Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.35
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. 5(paf 11:3): static const char* IDENT_STRING_H="$Date: 2003/03/25 10:21:08 $";
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) {}
5(paf 53:3): StringBody(CORD abody): body(abody) {
54:3): #ifdef _DEBUG
55:3): if(body && !*body)
56:3): switch(*(char*)(body+1)) {
57:3): case 1: case 4: case 6: break;
58:3): default: _asm int 3; break;
59:3): }
60:3): #endif
61:3): }
62:3): StringBody(const char* str, size_t length): body(CORD_EMPTY) {
2(paf 63:3): append(str, length);
64:3): }
4(paf 65:3): static StringBody Format(int value);
7(paf 66:3):
67:3): void clear() { body=CORD_EMPTY; }
68:3):
69:3): bool operator! () const { return is_empty(); }
70:3):
71:3): uint hash_code() const;
72:3):
0(paf 73:3): const char* cstr() const { return CORD_to_const_char_star(body); }
1(paf 74:3): char* cstrm() const { return CORD_to_char_star(body); }
0(paf 75:3):
7(paf 76:3): size_t length() const { return CORD_len(body); }
77:3):
78:3): bool is_empty() const { return body==CORD_EMPTY; }
79:3):
80:3): void append(const char *str, size_t length) {
5(paf 81:3): #ifdef _DEBUG
82:3): if(!*str)
83:3): _asm int 3;
84:3): bool were_empty=cstr==CORD_EMPTY;
85:3): #endif
2(paf 86:3): body=CORD_cat_char_star(body, str, length);
5(paf 87:3): #ifdef _DEBUG
88:3): if(were_empty && !*str)
89:3): _asm int 3;
90:3): #endif
7(paf 91:3): }
2(paf 92:3): void append(char c) { body=CORD_cat_char(body, c); }
93:3): void append(StringBody src) { body=CORD_cat(body, src.body); }
7(paf 94:3):
95:3): // could not figure out why this operator is needed [should do this chain: string->simple->==]
8(paf 96:3): bool operator < (const StringBody src) const { return CORD_cmp(body, src.body)<0; }
97:3): bool operator > (const StringBody src) const { return CORD_cmp(body, src.body)>0; }
98:3): bool operator <= (const StringBody src) const { return CORD_cmp(body, src.body)<=0; }
99:3): bool operator >= (const StringBody src) const { return CORD_cmp(body, src.body)>=0; }
100:3): bool operator != (const StringBody src) const { return CORD_cmp(body, src.body)!=0; }
101:3): bool operator == (const StringBody src) const { return CORD_cmp(body, src.body)==0; }
7(paf 102:3):
103:3): int ncmp(size_t x_begin, const StringBody y, size_t y_begin, size_t size) const {
104:3): return CORD_ncmp(body, x_begin, y.body, y_begin, size);
105:3): }
106:3):
107:3): char fetch(size_t index) const { return CORD_fetch(body, index); }
1(paf 108:3): StringBody mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
3(paf 109:3): size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
1(paf 110:3): size_t pos(const StringBody substr, size_t offset=0) const {
111:3): if(!substr.length())
112:3): return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
113:3): return CORD_str(body, offset, substr.body);
114:3): }
2(paf 115:3): size_t pos(char c,
116:3): size_t offset=0) const {
117:3): return CORD_chr(body, offset, c);
118:3): }
1(paf 119:3):
120:3): template<typename I> void for_each(int (*callback)(const char* s, I), I info) const {
121:3): CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info);
122:3): }
123:3):
124:3): void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
7(paf 125:3):
126:3): };
(paf 127:):
1.144.2.2 paf 128: /**
129: String which knows the language of all it's fragments.
1.41 paf 130:
131: All pieces remember
132: - whether they are tainted or not,
133: and the language which should be used to detaint them
134: */
1.144.2.28.2. (paf 135:): class String: public PA_Object {
136:):
2(paf 137:3): // friend class StringBody;
7(paf 138:3):
1.144.2.2 paf 139: public:
1.1 paf 140:
1.106 parser 141: /** piece is tainted or not. the language to use when detaint
142: remember to change String_Untaint_lang_name @ untaint.C along
143: */
1.144.2.28.2. 7(paf 144:3): enum Language {
145:3): L_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
146:3): // these two must go before others, there are checks for >L_AS_IS
147:3): L_CLEAN, ///< clean
148:3): L_AS_IS, ///< leave all characters intact
1.122 paf 149:
1.144.2.28.2. 7(paf 150:3): L_PASS_APPENDED,
1.41 paf 151: /**<
152: leave language built into string being appended.
153: just a flag, that value not stored
154: */
1.144.2.28.2. 7(paf 155:3): L_TAINTED, ///< tainted, untaint language as assigned later
1.122 paf 156: // untaint languages. assigned by ^untaint[lang]{...}
1.144.2.28.2. 7(paf 157:3): L_FILE_SPEC, ///< file specification
158:3): L_HTTP_HEADER, ///< text in HTTP response header
159:3): L_MAIL_HEADER, ///< text in mail header
160:3): L_URI, ///< text in uri
161:3): L_TABLE, ///< ^table:set body
162:3): L_SQL, ///< ^table:sql body
163:3): L_JS, ///< JavaScript code
164:3): L_XML, ///< ^dom:set xml
165:3): L_HTML, ///< HTML code (for editing)
166:3): L_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 167: };
168:
1.144.2.28.2. (paf 169:): struct Fragment {
7(paf 170:3): Language lang; ///< untaint flag, later untaint language
0(paf 171:3): size_t length; ///< length
9(paf 172:3): Fragment(Language alang, size_t asize): lang(alang), length(asize) {
0(paf 173:3): #ifdef _DEBUG
9(paf 174:3): if(alang==L_UNSPECIFIED)
0(paf 175:3): _asm int 3; // MUST NOT
176:3): if(asize==(size_t)-1)
177:3): _asm int 3; // MUST NOT
178:3): #endif
9(paf 179:3): }
(paf 180:): };
181:):
182:): class ArrayFragment: public Array<Fragment> {
6(paf 183:3): void append(element_type src) {
(paf 184:): *static_cast<Array<Fragment> *>(this)+=src;
185:): }
6(paf 186:3): /// hiding from accidental USE, use append_positions
187:3): void append(const ArrayFragment& src, int offset, int limit) {
188:3): static_cast<Array<Fragment> *>(this)->append(src, offset, limit);
189:3): }
(paf 190:): public:
191:): ArrayFragment& operator += (element_type src) {
192:): if(count()) { // not empty?
193:): // try to join with last
194:): Fragment& last=get(count()-1);
195:): if(last.lang==src.lang) {
0(paf 196:3): last.length+=src.length;
(paf 197:): return *this;
198:): }
199:): }
6(paf 200:3): append(src);
(paf 201:): return *this;
202:): }
6(paf 203:3): void append(const ArrayFragment& src) { append(src, 0, 0); }
204:3): void append_positions(const ArrayFragment& src, size_t substr_begin, size_t substr_end);
(paf 205:): };
206:):
207:): struct C {
208:): char *str;
0(paf 209:3): size_t length;
(paf 210:): operator char *() { return str; }
0(paf 211:3): C(char *astr, size_t asize): str(astr), length(asize) {}
(paf 212:): };
213:):
214:): private:
215:):
7(paf 216:3): StringBody body; ///< all characters of string
0(paf 217:3): ArrayFragment fragments; ///< fragment language+length info
(paf 218:):
1.8 paf 219: public:
220:
1.144.2.28.2. 1(paf 221:3): /// @todo check all places for length to be only HELPER length, and never be !=strlen(cstr)
0(paf 222:3): explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
1.14 paf 223: String(const String& src);
1.144.2.28.2. 7(paf 224:3): String(StringBody abody, Language alang): body(abody) {
225:3): fragments+=Fragment(alang, abody.length());
226:3): }
227:3):
228:3): /// for convinient hash lookup
2(paf 229:3): operator const StringBody() const { return body; }
7(paf 230:3):
231:3): bool is_empty() const { return body.is_empty(); }
232:3): size_t length() const { return body.length(); }
(paf 233:):
2(paf 234:3): /// convert to CORD. if 'lang' known, forcing 'lang' to it
1(paf 235:3): StringBody cstr_to_string_body(Language lang=L_AS_IS,
7(paf 236:3): SQL_Connection* connection=0,
(paf 237:): const Request_charsets *charsets=0) const;
238:):
2(paf 239:3): /// convert to constant C string. if 'lang' known, forcing 'lang' to it
7(paf 240:3): const char* cstr(Language lang=L_AS_IS,
241:3): SQL_Connection* connection=0,
1.144.2.14 paf 242: const Request_charsets *charsets=0) const {
1.144.2.28.2. 1(paf 243:3): return cstr_to_string_body(lang, connection, charsets).cstr();
(paf 244:): }
2(paf 245:3): /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
7(paf 246:3): char *cstrm(Language lang=L_AS_IS,
247:3): SQL_Connection* connection=0,
(paf 248:): const Request_charsets *charsets=0) const {
1(paf 249:3): return cstr_to_string_body(lang, connection, charsets).cstrm();
1.144.2.9 paf 250: }
1.108 parser 251: /// puts pieces to buf
1.144.2.28.2. (paf 252:): C serialize(size_t prolog_size) const;
1.108 parser 253: /// appends pieces from buf to self
1.144.2.28.2. 4(paf 254:3): bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
6(paf 255:3): /** append fragment,
9(paf 256:3): @c length can be zero, it will be autocalced
6(paf 257:3): when length<strlen(str) makes heap copy of @c str piece
1.46 paf 258: */
1.144.2.28.2. 9(paf 259:3): String& append(const char* str, size_t length, Language lang);
(paf 260:):
7(paf 261:3): bool operator == (const char* y) const { return body==StringBody(y); }
0(paf 262:3): bool operator != (const char* y) const { return body!=StringBody(y); }
(paf 263:):
264:): /// this starts with y
7(paf 265:3): bool starts_with(const char* y) const {
266:3): return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
1.80 paf 267: }
1.144.2.28.2. 9(paf 268:3): /// x starts with this
269:3): bool this_starts(const char* x) const {
270:3): return StringBody(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
(paf 271:): }
1.80 paf 272:
1.144.2.28.2. 7(paf 273:3): String& append_to(String& dest, Language lang, bool forced) const;
274:3): String& append(const String& src, Language lang, bool forced=false) {
1.144.2.11 paf 275: return src.append_to(*this, lang, forced);
276: }
1.144.2.28.2. 7(paf 277:3): String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
278:3): String& operator << (const char* src) { return append(src, 0, L_AS_IS); }
8(paf 279:3): String& operator << (const StringBody src) {
9(paf 280:3): body.append(src);
281:3): fragments+=Fragment(L_AS_IS, src.length());
282:3): return *this;
283:3): }
1.100 parser 284:
1.142 paf 285: /// extracts first char of a string, if any
286: char first_char() const {
1.144.2.28.2. 7(paf 287:3): return is_empty()?0:body.fetch(0);
1.142 paf 288: }
1.144.2.28.2. 8(paf 289:3):
290:3): bool operator < (const String& src) const { return body<src.body; }
291:3): bool operator > (const String& src) const { return body>src.body; }
292:3): bool operator <= (const String& src) const { return body<=src.body; }
293:3): bool operator >= (const String& src) const { return body>=src.body; }
294:3): bool operator != (const String& src) const { return body!=src.body; }
295:3): bool operator == (const String& src) const { return body==src.body; }
1.54 paf 296:
297: /// extracts [start, finish) piece of string
1.144.2.28.2. 2(paf 298:3): String& mid(size_t substr_begin, size_t substr_end) const;
(paf 299:):
300:): /**
7(paf 301:3): ignore lang if it's L_UNSPECIFIED
(paf 302:): but when specified: look for substring that lies in ONE fragment in THAT lang
303:): @return position of substr in string, -1 means "not found" [const char* version]
304:): */
1(paf 305:3): size_t pos(const StringBody substr,
7(paf 306:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
307:3): /// String version of @see pos(const char*, int, Language)
(paf 308:): size_t pos(const String& substr,
7(paf 309:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
2(paf 310:3): size_t pos(char c,
311:3): size_t this_offset=0) const {
312:3): return body.pos(c, this_offset);
313:3): }
1.62 paf 314:
1.144.2.22 paf 315: void split(ArrayString& result,
1.144.2.28.2. 2(paf 316:3): size_t& pos_after,
(paf 317:): const char* delim,
2(paf 318:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.144.2.22 paf 319: void split(ArrayString& result,
1.144.2.28.2. 5(paf 320:3): size_t& pos_after,
1.62 paf 321: const String& delim,
1.144.2.28.2. 7(paf 322:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 323:
1.144.2.28.2. 6(paf 324:3): typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 325: int prestart, int prefinish,
326: int poststart, int postfinish,
1.144.2.9 paf 327: void *info);
1.87 parser 328: /**
1.144.2.5 paf 329: @return table of found items, if any.
1.87 parser 330: table format is defined and fixed[can be used by others]:
331: @verbatim
332: prematch/match/postmatch/1/2/3/...
333: @endverbatim
334: */
1.144.2.28.2. (paf 335:): Table* match(Charset& source_charset,
1.64 paf 336: const String& regexp,
1.144.2.28.2. 5(paf 337:3): const String* options,
1.99 parser 338: Row_action row_action, void *info,
1.144.2.27 paf 339: bool& just_matched) const;
1.87 parser 340: enum Change_case_kind {
341: CC_UPPER,
342: CC_LOWER
343: };
1.144.2.28.2. (paf 344:): String& change_case(Charset& source_charset,
5(paf 345:3): Change_case_kind kind) const;
(paf 346:): const String& replace(const Dictionary& dict) const;
347:): double as_double() const;
348:): int as_int() const;
1.137 paf 349:
1.7 paf 350: private: //disabled
351:
1.12 paf 352: String& operator = (const String&) { return *this; }
1.7 paf 353:
1.1 paf 354: };
1.144.2.7 paf 355:
356: /// simple hash code of string. used by Hash
1.144.2.28.2. 8(paf 357:3): inline uint hash_code(const StringBody self) {
1.144.2.7 paf 358: return self.hash_code();
359: }
1.1 paf 360:
361: #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