Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.8
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: 2003/03/20 09:41:45 $";
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:): extern "C" { // cord's author forgot to do that
19:): #include "cord.h"
20:): };
21:):
1.144.2.12 paf 22: // forwards
1.144.2.10 paf 23:
1.144.2.28 paf 24: class Charset;
1.144.2.28.2. (paf 25:): class Table;
1.144.2.5 paf 26: class SQL_Connection;
27: class Dictionary;
1.144.2.15 paf 28: class Request_charsets;
1.144.2.28.2. (paf 29:): class String;
30:): typedef Array<const String* > ArrayString;
1.144.2.5 paf 31:
32: // helpers
33:
1.9 paf 34: #ifndef NO_STRING_ORIGIN
1.33 paf 35: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 36: const char* src, size_t size, \
1.144.2.28.2. (paf 37:): Untaint_lang lang, \
1.144.2.18 paf 38: const char* file, uint line
1.54 paf 39: /// appends piece to String @see String::real_append
40: # define APPEND(src, size, lang, file, line) \
41: real_append(src, size, lang, file, line)
1.9 paf 42: #else
1.33 paf 43: # define STRING_APPEND_PARAMS \
1.144.2.18 paf 44: const char* src, \
1.33 paf 45: size_t size, \
1.144.2.28.2. (paf 46:): Untaint_lang lang
1.54 paf 47: /// appends piece to String @see String::real_append
48: # define APPEND(src, size, lang, file, line) \
49: real_append(src, size, lang)
50: #endif
1.83 paf 51: /// appends clean piece to String @see String::real_append
1.54 paf 52: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 53: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 54: /// appends piece to String as-is @see String::real_append
55: #define APPEND_AS_IS(src, size, file, line) \
56: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 57: /// appends tainted piece to String @see String::real_append
1.54 paf 58: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 59: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 60: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 61: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 62:
1.144.2.28.2. (paf 63:): /// this is result of pos functions which mean that substr were not found
64:): #define STRING_NOT_FOUND ((size_t)-1)
65:):
66:):
1.144.2.2 paf 67: /**
68: String which knows the language of all it's fragments.
1.41 paf 69:
70: All pieces remember
71: - whether they are tainted or not,
72: and the language which should be used to detaint them
73: */
1.144.2.28.2. (paf 74:): class String: public PA_Object {
75:):
1.144.2.2 paf 76: public:
1.1 paf 77:
1.106 parser 78: /** piece is tainted or not. the language to use when detaint
79: remember to change String_Untaint_lang_name @ untaint.C along
80: */
1.27 paf 81: enum Untaint_lang {
1.72 paf 82: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 83: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 84: UL_CLEAN, ///< clean
1.122 paf 85: UL_AS_IS, ///< leave all characters intact
86:
1.48 paf 87: UL_PASS_APPENDED,
1.41 paf 88: /**<
89: leave language built into string being appended.
90: just a flag, that value not stored
91: */
1.122 paf 92: UL_TAINTED, ///< tainted, untaint language as assigned later
93: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 94: UL_FILE_SPEC, ///< file specification
1.74 paf 95: UL_HTTP_HEADER, ///< text in HTTP response header
96: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 97: UL_URI, ///< text in uri
98: UL_TABLE, ///< ^table:set body
99: UL_SQL, ///< ^table:sql body
100: UL_JS, ///< JavaScript code
1.102 parser 101: UL_XML, ///< ^dom:set xml
1.48 paf 102: UL_HTML, ///< HTML code (for editing)
1.144.2.7 paf 103: UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 104: };
105:
1.144.2.28.2. (paf 106:): struct Fragment {
107:): Untaint_lang lang; ///< untaint flag, later untaint language
108:): size_t size; ///< size
109:): Fragment(Untaint_lang alang, size_t asize): lang(alang), size(asize) {}
110:): };
111:):
112:): class ArrayFragment: public Array<Fragment> {
113:): void simple_append(element_type src) {
114:): *static_cast<Array<Fragment> *>(this)+=src;
115:): }
116:): public:
117:): ArrayFragment& operator += (element_type src) {
118:): if(count()) { // not empty?
119:): // try to join with last
120:): Fragment& last=get(count()-1);
121:): if(last.lang==src.lang) {
122:): last.size+=src.size;
123:): return *this;
124:): }
125:): }
126:): simple_append(src);
127:): return *this;
128:): }
129:): ArrayFragment& append(const ArrayFragment& src, size_t substr_start, size_t substr_end);
130:): };
131:):
132:): struct C {
133:): char *str;
134:): size_t size;
135:): operator char *() { return str; }
136:): C(char *astr, size_t asize): str(astr), size(asize) {}
137:): };
138:):
139:): private:
140:):
141:): CORD body; ///< all characters of string
142:): ArrayFragment fragments; ///< fragment language+size info
143:):
1.8 paf 144: public:
145:
1.144.2.18 paf 146: explicit String(const char* src=0, size_t src_size=0, bool tainted=false);
1.14 paf 147: String(const String& src);
1.144.2.28.2. (paf 148:): operator bool() const { return body!=CORD_EMPTY; }
149:): size_t size() const { return CORD_len(body); }
150:):
151:): /// convert to C string. if 'lang' known, forcing 'lang' to it
152:): CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS,
153:): SQL_Connection *connection=0,
154:): const Request_charsets *charsets=0) const;
155:):
1.69 paf 156: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.28.2. (paf 157:): const char* cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 158: SQL_Connection *connection=0,
1.144.2.14 paf 159: const Request_charsets *charsets=0) const {
1.144.2.28.2. (paf 160:): return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets));
161:): }
162:): /// convert to C string. if 'lang' known, forcing 'lang' to it
163:): char *cstrm(Untaint_lang lang=UL_AS_IS,
164:): SQL_Connection *connection=0,
165:): const Request_charsets *charsets=0) const {
166:): return CORD_to_char_star(cstr_to_cord(lang, connection, charsets));
1.144.2.9 paf 167: }
1.108 parser 168: /// puts pieces to buf
1.144.2.28.2. (paf 169:): C serialize(size_t prolog_size) const;
1.108 parser 170: /// appends pieces from buf to self
1.144.2.18 paf 171: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.46 paf 172: /** append fragment
1.83 paf 173: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 174: */
1.9 paf 175: String& real_append(STRING_APPEND_PARAMS);
1.50 paf 176: /**
177: @param partial
178: returns partial match status.
1.51 paf 179: - -1: strings too different
180: - 0: full match
181: - 1: means @c this starts @c src
182: - 2: means @src starts @this
1.50 paf 183: */
1.144.2.28.2. (paf 184:): //int cmp(CORD substr, size_t this_offset=0) const;
185:): /// String version of @see cmp(CORD, size_t)
186:): //int cmp(const String& src, size_t this_offset=0) const;
187:):
188:): bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; }
189:): bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; }
190:): bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; }
191:): bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; }
192:): bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; }
193:): bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; }
194:):
195:): /// this starts with y
196:): bool starts_with(CORD y) const {
197:): return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
1.80 paf 198: }
1.144.2.28.2. (paf 199:): /// y starts with this
200:): bool this_starts(CORD y) const {
201:): return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
202:): }
203:): bool operator == (CORD y) const { return CORD_cmp(body, y)==0; }
204:): bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; }
1.80 paf 205:
1.144.2.28.2. (paf 206:): String& append_to(String& dest, Untaint_lang lang, bool forced) const;
207:): String& append(const String& src, Untaint_lang lang, bool forced=false) {
1.144.2.11 paf 208: return src.append_to(*this, lang, forced);
209: }
1.76 paf 210: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.28.2. (paf 211:): //String& operator << (const String* src) { return append(*src, UL_PASS_APPENDED); }
1.144.2.18 paf 212: String& operator << (const char* src) { return APPEND_CONST(src); }
1.8 paf 213:
1.144.2.7 paf 214: /// simple hash code of string
1.14 paf 215: uint hash_code() const;
1.100 parser 216:
1.142 paf 217: /// extracts first char of a string, if any
218: char first_char() const {
1.144.2.28.2. (paf 219:): return operator bool()?CORD_fetch(body, 0):0;
1.142 paf 220: }
1.54 paf 221:
222: /// extracts [start, finish) piece of string
1.144.2.28.2. (paf 223:): const String& mid(size_t substr_start, size_t substr_end) const;
224:):
225:): /**
226:): ignore lang if it's UL_UNSPECIFIED
227:): but when specified: look for substring that lies in ONE fragment in THAT lang
228:): @return position of substr in string, -1 means "not found" [const char* version]
229:): */
230:): size_t pos(CORD substr,
231:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
232:): /// String version of @see pos(const char*, int, Untaint_lang)
233:): size_t pos(const String& substr,
234:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 235:
1.144.2.22 paf 236: void split(ArrayString& result,
1.144.2.28.2. (paf 237:): size_t& pos_after_ref,
238:): const char* delim,
1.144.2.17 paf 239: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 240: void split(ArrayString& result,
1.144.2.28.2. (paf 241:): size_t& pos_after_ref,
1.62 paf 242: const String& delim,
1.144.2.28.2. (paf 243:): Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 244:
1.144.2.28.2. (paf 245:): typedef void (*Row_action)(Table* table, ArrayString* row,
1.136 paf 246: int prestart, int prefinish,
247: int poststart, int postfinish,
1.144.2.9 paf 248: void *info);
1.87 parser 249: /**
1.144.2.5 paf 250: @return table of found items, if any.
1.87 parser 251: table format is defined and fixed[can be used by others]:
252: @verbatim
253: prematch/match/postmatch/1/2/3/...
254: @endverbatim
255: */
1.144.2.28.2. (paf 256:): Table* match(Charset& source_charset,
1.64 paf 257: const String& regexp,
1.144.2.28.2. (paf 258:): const String& options,
1.99 parser 259: Row_action row_action, void *info,
1.144.2.27 paf 260: bool& just_matched) const;
1.87 parser 261: enum Change_case_kind {
262: CC_UPPER,
263: CC_LOWER
264: };
1.144.2.28.2. (paf 265:): String& change_case(Charset& source_charset,
1.144.2.17 paf 266: Change_case_kind kind);
1.144.2.28.2. (paf 267:): const String& replace(const Dictionary& dict) const;
268:): double as_double() const;
269:): int as_int() const;
1.137 paf 270:
1.7 paf 271: private: //disabled
272:
1.12 paf 273: String& operator = (const String&) { return *this; }
1.7 paf 274:
1.1 paf 275: };
1.144.2.7 paf 276:
277: /// simple hash code of string. used by Hash
278: inline uint hash_code(const String& self) {
279: return self.hash_code();
280: }
1.1 paf 281:
282: #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