Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.10
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. 0(paf 11:3): static const char* IDENT_STRING_H="$Date: 2003/03/20 13:09:54 $";
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
1.144.2.28.2. 0(paf 33:3): /*
(paf 34:): /// appends clean piece to String @see String::append
0(paf 35:3): #define APPEND_CLEAN(src, length) append(src, length, String::UL_CLEAN)
(paf 36:): /// appends piece to String as-is @see String::append
0(paf 37:3): #define APPEND_AS_IS(src, length) append(src, length, String::UL_AS_IS)
(paf 38:): /// appends tainted piece to String @see String::append
0(paf 39:3): #define APPEND_TAINTED(src, length) append(src, length, String::UL_TAINTED)
(paf 40:): /// handy: appends const char* piece to String @see String::append
41:): #define APPEND_CONST(src) APPEND_AS_IS(src, 0)
0(paf 42:3): */
(paf 43:): /// this is result of pos functions which mean that substr were not found
44:): #define STRING_NOT_FOUND ((size_t)-1)
45:):
46:):
1.144.2.2 paf 47: /**
48: String which knows the language of all it's fragments.
1.41 paf 49:
50: All pieces remember
51: - whether they are tainted or not,
52: and the language which should be used to detaint them
53: */
1.144.2.28.2. (paf 54:): class String: public PA_Object {
55:):
1.144.2.2 paf 56: public:
1.1 paf 57:
1.106 parser 58: /** piece is tainted or not. the language to use when detaint
59: remember to change String_Untaint_lang_name @ untaint.C along
60: */
1.27 paf 61: enum Untaint_lang {
1.72 paf 62: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 63: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 64: UL_CLEAN, ///< clean
1.122 paf 65: UL_AS_IS, ///< leave all characters intact
66:
1.48 paf 67: UL_PASS_APPENDED,
1.41 paf 68: /**<
69: leave language built into string being appended.
70: just a flag, that value not stored
71: */
1.122 paf 72: UL_TAINTED, ///< tainted, untaint language as assigned later
73: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 74: UL_FILE_SPEC, ///< file specification
1.74 paf 75: UL_HTTP_HEADER, ///< text in HTTP response header
76: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 77: UL_URI, ///< text in uri
78: UL_TABLE, ///< ^table:set body
79: UL_SQL, ///< ^table:sql body
80: UL_JS, ///< JavaScript code
1.102 parser 81: UL_XML, ///< ^dom:set xml
1.48 paf 82: UL_HTML, ///< HTML code (for editing)
1.144.2.7 paf 83: UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 84: };
85:
1.144.2.28.2. (paf 86:): struct Fragment {
87:): Untaint_lang lang; ///< untaint flag, later untaint language
0(paf 88:3): size_t length; ///< length
89:3): Fragment(Untaint_lang alang, size_t asize): lang(alang), length(asize) {}
(paf 90:): };
91:):
92:): class ArrayFragment: public Array<Fragment> {
93:): void simple_append(element_type src) {
94:): *static_cast<Array<Fragment> *>(this)+=src;
95:): }
96:): public:
97:): ArrayFragment& operator += (element_type src) {
98:): if(count()) { // not empty?
99:): // try to join with last
100:): Fragment& last=get(count()-1);
101:): if(last.lang==src.lang) {
0(paf 102:3): last.length+=src.length;
(paf 103:): return *this;
104:): }
105:): }
106:): simple_append(src);
107:): return *this;
108:): }
109:): ArrayFragment& append(const ArrayFragment& src, size_t substr_start, size_t substr_end);
110:): };
111:):
112:): struct C {
113:): char *str;
0(paf 114:3): size_t length;
(paf 115:): operator char *() { return str; }
0(paf 116:3): C(char *astr, size_t asize): str(astr), length(asize) {}
(paf 117:): };
118:):
119:): private:
120:):
121:): CORD body; ///< all characters of string
0(paf 122:3): ArrayFragment fragments; ///< fragment language+length info
(paf 123:):
1.8 paf 124: public:
125:
1.144.2.28.2. 0(paf 126:3): explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
1.14 paf 127: String(const String& src);
1.144.2.28.2. (paf 128:): operator bool() const { return body!=CORD_EMPTY; }
0(paf 129:3): size_t length() const { return CORD_len(body); }
(paf 130:):
131:): /// convert to C string. if 'lang' known, forcing 'lang' to it
132:): CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS,
133:): SQL_Connection *connection=0,
134:): const Request_charsets *charsets=0) const;
135:):
1.69 paf 136: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.28.2. (paf 137:): const char* cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 138: SQL_Connection *connection=0,
1.144.2.14 paf 139: const Request_charsets *charsets=0) const {
1.144.2.28.2. (paf 140:): return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets));
141:): }
142:): /// convert to C string. if 'lang' known, forcing 'lang' to it
143:): char *cstrm(Untaint_lang lang=UL_AS_IS,
144:): SQL_Connection *connection=0,
145:): const Request_charsets *charsets=0) const {
146:): return CORD_to_char_star(cstr_to_cord(lang, connection, charsets));
1.144.2.9 paf 147: }
1.108 parser 148: /// puts pieces to buf
1.144.2.28.2. (paf 149:): C serialize(size_t prolog_size) const;
1.108 parser 150: /// appends pieces from buf to self
1.144.2.18 paf 151: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.144.2.28.2. 0(paf 152:3): /** append fragment,
153:3): @c heap_variable considered as cstr, that is zeroterminated
154:3): */
155:3): String& append_cstr(const char* heap_variable, size_t helper_length, Untaint_lang lang);
156:3): /** append fragment, making heap copy of @c auto_variable
157:3): @c auto_variable considered MAY BE NOT zeroterminated
158:3): if @c length=0 @c auto_variable considered zeroterminated
1.46 paf 159: */
1.144.2.28.2. 0(paf 160:3): String& append_copy(const char* auto_variable, size_t helper_length, Untaint_lang lang);
1.50 paf 161: /**
162: @param partial
163: returns partial match status.
1.51 paf 164: - -1: strings too different
165: - 0: full match
166: - 1: means @c this starts @c src
167: - 2: means @src starts @this
1.50 paf 168: */
1.144.2.28.2. (paf 169:): //int cmp(CORD substr, size_t this_offset=0) const;
170:): /// String version of @see cmp(CORD, size_t)
171:): //int cmp(const String& src, size_t this_offset=0) const;
172:):
173:): bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; }
174:): bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; }
175:): bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; }
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:):
180:): /// this starts with y
181:): bool starts_with(CORD y) const {
182:): return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
1.80 paf 183: }
1.144.2.28.2. (paf 184:): /// y starts with this
185:): bool this_starts(CORD y) const {
186:): return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0;
187:): }
188:): bool operator == (CORD y) const { return CORD_cmp(body, y)==0; }
189:): bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; }
1.80 paf 190:
1.144.2.28.2. (paf 191:): String& append_to(String& dest, Untaint_lang lang, bool forced) const;
192:): String& append(const String& src, Untaint_lang 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); }
1.144.2.28.2. 0(paf 196:3): String& operator << (const char* src) { return append_cstr(src, 0, UL_AS_IS); }
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:): /**
210:): ignore lang if it's UL_UNSPECIFIED
211:): but when specified: look for substring that lies in ONE fragment in THAT lang
212:): @return position of substr in string, -1 means "not found" [const char* version]
213:): */
214:): size_t pos(CORD substr,
215:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
216:): /// String version of @see pos(const char*, int, Untaint_lang)
217:): size_t pos(const String& substr,
218:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 219:
1.144.2.22 paf 220: void split(ArrayString& result,
1.144.2.28.2. (paf 221:): size_t& pos_after_ref,
222:): const char* delim,
1.144.2.17 paf 223: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 224: void split(ArrayString& result,
1.144.2.28.2. (paf 225:): size_t& pos_after_ref,
1.62 paf 226: const String& delim,
1.144.2.28.2. (paf 227:): Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 228:
1.144.2.28.2. (paf 229:): typedef void (*Row_action)(Table* table, ArrayString* row,
1.136 paf 230: int prestart, int prefinish,
231: int poststart, int postfinish,
1.144.2.9 paf 232: void *info);
1.87 parser 233: /**
1.144.2.5 paf 234: @return table of found items, if any.
1.87 parser 235: table format is defined and fixed[can be used by others]:
236: @verbatim
237: prematch/match/postmatch/1/2/3/...
238: @endverbatim
239: */
1.144.2.28.2. (paf 240:): Table* match(Charset& source_charset,
1.64 paf 241: const String& regexp,
1.144.2.28.2. (paf 242:): const String& options,
1.99 parser 243: Row_action row_action, void *info,
1.144.2.27 paf 244: bool& just_matched) const;
1.87 parser 245: enum Change_case_kind {
246: CC_UPPER,
247: CC_LOWER
248: };
1.144.2.28.2. (paf 249:): String& change_case(Charset& source_charset,
1.144.2.17 paf 250: Change_case_kind kind);
1.144.2.28.2. (paf 251:): const String& replace(const Dictionary& dict) const;
252:): double as_double() const;
253:): int as_int() const;
1.137 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: };
1.144.2.7 paf 260:
261: /// simple hash code of string. used by Hash
262: inline uint hash_code(const String& self) {
263: return self.hash_code();
264: }
1.1 paf 265:
266: #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