Annotation of parser3/src/include/pa_string.h, revision 1.163
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.156 paf 4: Copyright (c) 2001-2004 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:
1.163 ! paf 11: static const char * const IDENT_STRING_H="$Date: 2004/05/24 12:21:18 $";
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);
1.148 paf 27: size_t CORD_block_count(CORD x);
1.146 paf 28:
1.145 paf 29: // forwards
1.9 paf 30:
1.145 paf 31: class Charset;
1.135 paf 32: class Table;
1.71 paf 33: class SQL_Connection;
1.101 parser 34: class Dictionary;
1.145 paf 35: class Request_charsets;
36: class String;
37: typedef Array<const String*> ArrayString;
38:
1.155 paf 39: // generally useful
40:
41: int pa_atoi(const char* str, const String* problem_source=0);
42: double pa_atod(const char* str, const String* problem_source=0);
43:
1.145 paf 44: /// this is result of pos functions which mean that substr were not found
45: #define STRING_NOT_FOUND ((size_t)-1)
46:
1.146 paf 47: template<typename T>
48: inline size_t get_length(T current) {
49: return current;
50: }
1.147 paf 51:
1.42 paf 52: /**
1.146 paf 53: String which knows the lang of all it's langs.
1.41 paf 54:
55: All pieces remember
56: - whether they are tainted or not,
1.146 paf 57: and the lang which should be used to detaint them
1.41 paf 58: */
1.145 paf 59: class String: public PA_Object {
1.1 paf 60: public:
1.48 paf 61:
1.146 paf 62: /** piece is tainted or not. the lang to use when detaint
1.106 parser 63: remember to change String_Untaint_lang_name @ untaint.C along
1.148 paf 64:
65: WARNING WARNING WARNING WARNING WARNING WARNING
66:
67: pos function compares(<=) languages, that is used in searching
68: for table column separator being L_CLEAN or L_AS_IS.
69: they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check].
70:
71: letters assigned for debugging, but it's important for no language-letter
72: come before L_AS_IS other then L_CLEAN
73:
74: WARNING WARNING WARNING WARNING WARNING WARNING
1.106 parser 75: */
1.145 paf 76: enum Language {
1.146 paf 77: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 78: // these two must go before others, there are checks for >L_AS_IS
1.148 paf 79: L_CLEAN='0', ///< clean WARNING: read above warning before changing
80: L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing
1.122 paf 81:
1.148 paf 82: L_PASS_APPENDED='P',
1.41 paf 83: /**<
1.146 paf 84: leave lang built into string being appended.
1.41 paf 85: just a flag, that value not stored
86: */
1.148 paf 87: L_TAINTED='T', ///< tainted, untaint lang as assigned later
1.146 paf 88: // untaint langs. assigned by ^untaint[lang]{...}
1.148 paf 89: L_FILE_SPEC='F', ///< file specification
90: L_HTTP_HEADER='h', ///< text in HTTP response header
91: L_MAIL_HEADER='m', ///< text in mail header
92: L_URI='U', ///< text in uri
93: L_SQL='Q', ///< ^table:sql body
94: L_JS='J', ///< JavaScript code
95: L_XML='X', ///< ^dom:set xml
96: L_HTML='H', ///< HTML code (for editing)
97: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 98: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 99: };
100:
1.157 paf 101: enum Trim_kind {
102: TRIM_BOTH,
103: TRIM_START,
104: TRIM_END
105: };
106:
1.147 paf 107: union Languages {
1.146 paf 108:
1.147 paf 109: struct {
1.160 paf 110: #ifdef PA_LITTLE_ENDIAN
1.147 paf 111: Language lang:8;
112: int is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 113: #elif defined(PA_BIG_ENDIAN)
114: int is_not_just_lang:sizeof(CORD)*8-8;
115: Language lang:8;
116: #else
117: # error word endianness not determined for some obscure reason
118: #endif
1.147 paf 119: } opt;
120: CORD langs;
1.146 paf 121:
122: template<typename C>
123: CORD make_langs(C current) const {
1.147 paf 124: return opt.is_not_just_lang?
1.146 paf 125: langs
1.162 paf 126: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 127: }
128:
129: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 130: return opt.is_not_just_lang?
1.146 paf 131: CORD_substr(langs, aoffset, alength)
1.162 paf 132: :CORD_chars((char)opt.lang, alength);
1.145 paf 133: }
134:
1.157 paf 135: /// appending when 'langs' already contain something [simple cases handled elsewhere]
1.146 paf 136: template<typename C>
137: void append(C current,
138: const CORD to_nonempty_target_langs) {
139: assert(langs);
140:
1.147 paf 141: if(opt.is_not_just_lang)
1.146 paf 142: langs=CORD_cat(langs, to_nonempty_target_langs);
143: else { // we were "just lang"
144: size_t current_size=get_length(current);
145: assert(current_size);
146: langs=CORD_cat(
1.162 paf 147: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 148: to_nonempty_target_langs); // new piece
149: }
1.145 paf 150: }
1.146 paf 151:
1.145 paf 152: public:
1.146 paf 153:
154: const char* v() const;
155:
156: Languages(): langs(0) {}
1.147 paf 157: Languages(Language alang) {
158: opt.lang=alang;
159: opt.is_not_just_lang=0;
160: }
1.146 paf 161:
162: /// MUST be called exactly prior to modification of current [uses it's length]
163: template<typename C>
164: void append(C current, Language alang, size_t asize) {
165: assert(alang);
166: assert(asize);
167:
1.147 paf 168: if(!opt.is_not_just_lang)
169: if(opt.lang) {
170: if(opt.lang==alang) // same length? ignoring
1.146 paf 171: return;
172: } else {
1.147 paf 173: opt.lang=alang; // to uninitialized
1.146 paf 174: return;
1.145 paf 175: }
1.146 paf 176:
1.162 paf 177: append(current, CORD_chars((char)alang, asize));
1.146 paf 178: }
179:
180: /// MUST be called exactly prior to modification of current [uses it's length]
181: template<typename C>
182: void append(C current, size_t appending_length,
183: const Languages src) {
184: assert(appending_length);
185:
186: if(!langs)
187: langs=src.langs; // to uninitialized
1.147 paf 188: else if(!src.opt.is_not_just_lang)
189: append(current, src.opt.lang, appending_length); // simplifying when simple source
1.146 paf 190: else
191: append(current, src.make_langs(appending_length));
192: }
193:
194: /// MUST be called exactly prior to modification of current [uses it's length]
195: template<typename C>
196: void append(C current,
197: const Languages src, size_t aoffset, size_t alength) {
198: assert(alength);
199:
200: if(!langs) // to uninitialized?
1.147 paf 201: if(src.opt.is_not_just_lang)
1.146 paf 202: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
203: else
1.147 paf 204: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 205: else
1.147 paf 206: 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 207: return; // ignoring
208: else
209: append(current, src.make_langs(aoffset, alength));
210: }
211:
212: /// checks if we have lang<=alang all from aoffset to aoffset+alength
213: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
214: if(alang==L_UNSPECIFIED) // ignore lang?
215: return true;
216:
1.147 paf 217: if(opt.is_not_just_lang)
1.146 paf 218: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
219: else
1.147 paf 220: return opt.lang<=alang;
1.146 paf 221: }
222:
1.148 paf 223: /// @returns count of blocks
224: /// @todo currently there can be adjucent blocks of same language. someday merge them
225: size_t count() const {
226: return opt.is_not_just_lang?
227: CORD_block_count(langs)
228: : opt.lang?
229: 1
230: : 0;
231: };
232:
1.146 paf 233: template<typename C, typename I>
234: void for_each(C current,
235: int callback(char, size_t, I), I info) const {
236:
1.147 paf 237: if(opt.is_not_just_lang)
1.146 paf 238: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
239: else
1.147 paf 240: callback(opt.lang, get_length(current), info);
1.146 paf 241: }
242:
243: bool invariant(size_t current_length) {
244: if(!langs)
245: return current_length==0;
1.147 paf 246: if(opt.is_not_just_lang)
1.146 paf 247: return CORD_len(langs)==current_length;
248: return true; // uncheckable, actually
249: }
250: };
251:
252: class Body {
253:
254: CORD body;
255:
256: public:
257:
258: const char* v() const;
259:
260: Body(): body(CORD_EMPTY) {}
261: Body(CORD abody): body(abody) {
262: assert(!body // no body
263: || *body // ordinary string
264: || body[1]==1 // CONCAT_HDR
265: || body[1]==4 // FN_HDR
266: || body[1]==6 // SUBSTR_HDR
267: );
268: }
269: /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
270: Body(const char* str, size_t helper_length): body(CORD_EMPTY) {
271: append_know_length(str, helper_length?helper_length:strlen(str));
272: }
273: static Body Format(int value);
274:
275: void clear() { body=CORD_EMPTY; }
276:
277: bool operator! () const { return is_empty(); }
278:
279: uint hash_code() const;
280:
281: const char* cstr() const { return CORD_to_const_char_star(body); }
282: char* cstrm() const { return CORD_to_char_star(body); }
283:
284: size_t length() const { return CORD_len(body); }
285:
286: bool is_empty() const { return body==CORD_EMPTY; }
287:
288: void append_know_length(const char *str, size_t known_length) {
289: if(known_length)
290: body=CORD_cat_char_star(body, str, known_length);
291: }
292: void append_strdup_know_length(const char* str, size_t known_length) {
293: if(known_length)
294: append_know_length(pa_strdup(str, known_length), known_length);
295: }
296: void append(char c) { body=CORD_cat_char(body, c); }
297: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
298: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
299:
300: // could not figure out why this operator is needed [should do this chain: string->simple->==]
301: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
302: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
303: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
304: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
305: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
306: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
307:
308: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
309: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 310: }
311:
1.146 paf 312: char fetch(size_t index) const { return CORD_fetch(body, index); }
313: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
314: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
315: size_t pos(const Body substr, size_t offset=0) const {
316: if(!substr.length())
317: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 318:
319: // CORD_str checks for bad offset [CORD_chr does not]
1.146 paf 320: return CORD_str(body, offset, substr.body);
321: }
322: size_t pos(char c,
323: size_t offset=0) const {
1.150 paf 324: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
325: return STRING_NOT_FOUND;
326:
1.146 paf 327: return CORD_chr(body, offset, c);
1.145 paf 328: }
1.146 paf 329:
1.159 paf 330: template<typename I> int for_each(
331: int (*f)(char c, I),
332: I info) const {
333: return CORD_iter(body, (CORD_iter_fn)f, (void*)info);
334: }
335:
336: template<typename I> int for_each(
1.149 paf 337: int (*f1)(char c, I),
338: int (*f2)(const char* s, I),
339: I info) const {
1.159 paf 340: return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 341: }
1.146 paf 342:
343: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
344:
345: /*Body normalize() const {
346: return Body(CORD_balance(body));
347: }*/
1.157 paf 348:
349: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled
350: Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0,
351: size_t* out_start=0, size_t* out_length=0) const;
1.145 paf 352: };
353:
354: struct C {
355: const char *str;
356: size_t length;
357: operator const char *() { return str; }
1.158 paf 358: C(): str(0), length(0) {}
1.145 paf 359: C(const char *astr, size_t asize): str(astr), length(asize) {}
360: };
361:
362: struct Cm {
363: char *str;
364: size_t length;
365: //operator char *() { return str; }
1.158 paf 366: Cm(): str(0), length(0) {}
1.145 paf 367: Cm(char *astr, size_t asize): str(astr), length(asize) {}
368: };
369:
370: private:
371:
1.152 paf 372: Body body; ///< all characters of string
1.146 paf 373: Languages langs; ///< string characters lang
374:
375: const char* v() const;
376:
377: #define ASSERT_STRING_INVARIANT(string) \
378: assert((string).langs.invariant((string).body.length()))
1.145 paf 379:
1.8 paf 380: public:
1.151 paf 381:
382: static const String Empty;
1.8 paf 383:
1.145 paf 384: explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
385: explicit String(const C cstr, bool tainted=false);
1.146 paf 386: String(Body abody, Language alang): body(abody), langs(alang) {
387: ASSERT_STRING_INVARIANT(*this);
388: }
389: String(const String& src): body(src.body), langs(src.langs) {
390: ASSERT_STRING_INVARIANT(*this);
1.145 paf 391: }
392:
393: /// for convinient hash lookup
1.146 paf 394: operator const Body() const { return body; }
1.145 paf 395:
396: bool is_empty() const { return body.is_empty(); }
397: size_t length() const { return body.length(); }
398:
399: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 400: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 401: SQL_Connection* connection=0,
402: const Request_charsets *charsets=0) const;
403:
404: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
405: const char* cstr(Language lang=L_AS_IS,
406: SQL_Connection* connection=0,
407: const Request_charsets *charsets=0) const {
408: return cstr_to_string_body(lang, connection, charsets).cstr();
409: }
410: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
411: char *cstrm(Language lang=L_AS_IS,
412: SQL_Connection* connection=0,
413: const Request_charsets *charsets=0) const {
414: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 415: }
1.108 parser 416: /// puts pieces to buf
1.145 paf 417: Cm serialize(size_t prolog_size) const;
1.108 parser 418: /// appends pieces from buf to self
1.145 paf 419: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 420: /// @see Body::append_know_length
1.145 paf 421: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 422: /// @see Body::append_help_length
1.145 paf 423: String& append_help_length(const char* str, size_t helper_length, Language lang);
424: String& append_strdup(const char* str, size_t helper_length, Language lang);
425:
1.146 paf 426: bool operator == (const char* y) const { return body==Body(y); }
427: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 428:
429: /// this starts with y
430: bool starts_with(const char* y) const {
1.146 paf 431: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 432: }
433: /// x starts with this
434: bool this_starts(const char* x) const {
1.146 paf 435: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 436: }
437:
1.145 paf 438: String& append_to(String& dest, Language lang, bool forced) const;
439: String& append(const String& src, Language lang, bool forced=false) {
440: return src.append_to(*this, lang, forced);
441: }
442: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
443: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 paf 444: String& operator << (const Body src);
1.100 parser 445:
1.142 paf 446: /// extracts first char of a string, if any
447: char first_char() const {
1.145 paf 448: return is_empty()?0:body.fetch(0);
1.142 paf 449: }
1.54 paf 450:
1.145 paf 451: bool operator < (const String& src) const { return body<src.body; }
452: bool operator > (const String& src) const { return body>src.body; }
453: bool operator <= (const String& src) const { return body<=src.body; }
454: bool operator >= (const String& src) const { return body>=src.body; }
455: bool operator != (const String& src) const { return body!=src.body; }
456: bool operator == (const String& src) const { return body==src.body; }
457:
1.54 paf 458: /// extracts [start, finish) piece of string
1.145 paf 459: String& mid(size_t substr_begin, size_t substr_end) const;
460:
461: /**
462: ignore lang if it's L_UNSPECIFIED
463: but when specified: look for substring that lies in ONE fragment in THAT lang
464: @return position of substr in string, -1 means "not found" [const char* version]
465: */
1.146 paf 466: size_t pos(const Body substr,
1.145 paf 467: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
468: /// String version of @see pos(const char*, int, Language)
469: size_t pos(const String& substr,
470: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
471: size_t pos(char c,
472: size_t this_offset=0) const {
473: return body.pos(c, this_offset);
474: }
1.55 paf 475:
1.145 paf 476: void split(ArrayString& result,
477: size_t& pos_after,
478: const char* delim,
479: Language lang=L_UNSPECIFIED, int limit=-1) const;
480: void split(ArrayString& result,
481: size_t& pos_after,
1.62 paf 482: const String& delim,
1.145 paf 483: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 484:
1.145 paf 485: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 486: int prestart, int prefinish,
487: int poststart, int postfinish,
1.68 paf 488: void *info);
1.87 parser 489: /**
1.145 paf 490: @return table of found items, if any.
1.87 parser 491: table format is defined and fixed[can be used by others]:
492: @verbatim
493: prematch/match/postmatch/1/2/3/...
494: @endverbatim
495: */
1.145 paf 496: Table* match(Charset& source_charset,
1.64 paf 497: const String& regexp,
1.145 paf 498: const String* options,
1.99 parser 499: Row_action row_action, void *info,
1.145 paf 500: bool& just_matched) const;
1.87 parser 501: enum Change_case_kind {
502: CC_UPPER,
503: CC_LOWER
504: };
1.145 paf 505: String& change_case(Charset& source_charset,
1.87 parser 506: Change_case_kind kind) const;
1.145 paf 507: const String& replace(const Dictionary& dict) const;
1.157 paf 508: const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0) const;
1.155 paf 509: double as_double() const { return pa_atod(cstr(), this); }
510: int as_int() const { return pa_atoi(cstr(), this); }
1.137 paf 511:
1.7 paf 512: private: //disabled
513:
1.12 paf 514: String& operator = (const String&) { return *this; }
1.7 paf 515:
1.1 paf 516: };
1.119 paf 517:
1.146 paf 518: template<>
519: inline size_t get_length<String::Body>(String::Body body) {
520: return body.length();
521: }
522:
1.145 paf 523: /// simple hash code of string. used by Hash
1.146 paf 524: inline uint hash_code(const String::Body self) {
1.145 paf 525: return self.hash_code();
1.119 paf 526: }
1.147 paf 527:
528:
529: /// now that we've declared specialization we can use it
530: inline String& String::operator << (const String::Body src) {
531: langs.append(body, L_AS_IS, src.length());
532: body<<src;
533: return *this;
534: }
535:
1.1 paf 536:
537: #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