Annotation of parser3/src/include/pa_string.h, revision 1.160
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:
1.160 ! paf 11: static const char * const IDENT_STRING_H="$Date: 2004/04/06 07:53:57 $";
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_TABLE='L', ///< ^table:set body
94: L_SQL='Q', ///< ^table:sql body
95: L_JS='J', ///< JavaScript code
96: L_XML='X', ///< ^dom:set xml
97: L_HTML='H', ///< HTML code (for editing)
98: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 99: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 100: };
101:
1.157 paf 102: enum Trim_kind {
103: TRIM_BOTH,
104: TRIM_START,
105: TRIM_END
106: };
107:
1.147 paf 108: union Languages {
1.146 paf 109:
1.147 paf 110: struct {
1.160 ! paf 111: #ifdef PA_LITTLE_ENDIAN
1.147 paf 112: Language lang:8;
113: int is_not_just_lang:sizeof(CORD)*8-8;
1.160 ! paf 114: #elif defined(PA_BIG_ENDIAN)
! 115: int is_not_just_lang:sizeof(CORD)*8-8;
! 116: Language lang:8;
! 117: #else
! 118: # error word endianness not determined for some obscure reason
! 119: #endif
1.147 paf 120: } opt;
121: CORD langs;
1.146 paf 122:
123: template<typename C>
124: CORD make_langs(C current) const {
1.147 paf 125: return opt.is_not_just_lang?
1.146 paf 126: langs
1.147 paf 127: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 128: }
129:
130: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 131: return opt.is_not_just_lang?
1.146 paf 132: CORD_substr(langs, aoffset, alength)
1.147 paf 133: :CORD_chars((char)opt.lang, alength);
1.145 paf 134: }
135:
1.157 paf 136: /// appending when 'langs' already contain something [simple cases handled elsewhere]
1.146 paf 137: template<typename C>
138: void append(C current,
139: const CORD to_nonempty_target_langs) {
140: assert(langs);
141:
1.147 paf 142: if(opt.is_not_just_lang)
1.146 paf 143: langs=CORD_cat(langs, to_nonempty_target_langs);
144: else { // we were "just lang"
145: size_t current_size=get_length(current);
146: assert(current_size);
147: langs=CORD_cat(
1.147 paf 148: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 149: to_nonempty_target_langs); // new piece
150: }
1.145 paf 151: }
1.146 paf 152:
1.145 paf 153: public:
1.146 paf 154:
155: const char* v() const;
156:
157: Languages(): langs(0) {}
1.147 paf 158: Languages(Language alang) {
159: opt.lang=alang;
160: opt.is_not_just_lang=0;
161: }
1.146 paf 162:
163: /// MUST be called exactly prior to modification of current [uses it's length]
164: template<typename C>
165: void append(C current, Language alang, size_t asize) {
166: assert(alang);
167: assert(asize);
168:
1.147 paf 169: if(!opt.is_not_just_lang)
170: if(opt.lang) {
171: if(opt.lang==alang) // same length? ignoring
1.146 paf 172: return;
173: } else {
1.147 paf 174: opt.lang=alang; // to uninitialized
1.146 paf 175: return;
1.145 paf 176: }
1.146 paf 177:
178: append(current, CORD_chars((char)alang, asize));
179: }
180:
181: /// MUST be called exactly prior to modification of current [uses it's length]
182: template<typename C>
183: void append(C current, size_t appending_length,
184: const Languages src) {
185: assert(appending_length);
186:
187: if(!langs)
188: langs=src.langs; // to uninitialized
1.147 paf 189: else if(!src.opt.is_not_just_lang)
190: append(current, src.opt.lang, appending_length); // simplifying when simple source
1.146 paf 191: else
192: append(current, src.make_langs(appending_length));
193: }
194:
195: /// MUST be called exactly prior to modification of current [uses it's length]
196: template<typename C>
197: void append(C current,
198: const Languages src, size_t aoffset, size_t alength) {
199: assert(alength);
200:
201: if(!langs) // to uninitialized?
1.147 paf 202: if(src.opt.is_not_just_lang)
1.146 paf 203: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
204: else
1.147 paf 205: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 206: else
1.147 paf 207: 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 208: return; // ignoring
209: else
210: append(current, src.make_langs(aoffset, alength));
211: }
212:
213: /// checks if we have lang<=alang all from aoffset to aoffset+alength
214: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
215: if(alang==L_UNSPECIFIED) // ignore lang?
216: return true;
217:
1.147 paf 218: if(opt.is_not_just_lang)
1.146 paf 219: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
220: else
1.147 paf 221: return opt.lang<=alang;
1.146 paf 222: }
223:
1.148 paf 224: /// @returns count of blocks
225: /// @todo currently there can be adjucent blocks of same language. someday merge them
226: size_t count() const {
227: return opt.is_not_just_lang?
228: CORD_block_count(langs)
229: : opt.lang?
230: 1
231: : 0;
232: };
233:
1.146 paf 234: template<typename C, typename I>
235: void for_each(C current,
236: int callback(char, size_t, I), I info) const {
237:
1.147 paf 238: if(opt.is_not_just_lang)
1.146 paf 239: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
240: else
1.147 paf 241: callback(opt.lang, get_length(current), info);
1.146 paf 242: }
243:
244: bool invariant(size_t current_length) {
245: if(!langs)
246: return current_length==0;
1.147 paf 247: if(opt.is_not_just_lang)
1.146 paf 248: return CORD_len(langs)==current_length;
249: return true; // uncheckable, actually
250: }
251: };
252:
253: class Body {
254:
255: CORD body;
256:
257: public:
258:
259: const char* v() const;
260:
261: Body(): body(CORD_EMPTY) {}
262: Body(CORD abody): body(abody) {
263: assert(!body // no body
264: || *body // ordinary string
265: || body[1]==1 // CONCAT_HDR
266: || body[1]==4 // FN_HDR
267: || body[1]==6 // SUBSTR_HDR
268: );
269: }
270: /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
271: Body(const char* str, size_t helper_length): body(CORD_EMPTY) {
272: append_know_length(str, helper_length?helper_length:strlen(str));
273: }
274: static Body Format(int value);
275:
276: void clear() { body=CORD_EMPTY; }
277:
278: bool operator! () const { return is_empty(); }
279:
280: uint hash_code() const;
281:
282: const char* cstr() const { return CORD_to_const_char_star(body); }
283: char* cstrm() const { return CORD_to_char_star(body); }
284:
285: size_t length() const { return CORD_len(body); }
286:
287: bool is_empty() const { return body==CORD_EMPTY; }
288:
289: void append_know_length(const char *str, size_t known_length) {
290: if(known_length)
291: body=CORD_cat_char_star(body, str, known_length);
292: }
293: void append_strdup_know_length(const char* str, size_t known_length) {
294: if(known_length)
295: append_know_length(pa_strdup(str, known_length), known_length);
296: }
297: void append(char c) { body=CORD_cat_char(body, c); }
298: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
299: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
300:
301: // could not figure out why this operator is needed [should do this chain: string->simple->==]
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: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
308:
309: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
310: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 311: }
312:
1.146 paf 313: char fetch(size_t index) const { return CORD_fetch(body, index); }
314: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
315: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
316: size_t pos(const Body substr, size_t offset=0) const {
317: if(!substr.length())
318: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 319:
320: // CORD_str checks for bad offset [CORD_chr does not]
1.146 paf 321: return CORD_str(body, offset, substr.body);
322: }
323: size_t pos(char c,
324: size_t offset=0) const {
1.150 paf 325: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
326: return STRING_NOT_FOUND;
327:
1.146 paf 328: return CORD_chr(body, offset, c);
1.145 paf 329: }
1.146 paf 330:
1.159 paf 331: template<typename I> int for_each(
332: int (*f)(char c, I),
333: I info) const {
334: return CORD_iter(body, (CORD_iter_fn)f, (void*)info);
335: }
336:
337: template<typename I> int for_each(
1.149 paf 338: int (*f1)(char c, I),
339: int (*f2)(const char* s, I),
340: I info) const {
1.159 paf 341: return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 342: }
1.146 paf 343:
344: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
345:
346: /*Body normalize() const {
347: return Body(CORD_balance(body));
348: }*/
1.157 paf 349:
350: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled
351: Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0,
352: size_t* out_start=0, size_t* out_length=0) const;
1.145 paf 353: };
354:
355: struct C {
356: const char *str;
357: size_t length;
358: operator const char *() { return str; }
1.158 paf 359: C(): str(0), length(0) {}
1.145 paf 360: C(const char *astr, size_t asize): str(astr), length(asize) {}
361: };
362:
363: struct Cm {
364: char *str;
365: size_t length;
366: //operator char *() { return str; }
1.158 paf 367: Cm(): str(0), length(0) {}
1.145 paf 368: Cm(char *astr, size_t asize): str(astr), length(asize) {}
369: };
370:
371: private:
372:
1.152 paf 373: Body body; ///< all characters of string
1.146 paf 374: Languages langs; ///< string characters lang
375:
376: const char* v() const;
377:
378: #define ASSERT_STRING_INVARIANT(string) \
379: assert((string).langs.invariant((string).body.length()))
1.145 paf 380:
1.8 paf 381: public:
1.151 paf 382:
383: static const String Empty;
1.8 paf 384:
1.145 paf 385: explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
386: explicit String(const C cstr, bool tainted=false);
1.146 paf 387: String(Body abody, Language alang): body(abody), langs(alang) {
388: ASSERT_STRING_INVARIANT(*this);
389: }
390: String(const String& src): body(src.body), langs(src.langs) {
391: ASSERT_STRING_INVARIANT(*this);
1.145 paf 392: }
393:
394: /// for convinient hash lookup
1.146 paf 395: operator const Body() const { return body; }
1.145 paf 396:
397: bool is_empty() const { return body.is_empty(); }
398: size_t length() const { return body.length(); }
399:
400: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 401: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 402: SQL_Connection* connection=0,
403: const Request_charsets *charsets=0) const;
404:
405: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
406: const char* cstr(Language lang=L_AS_IS,
407: SQL_Connection* connection=0,
408: const Request_charsets *charsets=0) const {
409: return cstr_to_string_body(lang, connection, charsets).cstr();
410: }
411: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
412: char *cstrm(Language lang=L_AS_IS,
413: SQL_Connection* connection=0,
414: const Request_charsets *charsets=0) const {
415: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 416: }
1.108 parser 417: /// puts pieces to buf
1.145 paf 418: Cm serialize(size_t prolog_size) const;
1.108 parser 419: /// appends pieces from buf to self
1.145 paf 420: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 421: /// @see Body::append_know_length
1.145 paf 422: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 423: /// @see Body::append_help_length
1.145 paf 424: String& append_help_length(const char* str, size_t helper_length, Language lang);
425: String& append_strdup(const char* str, size_t helper_length, Language lang);
426:
1.146 paf 427: bool operator == (const char* y) const { return body==Body(y); }
428: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 429:
430: /// this starts with y
431: bool starts_with(const char* y) const {
1.146 paf 432: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 433: }
434: /// x starts with this
435: bool this_starts(const char* x) const {
1.146 paf 436: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 437: }
438:
1.145 paf 439: String& append_to(String& dest, Language lang, bool forced) const;
440: String& append(const String& src, Language lang, bool forced=false) {
441: return src.append_to(*this, lang, forced);
442: }
443: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
444: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 paf 445: String& operator << (const Body src);
1.100 parser 446:
1.142 paf 447: /// extracts first char of a string, if any
448: char first_char() const {
1.145 paf 449: return is_empty()?0:body.fetch(0);
1.142 paf 450: }
1.54 paf 451:
1.145 paf 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: bool operator == (const String& src) const { return body==src.body; }
458:
1.54 paf 459: /// extracts [start, finish) piece of string
1.145 paf 460: String& mid(size_t substr_begin, size_t substr_end) const;
461:
462: /**
463: ignore lang if it's L_UNSPECIFIED
464: but when specified: look for substring that lies in ONE fragment in THAT lang
465: @return position of substr in string, -1 means "not found" [const char* version]
466: */
1.146 paf 467: size_t pos(const Body substr,
1.145 paf 468: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
469: /// String version of @see pos(const char*, int, Language)
470: size_t pos(const String& substr,
471: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
472: size_t pos(char c,
473: size_t this_offset=0) const {
474: return body.pos(c, this_offset);
475: }
1.55 paf 476:
1.145 paf 477: void split(ArrayString& result,
478: size_t& pos_after,
479: const char* delim,
480: Language lang=L_UNSPECIFIED, int limit=-1) const;
481: void split(ArrayString& result,
482: size_t& pos_after,
1.62 paf 483: const String& delim,
1.145 paf 484: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 485:
1.145 paf 486: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 487: int prestart, int prefinish,
488: int poststart, int postfinish,
1.68 paf 489: void *info);
1.87 parser 490: /**
1.145 paf 491: @return table of found items, if any.
1.87 parser 492: table format is defined and fixed[can be used by others]:
493: @verbatim
494: prematch/match/postmatch/1/2/3/...
495: @endverbatim
496: */
1.145 paf 497: Table* match(Charset& source_charset,
1.64 paf 498: const String& regexp,
1.145 paf 499: const String* options,
1.99 parser 500: Row_action row_action, void *info,
1.145 paf 501: bool& just_matched) const;
1.87 parser 502: enum Change_case_kind {
503: CC_UPPER,
504: CC_LOWER
505: };
1.145 paf 506: String& change_case(Charset& source_charset,
1.87 parser 507: Change_case_kind kind) const;
1.145 paf 508: const String& replace(const Dictionary& dict) const;
1.157 paf 509: const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0) const;
1.155 paf 510: double as_double() const { return pa_atod(cstr(), this); }
511: int as_int() const { return pa_atoi(cstr(), this); }
1.137 paf 512:
1.7 paf 513: private: //disabled
514:
1.12 paf 515: String& operator = (const String&) { return *this; }
1.7 paf 516:
1.1 paf 517: };
1.119 paf 518:
1.146 paf 519: template<>
520: inline size_t get_length<String::Body>(String::Body body) {
521: return body.length();
522: }
523:
1.145 paf 524: /// simple hash code of string. used by Hash
1.146 paf 525: inline uint hash_code(const String::Body self) {
1.145 paf 526: return self.hash_code();
1.119 paf 527: }
1.147 paf 528:
529:
530: /// now that we've declared specialization we can use it
531: inline String& String::operator << (const String::Body src) {
532: langs.append(body, L_AS_IS, src.length());
533: body<<src;
534: return *this;
535: }
536:
1.1 paf 537:
538: #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