Annotation of parser3/src/include/pa_string.h, revision 1.240
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.234 moko 4: Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com)
1.229 moko 5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
1.1 paf 6: */
7:
8: #ifndef PA_STRING_H
9: #define PA_STRING_H
1.140 paf 10:
1.240 ! moko 11: #define IDENT_PA_STRING_H "$Id: pa_string.h,v 1.239 2026/01/06 13:27:59 moko Exp $"
1.145 paf 12:
13: // includes
1.4 paf 14: #include "pa_types.h"
1.239 moko 15: #include "pa_int.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
1.224 moko 20: #include "../lib/cord/include/cord.h"
1.186 misha 21:
22: #ifdef CORD_CAT_OPTIMIZATION
23: #define CORD_cat(x, y) CORD_cat_optimized(x, y)
24: #define CORD_cat_char_star(x, y, leny) CORD_cat_char_star_optimized(x, y, leny)
25: #endif
1.209 moko 26: }
1.4 paf 27:
1.164 paf 28: // defines
29:
1.183 misha 30: // cache hash code in String::Body for faster hash access
31: #define HASH_CODE_CACHING
32:
1.187 misha 33: // cache String::Body.length() for char* strings only, CORDs have own length
34: #define STRING_LENGTH_CACHING
35:
36:
1.146 paf 37: // cord extension
38: /* Returns true if x does contain */
39: /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */
40: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c);
1.148 paf 41: size_t CORD_block_count(CORD x);
1.146 paf 42:
1.145 paf 43: // forwards
1.9 paf 44:
1.145 paf 45: class Charset;
1.135 paf 46: class Table;
1.71 paf 47: class SQL_Connection;
1.101 parser 48: class Dictionary;
1.145 paf 49: class Request_charsets;
50: class String;
1.231 moko 51: class VRegex;
52:
53: #ifdef NDEBUG
1.145 paf 54: typedef Array<const String*> ArrayString;
1.231 moko 55: #else
56: class ArrayString : public Array<const String*> {
57: public:
58: inline ArrayString(size_t initial=0) : Array(initial){
59: }
1.232 moko 60: inline Array& operator+=(element_type src) {
1.231 moko 61: assert(src != NULL);
62: return Array::operator+=(src);
63: }
1.232 moko 64:
65: inline element_type get(size_t index) const {
66: element_type result=Array::get(index);
67: assert(result != NULL);
68: return result;
69: }
1.231 moko 70: };
71: #endif
1.145 paf 72:
1.155 paf 73: // generally useful
74:
1.145 paf 75: /// this is result of pos functions which mean that substr were not found
76: #define STRING_NOT_FOUND ((size_t)-1)
77:
1.211 moko 78: /// CORD can't be empty string, thus checking it in assigment
79: #define AS_CORD(v) ((v) && *(v) ? (CORD)(v):0)
80:
1.42 paf 81: /**
1.146 paf 82: String which knows the lang of all it's langs.
1.41 paf 83:
84: All pieces remember
85: - whether they are tainted or not,
1.146 paf 86: and the lang which should be used to detaint them
1.41 paf 87: */
1.176 misha 88:
89:
1.145 paf 90: class String: public PA_Object {
1.218 moko 91: friend class StringSplitHelper;
1.1 paf 92: public:
1.48 paf 93:
1.146 paf 94: /** piece is tainted or not. the lang to use when detaint
1.106 parser 95: remember to change String_Untaint_lang_name @ untaint.C along
1.148 paf 96:
97: WARNING WARNING WARNING WARNING WARNING WARNING
98:
99: pos function compares(<=) languages, that is used in searching
100: for table column separator being L_CLEAN or L_AS_IS.
101: they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check].
102:
103: letters assigned for debugging, but it's important for no language-letter
104: come before L_AS_IS other then L_CLEAN
105:
106: WARNING WARNING WARNING WARNING WARNING WARNING
1.106 parser 107: */
1.225 moko 108:
109: #if _MSC_VER >= 1900
110: /// required for VS2015+ to make sizeof(Languages::opt) == sizeof(CORD), will be 16 byte under x64 without it
111: enum Language : size_t {
112: #else
1.145 paf 113: enum Language {
1.225 moko 114: #endif
115: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 116: // these two must go before others, there are checks for >L_AS_IS
1.225 moko 117: L_CLEAN='0', ///< clean WARNING: read above warning before changing
118: L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing
1.122 paf 119:
1.225 moko 120: L_TAINTED='T', ///< tainted, untaint lang as assigned later
1.146 paf 121: // untaint langs. assigned by ^untaint[lang]{...}
1.170 misha 122: L_FILE_SPEC='F', ///< file specification
123: L_HTTP_HEADER='h', ///< text in HTTP response header
124: L_MAIL_HEADER='m', ///< text in mail header
1.225 moko 125: L_URI='U', ///< text in uri
126: L_SQL='Q', ///< ^table:sql body
127: L_JS='J', ///< JavaScript code
128: L_XML='X', ///< ^xdoc:create xml
129: L_HTML='H', ///< HTML code
1.198 misha 130: L_REGEX='R', ///< RegExp
1.225 moko 131: L_JSON='S', ///< JSON code
1.170 misha 132: L_HTTP_COOKIE='C', ///< cookies encoded as %uXXXX for compartibility with js functions encode/decode
1.192 misha 133: L_PARSER_CODE='p', ///< ^process body
1.148 paf 134: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 135: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 136: };
137:
1.157 paf 138: enum Trim_kind {
139: TRIM_BOTH,
140: TRIM_START,
141: TRIM_END
142: };
143:
1.187 misha 144: class Body;
1.176 misha 145:
1.147 paf 146: union Languages {
1.146 paf 147:
1.147 paf 148: struct {
1.160 paf 149: #ifdef PA_LITTLE_ENDIAN
1.147 paf 150: Language lang:8;
1.179 misha 151: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 152: #elif defined(PA_BIG_ENDIAN)
1.179 misha 153: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 154: Language lang:8;
155: #else
156: # error word endianness not determined for some obscure reason
157: #endif
1.147 paf 158: } opt;
159: CORD langs;
1.146 paf 160:
1.187 misha 161: CORD make_langs(const Body& current) const {
162: return opt.is_not_just_lang?langs:CORD_chars((char)opt.lang, current.length());
1.146 paf 163: }
164:
165: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 166: return opt.is_not_just_lang?
1.194 misha 167: CORD_substr(langs, aoffset, alength, 0)
1.162 paf 168: :CORD_chars((char)opt.lang, alength);
1.145 paf 169: }
170:
1.157 paf 171: /// appending when 'langs' already contain something [simple cases handled elsewhere]
1.187 misha 172: void append(size_t current, const CORD to_nonempty_target_langs) {
1.146 paf 173: assert(langs);
1.187 misha 174: if(opt.is_not_just_lang)
175: langs=CORD_cat(langs, to_nonempty_target_langs);
176: else {
177: assert(current);
178: langs=CORD_cat(CORD_chars((char)opt.lang, current), to_nonempty_target_langs);
179: }
180: }
1.146 paf 181:
1.187 misha 182: void append(const Body& current, const CORD to_nonempty_target_langs) {
183: assert(langs);
1.147 paf 184: if(opt.is_not_just_lang)
1.146 paf 185: langs=CORD_cat(langs, to_nonempty_target_langs);
1.187 misha 186: else {
187: size_t current_size=current.length();
1.146 paf 188: assert(current_size);
1.187 misha 189: langs=CORD_cat(CORD_chars((char)opt.lang, current_size), to_nonempty_target_langs);
1.146 paf 190: }
1.145 paf 191: }
1.146 paf 192:
1.145 paf 193: public:
1.146 paf 194:
1.222 moko 195: const char* visualize() const;
1.164 paf 196: void dump() const;
1.146 paf 197:
198: Languages(): langs(0) {}
1.147 paf 199: Languages(Language alang) {
200: opt.lang=alang;
201: opt.is_not_just_lang=0;
202: }
1.146 paf 203:
204: /// MUST be called exactly prior to modification of current [uses it's length]
1.187 misha 205: void append(size_t current, Language alang, size_t length) {
1.146 paf 206: assert(alang);
1.187 misha 207: assert(length);
1.146 paf 208:
1.208 moko 209: if(!opt.is_not_just_lang) {
1.147 paf 210: if(opt.lang) {
1.181 misha 211: if(opt.lang==alang) // same language? ignoring
1.146 paf 212: return;
213: } else {
1.147 paf 214: opt.lang=alang; // to uninitialized
1.146 paf 215: return;
1.145 paf 216: }
1.208 moko 217: }
1.146 paf 218:
1.187 misha 219: append(current, CORD_chars((char)alang, length));
1.146 paf 220: }
221:
1.187 misha 222: void append(const Body ¤t, Language alang, size_t length) {
1.176 misha 223: assert(alang);
1.187 misha 224: assert(length);
1.176 misha 225:
1.208 moko 226: if(!opt.is_not_just_lang) {
1.176 misha 227: if(opt.lang) {
1.183 misha 228: if(opt.lang==alang) // same language? ignoring
1.176 misha 229: return;
230: } else {
231: opt.lang=alang; // to uninitialized
232: return;
233: }
1.208 moko 234: }
1.176 misha 235:
1.187 misha 236: append(current, CORD_chars((char)alang, length));
1.176 misha 237: }
238:
1.187 misha 239: void appendHelper(const Body& current, Language alang, const Body &length_helper) {
240: assert(alang);
241:
1.208 moko 242: if(!opt.is_not_just_lang) {
1.187 misha 243: if(opt.lang) {
244: if(opt.lang==alang) // same language? ignoring
245: return;
246: } else {
247: opt.lang=alang; // to uninitialized
248: return;
249: }
1.208 moko 250: }
1.187 misha 251:
252: append(current, CORD_chars((char)alang, length_helper.length()));
253: }
254:
255: void appendHelper(const Body& current, const Languages &src, const Body& length_helper) {
1.186 misha 256: if(!langs){
1.176 misha 257: langs=src.langs; // to uninitialized
1.186 misha 258: #ifdef CORD_CAT_OPTIMIZATION
259: if(opt.is_not_just_lang && !CORD_IS_STRING(langs))
260: CORD_concatenation_protect(langs);
261: #endif
262: }
1.176 misha 263: else if(!src.opt.is_not_just_lang)
264: appendHelper(current, src.opt.lang, length_helper); // simplifying when simple source
265: else
266: append(current, src.make_langs(length_helper));
267: }
268:
1.146 paf 269: /// MUST be called exactly prior to modification of current [uses it's length]
1.187 misha 270: void append(const Body& current, const Languages src, size_t aoffset, size_t alength) {
1.146 paf 271: assert(alength);
272:
273: if(!langs) // to uninitialized?
1.147 paf 274: if(src.opt.is_not_just_lang)
1.194 misha 275: langs=CORD_substr(src.langs, aoffset, alength, 0); // to uninitialized complex
1.146 paf 276: else
1.147 paf 277: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 278: else
1.147 paf 279: 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 280: return; // ignoring
281: else
282: append(current, src.make_langs(aoffset, alength));
283: }
284:
285: /// checks if we have lang<=alang all from aoffset to aoffset+alength
286: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
287: if(alang==L_UNSPECIFIED) // ignore lang?
288: return true;
289:
1.147 paf 290: if(opt.is_not_just_lang)
1.146 paf 291: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
292: else
1.169 misha 293: return (unsigned)opt.lang<=(unsigned)alang;
1.146 paf 294: }
295:
1.148 paf 296: /// @returns count of blocks
297: /// @todo currently there can be adjucent blocks of same language. someday merge them
298: size_t count() const {
299: return opt.is_not_just_lang?
300: CORD_block_count(langs)
301: : opt.lang?
302: 1
303: : 0;
304: };
305:
1.187 misha 306: template<typename I>
307: void for_each(size_t current, int callback(char, size_t, I), I info) const {
308: if(opt.is_not_just_lang)
309: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
310: else
311: callback(opt.lang, current, info);
312: }
313:
314: template<typename I>
315: void for_each(const Body& current, int callback(char, size_t, I), I info) const {
1.147 paf 316: if(opt.is_not_just_lang)
1.146 paf 317: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
318: else
1.187 misha 319: callback(opt.lang, current.length(), info);
1.146 paf 320: }
321:
1.164 paf 322: bool invariant(size_t current_length) const {
1.146 paf 323: if(!langs)
324: return current_length==0;
1.147 paf 325: if(opt.is_not_just_lang)
1.146 paf 326: return CORD_len(langs)==current_length;
327: return true; // uncheckable, actually
328: }
329: };
330:
1.213 moko 331: struct C {
332: const char *str;
333: size_t length;
334: C(): str(0), length(0) {}
335: C(const char *astr, size_t asize): str(astr), length(asize) {}
1.221 moko 336: explicit C(Body abody): str(abody.cstr()), length(abody.length()) {}
1.213 moko 337: };
338:
339: struct Cm {
340: char *str;
341: size_t length;
342: Cm(): str(0), length(0) {}
343: Cm(char *astr, size_t asize): str(astr), length(asize) {}
1.221 moko 344: explicit Cm(Body abody): str(abody.cstrm()), length(abody.length()) {}
1.213 moko 345: };
346:
1.146 paf 347: class Body {
348:
349: CORD body;
350:
1.183 misha 351: #ifdef HASH_CODE_CACHING
352: // cached hash code is not reseted on write operations as test shows
353: // that string body does not change after it is stored as a hash key
354: mutable uint hash_code;
1.213 moko 355: #define INIT_HASH_CODE(c) ,hash_code(c)
1.215 moko 356: #define ZERO_HASH_CODE hash_code=0;
1.213 moko 357: #else
358: #define INIT_HASH_CODE(c)
1.215 moko 359: #define ZERO_HASH_CODE
1.183 misha 360: #endif
361:
1.187 misha 362: #ifdef STRING_LENGTH_CACHING
363: // cached length is reseted on modification, used only for char*, not CORD
364: mutable size_t string_length;
1.213 moko 365: #define INIT_LENGTH(l) ,string_length(l)
1.187 misha 366: #define ZERO_LENGTH string_length=0;
367: #else
1.213 moko 368: #define INIT_LENGTH(l)
1.187 misha 369: #define ZERO_LENGTH
370: #endif
371:
1.146 paf 372: public:
373:
1.164 paf 374: void dump() const;
1.146 paf 375:
1.213 moko 376: Body(): body(CORD_EMPTY) INIT_HASH_CODE(0) INIT_LENGTH(0) {}
1.237 moko 377: explicit Body(const char *abody): body(AS_CORD(abody)) INIT_HASH_CODE(0) INIT_LENGTH(0) {}
378: explicit Body(CORD abody, uint ahash_code): body(abody) INIT_HASH_CODE(ahash_code) INIT_LENGTH(0) {}
1.213 moko 379: explicit Body(C ac): body(AS_CORD(ac.str)) INIT_HASH_CODE(0) INIT_LENGTH(ac.length) {}
380: explicit Body(CORD abody): body(abody) INIT_HASH_CODE(0) INIT_LENGTH(0) {
1.186 misha 381: #ifdef CORD_CAT_OPTIMIZATION
382: assert(!body // no body
383: || *body // ordinary string
384: || body[1]==1 // CONCAT_HDR
385: || body[1]==3 // CONCAT_HDR_READ_ONLY
386: || body[1]==4 // FN_HDR
387: || body[1]==6 // SUBSTR_HDR
388: );
389: #else
1.146 paf 390: assert(!body // no body
391: || *body // ordinary string
392: || body[1]==1 // CONCAT_HDR
393: || body[1]==4 // FN_HDR
394: || body[1]==6 // SUBSTR_HDR
395: );
1.186 misha 396: #endif
1.146 paf 397: }
1.181 misha 398:
1.236 moko 399: static Body uitoa(size_t aindex);
1.213 moko 400:
1.215 moko 401: void clear() { ZERO_LENGTH ZERO_HASH_CODE body=CORD_EMPTY; }
1.146 paf 402:
403: bool operator! () const { return is_empty(); }
404:
1.211 moko 405: inline CORD get_cord() const { return body; }
1.183 misha 406: uint get_hash_code() const;
1.146 paf 407:
1.221 moko 408: // never null
1.197 misha 409: const char* cstr() const {
410: #ifdef STRING_LENGTH_CACHING
411: string_length = length();
1.211 moko 412: if(string_length){
413: const char *result=CORD_to_const_char_star(body, string_length);
414: const_cast<Body*>(this)->body=(CORD)result;
415: return result;
416: }
1.197 misha 417: #endif
418: return CORD_to_const_char_star(body, length());
419: }
420:
1.221 moko 421: // never null
1.194 misha 422: char* cstrm() const { return CORD_to_char_star(body, length()); }
1.146 paf 423:
1.187 misha 424: #ifdef STRING_LENGTH_CACHING
425: void set_length(size_t alength){ string_length = alength; }
1.211 moko 426: size_t length() const { return body ? CORD_IS_STRING(body) ? string_length ? string_length : (string_length=strlen((const char *)body)) : CORD_len(body) : 0; }
1.187 misha 427: #else
1.146 paf 428: size_t length() const { return CORD_len(body); }
1.187 misha 429: #endif
1.146 paf 430:
1.212 moko 431: inline bool is_empty() const { return body==CORD_EMPTY; }
1.146 paf 432:
433: void append_know_length(const char *str, size_t known_length) {
1.187 misha 434: if(known_length){
435: if(body){
436: body = CORD_cat_char_star(body, str, known_length);
437: ZERO_LENGTH
438: } else {
1.211 moko 439: body=(CORD)str;
1.187 misha 440: #ifdef STRING_LENGTH_CACHING
441: string_length=known_length;
442: #endif
443: }
444: }
1.146 paf 445: }
446: void append_strdup_know_length(const char* str, size_t known_length) {
447: if(known_length)
448: append_know_length(pa_strdup(str, known_length), known_length);
449: }
1.187 misha 450: void append(char c) { ZERO_LENGTH body=CORD_cat_char(body, c); }
451: Body& operator << (const Body src) { ZERO_LENGTH body=CORD_cat(body, src.body); return *this; }
452:
1.146 paf 453: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
454:
455: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
456: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
457: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
458: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
1.207 moko 459:
1.146 paf 460: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
461: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
462:
1.211 moko 463: bool operator != (const char *src) const { return CORD_cmp(body, AS_CORD(src))!=0; }
464: bool operator == (const char *src) const { return CORD_cmp(body, AS_CORD(src))==0; }
1.207 moko 465:
1.146 paf 466: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
467: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 468: }
469:
1.146 paf 470: char fetch(size_t index) const { return CORD_fetch(body, index); }
1.211 moko 471: Body mid(size_t aindex, size_t alength) const { return Body(CORD_substr(body, aindex, alength, length())); }
472: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, AS_CORD(substr), length()); }
1.146 paf 473: size_t pos(const Body substr, size_t offset=0) const {
1.180 misha 474: if(substr.is_empty())
1.146 paf 475: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 476:
477: // CORD_str checks for bad offset [CORD_chr does not]
1.194 misha 478: return CORD_str(body, offset, substr.body, length());
1.146 paf 479: }
1.207 moko 480: size_t pos(char c, size_t offset=0) const {
1.150 paf 481: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
482: return STRING_NOT_FOUND;
483:
1.146 paf 484: return CORD_chr(body, offset, c);
1.145 paf 485: }
1.146 paf 486:
1.203 misha 487: size_t strrpbrk(const char* chars, size_t left, size_t right) const;
488:
489: size_t rskipchars(const char* chars, size_t left, size_t right) const;
490:
1.187 misha 491: template<typename I>
492: int for_each(int (*f)(char c, I), I info) const {
1.159 paf 493: return CORD_iter(body, (CORD_iter_fn)f, (void*)info);
494: }
495:
1.187 misha 496: template<typename I>
497: int for_each(int (*f1)(char c, I), int (*f2)(const char* s, I), I info) const {
1.159 paf 498: return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 499: }
1.146 paf 500:
501: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
502:
1.157 paf 503: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled
504: Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0,
1.188 misha 505: size_t* out_start=0, size_t* out_length=0, Charset* source_charset=0) const;
1.145 paf 506: };
507:
1.218 moko 508: private:
1.145 paf 509:
1.152 paf 510: Body body; ///< all characters of string
1.146 paf 511: Languages langs; ///< string characters lang
512:
1.164 paf 513: void dump() const;
514: #define ASSERT_STRING_INVARIANT(string) \
515: assert((string).langs.invariant((string).body.length()))
1.145 paf 516:
1.8 paf 517: public:
1.151 paf 518:
519: static const String Empty;
1.8 paf 520:
1.181 misha 521: explicit String(){};
1.211 moko 522: explicit String(const char* cstr, Language alang=L_CLEAN) : body(cstr){
523: if(body.get_cord()){
1.182 misha 524: langs=alang;
1.181 misha 525: }
526: }
1.213 moko 527: explicit String(C ac, Language alang=L_CLEAN) : body(ac){
1.211 moko 528: if(body.get_cord()){
529: langs=alang;
530: }
531: }
1.146 paf 532: String(Body abody, Language alang): body(abody), langs(alang) {
533: ASSERT_STRING_INVARIANT(*this);
534: }
535: String(const String& src): body(src.body), langs(src.langs) {
536: ASSERT_STRING_INVARIANT(*this);
1.145 paf 537: }
1.212 moko 538: String(int value, const char *format);
1.145 paf 539:
540: /// for convinient hash lookup
1.183 misha 541: #ifdef HASH_CODE_CACHING
542: operator const Body&() const { return body; }
543: #else
1.146 paf 544: operator const Body() const { return body; }
1.183 misha 545: #endif
1.145 paf 546:
547: bool is_empty() const { return body.is_empty(); }
548: size_t length() const { return body.length(); }
1.171 misha 549: size_t length(Charset& charset) const;
1.145 paf 550:
1.191 misha 551: /// convert to CORD forcing lang tainting
552: Body cstr_to_string_body_taint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const;
553: /// convert to CORD with tainting dirty to lang
554: Body cstr_to_string_body_untaint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const;
1.145 paf 555:
1.212 moko 556: /// from body
557: const char* cstr() const { return body.cstr(); }
558: char* cstrm() const { return body.cstrm(); }
1.189 misha 559:
560: /// convert to constant C string forcing lang tainting
1.190 misha 561: const char* taint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 562: return cstr_to_string_body_taint(lang, connection, charsets).cstr();
563: }
1.190 misha 564: char *taint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 565: return cstr_to_string_body_taint(lang, connection, charsets).cstrm();
566: }
567:
568: /// convert to constant C string with tainting dirty to lang
1.190 misha 569: const char* untaint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 570: return cstr_to_string_body_untaint(lang, connection, charsets).cstr();
571: }
1.190 misha 572: char *untaint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 573: return cstr_to_string_body_untaint(lang, connection, charsets).cstrm();
574: }
1.195 misha 575:
1.196 misha 576: const char* untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const;
1.195 misha 577:
1.202 moko 578: bool is_not_just_lang() const {
579: return langs.opt.is_not_just_lang !=0;
580: }
581:
582: Language just_lang() const {
583: return langs.opt.lang;
584: }
585:
1.222 moko 586: char* visualize_langs() const;
587:
1.108 parser 588: /// puts pieces to buf
1.145 paf 589: Cm serialize(size_t prolog_size) const;
1.108 parser 590: /// appends pieces from buf to self
1.145 paf 591: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 592: /// @see Body::append_know_length
1.145 paf 593: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 594: /// @see Body::append_help_length
1.145 paf 595: String& append_help_length(const char* str, size_t helper_length, Language lang);
596: String& append_strdup(const char* str, size_t helper_length, Language lang);
597:
1.146 paf 598: bool operator == (const char* y) const { return body==Body(y); }
599: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 600:
601: /// this starts with y
602: bool starts_with(const char* y) const {
1.146 paf 603: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 604: }
605: /// x starts with this
606: bool this_starts(const char* x) const {
1.146 paf 607: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 608: }
609:
1.220 moko 610: String& append_to(String& dest) const;
1.187 misha 611: String& append_to(String& dest, Language lang, bool forced=false) const;
1.145 paf 612: String& append(const String& src, Language lang, bool forced=false) {
613: return src.append_to(*this, lang, forced);
614: }
1.199 misha 615: String& append_quoted(const String* src, Language lang=L_JSON){
616: *this << "\"";
617: if(src)
618: this->append(*src, lang, true/*forced lang*/);
619: *this << "\"";
620: return *this;
621: }
622:
1.220 moko 623: String& operator << (const String& src) { return src.append_to(*this); }
1.145 paf 624: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.187 misha 625: String& operator << (const Body& src){
626: langs.appendHelper(body, L_AS_IS, src);
627: body<<src;
628: return *this;
629: }
1.100 parser 630:
1.142 paf 631: char first_char() const {
1.145 paf 632: return is_empty()?0:body.fetch(0);
1.142 paf 633: }
1.54 paf 634:
1.205 moko 635: char last_char() const {
636: return is_empty()?0:body.fetch(body.length()-1);
637: }
638:
1.145 paf 639: bool operator < (const String& src) const { return body<src.body; }
640: bool operator > (const String& src) const { return body>src.body; }
641: bool operator <= (const String& src) const { return body<=src.body; }
642: bool operator >= (const String& src) const { return body>=src.body; }
643: bool operator != (const String& src) const { return body!=src.body; }
644: bool operator == (const String& src) const { return body==src.body; }
645:
1.54 paf 646: /// extracts [start, finish) piece of string
1.145 paf 647: String& mid(size_t substr_begin, size_t substr_end) const;
1.172 misha 648: String& mid(Charset& charset, size_t from, size_t to, size_t helper_length=0) const;
1.145 paf 649:
1.235 moko 650: /// return position of substr in string, -1 means "not found" [const char* version]
651: size_t pos(const char* substr, size_t this_offset=0) const { return body.pos(substr, this_offset); }
652: size_t pos(const Body substr, size_t this_offset=0) const { return body.pos(substr, this_offset); }
653: size_t pos(const String& substr, size_t this_offset=0) const { return body.pos(substr.body, this_offset); }
654: size_t pos(char c, size_t this_offset=0) const { return body.pos(c, this_offset); }
655: /// ignore lang if it's L_UNSPECIFIED, otherwise look for substring that lies in ONE fragment in THAT lang
656: size_t pos(const Body substr, size_t this_offset, Language lang) const;
657: size_t pos(const String& substr, size_t this_offset, Language lang) const;
1.216 moko 658: size_t pos(Charset& charset, const String& substr, size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.55 paf 659:
1.203 misha 660: size_t strrpbrk(const char* chars, size_t left=0) const {
661: return (length()) ? body.strrpbrk(chars, left, length()-1) : STRING_NOT_FOUND;
662: }
663: size_t strrpbrk(const char* chars, size_t left, size_t right) const {
664: return body.strrpbrk(chars, left, right);
665: }
666:
667: size_t rskipchars(const char* chars, size_t left=0) const {
668: return (length()) ? body.rskipchars(chars, left, length()-1) : STRING_NOT_FOUND;
669: }
670: size_t rskipchars(const char* chars, size_t left, size_t right) const {
671: return body.rskipchars(chars, left, right);
672: }
673:
1.219 moko 674: void split(ArrayString& result, size_t pos_after, const char* delim, Language lang=L_UNSPECIFIED) const;
675: void split(ArrayString& result, size_t pos_after, const String& delim, Language lang=L_UNSPECIFIED) const;
1.216 moko 676:
677: typedef void (*Row_action)(Table& table, ArrayString* row, int prestart, int prefinish, int poststart, int postfinish, void *info);
678:
1.87 parser 679: /**
1.145 paf 680: @return table of found items, if any.
1.87 parser 681: table format is defined and fixed[can be used by others]:
682: @verbatim
683: prematch/match/postmatch/1/2/3/...
684: @endverbatim
685: */
1.216 moko 686: Table* match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const;
687:
1.87 parser 688: enum Change_case_kind {
689: CC_UPPER,
690: CC_LOWER
691: };
1.216 moko 692: String& change_case(Charset& source_charset, Change_case_kind kind) const;
693:
1.145 paf 694: const String& replace(const Dictionary& dict) const;
1.188 misha 695: const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0, Charset* source_charset=0) const;
1.155 paf 696: double as_double() const { return pa_atod(cstr(), this); }
1.226 moko 697: int as_int() const { return pa_atoi(cstr(), 0, this); }
1.240 ! moko 698: pa_wint as_wint() const { return pa_atowi(cstr(), 0, this); }
! 699: bool as_bool() const { return as_wint()!=0; }
1.174 misha 700: const String& escape(Charset& source_charset) const;
1.137 paf 701:
1.7 paf 702: private: //disabled
703:
1.12 paf 704: String& operator = (const String&) { return *this; }
1.7 paf 705:
1.1 paf 706: };
1.119 paf 707:
1.183 misha 708: #ifndef HASH_CODE_CACHING
1.145 paf 709: /// simple hash code of string. used by Hash
1.146 paf 710: inline uint hash_code(const String::Body self) {
1.184 misha 711: return self.get_hash_code();
1.119 paf 712: }
1.183 misha 713: #endif
1.147 paf 714:
715:
1.1 paf 716: #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