Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.31
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. 1(paf 11:3): static const char* IDENT_STRING_H="$Date: 2003/03/24 16:08:35 $";
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;
7(paf 30:3): 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
7(paf 35:3): #define APPEND_CLEAN(src, length) append(src, length, String::L_CLEAN)
(paf 36:): /// appends piece to String as-is @see String::append
7(paf 37:3): #define APPEND_AS_IS(src, length) append(src, length, String::L_AS_IS)
(paf 38:): /// appends tainted piece to String @see String::append
7(paf 39:3): #define APPEND_TAINTED(src, length) append(src, length, String::L_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:):
7(paf 46:3): class StringBody {
47:3):
48:3): CORD body;
49:3):
50:3): public:
51:3):
52:3): StringBody(): body(CORD_EMPTY) {}
53:3): StringBody(CORD abody): body(abody) {}
2(paf 54:3): explicit StringBody(const char* str, size_t length): body(CORD_EMPTY) {
55:3): append(str, length);
56:3): }
7(paf 57:3):
58:3): void clear() { body=CORD_EMPTY; }
59:3):
60:3): bool operator! () const { return is_empty(); }
61:3):
62:3): uint hash_code() const;
63:3):
0(paf 64:3): const char* cstr() const { return CORD_to_const_char_star(body); }
1(paf 65:3): char* cstrm() const { return CORD_to_char_star(body); }
0(paf 66:3):
7(paf 67:3): size_t length() const { return CORD_len(body); }
68:3):
69:3): bool is_empty() const { return body==CORD_EMPTY; }
70:3):
71:3): void append(const char *str, size_t length) {
2(paf 72:3): body=CORD_cat_char_star(body, str, length);
7(paf 73:3): }
2(paf 74:3): void append(char c) { body=CORD_cat_char(body, c); }
75:3): void append(StringBody src) { body=CORD_cat(body, src.body); }
7(paf 76:3):
77:3): // could not figure out why this operator is needed [should do this chain: string->simple->==]
8(paf 78:3): bool operator < (const StringBody src) const { return CORD_cmp(body, src.body)<0; }
79:3): bool operator > (const StringBody src) const { return CORD_cmp(body, src.body)>0; }
80:3): bool operator <= (const StringBody src) const { return CORD_cmp(body, src.body)<=0; }
81:3): bool operator >= (const StringBody src) const { return CORD_cmp(body, src.body)>=0; }
82:3): bool operator != (const StringBody src) const { return CORD_cmp(body, src.body)!=0; }
83:3): bool operator == (const StringBody src) const { return CORD_cmp(body, src.body)==0; }
7(paf 84:3):
85:3): int ncmp(size_t x_begin, const StringBody y, size_t y_begin, size_t size) const {
86:3): return CORD_ncmp(body, x_begin, y.body, y_begin, size);
87:3): }
88:3):
89:3): char fetch(size_t index) const { return CORD_fetch(body, index); }
1(paf 90:3): StringBody mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
3(paf 91:3): size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
1(paf 92:3): size_t pos(const StringBody substr, size_t offset=0) const {
93:3): if(!substr.length())
94:3): return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
95:3): return CORD_str(body, offset, substr.body);
96:3): }
97:3):
98:3): template<typename I> void for_each(int (*callback)(const char* s, I), I info) const {
99:3): CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info);
100:3): }
101:3):
102:3): void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
7(paf 103:3):
104:3): };
(paf 105:):
1.144.2.2 paf 106: /**
107: String which knows the language of all it's fragments.
1.41 paf 108:
109: All pieces remember
110: - whether they are tainted or not,
111: and the language which should be used to detaint them
112: */
1.144.2.28.2. (paf 113:): class String: public PA_Object {
114:):
2(paf 115:3): // friend class StringBody;
7(paf 116:3):
1.144.2.2 paf 117: public:
1.1 paf 118:
1.106 parser 119: /** piece is tainted or not. the language to use when detaint
120: remember to change String_Untaint_lang_name @ untaint.C along
121: */
1.144.2.28.2. 7(paf 122:3): enum Language {
123:3): L_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
124:3): // these two must go before others, there are checks for >L_AS_IS
125:3): L_CLEAN, ///< clean
126:3): L_AS_IS, ///< leave all characters intact
1.122 paf 127:
1.144.2.28.2. 7(paf 128:3): L_PASS_APPENDED,
1.41 paf 129: /**<
130: leave language built into string being appended.
131: just a flag, that value not stored
132: */
1.144.2.28.2. 7(paf 133:3): L_TAINTED, ///< tainted, untaint language as assigned later
1.122 paf 134: // untaint languages. assigned by ^untaint[lang]{...}
1.144.2.28.2. 7(paf 135:3): L_FILE_SPEC, ///< file specification
136:3): L_HTTP_HEADER, ///< text in HTTP response header
137:3): L_MAIL_HEADER, ///< text in mail header
138:3): L_URI, ///< text in uri
139:3): L_TABLE, ///< ^table:set body
140:3): L_SQL, ///< ^table:sql body
141:3): L_JS, ///< JavaScript code
142:3): L_XML, ///< ^dom:set xml
143:3): L_HTML, ///< HTML code (for editing)
144:3): L_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 145: };
146:
1.144.2.28.2. (paf 147:): struct Fragment {
7(paf 148:3): Language lang; ///< untaint flag, later untaint language
0(paf 149:3): size_t length; ///< length
9(paf 150:3): Fragment(Language alang, size_t asize): lang(alang), length(asize) {
0(paf 151:3): #ifdef _DEBUG
9(paf 152:3): if(alang==L_UNSPECIFIED)
0(paf 153:3): _asm int 3; // MUST NOT
154:3): if(asize==(size_t)-1)
155:3): _asm int 3; // MUST NOT
156:3): #endif
9(paf 157:3): }
(paf 158:): };
159:):
160:): class ArrayFragment: public Array<Fragment> {
6(paf 161:3): void append(element_type src) {
(paf 162:): *static_cast<Array<Fragment> *>(this)+=src;
163:): }
6(paf 164:3): /// hiding from accidental USE, use append_positions
165:3): void append(const ArrayFragment& src, int offset, int limit) {
166:3): static_cast<Array<Fragment> *>(this)->append(src, offset, limit);
167:3): }
(paf 168:): public:
169:): ArrayFragment& operator += (element_type src) {
170:): if(count()) { // not empty?
171:): // try to join with last
172:): Fragment& last=get(count()-1);
173:): if(last.lang==src.lang) {
0(paf 174:3): last.length+=src.length;
(paf 175:): return *this;
176:): }
177:): }
6(paf 178:3): append(src);
(paf 179:): return *this;
180:): }
6(paf 181:3): void append(const ArrayFragment& src) { append(src, 0, 0); }
182:3): void append_positions(const ArrayFragment& src, size_t substr_begin, size_t substr_end);
(paf 183:): };
184:):
185:): struct C {
186:): char *str;
0(paf 187:3): size_t length;
(paf 188:): operator char *() { return str; }
0(paf 189:3): C(char *astr, size_t asize): str(astr), length(asize) {}
(paf 190:): };
191:):
192:): private:
193:):
7(paf 194:3): StringBody body; ///< all characters of string
0(paf 195:3): ArrayFragment fragments; ///< fragment language+length info
(paf 196:):
1.8 paf 197: public:
198:
1.144.2.28.2. 1(paf 199:3): /// @todo check all places for length to be only HELPER length, and never be !=strlen(cstr)
0(paf 200:3): explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
1.14 paf 201: String(const String& src);
1.144.2.28.2. 7(paf 202:3): String(StringBody abody, Language alang): body(abody) {
203:3): fragments+=Fragment(alang, abody.length());
204:3): }
205:3):
206:3): /// for convinient hash lookup
2(paf 207:3): operator const StringBody() const { return body; }
7(paf 208:3):
209:3): bool is_empty() const { return body.is_empty(); }
210:3): size_t length() const { return body.length(); }
(paf 211:):
2(paf 212:3): /// convert to CORD. if 'lang' known, forcing 'lang' to it
1(paf 213:3): StringBody cstr_to_string_body(Language lang=L_AS_IS,
7(paf 214:3): SQL_Connection* connection=0,
(paf 215:): const Request_charsets *charsets=0) const;
216:):
2(paf 217:3): /// convert to constant C string. if 'lang' known, forcing 'lang' to it
7(paf 218:3): const char* cstr(Language lang=L_AS_IS,
219:3): SQL_Connection* connection=0,
1.144.2.14 paf 220: const Request_charsets *charsets=0) const {
1.144.2.28.2. 1(paf 221:3): return cstr_to_string_body(lang, connection, charsets).cstr();
(paf 222:): }
2(paf 223:3): /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
7(paf 224:3): char *cstrm(Language lang=L_AS_IS,
225:3): SQL_Connection* connection=0,
(paf 226:): const Request_charsets *charsets=0) const {
1(paf 227:3): return cstr_to_string_body(lang, connection, charsets).cstrm();
1.144.2.9 paf 228: }
1.108 parser 229: /// puts pieces to buf
1.144.2.28.2. (paf 230:): C serialize(size_t prolog_size) const;
1.108 parser 231: /// appends pieces from buf to self
1.144.2.28.2. 4(paf 232:3): bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
6(paf 233:3): /** append fragment,
9(paf 234:3): @c length can be zero, it will be autocalced
6(paf 235:3): when length<strlen(str) makes heap copy of @c str piece
1.46 paf 236: */
1.144.2.28.2. 9(paf 237:3): String& append(const char* str, size_t length, Language lang);
(paf 238:):
7(paf 239:3): bool operator == (const char* y) const { return body==StringBody(y); }
0(paf 240:3): bool operator != (const char* y) const { return body!=StringBody(y); }
(paf 241:):
242:): /// this starts with y
7(paf 243:3): bool starts_with(const char* y) const {
244:3): return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
1.80 paf 245: }
1.144.2.28.2. 9(paf 246:3): /// x starts with this
247:3): bool this_starts(const char* x) const {
248:3): return StringBody(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
(paf 249:): }
1.80 paf 250:
1.144.2.28.2. 7(paf 251:3): String& append_to(String& dest, Language lang, bool forced) const;
252:3): String& append(const String& src, Language lang, bool forced=false) {
1.144.2.11 paf 253: return src.append_to(*this, lang, forced);
254: }
1.144.2.28.2. 7(paf 255:3): String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
256:3): String& operator << (const char* src) { return append(src, 0, L_AS_IS); }
8(paf 257:3): String& operator << (const StringBody src) {
9(paf 258:3): body.append(src);
259:3): fragments+=Fragment(L_AS_IS, src.length());
260:3): return *this;
261:3): }
1.100 parser 262:
1.142 paf 263: /// extracts first char of a string, if any
264: char first_char() const {
1.144.2.28.2. 7(paf 265:3): return is_empty()?0:body.fetch(0);
1.142 paf 266: }
1.144.2.28.2. 8(paf 267:3):
268:3): bool operator < (const String& src) const { return body<src.body; }
269:3): bool operator > (const String& src) const { return body>src.body; }
270:3): bool operator <= (const String& src) const { return body<=src.body; }
271:3): bool operator >= (const String& src) const { return body>=src.body; }
272:3): bool operator != (const String& src) const { return body!=src.body; }
273:3): bool operator == (const String& src) const { return body==src.body; }
1.54 paf 274:
275: /// extracts [start, finish) piece of string
1.144.2.28.2. 2(paf 276:3): String& mid(size_t substr_begin, size_t substr_end) const;
(paf 277:):
278:): /**
7(paf 279:3): ignore lang if it's L_UNSPECIFIED
(paf 280:): but when specified: look for substring that lies in ONE fragment in THAT lang
281:): @return position of substr in string, -1 means "not found" [const char* version]
282:): */
1(paf 283:3): size_t pos(const StringBody substr,
7(paf 284:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
285:3): /// String version of @see pos(const char*, int, Language)
(paf 286:): size_t pos(const String& substr,
7(paf 287:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.62 paf 288:
1.144.2.22 paf 289: void split(ArrayString& result,
1.144.2.28.2. 2(paf 290:3): size_t& pos_after,
(paf 291:): const char* delim,
2(paf 292:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.144.2.22 paf 293: void split(ArrayString& result,
1.144.2.28.2. 5(paf 294:3): size_t& pos_after,
1.62 paf 295: const String& delim,
1.144.2.28.2. 7(paf 296:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 297:
1.144.2.28.2. 6(paf 298:3): typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 299: int prestart, int prefinish,
300: int poststart, int postfinish,
1.144.2.9 paf 301: void *info);
1.87 parser 302: /**
1.144.2.5 paf 303: @return table of found items, if any.
1.87 parser 304: table format is defined and fixed[can be used by others]:
305: @verbatim
306: prematch/match/postmatch/1/2/3/...
307: @endverbatim
308: */
1.144.2.28.2. (paf 309:): Table* match(Charset& source_charset,
1.64 paf 310: const String& regexp,
1.144.2.28.2. 5(paf 311:3): const String* options,
1.99 parser 312: Row_action row_action, void *info,
1.144.2.27 paf 313: bool& just_matched) const;
1.87 parser 314: enum Change_case_kind {
315: CC_UPPER,
316: CC_LOWER
317: };
1.144.2.28.2. (paf 318:): String& change_case(Charset& source_charset,
5(paf 319:3): Change_case_kind kind) const;
(paf 320:): const String& replace(const Dictionary& dict) const;
321:): double as_double() const;
322:): int as_int() const;
1.137 paf 323:
1.7 paf 324: private: //disabled
325:
1.12 paf 326: String& operator = (const String&) { return *this; }
1.7 paf 327:
1.1 paf 328: };
1.144.2.7 paf 329:
330: /// simple hash code of string. used by Hash
1.144.2.28.2. 8(paf 331:3): inline uint hash_code(const StringBody self) {
1.144.2.7 paf 332: return self.hash_code();
333: }
1.1 paf 334:
335: #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