Annotation of parser3/src/include/pa_string.h, revision 1.156
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:
1.156 ! paf 11: static const char * const IDENT_STRING_H="$Date: 2004/01/30 09:56:49 $";
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:
1.147 paf 102: union Languages {
1.146 paf 103:
1.147 paf 104: struct {
105: Language lang:8;
106: int is_not_just_lang:sizeof(CORD)*8-8;
107: } opt;
108: CORD langs;
1.146 paf 109:
110: template<typename C>
111: CORD make_langs(C current) const {
1.147 paf 112: return opt.is_not_just_lang?
1.146 paf 113: langs
1.147 paf 114: :CORD_chars((char)opt.lang, get_length(current));
1.146 paf 115: }
116:
117: CORD make_langs(size_t aoffset, size_t alength) const {
1.147 paf 118: return opt.is_not_just_lang?
1.146 paf 119: CORD_substr(langs, aoffset, alength)
1.147 paf 120: :CORD_chars((char)opt.lang, alength);
1.145 paf 121: }
122:
1.146 paf 123: /// appending when 'langs' already contain something [simple cases hanled elsewhere]
124: template<typename C>
125: void append(C current,
126: const CORD to_nonempty_target_langs) {
127: assert(langs);
128:
1.147 paf 129: if(opt.is_not_just_lang)
1.146 paf 130: langs=CORD_cat(langs, to_nonempty_target_langs);
131: else { // we were "just lang"
132: size_t current_size=get_length(current);
133: assert(current_size);
134: langs=CORD_cat(
1.147 paf 135: CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang']
1.146 paf 136: to_nonempty_target_langs); // new piece
137: }
1.145 paf 138: }
1.146 paf 139:
1.145 paf 140: public:
1.146 paf 141:
142: const char* v() const;
143:
144: Languages(): langs(0) {}
1.147 paf 145: Languages(Language alang) {
146: opt.lang=alang;
147: opt.is_not_just_lang=0;
148: }
1.146 paf 149:
150: /// MUST be called exactly prior to modification of current [uses it's length]
151: template<typename C>
152: void append(C current, Language alang, size_t asize) {
153: assert(alang);
154: assert(asize);
155:
1.147 paf 156: if(!opt.is_not_just_lang)
157: if(opt.lang) {
158: if(opt.lang==alang) // same length? ignoring
1.146 paf 159: return;
160: } else {
1.147 paf 161: opt.lang=alang; // to uninitialized
1.146 paf 162: return;
1.145 paf 163: }
1.146 paf 164:
165: append(current, CORD_chars((char)alang, asize));
166: }
167:
168: /// MUST be called exactly prior to modification of current [uses it's length]
169: template<typename C>
170: void append(C current, size_t appending_length,
171: const Languages src) {
172: assert(appending_length);
173:
174: if(!langs)
175: langs=src.langs; // to uninitialized
1.147 paf 176: else if(!src.opt.is_not_just_lang)
177: append(current, src.opt.lang, appending_length); // simplifying when simple source
1.146 paf 178: else
179: append(current, src.make_langs(appending_length));
180: }
181:
182: /// MUST be called exactly prior to modification of current [uses it's length]
183: template<typename C>
184: void append(C current,
185: const Languages src, size_t aoffset, size_t alength) {
186: assert(alength);
187:
188: if(!langs) // to uninitialized?
1.147 paf 189: if(src.opt.is_not_just_lang)
1.146 paf 190: langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex
191: else
1.147 paf 192: opt.lang=src.opt.lang; // to uninitialized simple
1.146 paf 193: else
1.147 paf 194: 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 195: return; // ignoring
196: else
197: append(current, src.make_langs(aoffset, alength));
198: }
199:
200: /// checks if we have lang<=alang all from aoffset to aoffset+alength
201: bool check_lang(Language alang, size_t aoffset, size_t alength) const {
202: if(alang==L_UNSPECIFIED) // ignore lang?
203: return true;
204:
1.147 paf 205: if(opt.is_not_just_lang)
1.146 paf 206: return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0;
207: else
1.147 paf 208: return opt.lang<=alang;
1.146 paf 209: }
210:
1.148 paf 211: /// @returns count of blocks
212: /// @todo currently there can be adjucent blocks of same language. someday merge them
213: size_t count() const {
214: return opt.is_not_just_lang?
215: CORD_block_count(langs)
216: : opt.lang?
217: 1
218: : 0;
219: };
220:
1.146 paf 221: template<typename C, typename I>
222: void for_each(C current,
223: int callback(char, size_t, I), I info) const {
224:
1.147 paf 225: if(opt.is_not_just_lang)
1.146 paf 226: CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info);
227: else
1.147 paf 228: callback(opt.lang, get_length(current), info);
1.146 paf 229: }
230:
231: bool invariant(size_t current_length) {
232: if(!langs)
233: return current_length==0;
1.147 paf 234: if(opt.is_not_just_lang)
1.146 paf 235: return CORD_len(langs)==current_length;
236: return true; // uncheckable, actually
237: }
238: };
239:
240: class Body {
241:
242: CORD body;
243:
244: public:
245:
246: const char* v() const;
247:
248: Body(): body(CORD_EMPTY) {}
249: Body(CORD abody): body(abody) {
250: assert(!body // no body
251: || *body // ordinary string
252: || body[1]==1 // CONCAT_HDR
253: || body[1]==4 // FN_HDR
254: || body[1]==6 // SUBSTR_HDR
255: );
256: }
257: /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated
258: Body(const char* str, size_t helper_length): body(CORD_EMPTY) {
259: append_know_length(str, helper_length?helper_length:strlen(str));
260: }
261: static Body Format(int value);
262:
263: void clear() { body=CORD_EMPTY; }
264:
265: bool operator! () const { return is_empty(); }
266:
267: uint hash_code() const;
268:
269: const char* cstr() const { return CORD_to_const_char_star(body); }
270: char* cstrm() const { return CORD_to_char_star(body); }
271:
272: size_t length() const { return CORD_len(body); }
273:
274: bool is_empty() const { return body==CORD_EMPTY; }
275:
276: void append_know_length(const char *str, size_t known_length) {
277: if(known_length)
278: body=CORD_cat_char_star(body, str, known_length);
279: }
280: void append_strdup_know_length(const char* str, size_t known_length) {
281: if(known_length)
282: append_know_length(pa_strdup(str, known_length), known_length);
283: }
284: void append(char c) { body=CORD_cat_char(body, c); }
285: Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; }
286: Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; }
287:
288: // could not figure out why this operator is needed [should do this chain: string->simple->==]
289: bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; }
290: bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; }
291: bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; }
292: bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; }
293: bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; }
294: bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; }
295:
296: int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const {
297: return CORD_ncmp(body, x_begin, y.body, y_begin, size);
1.145 paf 298: }
299:
1.146 paf 300: char fetch(size_t index) const { return CORD_fetch(body, index); }
301: Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); }
302: size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); }
303: size_t pos(const Body substr, size_t offset=0) const {
304: if(!substr.length())
305: return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1]
1.150 paf 306:
307: // CORD_str checks for bad offset [CORD_chr does not]
1.146 paf 308: return CORD_str(body, offset, substr.body);
309: }
310: size_t pos(char c,
311: size_t offset=0) const {
1.150 paf 312: if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case]
313: return STRING_NOT_FOUND;
314:
1.146 paf 315: return CORD_chr(body, offset, c);
1.145 paf 316: }
1.146 paf 317:
1.149 paf 318: template<typename I> void for_each(
319: int (*f1)(char c, I),
320: int (*f2)(const char* s, I),
321: I info) const {
322: CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info);
1.148 paf 323: }
1.146 paf 324:
325: void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); }
326:
327: /*Body normalize() const {
328: return Body(CORD_balance(body));
329: }*/
1.145 paf 330: };
331:
332: struct C {
333: const char *str;
334: size_t length;
335: operator const char *() { return str; }
336: C(const char *astr, size_t asize): str(astr), length(asize) {}
337: };
338:
339: struct Cm {
340: char *str;
341: size_t length;
342: //operator char *() { return str; }
343: Cm(char *astr, size_t asize): str(astr), length(asize) {}
344: };
345:
346: private:
347:
1.152 paf 348: Body body; ///< all characters of string
1.146 paf 349: Languages langs; ///< string characters lang
350:
351: const char* v() const;
352:
353: #define ASSERT_STRING_INVARIANT(string) \
354: assert((string).langs.invariant((string).body.length()))
1.145 paf 355:
1.8 paf 356: public:
1.151 paf 357:
358: static const String Empty;
1.8 paf 359:
1.145 paf 360: explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false);
361: explicit String(const C cstr, bool tainted=false);
1.146 paf 362: String(Body abody, Language alang): body(abody), langs(alang) {
363: ASSERT_STRING_INVARIANT(*this);
364: }
365: String(const String& src): body(src.body), langs(src.langs) {
366: ASSERT_STRING_INVARIANT(*this);
1.145 paf 367: }
368:
369: /// for convinient hash lookup
1.146 paf 370: operator const Body() const { return body; }
1.145 paf 371:
372: bool is_empty() const { return body.is_empty(); }
373: size_t length() const { return body.length(); }
374:
375: /// convert to CORD. if 'lang' known, forcing 'lang' to it
1.146 paf 376: Body cstr_to_string_body(Language lang=L_AS_IS,
1.145 paf 377: SQL_Connection* connection=0,
378: const Request_charsets *charsets=0) const;
379:
380: /// convert to constant C string. if 'lang' known, forcing 'lang' to it
381: const char* cstr(Language lang=L_AS_IS,
382: SQL_Connection* connection=0,
383: const Request_charsets *charsets=0) const {
384: return cstr_to_string_body(lang, connection, charsets).cstr();
385: }
386: /// convert to Modifiable C string. if 'lang' known, forcing 'lang' to it
387: char *cstrm(Language lang=L_AS_IS,
388: SQL_Connection* connection=0,
389: const Request_charsets *charsets=0) const {
390: return cstr_to_string_body(lang, connection, charsets).cstrm();
1.50 paf 391: }
1.108 parser 392: /// puts pieces to buf
1.145 paf 393: Cm serialize(size_t prolog_size) const;
1.108 parser 394: /// appends pieces from buf to self
1.145 paf 395: bool deserialize(size_t prolog_size, void *buf, size_t buf_size);
1.146 paf 396: /// @see Body::append_know_length
1.145 paf 397: String& append_know_length(const char* str, size_t known_length, Language lang);
1.146 paf 398: /// @see Body::append_help_length
1.145 paf 399: String& append_help_length(const char* str, size_t helper_length, Language lang);
400: String& append_strdup(const char* str, size_t helper_length, Language lang);
401:
1.146 paf 402: bool operator == (const char* y) const { return body==Body(y); }
403: bool operator != (const char* y) const { return body!=Body(y); }
1.145 paf 404:
405: /// this starts with y
406: bool starts_with(const char* y) const {
1.146 paf 407: return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0;
1.145 paf 408: }
409: /// x starts with this
410: bool this_starts(const char* x) const {
1.146 paf 411: return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0;
1.26 paf 412: }
413:
1.145 paf 414: String& append_to(String& dest, Language lang, bool forced) const;
415: String& append(const String& src, Language lang, bool forced=false) {
416: return src.append_to(*this, lang, forced);
417: }
418: String& operator << (const String& src) { return append(src, L_PASS_APPENDED); }
419: String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); }
1.147 paf 420: String& operator << (const Body src);
1.100 parser 421:
1.142 paf 422: /// extracts first char of a string, if any
423: char first_char() const {
1.145 paf 424: return is_empty()?0:body.fetch(0);
1.142 paf 425: }
1.54 paf 426:
1.145 paf 427: bool operator < (const String& src) const { return body<src.body; }
428: bool operator > (const String& src) const { return body>src.body; }
429: bool operator <= (const String& src) const { return body<=src.body; }
430: bool operator >= (const String& src) const { return body>=src.body; }
431: bool operator != (const String& src) const { return body!=src.body; }
432: bool operator == (const String& src) const { return body==src.body; }
433:
1.54 paf 434: /// extracts [start, finish) piece of string
1.145 paf 435: String& mid(size_t substr_begin, size_t substr_end) const;
436:
437: /**
438: ignore lang if it's L_UNSPECIFIED
439: but when specified: look for substring that lies in ONE fragment in THAT lang
440: @return position of substr in string, -1 means "not found" [const char* version]
441: */
1.146 paf 442: size_t pos(const Body substr,
1.145 paf 443: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
444: /// String version of @see pos(const char*, int, Language)
445: size_t pos(const String& substr,
446: size_t this_offset=0, Language lang=L_UNSPECIFIED) const;
447: size_t pos(char c,
448: size_t this_offset=0) const {
449: return body.pos(c, this_offset);
450: }
1.55 paf 451:
1.145 paf 452: void split(ArrayString& result,
453: size_t& pos_after,
454: const char* delim,
455: Language lang=L_UNSPECIFIED, int limit=-1) const;
456: void split(ArrayString& result,
457: size_t& pos_after,
1.62 paf 458: const String& delim,
1.145 paf 459: Language lang=L_UNSPECIFIED, int limit=-1) const;
1.62 paf 460:
1.145 paf 461: typedef void (*Row_action)(Table& table, ArrayString* row,
1.136 paf 462: int prestart, int prefinish,
463: int poststart, int postfinish,
1.68 paf 464: void *info);
1.87 parser 465: /**
1.145 paf 466: @return table of found items, if any.
1.87 parser 467: table format is defined and fixed[can be used by others]:
468: @verbatim
469: prematch/match/postmatch/1/2/3/...
470: @endverbatim
471: */
1.145 paf 472: Table* match(Charset& source_charset,
1.64 paf 473: const String& regexp,
1.145 paf 474: const String* options,
1.99 parser 475: Row_action row_action, void *info,
1.145 paf 476: bool& just_matched) const;
1.87 parser 477: enum Change_case_kind {
478: CC_UPPER,
479: CC_LOWER
480: };
1.145 paf 481: String& change_case(Charset& source_charset,
1.87 parser 482: Change_case_kind kind) const;
1.145 paf 483: const String& replace(const Dictionary& dict) const;
1.155 paf 484: double as_double() const { return pa_atod(cstr(), this); }
485: int as_int() const { return pa_atoi(cstr(), this); }
1.137 paf 486:
1.7 paf 487: private: //disabled
488:
1.12 paf 489: String& operator = (const String&) { return *this; }
1.7 paf 490:
1.1 paf 491: };
1.119 paf 492:
1.146 paf 493: template<>
494: inline size_t get_length<String::Body>(String::Body body) {
495: return body.length();
496: }
497:
1.145 paf 498: /// simple hash code of string. used by Hash
1.146 paf 499: inline uint hash_code(const String::Body self) {
1.145 paf 500: return self.hash_code();
1.119 paf 501: }
1.147 paf 502:
503:
504: /// now that we've declared specialization we can use it
505: inline String& String::operator << (const String::Body src) {
506: langs.append(body, L_AS_IS, src.length());
507: body<<src;
508: return *this;
509: }
510:
1.1 paf 511:
512: #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