Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.24
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. 4(paf 11:3): static const char* IDENT_STRING_H="$Date: 2003/03/24 10:21:28 $";
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;
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): uint hash_code() const;
61:3):
0(paf 62:3): const char* cstr() const { return CORD_to_const_char_star(body); }
1(paf 63:3): char* cstrm() const { return CORD_to_char_star(body); }
0(paf 64:3):
7(paf 65:3): size_t length() const { return CORD_len(body); }
66:3):
67:3): bool is_empty() const { return body==CORD_EMPTY; }
68:3):
69:3): void append(const char *str, size_t length) {
2(paf 70:3): body=CORD_cat_char_star(body, str, length);
7(paf 71:3): }
2(paf 72:3): void append(char c) { body=CORD_cat_char(body, c); }
73:3): void append(StringBody src) { body=CORD_cat(body, src.body); }
7(paf 74:3):
75:3): // could not figure out why this operator is needed [should do this chain: string->simple->==]
76:3): bool operator < (const StringBody& src) const { return CORD_cmp(body, src.body)<0; }
77:3): bool operator > (const StringBody& src) const { return CORD_cmp(body, src.body)>0; }
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):
83:3): int ncmp(size_t x_begin, const StringBody y, size_t y_begin, size_t size) const {
84:3): return CORD_ncmp(body, x_begin, y.body, y_begin, size);
85:3): }
86:3):
87:3): char fetch(size_t index) const { return CORD_fetch(body, index); }
1(paf 88:3): StringBody mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
3(paf 89:3): size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
90:3): size_t pos(const StringBody substr, size_t offset=0) const { return CORD_str(body, offset, substr.body); }
1(paf 91:3):
92:3): template<typename I> void for_each(int (*callback)(const char* s, I), I info) const {
93:3): CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info);
94:3): }
95:3):
96:3): void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
7(paf 97:3):
98:3): };
(paf 99:):
1.144.2.2 paf 100: /**
101: String which knows the language of all it's fragments.
1.41 paf 102:
103: All pieces remember
104: - whether they are tainted or not,
105: and the language which should be used to detaint them
106: */
1.144.2.28.2. (paf 107:): class String: public PA_Object {
108:):
2(paf 109:3): // friend class StringBody;
7(paf 110:3):
1.144.2.2 paf 111: public:
1.1 paf 112:
1.106 parser 113: /** piece is tainted or not. the language to use when detaint
114: remember to change String_Untaint_lang_name @ untaint.C along
115: */
1.144.2.28.2. 7(paf 116:3): enum Language {
117:3): L_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
118:3): // these two must go before others, there are checks for >L_AS_IS
119:3): L_CLEAN, ///< clean
120:3): L_AS_IS, ///< leave all characters intact
1.122 paf 121:
1.144.2.28.2. 7(paf 122:3): L_PASS_APPENDED,
1.41 paf 123: /**<
124: leave language built into string being appended.
125: just a flag, that value not stored
126: */
1.144.2.28.2. 7(paf 127:3): L_TAINTED, ///< tainted, untaint language as assigned later
1.122 paf 128: // untaint languages. assigned by ^untaint[lang]{...}
1.144.2.28.2. 7(paf 129:3): L_FILE_SPEC, ///< file specification
130:3): L_HTTP_HEADER, ///< text in HTTP response header
131:3): L_MAIL_HEADER, ///< text in mail header
132:3): L_URI, ///< text in uri
133:3): L_TABLE, ///< ^table:set body
134:3): L_SQL, ///< ^table:sql body
135:3): L_JS, ///< JavaScript code
136:3): L_XML, ///< ^dom:set xml
137:3): L_HTML, ///< HTML code (for editing)
138:3): L_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization
1.27 paf 139: };
140:
1.144.2.28.2. (paf 141:): struct Fragment {
7(paf 142:3): Language lang; ///< untaint flag, later untaint language
0(paf 143:3): size_t length; ///< length
7(paf 144:3): Fragment(Language alang, size_t asize): lang(alang), length(asize) {}
(paf 145:): };
146:):
147:): class ArrayFragment: public Array<Fragment> {
6(paf 148:3): void append(element_type src) {
(paf 149:): *static_cast<Array<Fragment> *>(this)+=src;
150:): }
6(paf 151:3): /// hiding from accidental USE, use append_positions
152:3): void append(const ArrayFragment& src, int offset, int limit) {
153:3): static_cast<Array<Fragment> *>(this)->append(src, offset, limit);
154:3): }
(paf 155:): public:
156:): ArrayFragment& operator += (element_type src) {
157:): if(count()) { // not empty?
158:): // try to join with last
159:): Fragment& last=get(count()-1);
160:): if(last.lang==src.lang) {
0(paf 161:3): last.length+=src.length;
(paf 162:): return *this;
163:): }
164:): }
6(paf 165:3): append(src);
(paf 166:): return *this;
167:): }
6(paf 168:3): void append(const ArrayFragment& src) { append(src, 0, 0); }
169:3): void append_positions(const ArrayFragment& src, size_t substr_begin, size_t substr_end);
(paf 170:): };
171:):
172:): struct C {
173:): char *str;
0(paf 174:3): size_t length;
(paf 175:): operator char *() { return str; }
0(paf 176:3): C(char *astr, size_t asize): str(astr), length(asize) {}
(paf 177:): };
178:):
179:): private:
180:):
7(paf 181:3): StringBody body; ///< all characters of string
0(paf 182:3): ArrayFragment fragments; ///< fragment language+length info
(paf 183:):
1.8 paf 184: public:
185:
1.144.2.28.2. 1(paf 186:3): /// @todo check all places for length to be only HELPER length, and never be !=strlen(cstr)
0(paf 187:3): explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
1.14 paf 188: String(const String& src);
1.144.2.28.2. 7(paf 189:3): String(StringBody abody, Language alang): body(abody) {
190:3): fragments+=Fragment(alang, abody.length());
191:3): }
192:3):
193:3): /// for convinient hash lookup
2(paf 194:3): operator const StringBody() const { return body; }
7(paf 195:3):
196:3): bool is_empty() const { return body.is_empty(); }
197:3): size_t length() const { return body.length(); }
(paf 198:):
2(paf 199:3): /// convert to CORD. if 'lang' known, forcing 'lang' to it
1(paf 200:3): StringBody cstr_to_string_body(Language lang=L_AS_IS,
7(paf 201:3): SQL_Connection* connection=0,
(paf 202:): const Request_charsets *charsets=0) const;
203:):
2(paf 204:3): /// convert to constant C string. if 'lang' known, forcing 'lang' to it
7(paf 205:3): const char* cstr(Language lang=L_AS_IS,
206:3): SQL_Connection* connection=0,
1.144.2.14 paf 207: const Request_charsets *charsets=0) const {
1.144.2.28.2. 1(paf 208:3): return cstr_to_string_body(lang, connection, charsets).cstr();
(paf 209:): }
2(paf 210:3): /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
7(paf 211:3): char *cstrm(Language lang=L_AS_IS,
212:3): SQL_Connection* connection=0,
(paf 213:): const Request_charsets *charsets=0) const {
1(paf 214:3): return cstr_to_string_body(lang, connection, charsets).cstrm();
1.144.2.9 paf 215: }
1.108 parser 216: /// puts pieces to buf
1.144.2.28.2. (paf 217:): C serialize(size_t prolog_size) const;
1.108 parser 218: /// appends pieces from buf to self
1.144.2.28.2. 4(paf 219:3): bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
6(paf 220:3): /** append fragment,
9(paf 221:3): @c length can be zero, it will be autocalced
6(paf 222:3): when length<strlen(str) makes heap copy of @c str piece
1.46 paf 223: */
1.144.2.28.2. 9(paf 224:3): String& append(const char* str, size_t length, Language lang);
(paf 225:):
7(paf 226:3): bool operator == (const char* y) const { return body==StringBody(y); }
0(paf 227:3): bool operator != (const char* y) const { return body!=StringBody(y); }
(paf 228:):
229:): /// this starts with y
7(paf 230:3): bool starts_with(const char* y) const {
231:3): return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
1.80 paf 232: }
1.144.2.28.2. (paf 233:): /// y starts with this
7(paf 234:3): bool this_starts(const char* y) const {
235:3): return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0;
(paf 236:): }
1.80 paf 237:
1.144.2.28.2. 7(paf 238:3): String& append_to(String& dest, Language lang, bool forced) const;
239:3): String& append(const String& src, Language lang, bool forced=false) {
1.144.2.11 paf 240: return src.append_to(*this, lang, forced);
241: }
1.144.2.28.2. 7(paf 242:3): String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
243:3): String& operator << (const char* src) { return append(src, 0, L_AS_IS); }
9(paf 244:3): String& operator << (const StringBody& src) {
245:3): body.append(src);
246:3): fragments+=Fragment(L_AS_IS, src.length());
247:3): return *this;
248:3): }
1.100 parser 249:
1.142 paf 250: /// extracts first char of a string, if any
251: char first_char() const {
1.144.2.28.2. 7(paf 252:3): return is_empty()?0:body.fetch(0);
1.142 paf 253: }
1.144.2.28.2. 8(paf 254:3):
255:3): bool operator < (const String& src) const { return body<src.body; }
256:3): bool operator > (const String& src) const { return body>src.body; }
257:3): bool operator <= (const String& src) const { return body<=src.body; }
258:3): bool operator >= (const String& src) const { return body>=src.body; }
259:3): bool operator != (const String& src) const { return body!=src.body; }
260:3): bool operator == (const String& src) const { return body==src.body; }
1.54 paf 261:
262: /// extracts [start, finish) piece of string
1.144.2.28.2. 2(paf 263:3): String& mid(size_t substr_begin, size_t substr_end) const;
(paf 264:):
265:): /**
7(paf 266:3): ignore lang if it's L_UNSPECIFIED
(paf 267:): but when specified: look for substring that lies in ONE fragment in THAT lang
268:): @return position of substr in string, -1 means "not found" [const char* version]
269:): */
1(paf 270:3): size_t pos(const StringBody substr,
7(paf 271:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
272:3): /// String version of @see pos(const char*, int, Language)
(paf 273:): size_t pos(const String& substr,
7(paf 274:3): size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.62 paf 275:
1.144.2.22 paf 276: void split(ArrayString& result,
1.144.2.28.2. 2(paf 277:3): size_t& pos_after,
(paf 278:): const char* delim,
2(paf 279:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.144.2.22 paf 280: void split(ArrayString& result,
1.144.2.28.2. 5(paf 281:3): size_t& pos_after,
1.62 paf 282: const String& delim,
1.144.2.28.2. 7(paf 283:3): Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 284:
1.144.2.28.2. (paf 285:): typedef void (*Row_action)(Table* table, ArrayString* row,
1.136 paf 286: int prestart, int prefinish,
287: int poststart, int postfinish,
1.144.2.9 paf 288: void *info);
1.87 parser 289: /**
1.144.2.5 paf 290: @return table of found items, if any.
1.87 parser 291: table format is defined and fixed[can be used by others]:
292: @verbatim
293: prematch/match/postmatch/1/2/3/...
294: @endverbatim
295: */
1.144.2.28.2. (paf 296:): Table* match(Charset& source_charset,
1.64 paf 297: const String& regexp,
1.144.2.28.2. (paf 298:): const String& options,
1.99 parser 299: Row_action row_action, void *info,
1.144.2.27 paf 300: bool& just_matched) const;
1.87 parser 301: enum Change_case_kind {
302: CC_UPPER,
303: CC_LOWER
304: };
1.144.2.28.2. (paf 305:): String& change_case(Charset& source_charset,
1.144.2.17 paf 306: Change_case_kind kind);
1.144.2.28.2. (paf 307:): const String& replace(const Dictionary& dict) const;
308:): double as_double() const;
309:): int as_int() const;
1.137 paf 310:
1.7 paf 311: private: //disabled
312:
1.12 paf 313: String& operator = (const String&) { return *this; }
1.7 paf 314:
1.1 paf 315: };
1.144.2.7 paf 316:
317: /// simple hash code of string. used by Hash
1.144.2.28.2. 7(paf 318:3): inline uint hash_code(const StringBody& self) {
1.144.2.7 paf 319: return self.hash_code();
320: }
1.1 paf 321:
322: #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