Annotation of parser3/src/include/pa_string.h, revision 1.150
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:
1.150 ! paf 11: static const char* IDENT_STRING_H="$Date: 2003/09/26 13:36:35 $";
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:
39: /// this is result of pos functions which mean that substr were not found
40: #define STRING_NOT_FOUND ((size_t)-1)
41:
1.146 paf 42: template<typename T>
43: inline size_t get_length(T current) {
44: return current;
45: }
1.147 paf 46:
1.42 paf 47: /**
1.146 paf 48: String which knows the lang of all it's langs.
1.41 paf 49:
50: All pieces remember
51: - whether they are tainted or not,
1.146 paf 52: and the lang which should be used to detaint them
1.41 paf 53: */
1.145 paf 54: class String: public PA_Object {
1.1 paf 55: public:
1.48 paf 56:
1.146 paf 57: /** piece is tainted or not. the lang to use when detaint
1.106 parser 58: remember to change String_Untaint_lang_name @ untaint.C along
1.148 paf 59:
60: WARNING WARNING WARNING WARNING WARNING WARNING
61:
62: pos function compares(<=) languages, that is used in searching
63: for table column separator being L_CLEAN or L_AS_IS.
64: they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check].
65:
66: letters assigned for debugging, but it's important for no language-letter
67: come before L_AS_IS other then L_CLEAN
68:
69: WARNING WARNING WARNING WARNING WARNING WARNING
1.106 parser 70: */
1.145 paf 71: enum Language {
1.146 paf 72: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 73: // these two must go before others, there are checks for >L_AS_IS
1.148 paf 74: L_CLEAN='0', ///< clean WARNING: read above warning before changing
75: L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing
1.122 paf 76:
1.148 paf 77: L_PASS_APPENDED='P',
1.41 paf 78: /**<
1.146 paf 79: leave lang built into string being appended.
1.41 paf 80: just a flag, that value not stored
81: */
1.148 paf 82: L_TAINTED='T', ///< tainted, untaint lang as assigned later
1.146 paf 83: // untaint langs. assigned by ^untaint[lang]{...}
1.148 paf 84: L_FILE_SPEC='F', ///< file specification
85: L_HTTP_HEADER='h', ///< text in HTTP response header
86: L_MAIL_HEADER='m', ///< text in mail header
87: L_URI='U', ///< text in uri
88: L_TABLE='L', ///< ^table:set body
89: L_SQL='Q', ///< ^table:sql body
90: L_JS='J', ///< JavaScript code
91: L_XML='X', ///< ^dom:set xml
92: L_HTML='H', ///< HTML code (for editing)
93: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 94: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 95: };
96:
1.147 paf 97: union Languages {
1.146 paf 98:
1.147 paf 99: struct {
100: Language lang:8;
101: int is_not_just_lang:sizeof(CORD)*8-8;
102: } opt;
103: CORD langs;
1.146 paf 104:
105: template<typename C>
106: CORD make_langs(C current) const {
1.147 paf 107: return opt.is_not_just_lang?
1.146 paf 108: langs
1.147 paf 109: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 110: }
111:
112: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 113: return opt.is_not_just_lang?
1.146 paf 114: CORD_substr(langs, aoffset, alength)
1.147 paf 115: :CORD_chars((char)opt.lang, alength);
1.145 paf 116: }
117:
1.146 paf 118: /// appending when 'langs' already contain something [simple cases hanled elsewhere]
119: template<typename C>
120: void append(C current,
121: const CORD to_nonempty_target_langs) {
122: assert(langs);
123:
1.147 paf 124: if(opt.is_not_just_lang)
1.146 paf 125: langs=CORD_cat(langs, to_nonempty_target_langs);
126: else { // we were "just lang"
127: size_t current_size=get_length(current);
128: assert(current_size);
129: langs=CORD_cat(
1.147 paf 130: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 131: to_nonempty_target_langs); // new piece
132: }
1.145 paf 133: }
1.146 paf 134:
1.145 paf 135: public:
1.146 paf 136:
137: const char* v() const;
138:
139: Languages(): langs(0) {}
1.147 paf 140: Languages(Language alang) {
141: opt.lang=alang;
142: opt.is_not_just_lang=0;
143: }
1.146 paf 144:
145: /// MUST be called exactly prior to modification of current [uses it's length]
146: template<typename C>
147: void append(C current, Language alang, size_t asize) {
148: assert(alang);
149: assert(asize);
150:
1.147 paf 151: if(!opt.is_not_just_lang)
152: if(opt.lang) {
153: if(opt.lang==alang) // same length? ignoring
1.146 paf 154: return;
155: } else {
1.147 paf 156: opt.lang=alang; // to uninitialized
1.146 paf 157: return;
1.145 paf 158: }
1.146 paf 159:
160: append(current, CORD_chars((char)alang, asize));
161: }
162:
163: /// MUST be called exactly prior to modification of current [uses it's length]
164: template<typename C>
165: void append(C current, size_t appending_length,
166: const Languages src) {
167: assert(appending_length);
168:
169: if(!langs)
170: langs=src.langs; // to uninitialized
1.147 paf 171: else if(!src.opt.is_not_just_lang)
172: append(current, src.opt.lang, appending_length); // simplifying when simple source
1.146 paf 173: else
174: append(current, src.make_langs(appending_length));
175: }
176:
177: /// MUST be called exactly prior to modification of current [uses it's length]
178: template<typename C>
179: void append(C current,
180: const Languages src, size_t aoffset, size_t alength) {
181: assert(alength);
182:
183: if(!langs) // to uninitialized?
1.147 paf 184: if(src.opt.is_not_just_lang)
1.146 paf 185: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
186: else
1.147 paf 187: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 188: else
1.147 paf 189: 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 190: return; // ignoring
191: else
192: append(current, src.make_langs(aoffset, alength));
193: }
194:
195: /// checks if we have lang<=alang all from aoffset to aoffset+alength
196: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
197: if(alang==L_UNSPECIFIED) // ignore lang?
198: return true;
199:
1.147 paf 200: if(opt.is_not_just_lang)
1.146 paf 201: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
202: else
1.147 paf 203: return opt.lang<=alang;
1.146 paf 204: }
205:
1.148 paf 206: /// @returns count of blocks
207: /// @todo currently there can be adjucent blocks of same language. someday merge them
208: size_t count() const {
209: return opt.is_not_just_lang?
210: CORD_block_count(langs)
211: : opt.lang?
212: 1
213: : 0;
214: };
215:
1.146 paf 216: template<typename C, typename I>
217: void for_each(C current,
218: int callback(char, size_t, I), I info) const {
219:
1.147 paf 220: if(opt.is_not_just_lang)
1.146 paf 221: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
222: else
1.147 paf 223: callback(opt.lang, get_length(current), info);
1.146 paf 224: }
225:
226: bool invariant(size_t current_length) {
227: if(!langs)
228: return current_length==0;
1.147 paf 229: if(opt.is_not_just_lang)
1.146 paf 230: return CORD_len(langs)==current_length;
231: return true; // uncheckable, actually
232: }
233: };
234:
235: class Body {
236:
237: CORD body;
238:
239: public:
240:
241: const char* v() const;
242:
243: Body(): body(CORD_EMPTY) {}
244: Body(CORD abody): body(abody) {
245: assert(!body // no body
246: || *body // ordinary string
247: || body[1]==1 // CONCAT_HDR
248: || body[1]==4 // FN_HDR
249: || body[1]==6 // SUBSTR_HDR
250: );
251: }
252: /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
253: Body(const char* str, size_t helper_length): body(CORD_EMPTY) {
254: append_know_length(str, helper_length?helper_length:strlen(str));
255: }
256: static Body Format(int value);
257:
258: void clear() { body=CORD_EMPTY; }
259:
260: bool operator! () const { return is_empty(); }
261:
262: uint hash_code() const;
263:
264: const char* cstr() const { return CORD_to_const_char_star(body); }
265: char* cstrm() const { return CORD_to_char_star(body); }
266:
267: size_t length() const { return CORD_len(body); }
268:
269: bool is_empty() const { return body==CORD_EMPTY; }
270:
271: void append_know_length(const char *str, size_t known_length) {
272: if(known_length)
273: body=CORD_cat_char_star(body, str, known_length);
274: }
275: void append_strdup_know_length(const char* str, size_t known_length) {
276: if(known_length)
277: append_know_length(pa_strdup(str, known_length), known_length);
278: }
279: void append(char c) { body=CORD_cat_char(body, c); }
280: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
281: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
282:
283: // could not figure out why this operator is needed [should do this chain: string->simple->==]
284: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
285: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
286: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
287: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
288: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
289: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
290:
291: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
292: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 293: }
294:
1.146 paf 295: char fetch(size_t index) const { return CORD_fetch(body, index); }
296: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
297: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
298: size_t pos(const Body substr, size_t offset=0) const {
299: if(!substr.length())
300: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 ! paf 301:
! 302: // CORD_str checks for bad offset [CORD_chr does not]
1.146 paf 303: return CORD_str(body, offset, substr.body);
304: }
305: size_t pos(char c,
306: size_t offset=0) const {
1.150 ! paf 307: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
! 308: return STRING_NOT_FOUND;
! 309:
1.146 paf 310: return CORD_chr(body, offset, c);
1.145 paf 311: }
1.146 paf 312:
1.149 paf 313: template<typename I> void for_each(
314: int (*f1)(char c, I),
315: int (*f2)(const char* s, I),
316: I info) const {
317: CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 318: }
1.146 paf 319:
320: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
321:
322: /*Body normalize() const {
323: return Body(CORD_balance(body));
324: }*/
1.145 paf 325: };
326:
327: struct C {
328: const char *str;
329: size_t length;
330: operator const char *() { return str; }
331: C(const char *astr, size_t asize): str(astr), length(asize) {}
332: };
333:
334: struct Cm {
335: char *str;
336: size_t length;
337: //operator char *() { return str; }
338: Cm(char *astr, size_t asize): str(astr), length(asize) {}
339: };
340:
341: private:
342:
1.146 paf 343: Languages langs; ///< string characters lang
344: Body body; ///< all characters of string
345:
346: const char* v() const;
347:
348: #define ASSERT_STRING_INVARIANT(string) \
349: assert((string).langs.invariant((string).body.length()))
1.145 paf 350:
1.8 paf 351: public:
352:
1.145 paf 353: explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
354: explicit String(const C cstr, bool tainted=false);
1.146 paf 355: String(Body abody, Language alang): body(abody), langs(alang) {
356: assert(!body.is_empty());
357: ASSERT_STRING_INVARIANT(*this);
358: }
359: String(const String& src): body(src.body), langs(src.langs) {
360: ASSERT_STRING_INVARIANT(*this);
1.145 paf 361: }
362:
363: /// for convinient hash lookup
1.146 paf 364: operator const Body() const { return body; }
1.145 paf 365:
366: bool is_empty() const { return body.is_empty(); }
367: size_t length() const { return body.length(); }
368:
369: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 370: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 371: SQL_Connection* connection=0,
372: const Request_charsets *charsets=0) const;
373:
374: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
375: const char* cstr(Language lang=L_AS_IS,
376: SQL_Connection* connection=0,
377: const Request_charsets *charsets=0) const {
378: return cstr_to_string_body(lang, connection, charsets).cstr();
379: }
380: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
381: char *cstrm(Language lang=L_AS_IS,
382: SQL_Connection* connection=0,
383: const Request_charsets *charsets=0) const {
384: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 385: }
1.108 parser 386: /// puts pieces to buf
1.145 paf 387: Cm serialize(size_t prolog_size) const;
1.108 parser 388: /// appends pieces from buf to self
1.145 paf 389: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 390: /// @see Body::append_know_length
1.145 paf 391: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 392: /// @see Body::append_help_length
1.145 paf 393: String& append_help_length(const char* str, size_t helper_length, Language lang);
394: String& append_strdup(const char* str, size_t helper_length, Language lang);
395:
1.146 paf 396: bool operator == (const char* y) const { return body==Body(y); }
397: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 398:
399: /// this starts with y
400: bool starts_with(const char* y) const {
1.146 paf 401: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 402: }
403: /// x starts with this
404: bool this_starts(const char* x) const {
1.146 paf 405: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 406: }
407:
1.145 paf 408: String& append_to(String& dest, Language lang, bool forced) const;
409: String& append(const String& src, Language lang, bool forced=false) {
410: return src.append_to(*this, lang, forced);
411: }
412: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
413: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 paf 414: String& operator << (const Body src);
1.100 parser 415:
1.142 paf 416: /// extracts first char of a string, if any
417: char first_char() const {
1.145 paf 418: return is_empty()?0:body.fetch(0);
1.142 paf 419: }
1.54 paf 420:
1.145 paf 421: bool operator < (const String& src) const { return body<src.body; }
422: bool operator > (const String& src) const { return body>src.body; }
423: bool operator <= (const String& src) const { return body<=src.body; }
424: bool operator >= (const String& src) const { return body>=src.body; }
425: bool operator != (const String& src) const { return body!=src.body; }
426: bool operator == (const String& src) const { return body==src.body; }
427:
1.54 paf 428: /// extracts [start, finish) piece of string
1.145 paf 429: String& mid(size_t substr_begin, size_t substr_end) const;
430:
431: /**
432: ignore lang if it's L_UNSPECIFIED
433: but when specified: look for substring that lies in ONE fragment in THAT lang
434: @return position of substr in string, -1 means "not found" [const char* version]
435: */
1.146 paf 436: size_t pos(const Body substr,
1.145 paf 437: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
438: /// String version of @see pos(const char*, int, Language)
439: size_t pos(const String& substr,
440: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
441: size_t pos(char c,
442: size_t this_offset=0) const {
443: return body.pos(c, this_offset);
444: }
1.55 paf 445:
1.145 paf 446: void split(ArrayString& result,
447: size_t& pos_after,
448: const char* delim,
449: Language lang=L_UNSPECIFIED, int limit=-1) const;
450: void split(ArrayString& result,
451: size_t& pos_after,
1.62 paf 452: const String& delim,
1.145 paf 453: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 454:
1.145 paf 455: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 456: int prestart, int prefinish,
457: int poststart, int postfinish,
1.68 paf 458: void *info);
1.87 parser 459: /**
1.145 paf 460: @return table of found items, if any.
1.87 parser 461: table format is defined and fixed[can be used by others]:
462: @verbatim
463: prematch/match/postmatch/1/2/3/...
464: @endverbatim
465: */
1.145 paf 466: Table* match(Charset& source_charset,
1.64 paf 467: const String& regexp,
1.145 paf 468: const String* options,
1.99 parser 469: Row_action row_action, void *info,
1.145 paf 470: bool& just_matched) const;
1.87 parser 471: enum Change_case_kind {
472: CC_UPPER,
473: CC_LOWER
474: };
1.145 paf 475: String& change_case(Charset& source_charset,
1.87 parser 476: Change_case_kind kind) const;
1.145 paf 477: const String& replace(const Dictionary& dict) const;
1.96 parser 478: double as_double() const;
479: int as_int() const;
1.137 paf 480:
1.7 paf 481: private: //disabled
482:
1.12 paf 483: String& operator = (const String&) { return *this; }
1.7 paf 484:
1.1 paf 485: };
1.119 paf 486:
1.146 paf 487: template<>
488: inline size_t get_length<String::Body>(String::Body body) {
489: return body.length();
490: }
491:
1.145 paf 492: /// simple hash code of string. used by Hash
1.146 paf 493: inline uint hash_code(const String::Body self) {
1.145 paf 494: return self.hash_code();
1.119 paf 495: }
1.147 paf 496:
497:
498: /// now that we've declared specialization we can use it
499: inline String& String::operator << (const String::Body src) {
500: langs.append(body, L_AS_IS, src.length());
501: body<<src;
502: return *this;
503: }
504:
1.1 paf 505:
506: #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