Annotation of parser3/src/include/pa_string.h, revision 1.228
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.228 ! moko 4: Copyright (c) 2001-2020 Art. Lebedev Studio (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:
1.228 ! moko 11: #define IDENT_PA_STRING_H "$Id: pa_string.h,v 1.227 2020/12/02 15:25:41 moko Exp $"
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
1.224 moko 19: #include "../lib/cord/include/cord.h"
1.186 misha 20:
21: #ifdef CORD_CAT_OPTIMIZATION
22: #define CORD_cat(x, y) CORD_cat_optimized(x, y)
23: #define CORD_cat_char_star(x, y, leny) CORD_cat_char_star_optimized(x, y, leny)
24: #endif
1.209 moko 25: }
1.4 paf 26:
1.164 paf 27: // defines
28:
1.183 misha 29: // cache hash code in String::Body for faster hash access
30: #define HASH_CODE_CACHING
31:
1.187 misha 32: // cache String::Body.length() for char* strings only, CORDs have own length
33: #define STRING_LENGTH_CACHING
34:
35:
1.146 paf 36: // cord extension
37: /* Returns true if x does contain */
38: /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */
39: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c);
1.148 paf 40: size_t CORD_block_count(CORD x);
1.146 paf 41:
1.145 paf 42: // forwards
1.9 paf 43:
1.145 paf 44: class Charset;
1.135 paf 45: class Table;
1.71 paf 46: class SQL_Connection;
1.101 parser 47: class Dictionary;
1.145 paf 48: class Request_charsets;
49: class String;
50: typedef Array<const String*> ArrayString;
1.178 misha 51: class VRegex;
1.145 paf 52:
1.155 paf 53: // generally useful
54:
55: double pa_atod(const char* str, const String* problem_source=0);
1.226 moko 56: int pa_atoi(const char* str, int base=10, const String* problem_source=0);
57: unsigned int pa_atoui(const char *str, int base=10, const String* problem_source=0);
1.227 moko 58: uint64_t pa_atoul(const char *str, int base=10, const String* problem_source=0);
1.155 paf 59:
1.145 paf 60: /// this is result of pos functions which mean that substr were not found
61: #define STRING_NOT_FOUND ((size_t)-1)
62:
1.211 moko 63: /// CORD can't be empty string, thus checking it in assigment
64: #define AS_CORD(v) ((v) && *(v) ? (CORD)(v):0)
65:
1.42 paf 66: /**
1.146 paf 67: String which knows the lang of all it's langs.
1.41 paf 68:
69: All pieces remember
70: - whether they are tainted or not,
1.146 paf 71: and the lang which should be used to detaint them
1.41 paf 72: */
1.176 misha 73:
74:
1.145 paf 75: class String: public PA_Object {
1.218 moko 76: friend class StringSplitHelper;
1.1 paf 77: public:
1.48 paf 78:
1.146 paf 79: /** piece is tainted or not. the lang to use when detaint
1.106 parser 80: remember to change String_Untaint_lang_name @ untaint.C along
1.148 paf 81:
82: WARNING WARNING WARNING WARNING WARNING WARNING
83:
84: pos function compares(<=) languages, that is used in searching
85: for table column separator being L_CLEAN or L_AS_IS.
86: they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check].
87:
88: letters assigned for debugging, but it's important for no language-letter
89: come before L_AS_IS other then L_CLEAN
90:
91: WARNING WARNING WARNING WARNING WARNING WARNING
1.106 parser 92: */
1.225 moko 93:
94: #if _MSC_VER >= 1900
95: /// required for VS2015+ to make sizeof(Languages::opt) == sizeof(CORD), will be 16 byte under x64 without it
96: enum Language : size_t {
97: #else
1.145 paf 98: enum Language {
1.225 moko 99: #endif
100: L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty
1.145 paf 101: // these two must go before others, there are checks for >L_AS_IS
1.225 moko 102: L_CLEAN='0', ///< clean WARNING: read above warning before changing
103: L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing
1.122 paf 104:
1.225 moko 105: L_TAINTED='T', ///< tainted, untaint lang as assigned later
1.146 paf 106: // untaint langs. assigned by ^untaint[lang]{...}
1.170 misha 107: L_FILE_SPEC='F', ///< file specification
108: L_HTTP_HEADER='h', ///< text in HTTP response header
109: L_MAIL_HEADER='m', ///< text in mail header
1.225 moko 110: L_URI='U', ///< text in uri
111: L_SQL='Q', ///< ^table:sql body
112: L_JS='J', ///< JavaScript code
113: L_XML='X', ///< ^xdoc:create xml
114: L_HTML='H', ///< HTML code
1.198 misha 115: L_REGEX='R', ///< RegExp
1.225 moko 116: L_JSON='S', ///< JSON code
1.170 misha 117: L_HTTP_COOKIE='C', ///< cookies encoded as %uXXXX for compartibility with js functions encode/decode
1.192 misha 118: L_PARSER_CODE='p', ///< ^process body
1.148 paf 119: // READ WARNING ABOVE BEFORE ADDING ANYTHING
1.146 paf 120: L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 121: };
122:
1.157 paf 123: enum Trim_kind {
124: TRIM_BOTH,
125: TRIM_START,
126: TRIM_END
127: };
128:
1.187 misha 129: class Body;
1.176 misha 130:
1.147 paf 131: union Languages {
1.146 paf 132:
1.147 paf 133: struct {
1.160 paf 134: #ifdef PA_LITTLE_ENDIAN
1.147 paf 135: Language lang:8;
1.179 misha 136: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 137: #elif defined(PA_BIG_ENDIAN)
1.179 misha 138: size_t is_not_just_lang:sizeof(CORD)*8-8;
1.160 paf 139: Language lang:8;
140: #else
141: # error word endianness not determined for some obscure reason
142: #endif
1.147 paf 143: } opt;
144: CORD langs;
1.146 paf 145:
1.187 misha 146: CORD make_langs(const Body& current) const {
147: return opt.is_not_just_lang?langs:CORD_chars((char)opt.lang, current.length());
1.146 paf 148: }
149:
150: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 151: return opt.is_not_just_lang?
1.194 misha 152: CORD_substr(langs, aoffset, alength, 0)
1.162 paf 153: :CORD_chars((char)opt.lang, alength);
1.145 paf 154: }
155:
1.157 paf 156: /// appending when 'langs' already contain something [simple cases handled elsewhere]
1.187 misha 157: void append(size_t current, const CORD to_nonempty_target_langs) {
1.146 paf 158: assert(langs);
1.187 misha 159: if(opt.is_not_just_lang)
160: langs=CORD_cat(langs, to_nonempty_target_langs);
161: else {
162: assert(current);
163: langs=CORD_cat(CORD_chars((char)opt.lang, current), to_nonempty_target_langs);
164: }
165: }
1.146 paf 166:
1.187 misha 167: void append(const Body& current, const CORD to_nonempty_target_langs) {
168: assert(langs);
1.147 paf 169: if(opt.is_not_just_lang)
1.146 paf 170: langs=CORD_cat(langs, to_nonempty_target_langs);
1.187 misha 171: else {
172: size_t current_size=current.length();
1.146 paf 173: assert(current_size);
1.187 misha 174: langs=CORD_cat(CORD_chars((char)opt.lang, current_size), to_nonempty_target_langs);
1.146 paf 175: }
1.145 paf 176: }
1.146 paf 177:
1.145 paf 178: public:
1.146 paf 179:
1.222 moko 180: const char* visualize() const;
1.164 paf 181: void dump() const;
1.146 paf 182:
183: Languages(): langs(0) {}
1.147 paf 184: Languages(Language alang) {
185: opt.lang=alang;
186: opt.is_not_just_lang=0;
187: }
1.146 paf 188:
189: /// MUST be called exactly prior to modification of current [uses it's length]
1.187 misha 190: void append(size_t current, Language alang, size_t length) {
1.146 paf 191: assert(alang);
1.187 misha 192: assert(length);
1.146 paf 193:
1.208 moko 194: if(!opt.is_not_just_lang) {
1.147 paf 195: if(opt.lang) {
1.181 misha 196: if(opt.lang==alang) // same language? ignoring
1.146 paf 197: return;
198: } else {
1.147 paf 199: opt.lang=alang; // to uninitialized
1.146 paf 200: return;
1.145 paf 201: }
1.208 moko 202: }
1.146 paf 203:
1.187 misha 204: append(current, CORD_chars((char)alang, length));
1.146 paf 205: }
206:
1.187 misha 207: void append(const Body ¤t, Language alang, size_t length) {
1.176 misha 208: assert(alang);
1.187 misha 209: assert(length);
1.176 misha 210:
1.208 moko 211: if(!opt.is_not_just_lang) {
1.176 misha 212: if(opt.lang) {
1.183 misha 213: if(opt.lang==alang) // same language? ignoring
1.176 misha 214: return;
215: } else {
216: opt.lang=alang; // to uninitialized
217: return;
218: }
1.208 moko 219: }
1.176 misha 220:
1.187 misha 221: append(current, CORD_chars((char)alang, length));
1.176 misha 222: }
223:
1.187 misha 224: void appendHelper(const Body& current, Language alang, const Body &length_helper) {
225: assert(alang);
226:
1.208 moko 227: if(!opt.is_not_just_lang) {
1.187 misha 228: if(opt.lang) {
229: if(opt.lang==alang) // same language? ignoring
230: return;
231: } else {
232: opt.lang=alang; // to uninitialized
233: return;
234: }
1.208 moko 235: }
1.187 misha 236:
237: append(current, CORD_chars((char)alang, length_helper.length()));
238: }
239:
240: void appendHelper(const Body& current, const Languages &src, const Body& length_helper) {
1.186 misha 241: if(!langs){
1.176 misha 242: langs=src.langs; // to uninitialized
1.186 misha 243: #ifdef CORD_CAT_OPTIMIZATION
244: if(opt.is_not_just_lang && !CORD_IS_STRING(langs))
245: CORD_concatenation_protect(langs);
246: #endif
247: }
1.176 misha 248: else if(!src.opt.is_not_just_lang)
249: appendHelper(current, src.opt.lang, length_helper); // simplifying when simple source
250: else
251: append(current, src.make_langs(length_helper));
252: }
253:
1.146 paf 254: /// MUST be called exactly prior to modification of current [uses it's length]
1.187 misha 255: void append(const Body& current, const Languages src, size_t aoffset, size_t alength) {
1.146 paf 256: assert(alength);
257:
258: if(!langs) // to uninitialized?
1.147 paf 259: if(src.opt.is_not_just_lang)
1.194 misha 260: langs=CORD_substr(src.langs, aoffset, alength, 0); // to uninitialized complex
1.146 paf 261: else
1.147 paf 262: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 263: else
1.147 paf 264: 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 265: return; // ignoring
266: else
267: append(current, src.make_langs(aoffset, alength));
268: }
269:
270: /// checks if we have lang<=alang all from aoffset to aoffset+alength
271: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
272: if(alang==L_UNSPECIFIED) // ignore lang?
273: return true;
274:
1.147 paf 275: if(opt.is_not_just_lang)
1.146 paf 276: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
277: else
1.169 misha 278: return (unsigned)opt.lang<=(unsigned)alang;
1.146 paf 279: }
280:
1.148 paf 281: /// @returns count of blocks
282: /// @todo currently there can be adjucent blocks of same language. someday merge them
283: size_t count() const {
284: return opt.is_not_just_lang?
285: CORD_block_count(langs)
286: : opt.lang?
287: 1
288: : 0;
289: };
290:
1.187 misha 291: template<typename I>
292: void for_each(size_t current, int callback(char, size_t, I), I info) const {
293: if(opt.is_not_just_lang)
294: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
295: else
296: callback(opt.lang, current, info);
297: }
298:
299: template<typename I>
300: void for_each(const Body& current, int callback(char, size_t, I), I info) const {
1.147 paf 301: if(opt.is_not_just_lang)
1.146 paf 302: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
303: else
1.187 misha 304: callback(opt.lang, current.length(), info);
1.146 paf 305: }
306:
1.164 paf 307: bool invariant(size_t current_length) const {
1.146 paf 308: if(!langs)
309: return current_length==0;
1.147 paf 310: if(opt.is_not_just_lang)
1.146 paf 311: return CORD_len(langs)==current_length;
312: return true; // uncheckable, actually
313: }
314: };
315:
1.213 moko 316: struct C {
317: const char *str;
318: size_t length;
319: C(): str(0), length(0) {}
320: C(const char *astr, size_t asize): str(astr), length(asize) {}
1.221 moko 321: explicit C(Body abody): str(abody.cstr()), length(abody.length()) {}
1.213 moko 322: };
323:
324: struct Cm {
325: char *str;
326: size_t length;
327: Cm(): str(0), length(0) {}
328: Cm(char *astr, size_t asize): str(astr), length(asize) {}
1.221 moko 329: explicit Cm(Body abody): str(abody.cstrm()), length(abody.length()) {}
1.213 moko 330: };
331:
1.146 paf 332: class Body {
333:
334: CORD body;
335:
1.183 misha 336: #ifdef HASH_CODE_CACHING
337: // cached hash code is not reseted on write operations as test shows
338: // that string body does not change after it is stored as a hash key
339: mutable uint hash_code;
1.213 moko 340: #define INIT_HASH_CODE(c) ,hash_code(c)
1.215 moko 341: #define ZERO_HASH_CODE hash_code=0;
1.213 moko 342: #else
343: #define INIT_HASH_CODE(c)
1.215 moko 344: #define ZERO_HASH_CODE
1.183 misha 345: #endif
346:
1.187 misha 347: #ifdef STRING_LENGTH_CACHING
348: // cached length is reseted on modification, used only for char*, not CORD
349: mutable size_t string_length;
1.213 moko 350: #define INIT_LENGTH(l) ,string_length(l)
1.187 misha 351: #define ZERO_LENGTH string_length=0;
352: #else
1.213 moko 353: #define INIT_LENGTH(l)
1.187 misha 354: #define ZERO_LENGTH
355: #endif
356:
1.146 paf 357: public:
358:
1.164 paf 359: void dump() const;
1.146 paf 360:
1.213 moko 361: Body(): body(CORD_EMPTY) INIT_HASH_CODE(0) INIT_LENGTH(0) {}
362: Body(const char *abody): body(AS_CORD(abody)) INIT_HASH_CODE(0) INIT_LENGTH(0) {}
363: Body(CORD abody, uint ahash_code): body(abody) INIT_HASH_CODE(ahash_code) INIT_LENGTH(0) {}
364: explicit Body(C ac): body(AS_CORD(ac.str)) INIT_HASH_CODE(0) INIT_LENGTH(ac.length) {}
365: explicit Body(CORD abody): body(abody) INIT_HASH_CODE(0) INIT_LENGTH(0) {
1.186 misha 366: #ifdef CORD_CAT_OPTIMIZATION
367: assert(!body // no body
368: || *body // ordinary string
369: || body[1]==1 // CONCAT_HDR
370: || body[1]==3 // CONCAT_HDR_READ_ONLY
371: || body[1]==4 // FN_HDR
372: || body[1]==6 // SUBSTR_HDR
373: );
374: #else
1.146 paf 375: assert(!body // no body
376: || *body // ordinary string
377: || body[1]==1 // CONCAT_HDR
378: || body[1]==4 // FN_HDR
379: || body[1]==6 // SUBSTR_HDR
380: );
1.186 misha 381: #endif
1.146 paf 382: }
1.181 misha 383:
1.213 moko 384:
1.146 paf 385: static Body Format(int value);
386:
1.215 moko 387: void clear() { ZERO_LENGTH ZERO_HASH_CODE body=CORD_EMPTY; }
1.146 paf 388:
389: bool operator! () const { return is_empty(); }
390:
1.211 moko 391: inline CORD get_cord() const { return body; }
1.183 misha 392: uint get_hash_code() const;
1.146 paf 393:
1.221 moko 394: // never null
1.197 misha 395: const char* cstr() const {
396: #ifdef STRING_LENGTH_CACHING
397: string_length = length();
1.211 moko 398: if(string_length){
399: const char *result=CORD_to_const_char_star(body, string_length);
400: const_cast<Body*>(this)->body=(CORD)result;
401: return result;
402: }
1.197 misha 403: #endif
404: return CORD_to_const_char_star(body, length());
405: }
406:
1.221 moko 407: // never null
1.194 misha 408: char* cstrm() const { return CORD_to_char_star(body, length()); }
1.146 paf 409:
1.187 misha 410: #ifdef STRING_LENGTH_CACHING
411: void set_length(size_t alength){ string_length = alength; }
1.211 moko 412: 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 413: #else
1.146 paf 414: size_t length() const { return CORD_len(body); }
1.187 misha 415: #endif
1.146 paf 416:
1.212 moko 417: inline bool is_empty() const { return body==CORD_EMPTY; }
1.146 paf 418:
419: void append_know_length(const char *str, size_t known_length) {
1.187 misha 420: if(known_length){
421: if(body){
422: body = CORD_cat_char_star(body, str, known_length);
423: ZERO_LENGTH
424: } else {
1.211 moko 425: body=(CORD)str;
1.187 misha 426: #ifdef STRING_LENGTH_CACHING
427: string_length=known_length;
428: #endif
429: }
430: }
1.146 paf 431: }
432: void append_strdup_know_length(const char* str, size_t known_length) {
433: if(known_length)
434: append_know_length(pa_strdup(str, known_length), known_length);
435: }
1.187 misha 436: void append(char c) { ZERO_LENGTH body=CORD_cat_char(body, c); }
437: Body& operator << (const Body src) { ZERO_LENGTH body=CORD_cat(body, src.body); return *this; }
438:
1.146 paf 439: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
440:
441: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
442: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
443: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
444: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
1.207 moko 445:
1.146 paf 446: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
447: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
448:
1.211 moko 449: bool operator != (const char *src) const { return CORD_cmp(body, AS_CORD(src))!=0; }
450: bool operator == (const char *src) const { return CORD_cmp(body, AS_CORD(src))==0; }
1.207 moko 451:
1.146 paf 452: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
453: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 454: }
455:
1.146 paf 456: char fetch(size_t index) const { return CORD_fetch(body, index); }
1.211 moko 457: Body mid(size_t aindex, size_t alength) const { return Body(CORD_substr(body, aindex, alength, length())); }
458: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, AS_CORD(substr), length()); }
1.146 paf 459: size_t pos(const Body substr, size_t offset=0) const {
1.180 misha 460: if(substr.is_empty())
1.146 paf 461: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 462:
463: // CORD_str checks for bad offset [CORD_chr does not]
1.194 misha 464: return CORD_str(body, offset, substr.body, length());
1.146 paf 465: }
1.207 moko 466: size_t pos(char c, size_t offset=0) const {
1.150 paf 467: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
468: return STRING_NOT_FOUND;
469:
1.146 paf 470: return CORD_chr(body, offset, c);
1.145 paf 471: }
1.146 paf 472:
1.203 misha 473: size_t strrpbrk(const char* chars, size_t left, size_t right) const;
474:
475: size_t rskipchars(const char* chars, size_t left, size_t right) const;
476:
1.187 misha 477: template<typename I>
478: int for_each(int (*f)(char c, I), I info) const {
1.159 paf 479: return CORD_iter(body, (CORD_iter_fn)f, (void*)info);
480: }
481:
1.187 misha 482: template<typename I>
483: int for_each(int (*f1)(char c, I), int (*f2)(const char* s, I), I info) const {
1.159 paf 484: return CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 485: }
1.146 paf 486:
487: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
488:
1.157 paf 489: /// @returns this or 0 or mid. if returns this or 0 out_* are not filled
490: Body trim(Trim_kind kind=TRIM_BOTH, const char* chars=0,
1.188 misha 491: size_t* out_start=0, size_t* out_length=0, Charset* source_charset=0) const;
1.145 paf 492: };
493:
1.218 moko 494: private:
1.145 paf 495:
1.152 paf 496: Body body; ///< all characters of string
1.146 paf 497: Languages langs; ///< string characters lang
498:
1.164 paf 499: void dump() const;
500: #define ASSERT_STRING_INVARIANT(string) \
501: assert((string).langs.invariant((string).body.length()))
1.145 paf 502:
1.8 paf 503: public:
1.151 paf 504:
505: static const String Empty;
1.8 paf 506:
1.181 misha 507: explicit String(){};
1.211 moko 508: explicit String(const char* cstr, Language alang=L_CLEAN) : body(cstr){
509: if(body.get_cord()){
1.182 misha 510: langs=alang;
1.181 misha 511: }
512: }
1.213 moko 513: explicit String(C ac, Language alang=L_CLEAN) : body(ac){
1.211 moko 514: if(body.get_cord()){
515: langs=alang;
516: }
517: }
1.146 paf 518: String(Body abody, Language alang): body(abody), langs(alang) {
519: ASSERT_STRING_INVARIANT(*this);
520: }
521: String(const String& src): body(src.body), langs(src.langs) {
522: ASSERT_STRING_INVARIANT(*this);
1.145 paf 523: }
1.212 moko 524: String(int value, const char *format);
1.145 paf 525:
526: /// for convinient hash lookup
1.183 misha 527: #ifdef HASH_CODE_CACHING
528: operator const Body&() const { return body; }
529: #else
1.146 paf 530: operator const Body() const { return body; }
1.183 misha 531: #endif
1.145 paf 532:
533: bool is_empty() const { return body.is_empty(); }
534: size_t length() const { return body.length(); }
1.171 misha 535: size_t length(Charset& charset) const;
1.145 paf 536:
1.191 misha 537: /// convert to CORD forcing lang tainting
538: Body cstr_to_string_body_taint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const;
539: /// convert to CORD with tainting dirty to lang
540: Body cstr_to_string_body_untaint(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const;
1.145 paf 541:
1.212 moko 542: /// from body
543: const char* cstr() const { return body.cstr(); }
544: char* cstrm() const { return body.cstrm(); }
1.189 misha 545:
546: /// convert to constant C string forcing lang tainting
1.190 misha 547: const char* taint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 548: return cstr_to_string_body_taint(lang, connection, charsets).cstr();
549: }
1.190 misha 550: char *taint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 551: return cstr_to_string_body_taint(lang, connection, charsets).cstrm();
552: }
553:
554: /// convert to constant C string with tainting dirty to lang
1.190 misha 555: const char* untaint_cstr(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 556: return cstr_to_string_body_untaint(lang, connection, charsets).cstr();
557: }
1.190 misha 558: char *untaint_cstrm(Language lang, SQL_Connection* connection=0, const Request_charsets *charsets=0) const {
1.189 misha 559: return cstr_to_string_body_untaint(lang, connection, charsets).cstrm();
560: }
1.195 misha 561:
1.196 misha 562: const char* untaint_and_transcode_cstr(Language lang, const Request_charsets *charsets) const;
1.195 misha 563:
1.202 moko 564: bool is_not_just_lang() const {
565: return langs.opt.is_not_just_lang !=0;
566: }
567:
568: Language just_lang() const {
569: return langs.opt.lang;
570: }
571:
1.222 moko 572: char* visualize_langs() const;
573:
1.108 parser 574: /// puts pieces to buf
1.145 paf 575: Cm serialize(size_t prolog_size) const;
1.108 parser 576: /// appends pieces from buf to self
1.145 paf 577: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 578: /// @see Body::append_know_length
1.145 paf 579: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 580: /// @see Body::append_help_length
1.145 paf 581: String& append_help_length(const char* str, size_t helper_length, Language lang);
582: String& append_strdup(const char* str, size_t helper_length, Language lang);
583:
1.146 paf 584: bool operator == (const char* y) const { return body==Body(y); }
585: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 586:
587: /// this starts with y
588: bool starts_with(const char* y) const {
1.146 paf 589: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 590: }
591: /// x starts with this
592: bool this_starts(const char* x) const {
1.146 paf 593: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 594: }
595:
1.220 moko 596: String& append_to(String& dest) const;
1.187 misha 597: String& append_to(String& dest, Language lang, bool forced=false) const;
1.145 paf 598: String& append(const String& src, Language lang, bool forced=false) {
599: return src.append_to(*this, lang, forced);
600: }
1.199 misha 601: String& append_quoted(const String* src, Language lang=L_JSON){
602: *this << "\"";
603: if(src)
604: this->append(*src, lang, true/*forced lang*/);
605: *this << "\"";
606: return *this;
607: }
608:
1.220 moko 609: String& operator << (const String& src) { return src.append_to(*this); }
1.145 paf 610: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.187 misha 611: String& operator << (const Body& src){
612: langs.appendHelper(body, L_AS_IS, src);
613: body<<src;
614: return *this;
615: }
1.100 parser 616:
1.142 paf 617: char first_char() const {
1.145 paf 618: return is_empty()?0:body.fetch(0);
1.142 paf 619: }
1.54 paf 620:
1.205 moko 621: char last_char() const {
622: return is_empty()?0:body.fetch(body.length()-1);
623: }
624:
1.145 paf 625: bool operator < (const String& src) const { return body<src.body; }
626: bool operator > (const String& src) const { return body>src.body; }
627: bool operator <= (const String& src) const { return body<=src.body; }
628: bool operator >= (const String& src) const { return body>=src.body; }
629: bool operator != (const String& src) const { return body!=src.body; }
630: bool operator == (const String& src) const { return body==src.body; }
631:
1.54 paf 632: /// extracts [start, finish) piece of string
1.145 paf 633: String& mid(size_t substr_begin, size_t substr_end) const;
1.172 misha 634: String& mid(Charset& charset, size_t from, size_t to, size_t helper_length=0) const;
1.145 paf 635:
636: /**
637: ignore lang if it's L_UNSPECIFIED
638: but when specified: look for substring that lies in ONE fragment in THAT lang
639: @return position of substr in string, -1 means "not found" [const char* version]
640: */
1.216 moko 641: size_t pos(const Body substr, size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.145 paf 642: /// String version of @see pos(const char*, int, Language)
1.216 moko 643: size_t pos(const String& substr, size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
644: size_t pos(char c, size_t this_offset=0) const {
1.145 paf 645: return body.pos(c, this_offset);
646: }
1.216 moko 647: size_t pos(Charset& charset, const String& substr, size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
1.55 paf 648:
1.203 misha 649: size_t strrpbrk(const char* chars, size_t left=0) const {
650: return (length()) ? body.strrpbrk(chars, left, length()-1) : STRING_NOT_FOUND;
651: }
652: size_t strrpbrk(const char* chars, size_t left, size_t right) const {
653: return body.strrpbrk(chars, left, right);
654: }
655:
656: size_t rskipchars(const char* chars, size_t left=0) const {
657: return (length()) ? body.rskipchars(chars, left, length()-1) : STRING_NOT_FOUND;
658: }
659: size_t rskipchars(const char* chars, size_t left, size_t right) const {
660: return body.rskipchars(chars, left, right);
661: }
662:
1.219 moko 663: void split(ArrayString& result, size_t pos_after, const char* delim, Language lang=L_UNSPECIFIED) const;
664: void split(ArrayString& result, size_t pos_after, const String& delim, Language lang=L_UNSPECIFIED) const;
1.216 moko 665:
666: typedef void (*Row_action)(Table& table, ArrayString* row, int prestart, int prefinish, int poststart, int postfinish, void *info);
667:
1.87 parser 668: /**
1.145 paf 669: @return table of found items, if any.
1.87 parser 670: table format is defined and fixed[can be used by others]:
671: @verbatim
672: prematch/match/postmatch/1/2/3/...
673: @endverbatim
674: */
1.216 moko 675: Table* match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const;
676:
1.87 parser 677: enum Change_case_kind {
678: CC_UPPER,
679: CC_LOWER
680: };
1.216 moko 681: String& change_case(Charset& source_charset, Change_case_kind kind) const;
682:
1.145 paf 683: const String& replace(const Dictionary& dict) const;
1.188 misha 684: const String& trim(Trim_kind kind=TRIM_BOTH, const char* chars=0, Charset* source_charset=0) const;
1.155 paf 685: double as_double() const { return pa_atod(cstr(), this); }
1.226 moko 686: int as_int() const { return pa_atoi(cstr(), 0, this); }
1.167 misha 687: bool as_bool() const { return as_int()!=0; }
1.174 misha 688: const String& escape(Charset& source_charset) const;
1.137 paf 689:
1.7 paf 690: private: //disabled
691:
1.12 paf 692: String& operator = (const String&) { return *this; }
1.7 paf 693:
1.1 paf 694: };
1.119 paf 695:
1.183 misha 696: #ifndef HASH_CODE_CACHING
1.145 paf 697: /// simple hash code of string. used by Hash
1.146 paf 698: inline uint hash_code(const String::Body self) {
1.184 misha 699: return self.get_hash_code();
1.119 paf 700: }
1.183 misha 701: #endif
1.147 paf 702:
703:
1.1 paf 704: #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