Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.16
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. 6(paf 11:3): static const char* IDENT_STRING_H="$Date: 2003/03/20 15:58:28 $";
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> {
6(paf 93:3): void append(element_type src) {
(paf 94:): *static_cast<Array<Fragment> *>(this)+=src;
95:): }
6(paf 96:3): /// hiding from accidental USE, use append_positions
97:3): void append(const ArrayFragment& src, int offset, int limit) {
98:3): static_cast<Array<Fragment> *>(this)->append(src, offset, limit);
99:3): }
(paf 100:): public:
101:): ArrayFragment& operator += (element_type src) {
102:): if(count()) { // not empty?
103:): // try to join with last
104:): Fragment& last=get(count()-1);
105:): if(last.lang==src.lang) {
0(paf 106:3): last.length+=src.length;
(paf 107:): return *this;
108:): }
109:): }
6(paf 110:3): append(src);
(paf 111:): return *this;
112:): }
6(paf 113:3): void append(const ArrayFragment& src) { append(src, 0, 0); }
114:3): void append_positions(const ArrayFragment& src, size_t substr_begin, size_t substr_end);
(paf 115:): };
116:):
117:): struct C {
118:): char *str;
0(paf 119:3): size_t length;
(paf 120:): operator char *() { return str; }
0(paf 121:3): C(char *astr, size_t asize): str(astr), length(asize) {}
(paf 122:): };
123:):
124:): private:
125:):
126:): CORD body; ///< all characters of string
0(paf 127:3): ArrayFragment fragments; ///< fragment language+length info
(paf 128:):
1.8 paf 129: public:
130:
1.144.2.28.2. 1(paf 131:3): /// @todo check all places for length to be only HELPER length, and never be !=strlen(cstr)
0(paf 132:3): explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
1.14 paf 133: String(const String& src);
1.144.2.28.2. (paf 134:): operator bool() const { return body!=CORD_EMPTY; }
0(paf 135:3): size_t length() const { return CORD_len(body); }
(paf 136:):
2(paf 137:3): /// convert to CORD. if 'lang' known, forcing 'lang' to it
(paf 138:): CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS,
139:): SQL_Connection *connection=0,
140:): const Request_charsets *charsets=0) const;
141:):
2(paf 142:3): /// convert to constant C string. if 'lang' known, forcing 'lang' to it
(paf 143:): const char* cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 144: SQL_Connection *connection=0,
1.144.2.14 paf 145: const Request_charsets *charsets=0) const {
1.144.2.28.2. (paf 146:): return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets));
147:): }
2(paf 148:3): /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
(paf 149:): char *cstrm(Untaint_lang lang=UL_AS_IS,
150:): SQL_Connection *connection=0,
151:): const Request_charsets *charsets=0) const {
152:): return CORD_to_char_star(cstr_to_cord(lang, connection, charsets));
1.144.2.9 paf 153: }
1.108 parser 154: /// puts pieces to buf
1.144.2.28.2. (paf 155:): C serialize(size_t prolog_size) const;
1.108 parser 156: /// appends pieces from buf to self
1.144.2.18 paf 157: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.144.2.28.2. 6(paf 158:3): /** append fragment,
159:3): @c helper_length can be zero, it will be autocalced
160:3): when length<strlen(str) makes heap copy of @c str piece
1.46 paf 161: */
1.144.2.28.2. 6(paf 162:3): String& append(const char* str, size_t helper_length, Untaint_lang lang);
(paf 163:):
164:): bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; }
165:): bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; }
166:): bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; }
167:): bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; }
168:): bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; }
169:): bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; }
170:):
171:): /// this starts with y
172:): bool starts_with(CORD y) const {
4(paf 173:3): return CORD_ncmp(body, 0/*x_begin*/, y, 0/*y_begin*/, CORD_len(y))==0;
1.80 paf 174: }
1.144.2.28.2. (paf 175:): /// y starts with this
176:): bool this_starts(CORD y) const {
4(paf 177:3): return CORD_ncmp(body, 0/*x_begin*/, y, 0/*y_begin*/, CORD_len(body))==0;
(paf 178:): }
179:): bool operator == (CORD y) const { return CORD_cmp(body, y)==0; }
180:): bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; }
1.80 paf 181:
1.144.2.28.2. (paf 182:): String& append_to(String& dest, Untaint_lang lang, bool forced) const;
183:): String& append(const String& src, Untaint_lang lang, bool forced=false) {
1.144.2.11 paf 184: return src.append_to(*this, lang, forced);
185: }
1.76 paf 186: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.28.2. 6(paf 187:3): String& operator << (const char* src) { return append(src, 0, UL_AS_IS); }
1.8 paf 188:
1.144.2.7 paf 189: /// simple hash code of string
1.14 paf 190: uint hash_code() const;
1.100 parser 191:
1.142 paf 192: /// extracts first char of a string, if any
193: char first_char() const {
1.144.2.28.2. (paf 194:): return operator bool()?CORD_fetch(body, 0):0;
1.142 paf 195: }
1.54 paf 196:
197: /// extracts [start, finish) piece of string
1.144.2.28.2. 4(paf 198:3): const String& mid(size_t substr_begin, size_t substr_end) const;
(paf 199:):
200:): /**
201:): ignore lang if it's UL_UNSPECIFIED
202:): but when specified: look for substring that lies in ONE fragment in THAT lang
203:): @return position of substr in string, -1 means "not found" [const char* version]
204:): */
205:): size_t pos(CORD substr,
206:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
207:): /// String version of @see pos(const char*, int, Untaint_lang)
208:): size_t pos(const String& substr,
209:): size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 210:
1.144.2.22 paf 211: void split(ArrayString& result,
1.144.2.28.2. 5(paf 212:3): size_t& pos_after,
(paf 213:): const char* delim,
1.144.2.17 paf 214: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22 paf 215: void split(ArrayString& result,
1.144.2.28.2. 5(paf 216:3): size_t& pos_after,
1.62 paf 217: const String& delim,
1.144.2.28.2. (paf 218:): Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 219:
1.144.2.28.2. (paf 220:): typedef void (*Row_action)(Table* table, ArrayString* row,
1.136 paf 221: int prestart, int prefinish,
222: int poststart, int postfinish,
1.144.2.9 paf 223: void *info);
1.87 parser 224: /**
1.144.2.5 paf 225: @return table of found items, if any.
1.87 parser 226: table format is defined and fixed[can be used by others]:
227: @verbatim
228: prematch/match/postmatch/1/2/3/...
229: @endverbatim
230: */
1.144.2.28.2. (paf 231:): Table* match(Charset& source_charset,
1.64 paf 232: const String& regexp,
1.144.2.28.2. (paf 233:): const String& options,
1.99 parser 234: Row_action row_action, void *info,
1.144.2.27 paf 235: bool& just_matched) const;
1.87 parser 236: enum Change_case_kind {
237: CC_UPPER,
238: CC_LOWER
239: };
1.144.2.28.2. (paf 240:): String& change_case(Charset& source_charset,
1.144.2.17 paf 241: Change_case_kind kind);
1.144.2.28.2. (paf 242:): const String& replace(const Dictionary& dict) const;
243:): double as_double() const;
244:): int as_int() const;
1.137 paf 245:
1.7 paf 246: private: //disabled
247:
1.12 paf 248: String& operator = (const String&) { return *this; }
1.7 paf 249:
1.1 paf 250: };
1.144.2.7 paf 251:
252: /// simple hash code of string. used by Hash
253: inline uint hash_code(const String& self) {
254: return self.hash_code();
255: }
1.1 paf 256:
257: #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