Annotation of parser3/src/include/pa_string.h, revision 1.147
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.145 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:
1.147 ! paf 11: static const char* IDENT_STRING_H="$Date: 2003/09/25 09:15:02 $";
1.145 paf 12:
13: // includes
1.1 paf 14:
1.4 paf 15: #include "pa_types.h"
1.145 paf 16: #include "pa_array.h"
17:
18: extern "C" { // cord's author forgot to do that
19: #define CORD_NO_IO
20: #include "cord.h"
21: };
1.4 paf 22:
1.146 paf 23: // cord extension
24: /* Returns true if x does contain */
25: /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */
26: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c);
27:
1.145 paf 28: // forwards
1.9 paf 29:
1.145 paf 30: class Charset;
1.135 paf 31: class Table;
1.71 paf 32: class SQL_Connection;
1.101 parser 33: class Dictionary;
1.145 paf 34: class Request_charsets;
35: class String;
36: typedef Array<const String*> ArrayString;
37:
38: /// this is result of pos functions which mean that substr were not found
39: #define STRING_NOT_FOUND ((size_t)-1)
40:
1.146 paf 41: template<typename T>
42: inline size_t get_length(T current) {
43: return current;
44: }
1.147 ! paf 45:
1.42 paf 46: /**
1.146 paf 47: String which knows the lang of all it's langs.
1.41 paf 48:
49: All pieces remember
50: - whether they are tainted or not,
1.146 paf 51: and the lang which should be used to detaint them
1.41 paf 52: */
1.145 paf 53: class String: public PA_Object {
1.1 paf 54: public:
1.48 paf 55:
1.146 paf 56: /** piece is tainted or not. the lang to use when detaint
1.106 parser 57: remember to change String_Untaint_lang_name @ untaint.C along
58: */
1.145 paf 59: enum Language {
1.146 paf 60: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 61: // these two must go before others, there are checks for >L_AS_IS
1.146 paf 62: L_CLEAN='1', ///< clean
1.145 paf 63: L_AS_IS, ///< leave all characters intact
1.122 paf 64:
1.145 paf 65: L_PASS_APPENDED,
1.41 paf 66: /**<
1.146 paf 67: leave lang built into string being appended.
1.41 paf 68: just a flag, that value not stored
69: */
1.146 paf 70: L_TAINTED, ///< tainted, untaint lang as assigned later
71: // untaint langs. assigned by ^untaint[lang]{...}
1.145 paf 72: L_FILE_SPEC, ///< file specification
73: L_HTTP_HEADER, ///< text in HTTP response header
74: L_MAIL_HEADER, ///< text in mail header
75: L_URI, ///< text in uri
76: L_TABLE, ///< ^table:set body
77: L_SQL, ///< ^table:sql body
78: L_JS, ///< JavaScript code
79: L_XML, ///< ^dom:set xml
80: L_HTML, ///< HTML code (for editing)
1.146 paf 81: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 82: };
83:
1.147 ! paf 84: union Languages {
1.146 paf 85:
1.147 ! paf 86: struct {
! 87: Language lang:8;
! 88: int is_not_just_lang:sizeof(CORD)*8-8;
! 89: } opt;
! 90: CORD langs;
1.146 paf 91:
92: template<typename C>
93: CORD make_langs(C current) const {
1.147 ! paf 94: return opt.is_not_just_lang?
1.146 paf 95: langs
1.147 ! paf 96: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 97: }
98:
99: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 ! paf 100: return opt.is_not_just_lang?
1.146 paf 101: CORD_substr(langs, aoffset, alength)
1.147 ! paf 102: :CORD_chars((char)opt.lang, alength);
1.145 paf 103: }
104:
1.146 paf 105: /// appending when 'langs' already contain something [simple cases hanled elsewhere]
106: template<typename C>
107: void append(C current,
108: const CORD to_nonempty_target_langs) {
109: assert(langs);
110:
1.147 ! paf 111: if(opt.is_not_just_lang)
1.146 paf 112: langs=CORD_cat(langs, to_nonempty_target_langs);
113: else { // we were "just lang"
114: size_t current_size=get_length(current);
115: assert(current_size);
116: langs=CORD_cat(
1.147 ! paf 117: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 118: to_nonempty_target_langs); // new piece
119: }
1.145 paf 120: }
1.146 paf 121:
1.145 paf 122: public:
1.146 paf 123:
124: const char* v() const;
125:
126: Languages(): langs(0) {}
1.147 ! paf 127: Languages(Language alang) {
! 128: opt.lang=alang;
! 129: opt.is_not_just_lang=0;
! 130: }
1.146 paf 131:
132: /// MUST be called exactly prior to modification of current [uses it's length]
133: template<typename C>
134: void append(C current, Language alang, size_t asize) {
135: assert(alang);
136: assert(asize);
137:
1.147 ! paf 138: if(!opt.is_not_just_lang)
! 139: if(opt.lang) {
! 140: if(opt.lang==alang) // same length? ignoring
1.146 paf 141: return;
142: } else {
1.147 ! paf 143: opt.lang=alang; // to uninitialized
1.146 paf 144: return;
1.145 paf 145: }
1.146 paf 146:
147: append(current, CORD_chars((char)alang, asize));
148: }
149:
150: /// MUST be called exactly prior to modification of current [uses it's length]
151: template<typename C>
152: void append(C current, size_t appending_length,
153: const Languages src) {
154: assert(appending_length);
155:
156: if(!langs)
157: langs=src.langs; // to uninitialized
1.147 ! paf 158: else if(!src.opt.is_not_just_lang)
! 159: append(current, src.opt.lang, appending_length); // simplifying when simple source
1.146 paf 160: else
161: append(current, src.make_langs(appending_length));
162: }
163:
164: /// MUST be called exactly prior to modification of current [uses it's length]
165: template<typename C>
166: void append(C current,
167: const Languages src, size_t aoffset, size_t alength) {
168: assert(alength);
169:
170: if(!langs) // to uninitialized?
1.147 ! paf 171: if(src.opt.is_not_just_lang)
1.146 paf 172: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
173: else
1.147 ! paf 174: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 175: else
1.147 ! paf 176: if(!opt.is_not_just_lang && !src.opt.is_not_just_lang && opt.lang==src.opt.lang) // both simple & of same language?
1.146 paf 177: return; // ignoring
178: else
179: append(current, src.make_langs(aoffset, alength));
180: }
181:
182: /// checks if we have lang<=alang all from aoffset to aoffset+alength
183: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
184: if(alang==L_UNSPECIFIED) // ignore lang?
185: return true;
186:
1.147 ! paf 187: if(opt.is_not_just_lang)
1.146 paf 188: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
189: else
1.147 ! paf 190: return opt.lang<=alang;
1.146 paf 191: }
192:
193: template<typename C, typename I>
194: void for_each(C current,
195: int callback(char, size_t, I), I info) const {
196:
1.147 ! paf 197: if(opt.is_not_just_lang)
1.146 paf 198: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
199: else
1.147 ! paf 200: callback(opt.lang, get_length(current), info);
1.146 paf 201: }
202:
203: bool invariant(size_t current_length) {
204: if(!langs)
205: return current_length==0;
1.147 ! paf 206: if(opt.is_not_just_lang)
1.146 paf 207: return CORD_len(langs)==current_length;
208: return true; // uncheckable, actually
209: }
210: };
211:
212: class Body {
213:
214: CORD body;
215:
216: public:
217:
218: const char* v() const;
219:
220: Body(): body(CORD_EMPTY) {}
221: Body(CORD abody): body(abody) {
222: assert(!body // no body
223: || *body // ordinary string
224: || body[1]==1 // CONCAT_HDR
225: || body[1]==4 // FN_HDR
226: || body[1]==6 // SUBSTR_HDR
227: );
228: }
229: /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
230: Body(const char* str, size_t helper_length): body(CORD_EMPTY) {
231: append_know_length(str, helper_length?helper_length:strlen(str));
232: }
233: static Body Format(int value);
234:
235: void clear() { body=CORD_EMPTY; }
236:
237: bool operator! () const { return is_empty(); }
238:
239: uint hash_code() const;
240:
241: const char* cstr() const { return CORD_to_const_char_star(body); }
242: char* cstrm() const { return CORD_to_char_star(body); }
243:
244: size_t length() const { return CORD_len(body); }
245:
246: bool is_empty() const { return body==CORD_EMPTY; }
247:
248: void append_know_length(const char *str, size_t known_length) {
249: if(known_length)
250: body=CORD_cat_char_star(body, str, known_length);
251: }
252: void append_strdup_know_length(const char* str, size_t known_length) {
253: if(known_length)
254: append_know_length(pa_strdup(str, known_length), known_length);
255: }
256: void append(char c) { body=CORD_cat_char(body, c); }
257: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
258: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
259:
260: // could not figure out why this operator is needed [should do this chain: string->simple->==]
261: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
262: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
263: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
264: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
265: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
266: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
267:
268: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
269: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 270: }
271:
1.146 paf 272: char fetch(size_t index) const { return CORD_fetch(body, index); }
273: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
274: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
275: size_t pos(const Body substr, size_t offset=0) const {
276: if(!substr.length())
277: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
278: return CORD_str(body, offset, substr.body);
279: }
280: size_t pos(char c,
281: size_t offset=0) const {
282: return CORD_chr(body, offset, c);
1.145 paf 283: }
1.146 paf 284:
285: /* template<typename I> void for_each(int (*callback)(const char* s, I), I info) const {
286: CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info);
287: }*/
288:
289: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
290:
291: /*Body normalize() const {
292: return Body(CORD_balance(body));
293: }*/
1.145 paf 294: };
295:
296: struct C {
297: const char *str;
298: size_t length;
299: operator const char *() { return str; }
300: C(const char *astr, size_t asize): str(astr), length(asize) {}
301: };
302:
303: struct Cm {
304: char *str;
305: size_t length;
306: //operator char *() { return str; }
307: Cm(char *astr, size_t asize): str(astr), length(asize) {}
308: };
309:
310: private:
311:
1.146 paf 312: Languages langs; ///< string characters lang
313: Body body; ///< all characters of string
314:
315: const char* v() const;
316:
317: #define ASSERT_STRING_INVARIANT(string) \
318: assert((string).langs.invariant((string).body.length()))
1.145 paf 319:
1.8 paf 320: public:
321:
1.145 paf 322: explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
323: explicit String(const C cstr, bool tainted=false);
1.146 paf 324: String(Body abody, Language alang): body(abody), langs(alang) {
325: assert(!body.is_empty());
326: ASSERT_STRING_INVARIANT(*this);
327: }
328: String(const String& src): body(src.body), langs(src.langs) {
329: ASSERT_STRING_INVARIANT(*this);
1.145 paf 330: }
331:
332: /// for convinient hash lookup
1.146 paf 333: operator const Body() const { return body; }
1.145 paf 334:
335: bool is_empty() const { return body.is_empty(); }
336: size_t length() const { return body.length(); }
337:
338: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 339: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 340: SQL_Connection* connection=0,
341: const Request_charsets *charsets=0) const;
342:
343: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
344: const char* cstr(Language lang=L_AS_IS,
345: SQL_Connection* connection=0,
346: const Request_charsets *charsets=0) const {
347: return cstr_to_string_body(lang, connection, charsets).cstr();
348: }
349: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
350: char *cstrm(Language lang=L_AS_IS,
351: SQL_Connection* connection=0,
352: const Request_charsets *charsets=0) const {
353: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 354: }
1.108 parser 355: /// puts pieces to buf
1.145 paf 356: Cm serialize(size_t prolog_size) const;
1.108 parser 357: /// appends pieces from buf to self
1.145 paf 358: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 359: /// @see Body::append_know_length
1.145 paf 360: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 361: /// @see Body::append_help_length
1.145 paf 362: String& append_help_length(const char* str, size_t helper_length, Language lang);
363: String& append_strdup(const char* str, size_t helper_length, Language lang);
364:
1.146 paf 365: bool operator == (const char* y) const { return body==Body(y); }
366: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 367:
368: /// this starts with y
369: bool starts_with(const char* y) const {
1.146 paf 370: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 371: }
372: /// x starts with this
373: bool this_starts(const char* x) const {
1.146 paf 374: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 375: }
376:
1.145 paf 377: String& append_to(String& dest, Language lang, bool forced) const;
378: String& append(const String& src, Language lang, bool forced=false) {
379: return src.append_to(*this, lang, forced);
380: }
381: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
382: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 ! paf 383: String& operator << (const Body src);
1.100 parser 384:
1.142 paf 385: /// extracts first char of a string, if any
386: char first_char() const {
1.145 paf 387: return is_empty()?0:body.fetch(0);
1.142 paf 388: }
1.54 paf 389:
1.145 paf 390: bool operator < (const String& src) const { return body<src.body; }
391: bool operator > (const String& src) const { return body>src.body; }
392: bool operator <= (const String& src) const { return body<=src.body; }
393: bool operator >= (const String& src) const { return body>=src.body; }
394: bool operator != (const String& src) const { return body!=src.body; }
395: bool operator == (const String& src) const { return body==src.body; }
396:
1.54 paf 397: /// extracts [start, finish) piece of string
1.145 paf 398: String& mid(size_t substr_begin, size_t substr_end) const;
399:
400: /**
401: ignore lang if it's L_UNSPECIFIED
402: but when specified: look for substring that lies in ONE fragment in THAT lang
403: @return position of substr in string, -1 means "not found" [const char* version]
404: */
1.146 paf 405: size_t pos(const Body substr,
1.145 paf 406: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
407: /// String version of @see pos(const char*, int, Language)
408: size_t pos(const String& substr,
409: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
410: size_t pos(char c,
411: size_t this_offset=0) const {
412: return body.pos(c, this_offset);
413: }
1.55 paf 414:
1.145 paf 415: void split(ArrayString& result,
416: size_t& pos_after,
417: const char* delim,
418: Language lang=L_UNSPECIFIED, int limit=-1) const;
419: void split(ArrayString& result,
420: size_t& pos_after,
1.62 paf 421: const String& delim,
1.145 paf 422: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 423:
1.145 paf 424: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 425: int prestart, int prefinish,
426: int poststart, int postfinish,
1.68 paf 427: void *info);
1.87 parser 428: /**
1.145 paf 429: @return table of found items, if any.
1.87 parser 430: table format is defined and fixed[can be used by others]:
431: @verbatim
432: prematch/match/postmatch/1/2/3/...
433: @endverbatim
434: */
1.145 paf 435: Table* match(Charset& source_charset,
1.64 paf 436: const String& regexp,
1.145 paf 437: const String* options,
1.99 parser 438: Row_action row_action, void *info,
1.145 paf 439: bool& just_matched) const;
1.87 parser 440: enum Change_case_kind {
441: CC_UPPER,
442: CC_LOWER
443: };
1.145 paf 444: String& change_case(Charset& source_charset,
1.87 parser 445: Change_case_kind kind) const;
1.145 paf 446: const String& replace(const Dictionary& dict) const;
1.96 parser 447: double as_double() const;
448: int as_int() const;
1.137 paf 449:
1.7 paf 450: private: //disabled
451:
1.12 paf 452: String& operator = (const String&) { return *this; }
1.7 paf 453:
1.1 paf 454: };
1.119 paf 455:
1.146 paf 456: template<>
457: inline size_t get_length<String::Body>(String::Body body) {
458: return body.length();
459: }
460:
1.145 paf 461: /// simple hash code of string. used by Hash
1.146 paf 462: inline uint hash_code(const String::Body self) {
1.145 paf 463: return self.hash_code();
1.119 paf 464: }
1.147 ! paf 465:
! 466:
! 467: /// now that we've declared specialization we can use it
! 468: inline String& String::operator << (const String::Body src) {
! 469: langs.append(body, L_AS_IS, src.length());
! 470: body<<src;
! 471: return *this;
! 472: }
! 473:
1.1 paf 474:
475: #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