Annotation of parser3/src/include/pa_string.h, revision 1.182
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.179 misha 4: Copyright (c) 2001-2009 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:
1.182 ! misha 11: static const char * const IDENT_STRING_H="$Date: 2009-05-13 07:35:52 $";
1.145 paf 12:
13: // includes
1.4 paf 14: #include "pa_types.h"
1.145 paf 15: #include "pa_array.h"
16:
17: extern "C" { // cord's author forgot to do that
18: #define CORD_NO_IO
19: #include "cord.h"
20: };
1.4 paf 21:
1.164 paf 22: // defines
23:
1.146 paf 24: // cord extension
25: /* Returns true if x does contain */
26: /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */
27: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c);
1.148 paf 28: size_t CORD_block_count(CORD x);
1.146 paf 29:
1.145 paf 30: // forwards
1.9 paf 31:
1.145 paf 32: class Charset;
1.135 paf 33: class Table;
1.71 paf 34: class SQL_Connection;
1.101 parser 35: class Dictionary;
1.145 paf 36: class Request_charsets;
37: class String;
38: typedef Array<const String*> ArrayString;
1.178 misha 39: class VRegex;
1.145 paf 40:
1.155 paf 41: // generally useful
42:
43: int pa_atoi(const char* str, const String* problem_source=0);
44: double pa_atod(const char* str, const String* problem_source=0);
45:
1.145 paf 46: /// this is result of pos functions which mean that substr were not found
47: #define STRING_NOT_FOUND ((size_t)-1)
48:
1.146 paf 49: template<typename T>
50: inline size_t get_length(T current) {
51: return current;
52: }
1.147 paf 53:
1.42 paf 54: /**
1.146 paf 55: String which knows the lang of all it's langs.
1.41 paf 56:
57: All pieces remember
58: - whether they are tainted or not,
1.146 paf 59: and the lang which should be used to detaint them
1.41 paf 60: */
1.176 misha 61:
62:
1.145 paf 63: class String: public PA_Object {
1.1 paf 64: public:
1.48 paf 65:
1.146 paf 66: /** piece is tainted or not. the lang to use when detaint
1.106 parser 67: remember to change String_Untaint_lang_name @ untaint.C along
1.148 paf 68:
69: WARNING WARNING WARNING WARNING WARNING WARNING
70:
71: pos function compares(<=) languages, that is used in searching
72: for table column separator being L_CLEAN or L_AS_IS.
73: they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check].
74:
75: letters assigned for debugging, but it's important for no language-letter
76: come before L_AS_IS other then L_CLEAN
77:
78: WARNING WARNING WARNING WARNING WARNING WARNING
1.106 parser 79: */
1.145 paf 80: enum Language {
1.146 paf 81: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 82: // these two must go before others, there are checks for >L_AS_IS
1.170 misha 83: L_CLEAN='0', ///< clean WARNING: read above warning before changing
84: L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing
1.122 paf 85:
1.148 paf 86: L_PASS_APPENDED='P',
1.41 paf 87: /**<
1.146 paf 88: leave lang built into string being appended.
1.41 paf 89: just a flag, that value not stored
90: */
1.170 misha 91: L_TAINTED='T', ///< tainted, untaint lang as assigned later
1.146 paf 92: // untaint langs. assigned by ^untaint[lang]{...}
1.170 misha 93: L_FILE_SPEC='F', ///< file specification
94: L_HTTP_HEADER='h', ///< text in HTTP response header
95: L_MAIL_HEADER='m', ///< text in mail header
96: L_URI='U', ///< text in uri
97: L_SQL='Q', ///< ^table:sql body
98: L_JS='J', ///< JavaScript code
1.148 paf 99: L_XML='X', ///< ^dom:set xml
1.170 misha 100: L_HTML='H', ///< HTML code
101: L_REGEX='R', ///< RegEx expression
102: L_HTTP_COOKIE='C', ///< cookies encoded as %uXXXX for compartibility with js functions encode/decode
1.175 misha 103: L_FILE_POST='f', ///temporary escaping zero-char
1.148 paf 104: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 105: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 106: };
107:
1.157 paf 108: enum Trim_kind {
109: TRIM_BOTH,
110: TRIM_START,
111: TRIM_END
112: };
113:
1.176 misha 114:
1.147 paf 115: union Languages {
1.146 paf 116:
1.147 paf 117: struct {
1.160 paf 118: #ifdef PA_LITTLE_ENDIAN
1.147 paf 119: Language lang:8;
1.179 misha 120: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 121: #elif defined(PA_BIG_ENDIAN)
1.179 misha 122: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 123: Language lang:8;
124: #else
125: # error word endianness not determined for some obscure reason
126: #endif
1.147 paf 127: } opt;
128: CORD langs;
1.146 paf 129:
130: template<typename C>
131: CORD make_langs(C current) const {
1.147 paf 132: return opt.is_not_just_lang?
1.146 paf 133: langs
1.162 paf 134: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 135: }
136:
137: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 138: return opt.is_not_just_lang?
1.146 paf 139: CORD_substr(langs, aoffset, alength)
1.162 paf 140: :CORD_chars((char)opt.lang, alength);
1.145 paf 141: }
142:
1.157 paf 143: /// appending when 'langs' already contain something [simple cases handled elsewhere]
1.146 paf 144: template<typename C>
145: void append(C current,
146: const CORD to_nonempty_target_langs) {
147: assert(langs);
148:
1.147 paf 149: if(opt.is_not_just_lang)
1.146 paf 150: langs=CORD_cat(langs, to_nonempty_target_langs);
151: else { // we were "just lang"
152: size_t current_size=get_length(current);
153: assert(current_size);
154: langs=CORD_cat(
1.162 paf 155: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 156: to_nonempty_target_langs); // new piece
157: }
1.145 paf 158: }
1.146 paf 159:
1.145 paf 160: public:
1.146 paf 161:
162: const char* v() const;
1.164 paf 163: void dump() const;
1.146 paf 164:
165: Languages(): langs(0) {}
1.147 paf 166: Languages(Language alang) {
167: opt.lang=alang;
168: opt.is_not_just_lang=0;
169: }
1.146 paf 170:
171: /// MUST be called exactly prior to modification of current [uses it's length]
172: template<typename C>
173: void append(C current, Language alang, size_t asize) {
174: assert(alang);
175: assert(asize);
176:
1.147 paf 177: if(!opt.is_not_just_lang)
178: if(opt.lang) {
1.181 misha 179: if(opt.lang==alang) // same language? ignoring
1.146 paf 180: return;
181: } else {
1.147 paf 182: opt.lang=alang; // to uninitialized
1.146 paf 183: return;
1.145 paf 184: }
1.146 paf 185:
1.162 paf 186: append(current, CORD_chars((char)alang, asize));
1.146 paf 187: }
188:
1.176 misha 189: template<typename C>
190: void appendHelper(C current, Language alang, C asize_helper) {
191: assert(alang);
192:
193: if(!opt.is_not_just_lang)
194: if(opt.lang) {
195: if(opt.lang==alang) // same length? ignoring
196: return;
197: } else {
198: opt.lang=alang; // to uninitialized
199: return;
200: }
201:
202: append(current, CORD_chars((char)alang, asize_helper.length()));
203: }
204:
205: template<typename C>
206: void appendHelper(C current, C length_helper, const Languages src) {
207:
208: if(!langs)
209: langs=src.langs; // to uninitialized
210: else if(!src.opt.is_not_just_lang)
211: appendHelper(current, src.opt.lang, length_helper); // simplifying when simple source
212: else
213: append(current, src.make_langs(length_helper));
214: }
215:
216:
1.146 paf 217: /// MUST be called exactly prior to modification of current [uses it's length]
218: template<typename C>
219: void append(C current,
220: const Languages src, size_t aoffset, size_t alength) {
221: assert(alength);
222:
223: if(!langs) // to uninitialized?
1.147 paf 224: if(src.opt.is_not_just_lang)
1.146 paf 225: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
226: else
1.147 paf 227: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 228: else
1.147 paf 229: 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 230: return; // ignoring
231: else
232: append(current, src.make_langs(aoffset, alength));
233: }
234:
235: /// checks if we have lang<=alang all from aoffset to aoffset+alength
236: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
237: if(alang==L_UNSPECIFIED) // ignore lang?
238: return true;
239:
1.147 paf 240: if(opt.is_not_just_lang)
1.146 paf 241: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
242: else
1.169 misha 243: return (unsigned)opt.lang<=(unsigned)alang;
1.146 paf 244: }
245:
1.148 paf 246: /// @returns count of blocks
247: /// @todo currently there can be adjucent blocks of same language. someday merge them
248: size_t count() const {
249: return opt.is_not_just_lang?
250: CORD_block_count(langs)
251: : opt.lang?
252: 1
253: : 0;
254: };
255:
1.146 paf 256: template<typename C, typename I>
257: void for_each(C current,
258: int callback(char, size_t, I), I info) const {
259:
1.147 paf 260: if(opt.is_not_just_lang)
1.146 paf 261: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
262: else
1.147 paf 263: callback(opt.lang, get_length(current), info);
1.146 paf 264: }
265:
1.164 paf 266: bool invariant(size_t current_length) const {
1.146 paf 267: if(!langs)
268: return current_length==0;
1.147 paf 269: if(opt.is_not_just_lang)
1.146 paf 270: return CORD_len(langs)==current_length;
271: return true; // uncheckable, actually
272: }
273: };
274:
275: class Body {
276:
277: CORD body;
278:
279: public:
280:
281: const char* v() const;
1.164 paf 282: void dump() const;
1.146 paf 283:
284: Body(): body(CORD_EMPTY) {}
285: Body(CORD abody): body(abody) {
286: assert(!body // no body
287: || *body // ordinary string
288: || body[1]==1 // CONCAT_HDR
289: || body[1]==4 // FN_HDR
290: || body[1]==6 // SUBSTR_HDR
291: );
292: }
1.181 misha 293:
1.146 paf 294: static Body Format(int value);
295:
296: void clear() { body=CORD_EMPTY; }
297:
298: bool operator! () const { return is_empty(); }
299:
300: uint hash_code() const;
301:
302: const char* cstr() const { return CORD_to_const_char_star(body); }
303: char* cstrm() const { return CORD_to_char_star(body); }
304:
305: size_t length() const { return CORD_len(body); }
306:
307: bool is_empty() const { return body==CORD_EMPTY; }
308:
309: void append_know_length(const char *str, size_t known_length) {
310: if(known_length)
311: body=CORD_cat_char_star(body, str, known_length);
312: }
313: void append_strdup_know_length(const char* str, size_t known_length) {
314: if(known_length)
315: append_know_length(pa_strdup(str, known_length), known_length);
316: }
317: void append(char c) { body=CORD_cat_char(body, c); }
318: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
319: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
320:
321: // could not figure out why this operator is needed [should do this chain: string->simple->==]
322: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
323: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
324: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
325: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
326: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
327: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
328:
329: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
330: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 331: }
332:
1.146 paf 333: char fetch(size_t index) const { return CORD_fetch(body, index); }
334: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
335: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
336: size_t pos(const Body substr, size_t offset=0) const {
1.180 misha 337: if(substr.is_empty())
1.146 paf 338: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 339:
340: // CORD_str checks for bad offset [CORD_chr does not]
1.146 paf 341: return CORD_str(body, offset, substr.body);
342: }
343: size_t pos(char c,
344: size_t offset=0) const {
1.150 paf 345: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
346: return STRING_NOT_FOUND;
347:
1.146 paf 348: return CORD_chr(body, offset, c);
1.145 paf 349: }
1.146 paf 350:
1.159 paf 351: template<typename I> int for_each(
352: int (*f)(char c, I),
353: I info) const {
354: return CORD_iter(body, (CORD_iter_fn)f, (void*)info);
355: }
356:
357: template<typename I> int for_each(
1.149 paf 358: int (*f1)(char c, I),
359: int (*f2)(const char* s, I),
360: I info) const {
1.159 paf 361: return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 362: }
1.146 paf 363:
364: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
365:
366: /*Body normalize() const {
367: return Body(CORD_balance(body));
368: }*/
1.157 paf 369:
370: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled
371: Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0,
372: size_t* out_start=0, size_t* out_length=0) const;
1.145 paf 373: };
374:
375: struct C {
376: const char *str;
377: size_t length;
378: operator const char *() { return str; }
1.158 paf 379: C(): str(0), length(0) {}
1.145 paf 380: C(const char *astr, size_t asize): str(astr), length(asize) {}
381: };
382:
383: struct Cm {
384: char *str;
385: size_t length;
386: //operator char *() { return str; }
1.158 paf 387: Cm(): str(0), length(0) {}
1.145 paf 388: Cm(char *astr, size_t asize): str(astr), length(asize) {}
389: };
390:
391: private:
392:
1.152 paf 393: Body body; ///< all characters of string
1.146 paf 394: Languages langs; ///< string characters lang
395:
396: const char* v() const;
1.164 paf 397: void dump() const;
398: #define ASSERT_STRING_INVARIANT(string) \
399: assert((string).langs.invariant((string).body.length()))
1.145 paf 400:
1.8 paf 401: public:
1.151 paf 402:
403: static const String Empty;
1.8 paf 404:
1.181 misha 405: explicit String(){};
1.182 ! misha 406: explicit String(const char* cstr, Language alang=L_CLEAN){
1.181 misha 407: if(cstr && *cstr){
408: body=cstr;
1.182 ! misha 409: langs=alang;
1.181 misha 410: }
411: }
1.182 ! misha 412: explicit String(const String::C cstr, Language alang=L_CLEAN){
1.181 misha 413: if(cstr.length){
414: body=cstr.str;
1.182 ! misha 415: langs=alang;
1.181 misha 416: }
417: }
1.146 paf 418: String(Body abody, Language alang): body(abody), langs(alang) {
419: ASSERT_STRING_INVARIANT(*this);
420: }
421: String(const String& src): body(src.body), langs(src.langs) {
422: ASSERT_STRING_INVARIANT(*this);
1.145 paf 423: }
424:
425: /// for convinient hash lookup
1.146 paf 426: operator const Body() const { return body; }
1.145 paf 427:
428: bool is_empty() const { return body.is_empty(); }
429: size_t length() const { return body.length(); }
1.171 misha 430: size_t length(Charset& charset) const;
1.145 paf 431:
432: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 433: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 434: SQL_Connection* connection=0,
435: const Request_charsets *charsets=0) const;
436:
437: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
438: const char* cstr(Language lang=L_AS_IS,
439: SQL_Connection* connection=0,
440: const Request_charsets *charsets=0) const {
441: return cstr_to_string_body(lang, connection, charsets).cstr();
442: }
443: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
444: char *cstrm(Language lang=L_AS_IS,
445: SQL_Connection* connection=0,
446: const Request_charsets *charsets=0) const {
447: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 448: }
1.108 parser 449: /// puts pieces to buf
1.145 paf 450: Cm serialize(size_t prolog_size) const;
1.108 parser 451: /// appends pieces from buf to self
1.145 paf 452: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 453: /// @see Body::append_know_length
1.145 paf 454: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 455: /// @see Body::append_help_length
1.145 paf 456: String& append_help_length(const char* str, size_t helper_length, Language lang);
457: String& append_strdup(const char* str, size_t helper_length, Language lang);
458:
1.146 paf 459: bool operator == (const char* y) const { return body==Body(y); }
460: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 461:
462: /// this starts with y
463: bool starts_with(const char* y) const {
1.146 paf 464: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 465: }
466: /// x starts with this
467: bool this_starts(const char* x) const {
1.146 paf 468: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 469: }
470:
1.145 paf 471: String& append_to(String& dest, Language lang, bool forced) const;
472: String& append(const String& src, Language lang, bool forced=false) {
473: return src.append_to(*this, lang, forced);
474: }
475: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
476: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 paf 477: String& operator << (const Body src);
1.100 parser 478:
1.142 paf 479: /// extracts first char of a string, if any
480: char first_char() const {
1.145 paf 481: return is_empty()?0:body.fetch(0);
1.142 paf 482: }
1.54 paf 483:
1.145 paf 484: bool operator < (const String& src) const { return body<src.body; }
485: bool operator > (const String& src) const { return body>src.body; }
486: bool operator <= (const String& src) const { return body<=src.body; }
487: bool operator >= (const String& src) const { return body>=src.body; }
488: bool operator != (const String& src) const { return body!=src.body; }
489: bool operator == (const String& src) const { return body==src.body; }
490:
1.54 paf 491: /// extracts [start, finish) piece of string
1.145 paf 492: String& mid(size_t substr_begin, size_t substr_end) const;
1.172 misha 493: String& mid(Charset& charset, size_t from, size_t to, size_t helper_length=0) const;
1.145 paf 494:
495: /**
496: ignore lang if it's L_UNSPECIFIED
497: but when specified: look for substring that lies in ONE fragment in THAT lang
498: @return position of substr in string, -1 means "not found" [const char* version]
499: */
1.146 paf 500: size_t pos(const Body substr,
1.145 paf 501: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
502: /// String version of @see pos(const char*, int, Language)
503: size_t pos(const String& substr,
504: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
505: size_t pos(char c,
506: size_t this_offset=0) const {
507: return body.pos(c, this_offset);
508: }
1.173 misha 509: size_t pos(Charset& charset,
1.171 misha 510: const String& substr,
511: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.55 paf 512:
1.145 paf 513: void split(ArrayString& result,
514: size_t& pos_after,
515: const char* delim,
516: Language lang=L_UNSPECIFIED, int limit=-1) const;
517: void split(ArrayString& result,
518: size_t& pos_after,
1.62 paf 519: const String& delim,
1.145 paf 520: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 521:
1.145 paf 522: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 523: int prestart, int prefinish,
524: int poststart, int postfinish,
1.68 paf 525: void *info);
1.87 parser 526: /**
1.145 paf 527: @return table of found items, if any.
1.87 parser 528: table format is defined and fixed[can be used by others]:
529: @verbatim
530: prematch/match/postmatch/1/2/3/...
531: @endverbatim
532: */
1.178 misha 533: Table* match(VRegex* vregex,
1.99 parser 534: Row_action row_action, void *info,
1.168 misha 535: int& matches_count) const;
1.87 parser 536: enum Change_case_kind {
537: CC_UPPER,
538: CC_LOWER
539: };
1.145 paf 540: String& change_case(Charset& source_charset,
1.87 parser 541: Change_case_kind kind) const;
1.145 paf 542: const String& replace(const Dictionary& dict) const;
1.157 paf 543: const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0) const;
1.155 paf 544: double as_double() const { return pa_atod(cstr(), this); }
545: int as_int() const { return pa_atoi(cstr(), this); }
1.167 misha 546: bool as_bool() const { return as_int()!=0; }
1.174 misha 547: const String& escape(Charset& source_charset) const;
1.137 paf 548:
1.7 paf 549: private: //disabled
550:
1.12 paf 551: String& operator = (const String&) { return *this; }
1.7 paf 552:
1.1 paf 553: };
1.119 paf 554:
1.146 paf 555: template<>
556: inline size_t get_length<String::Body>(String::Body body) {
557: return body.length();
558: }
559:
1.145 paf 560: /// simple hash code of string. used by Hash
1.146 paf 561: inline uint hash_code(const String::Body self) {
1.145 paf 562: return self.hash_code();
1.119 paf 563: }
1.147 paf 564:
565:
566: /// now that we've declared specialization we can use it
567: inline String& String::operator << (const String::Body src) {
568: langs.append(body, L_AS_IS, src.length());
569: body<<src;
570: return *this;
571: }
572:
1.1 paf 573:
574: #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