139:): struct C {
140:): char *str;
141:): size_t size;
142:): operator char *() { return str; }
143:): C(char *astr, size_t asize): str(astr), size(asize) {}
144:): };
145:):
1.69 paf 146: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.28.2. (paf 147:): C cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 148: SQL_Connection *connection=0,
1.144.2.14 paf 149: const Request_charsets *charsets=0) const {
1.88 parser 150:
1.144.2.28.2. (paf 151:): char* cstr=new char[cstr_bufsize(lang, connection, charsets)];
152:): char* eol=store_to(cstr, lang, connection, charsets);
1.144.2.9 paf 153: *eol=0;
1.144.2.28.2. (paf 154:): return C(cstr, eol-cstr);
1.144.2.9 paf 155: }
1.108 parser 156: /// puts pieces to buf
1.144.2.28.2. (paf 157:): void serialize(size_t prolog_size, char *& buf, size_t& buf_size) const;
1.108 parser 158: /// appends pieces from buf to self
1.144.2.18 paf 159: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.46 paf 160: /** append fragment
1.83 paf 161: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 162: */
1.9 paf 163: String& real_append(STRING_APPEND_PARAMS);
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: */
1.144.2.28.2. (paf 172:): //int cmp(CORD substr, size_t this_offset=0) const;
173:): /// String version of @see cmp(CORD, size_t)
174:): //int cmp(const String& src, size_t this_offset=0) const;
175:):
176:): bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; }
177:): bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; }
178:): bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; }
179:): bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; }
180:): bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; }
181:): bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; }
182:):
183:): /// this starts with y
184:): bool starts_with(CORD y) const {
185:): return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
1.80 paf 186: }
1.144.2.28.2. (paf 187:): bool operator == (CORD y) const { return CORD_cmp(body, y)==0; }
188:): bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; }
1.80 paf 189:
1.144.2.28.2. (paf 190:): String& append_to(String& dest, Untaint_lang lang, bool forced) const;
191:): String& append(const String& src, Untaint_lang lang, bool forced=false) {
1.144.2.11 paf 192: return src.append_to(*this, lang, forced);
193: }
1.76 paf 194: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.28.2. (paf 195:): //String& operator << (const String* src) { return append(*src, UL_PASS_APPENDED); }
1.144.2.18 paf 196: String& operator << (const char* src) { return APPEND_CONST(src); }
1.8 paf 197:
1.144.2.7 paf 198: /// simple hash code of string
1.14 paf 199: uint hash_code() const;
1.100 parser 200:
1.142 paf 201: /// extracts first char of a string, if any
202: char first_char() const {
1.144.2.28.2. (paf 203:): return operator bool()?CORD_fetch(body, 0):0;
1.142 paf 204: }
1.54 paf 205:
206: /// extracts [start, finish) piece of string
1.144.2.28.2. (paf 207:): const String& mid(size_t substr_start, size_t substr_end) const;
208:):
209:): /// this is result of pos functions which mean that substr were not found
210:): static const size_t NOT_FOUND;
1.55 paf 211:
1.144.2.28.2. (paf 212:): /**
213:): ignore lang if it's UL_UNSPECIFIED
214:): but when specified: look for substring that lies in ONE fragment in THAT lang
215:): @return position of substr in string, -1 means "not found" [const char* version]
216:): */
217:): size_t pos(CORD substr,
218:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
219:): /// String version of @see pos(const char*, int, Untaint_lang)
220:): size_t pos(const String& substr,
221:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 222:
1.144.2.22 paf 223: void split(ArrayString& result,
1.144.2.28.2. (paf 224:): size_t& pos_after_ref,
225:): const char* delim,
1.144.2.17 paf 226: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 227: void split(ArrayString& result,
1.144.2.28.2. (paf 228:): size_t& pos_after_ref,
1.62 paf 229: const String& delim,
1.144.2.28.2. (paf 230:): Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 231:
1.144.2.28.2. (paf 232:): typedef void (*Row_action)(Table* table, ArrayString* row,
1.136 paf 233: int prestart, int prefinish,
234: int poststart, int postfinish,
1.144.2.9 paf 235: void *info);
1.87 parser 236: /**
1.144.2.5 paf 237: @return table of found items, if any.
1.87 parser 238: table format is defined and fixed[can be used by others]:
239: @verbatim
240: prematch/match/postmatch/1/2/3/...
241: @endverbatim
242: */
1.144.2.28.2. (paf 243:): Table* match(Charset& source_charset,
1.64 paf 244: const String& regexp,
1.144.2.28.2. (paf 245:): const String& options,
1.99 parser 246: Row_action row_action, void *info,
1.144.2.27 paf 247: bool& just_matched) const;
1.87 parser 248: enum Change_case_kind {
249: CC_UPPER,
250: CC_LOWER
251: };
1.144.2.28.2. (paf 252:): String& change_case(Charset& source_charset,
1.144.2.17 paf 253: Change_case_kind kind);
1.144.2.28.2. (paf 254:): const String& replace(const Dictionary& dict) const;
1.144.2.17 paf 255: double as_double();
256: int as_int();
1.137 paf 257:
1.144.2.5 paf 258: size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
259: SQL_Connection *connection=0,
1.144.2.14 paf 260: const Request_charsets *charsets=0) const;
1.79 paf 261: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 262: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
263: SQL_Connection *connection=0,
1.144.2.14 paf 264: const Request_charsets *charsets=0) const;
1.144.2.2 paf 265: /*
1.144.2.1 paf 266: void join_chain(
1.130 paf 267: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.144.2.28.2. (paf 268:): Untaint_lang& joined_lang, const char* & joined_ptr, size_t& joined_size) const;
1.144.2.2 paf 269: */
1.7 paf 270: private: //disabled
271:
1.12 paf 272: String& operator = (const String&) { return *this; }
1.7 paf 273:
1.1 paf 274: };
1.144.2.22 paf 275:
1.144.2.28.2. (paf 276:): const size_t String::NOT_FOUND=(size_t)-1;
1.144.2.7 paf 277:
278: /// simple hash code of string. used by Hash
279: inline uint hash_code(const String& self) {
280: return self.hash_code();
281: }
1.144.2.5 paf 282:
283: // helpers
284:
1.144.2.28.2. (paf 285:): /*
1.144.2.5 paf 286: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
287: for(; fragment<last; fragment++) { \
288: code; \
289: }
290:
291: #define STRING_FOREACH_FRAGMENT(code) \
292: { \
293: element_type *last = felements+fused; \
294: element_type *fragment = felements; \
295: STRING_PREPARED_FOREACH_FRAGMENT(code) \
296: }
1.144.2.28.2. (paf 297:): */
1.1 paf 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