Annotation of parser3/src/main/pa_string.C, revision 1.272
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.269 moko 4: Copyright (c) 2001-2023 Art. Lebedev Studio (http://www.artlebedev.com)
5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
1.164 paf 6: */
1.46 paf 7:
1.12 paf 8: #include "pa_string.h"
1.22 paf 9: #include "pa_exception.h"
1.61 paf 10: #include "pa_table.h"
1.101 parser 11: #include "pa_dictionary.h"
1.132 paf 12: #include "pa_charset.h"
1.222 misha 13: #include "pa_vregex.h"
1.215 misha 14:
1.272 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.271 2023/12/13 20:07:11 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.262 moko 19: #define COMPILE_ASSERT(x) extern int assert_checker[(x) ? 1 : -1]
1.261 moko 20: COMPILE_ASSERT(sizeof(String::Languages) == sizeof(CORD));
21:
1.242 moko 22: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
23:
1.249 moko 24: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
25: T result = 0;
1.242 moko 26: const char *pos = str;
27:
28: while (isspace(*pos)) /* skip leading whitespace */
29: ++pos;
30:
31: if (base == 16 && *pos == '0') { /* handle option prefix */
32: ++pos;
33: if (*pos == 'x' || *pos == 'X') {
34: ++pos;
35: }
36: }
37:
38: if (base == 0) { /* dynamic base */
39: base = 10; /* default is 10 */
40: if (*pos == '0') {
41: ++pos;
1.245 moko 42: if (*pos == 'x' || *pos == 'X'){
1.242 moko 43: ++pos;
44: base=16;
1.245 moko 45: }
1.242 moko 46: }
47: }
48:
49: if (base < 2 || base > 16) { /* illegal base */
50: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
51: }
52:
1.249 moko 53: T cutoff = max / base;
54: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 55:
56: while(true) {
57: int digit;
58:
59: if ((*pos >= '0') && (*pos <= '9')) {
60: digit = (*pos - '0');
61: } else if (*pos >= 'a') {
62: digit = (*pos - 'a' + 10);
63: } else if (*pos >= 'A') {
64: digit = (*pos - 'A' + 10);
65: } else break;
66:
67: if (digit >= base) {
68: break;
69: }
70:
71: ++pos;
72:
73: /* adjust number, with overflow check */
74: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
75: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
76: } else {
77: result = result * base + digit;
78: }
79: }
80:
81: while(char c=*pos++)
82: if(!isspace(c))
83: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
84:
85: return result;
86: }
87:
1.249 moko 88: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.263 moko 89: if(!str)
90: return 0;
91:
1.254 moko 92: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 93: }
94:
1.265 moko 95: uint64_t pa_atoul(const char *str, int base, const String* problem_source){
1.263 moko 96: if(!str)
97: return 0;
98:
1.265 moko 99: return pa_ato_any<uint64_t>(str, base, problem_source, ULLONG_MAX);
1.249 moko 100: }
101:
1.264 moko 102: int pa_atoi(const char* str, int base, const String* problem_source) {
1.193 paf 103: if(!str)
104: return 0;
105:
1.242 moko 106: while(isspace(*str))
1.193 paf 107: str++;
1.242 moko 108:
1.193 paf 109: if(!*str)
110: return 0;
111:
1.270 moko 112: const char *str_copy=str;
1.200 paf 113: bool negative=false;
114: if(str[0]=='-') {
115: negative=true;
116: str++;
1.270 moko 117: if(!*str || isspace(*str))
118: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str_copy);
1.200 paf 119: } else if(str[0]=='+') {
120: str++;
1.270 moko 121: if(!*str || isspace(*str))
122: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str_copy);
1.200 paf 123: }
1.193 paf 124:
1.264 moko 125: unsigned int result=pa_atoui(str, base, problem_source);
1.193 paf 126:
1.242 moko 127: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 128: return -(int)result;
1.242 moko 129:
130: if(result<=INT_MAX)
1.243 moko 131: return (int)result;
1.270 moko 132:
133: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str_copy);
1.193 paf 134: }
135:
1.270 moko 136: double pa_atod(const char* str, const String* problem_source /* never null */) {
1.193 paf 137: if(!str)
138: return 0;
139:
1.242 moko 140: while(isspace(*str))
1.193 paf 141: str++;
1.242 moko 142:
1.193 paf 143: if(!*str)
144: return 0;
145:
1.200 paf 146: bool negative=false;
147: if(str[0]=='-') {
148: negative=true;
149: str++;
1.270 moko 150: if(!*str || isspace(*str))
151: throw Exception("number.format", problem_source, "invalid number (double)");
1.200 paf 152: } else if(str[0]=='+') {
153: str++;
1.270 moko 154: if(!*str || isspace(*str))
155: throw Exception("number.format", problem_source, "invalid number (double)");
1.200 paf 156: }
1.242 moko 157:
158: double result;
1.247 moko 159: if(str[0]=='0') {
160: if(str[1]=='x' || str[1]=='X') {
1.242 moko 161: // 0xABC
1.249 moko 162: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 163: return negative ? -result : result;
164: } else {
1.200 paf 165: // skip leading 0000, to disable octal interpretation
1.242 moko 166: do str++; while(*str=='0');
1.200 paf 167: }
1.247 moko 168: }
1.242 moko 169:
170: char *error_pos;
171: result=strtod(str, &error_pos);
1.193 paf 172:
1.270 moko 173: while(const char c=*error_pos++)
174: if(!isspace(c))
175: throw Exception("number.format", problem_source, "invalid number (double)");
1.193 paf 176:
1.242 moko 177: return negative ? -result : result;
1.193 paf 178: }
179:
1.176 paf 180: // cord lib extension
181:
182: #ifndef DOXYGEN
183: typedef struct {
1.254 moko 184: ssize_t countdown;
185: int target; /* Character we're looking for */
1.176 paf 186: } chr_data;
187: #endif
1.248 moko 188:
1.176 paf 189: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
190: {
1.272 ! moko 191: chr_data * d = (chr_data *)client_data;
1.254 moko 192:
193: if (d -> countdown<=0) return(2);
194: d -> countdown -= size;
195: if (c > d -> target) return(1);
196: return(0);
1.176 paf 197: }
1.248 moko 198:
1.176 paf 199: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
200: {
1.254 moko 201: chr_data d;
1.176 paf 202:
1.254 moko 203: d.countdown = n;
204: d.target = c;
205: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
1.176 paf 206: }
207:
1.187 paf 208: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 209: {
1.254 moko 210: int* result=(int*)client_data;
211: (*result)++;
212: return(0); // 0=continue
1.178 paf 213: }
1.248 moko 214:
1.178 paf 215: size_t CORD_block_count(CORD x)
216: {
217: size_t result=0;
218: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 219: return result;
1.178 paf 220: }
221:
1.174 paf 222: // helpers
1.139 paf 223:
1.174 paf 224: /// String::match uses this as replace & global search table columns
1.139 paf 225:
1.174 paf 226: const int MAX_MATCH_GROUPS=100;
1.139 paf 227:
1.174 paf 228: class String_match_table_template_columns: public ArrayString {
229: public:
230: String_match_table_template_columns() {
231: *this+=new String("prematch");
232: *this+=new String("match");
233: *this+=new String("postmatch");
234: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 235: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 236: }
237: }
238: };
239:
240: Table string_match_table_template(new String_match_table_template_columns);
241:
1.176 paf 242: // String::Body methods
1.140 paf 243:
1.176 paf 244: String::Body String::Body::Format(int value) {
1.174 paf 245: char local[MAX_NUMBER];
246: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 247: return String::Body(pa_strdup(local, length));
1.120 paf 248: }
249:
1.254 moko 250: String::Body String::Body::trim(String::Trim_kind kind, const char* chars, size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 251: size_t our_length=length();
252: if(!our_length)
253: return *this;
1.229 misha 254:
255: // check if any UTF-8 in chars
256: bool fast=true;
257: if(chars && source_charset && source_charset->isUTF8()){
258: const char* pos=chars;
259: while(unsigned char c=*pos++)
260: if(c>127){
261: fast=false;
262: break;
263: }
264: }
265:
266: size_t start=0;
267: size_t end=our_length;
1.195 paf 268: if(!chars)
269: chars=" \t\n"; // white space
270:
1.229 misha 271: if(fast){
272: // from left...
273: if(kind!=TRIM_END) {
274: CORD_pos pos; set_pos(pos, 0);
275: while(true) {
276: char c=CORD_pos_fetch(pos);
277: if(strchr(chars, c)) {
278: if(++start==our_length)
279: return 0; // all chars are empty, just return empty string
280: } else
281: break;
282:
283: CORD_next(pos);
284: }
285: }
286:
287: // from right..
288: if(kind!=TRIM_START) {
289: CORD_pos pos; set_pos(pos, end-1);
290: while(true) {
291: char c=CORD_pos_fetch(pos);
292: if(strchr(chars, c)) {
293: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
294: return 0; // all chars are empty, just return empty string
295: } else
296: break;
297:
298: CORD_prev(pos);
299: }
300: }
301: } else {
1.234 misha 302: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 303: const XMLByte* src_end=src_begin+our_length;
304:
305: // from left...
306: if(kind!=TRIM_END) {
307: while(src_begin<src_end){
308: uint char_length=1;
309: const XMLByte* ptr=src_begin;
1.231 misha 310: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
311: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 312: char_length++;
313:
314: bool found=false;
315: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
316: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
317: found=true;
318: break;
319: }
320:
321: if(found){
322: start+=char_length;
323: if(start==our_length)
324: return 0; // all chars are empty, just return empty string
325: } else
326: break;
327: }
328: }
1.196 paf 329:
1.229 misha 330: // from right..
331: if(kind!=TRIM_START) {
332: while(src_begin<src_end){
333: uint char_length=1;
1.231 misha 334: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
335: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 336: char_length++;
337:
338: bool found=false;
339: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
340: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
341: found=true;
342: break;
343: }
344:
345: if(found){
346: end-=char_length;
347: if(end==0)
348: return 0; // all chars are empty, just return empty string
349: } else
350: break;
351: }
1.196 paf 352: }
353: }
1.195 paf 354:
355: if(start==0 && end==our_length) // nobody moved a thing
356: return *this;
357:
358: if(out_start)
359: *out_start=start;
360: size_t new_length=end-start;
361: if(out_length)
362: *out_length=new_length;
363:
364: return mid(start, new_length);
365: }
366:
1.174 paf 367: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
368: uint& result=*static_cast<uint*>(client_data);
369: generic_hash_code(result, c);
370: return 0;
371: }
1.248 moko 372:
1.174 paf 373: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
374: uint& result=*static_cast<uint*>(client_data);
375: generic_hash_code(result, s);
376: return 0;
1.248 moko 377: }
378:
1.225 misha 379: uint String::Body::get_hash_code() const {
380: #ifdef HASH_CODE_CACHING
381: if(hash_code)
382: return hash_code;
383: #else
384: uint hash_code=0;
385: #endif
1.220 misha 386: if (body && CORD_IS_STRING(body)){
1.250 moko 387: generic_hash_code(hash_code, (const char *)body);
1.220 misha 388: } else {
389: CORD_iter5(body, 0,
390: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 391: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 392: }
1.225 misha 393: return hash_code;
1.94 parser 394: }
395:
1.241 misha 396: struct CORD_pos_info {
397: const char* chars;
398: size_t left;
399: size_t pos;
400: };
401:
402: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
403: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
404: if(info->pos < info->left){
405: info->pos=STRING_NOT_FOUND;
406: return 1;
407: }
408: if(strchr(info->chars, c))
409: return 1;
410: --(info->pos);
411: return 0;
412: }
413:
414: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
415: if(is_empty() || !chars || !strlen(chars))
416: return STRING_NOT_FOUND;
417: CORD_pos_info info={chars, left, right};
418: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
419: return info.pos;
420: else
421: return STRING_NOT_FOUND;
422: }
423:
424:
425: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
426: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
427: if(info->pos < info->left) {
428: info->pos=STRING_NOT_FOUND;
429: return 1;
430: }
431: if(!strchr(info->chars, c))
432: return 1;
433: --(info->pos);
434: return 0;
435: }
436:
437: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
438: if(is_empty() || !chars || !strlen(chars))
439: return STRING_NOT_FOUND;
440: CORD_pos_info info={chars, left, right};
441: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
442: return info.pos;
443: else
444: return STRING_NOT_FOUND;
445: }
446:
1.174 paf 447: // String methods
448:
449: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
450: if(!known_length)
1.9 paf 451: return *this;
1.122 paf 452:
1.176 paf 453: // first: langs
454: langs.append(body, lang, known_length);
455: // next: letters themselves
1.174 paf 456: body.append_know_length(str, known_length);
1.1 paf 457:
1.174 paf 458: ASSERT_STRING_INVARIANT(*this);
1.1 paf 459: return *this;
460: }
1.248 moko 461:
1.174 paf 462: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
463: if(!str)
464: return *this;
465: size_t known_length=helper_length?helper_length:strlen(str);
466: if(!known_length)
467: return *this;
1.1 paf 468:
1.174 paf 469: return append_know_length(str, known_length, lang);
1.5 paf 470: }
1.248 moko 471:
1.244 moko 472: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 473: char buf[MAX_NUMBER];
474: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
475: }
1.248 moko 476:
1.174 paf 477: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
478: size_t known_length=helper_length?helper_length:strlen(str);
479: if(!known_length)
480: return *this;
1.5 paf 481:
1.176 paf 482: // first: langs
483: langs.append(body, lang, known_length);
484: // next: letters themselves
1.174 paf 485: body.append_strdup_know_length(str, known_length);
1.33 paf 486:
1.174 paf 487: ASSERT_STRING_INVARIANT(*this);
488: return *this;
1.5 paf 489: }
1.46 paf 490:
1.234 misha 491: struct CORD_length_info {
492: size_t len;
493: size_t skip;
494: };
495:
496: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 497: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
498: return 0;
1.221 misha 499: }
500:
1.234 misha 501: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
502: int CORD_batched_len(const char c, CORD_length_info* info){
503: if (info->skip==0){
504: info->len++;
505: info->skip = lengthUTF8Char(c)-1;
506: } else {
507: info->skip--;
508: }
1.220 misha 509: return 0;
510: }
511:
1.210 misha 512: size_t String::length(Charset& charset) const {
513: if(charset.isUTF8()){
1.234 misha 514: CORD_length_info info = {0, 0};
515: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
516: return info.len;
1.210 misha 517: } else
518: return body.length();
519: }
520:
1.174 paf 521: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
522: String& String::mid(size_t substr_begin, size_t substr_end) const {
523: String& result=*new String;
524:
525: size_t self_length=length();
526: substr_begin=min(substr_begin, self_length);
527: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 528: size_t substr_length=substr_end-substr_begin;
529: if(!substr_length)
1.107 parser 530: return result;
1.53 paf 531:
1.176 paf 532: // first: their langs
533: result.langs.append(result.body, langs, substr_begin, substr_length);
534: // next: letters themselves
535: result.body=body.mid(substr_begin, substr_length);
1.174 paf 536:
537: ASSERT_STRING_INVARIANT(result);
1.53 paf 538: return result;
1.54 paf 539: }
540:
1.211 misha 541: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
542: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 543: String& result=*new String;
544:
1.255 moko 545: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 546:
547: if(!self_length)
548: return result;
549:
550: from=min(min(to, from), self_length);
1.210 misha 551: to=min(max(to, from), self_length);
1.211 misha 552:
1.210 misha 553: size_t substr_length=to-from;
1.211 misha 554:
1.210 misha 555: if(!substr_length)
556: return result;
557:
558: if(charset.isUTF8()){
1.234 misha 559: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 560: const XMLByte* src_end=src_begin+body.length();
1.210 misha 561:
1.212 misha 562: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 563: from=getUTF8BytePos(src_begin, src_end, from);
564: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 565: if(!substr_length)
566: return result;
567: }
568:
569: // first: their langs
570: result.langs.append(result.body, langs, from, substr_length);
571: // next: letters themselves
572: result.body=body.mid(from, substr_length);
573:
574: ASSERT_STRING_INVARIANT(result);
575: return result;
576: }
577:
1.176 paf 578: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 579: size_t substr_length=substr.length();
580: while(true) {
581: size_t substr_begin=body.pos(substr, this_offset);
582:
583: if(substr_begin==CORD_NOT_FOUND)
584: return STRING_NOT_FOUND;
1.174 paf 585:
1.183 paf 586: if(langs.check_lang(lang, substr_begin, substr_length))
587: return substr_begin;
588:
589: this_offset=substr_begin+substr_length;
590: }
1.58 paf 591: }
592:
1.254 moko 593: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 594: return pos(substr.body, this_offset, lang);
1.60 paf 595: }
596:
1.254 moko 597: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 598:
599: if(charset.isUTF8()){
1.234 misha 600: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 601: const XMLByte* srcEnd=srcPtr+body.length();
602:
603: // convert 'this_offset' from 'characters' to 'bytes'
604: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
605:
1.229 misha 606: size_t result=pos(substr.body, this_offset, lang);
607: return (result==CORD_NOT_FOUND)
608: ? STRING_NOT_FOUND
609: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 610: } else {
1.229 misha 611: size_t result=pos(substr.body, this_offset, lang);
612: return (result==CORD_NOT_FOUND)
613: ? STRING_NOT_FOUND
614: : result;
1.210 misha 615: }
616: }
617:
1.256 moko 618: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 619: if(is_empty())
620: return;
1.174 paf 621: size_t self_length=length();
1.237 misha 622: if(size_t delim_length=strlen(delim)) {
1.186 paf 623: size_t pos_before;
1.60 paf 624: // while we have 'delim'...
1.256 moko 625: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 626: result+=&mid(pos_after, pos_before);
1.174 paf 627: pos_after=pos_before+delim_length;
1.60 paf 628: }
629: // last piece
1.256 moko 630: if(pos_after<self_length)
1.174 paf 631: result+=&mid(pos_after, self_length);
1.60 paf 632: } else { // empty delim
633: result+=this;
634: }
635: }
636:
1.256 moko 637: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 638: if(is_empty())
639: return;
640: if(!delim.is_empty()) {
1.186 paf 641: size_t pos_before;
1.60 paf 642: // while we have 'delim'...
1.256 moko 643: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 644: result+=&mid(pos_after, pos_before);
1.174 paf 645: pos_after=pos_before+delim.length();
1.60 paf 646: }
647: // last piece
1.256 moko 648: if(pos_after<length())
1.174 paf 649: result+=&mid(pos_after, length());
1.60 paf 650: } else { // empty delim
651: result+=this;
652: }
1.61 paf 653: }
654:
1.254 moko 655: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 656:
1.222 misha 657: // vregex->info(); // I have no idea what does it for?
1.63 paf 658:
1.222 misha 659: bool need_pre_post_match=vregex->is_pre_post_match_needed();
660: bool global=vregex->is_global_search();
1.63 paf 661:
1.174 paf 662: const char* subject=cstr();
1.234 misha 663: size_t subject_length=length();
1.271 moko 664: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */
1.222 misha 665: int ovector[ovector_size];
1.155 paf 666:
1.173 paf 667: Table::Action_options table_options;
1.174 paf 668: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 669:
1.154 paf 670: int prestart=0;
671: int poststart=0;
1.174 paf 672: int postfinish=length();
1.267 moko 673: int action_was_executed=-1;
1.63 paf 674: while(true) {
1.222 misha 675: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 676:
1.222 misha 677: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
678: break;
1.63 paf 679:
1.154 paf 680: int prefinish=ovector[0];
681: poststart=ovector[1];
1.236 moko 682:
1.267 moko 683: if (prestart==poststart && action_was_executed==1){
1.236 moko 684: prestart++;
1.267 moko 685: action_was_executed=0;
1.236 moko 686: continue;
687: }
688:
1.232 misha 689: ArrayString* row=new ArrayString(3);
1.174 paf 690: if(need_pre_post_match) {
691: *row+=&mid(0, prefinish); // .prematch column value
692: *row+=&mid(prefinish, poststart); // .match
693: *row+=&mid(poststart, postfinish); // .postmatch
694: } else {
1.185 paf 695: *row+=&Empty; // .prematch column value
696: *row+=&Empty; // .match
697: *row+=&Empty; // .postmatch
1.174 paf 698: }
1.63 paf 699:
1.222 misha 700: for(int i=1; i<exec_result; i++) {
1.69 paf 701: // -1:-1 case handled peacefully by mid() itself
1.232 misha 702: *row+=(ovector[i*2+0]>=0 && ovector[i*2+1]>0)?&mid(ovector[i*2+0], ovector[i*2+1]):new String; // .i column value
1.63 paf 703: }
704:
1.209 misha 705: matches_count++;
1.267 moko 706: row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info);
1.63 paf 707:
1.268 moko 708: if(!global || poststart>=subject_length) // last step, avoid prestart++ after last char
1.222 misha 709: break;
710:
1.154 paf 711: prestart=poststart;
1.267 moko 712: action_was_executed=1;
1.222 misha 713: }
1.63 paf 714:
1.222 misha 715: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
716: return vregex->is_just_count() ? 0 : &table;
1.82 parser 717: }
718:
1.174 paf 719: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
720: String& result=*new String();
721: if(is_empty())
722: return result;
723:
724: char* new_cstr=cstrm();
1.216 misha 725:
1.181 paf 726: if(source_charset.isUTF8()) {
1.216 misha 727: size_t new_cstr_len=length();
1.181 paf 728: switch(kind) {
729: case CC_UPPER:
1.192 paf 730: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 731: break;
732: case CC_LOWER:
1.192 paf 733: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 734: break;
735: default:
736: assert(!"unknown change case kind");
737: break; // never
738: }
739:
740: } else {
741: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 742:
1.181 paf 743: const unsigned char *a;
744: const unsigned char *b;
745: switch(kind) {
746: case CC_UPPER:
747: a=tables+lcc_offset;
748: b=tables+fcc_offset;
749: break;
750: case CC_LOWER:
751: a=tables+lcc_offset;
752: b=0;
753: break;
754: default:
755: assert(!"unknown change case kind");
756: a=b=0; // calm, compiler
757: break; // never
758: }
759:
1.192 paf 760: char *dest=new_cstr;
1.181 paf 761: unsigned char index;
1.190 paf 762: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 763: unsigned char c=a[index];
764: if(b)
765: c=b[c];
766:
767: *dest++=(char)c;
768: }
1.174 paf 769: }
1.176 paf 770: result.langs=langs;
1.174 paf 771: result.body=new_cstr;
1.89 parser 772:
1.101 parser 773: return result;
774: }
775:
1.213 misha 776: const String& String::escape(Charset& source_charset) const {
777: if(is_empty())
778: return *this;
779:
780: return Charset::escape(*this, source_charset);
781: }
782:
1.238 misha 783: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
784: result.langs.append(result.body, langs, langs_offset, length); \
785: result.body.append_strdup_know_length(from_cstr, length);
786:
1.174 paf 787: const String& String::replace(const Dictionary& dict) const {
1.238 misha 788: if(!dict.count() || is_empty())
789: return *this;
790:
1.174 paf 791: String& result=*new String();
792: const char* old_cstr=cstr();
793: const char* prematch_begin=old_cstr;
794:
1.238 misha 795: if(dict.count()==1) {
796: // optimized simple case
797:
798: Dictionary::Subst subst=dict.get(0);
1.239 moko 799: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 800: // prematch
1.238 misha 801: if(size_t prematch_length=p-prematch_begin) {
802: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 803: }
804:
1.174 paf 805: // match
1.238 misha 806: prematch_begin=p+subst.from_length;
1.174 paf 807:
1.184 paf 808: if(const String* b=subst.to) // are there any b?
1.174 paf 809: result<<*b;
1.238 misha 810: }
811:
812: } else {
813:
814: const char* current=old_cstr;
815: while(*current) {
816: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
817: // prematch
818: if(size_t prematch_length=current-prematch_begin) {
819: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
820: }
821:
822: // match
823: // skip 'a' in 'current'; move prematch_begin
824: current+=subst.from_length; prematch_begin=current;
825:
826: if(const String* b=subst.to) // are there any b?
827: result<<*b;
828: } else // simply advance
829: current++;
830: }
831:
1.174 paf 832: }
1.156 paf 833:
1.238 misha 834: if(prematch_begin==old_cstr) // not modified
835: return *this;
836:
1.174 paf 837: // postmatch
1.238 misha 838: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
839: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 840: }
1.156 paf 841:
1.174 paf 842: ASSERT_STRING_INVARIANT(result);
1.82 parser 843: return result;
1.61 paf 844: }
1.113 parser 845:
1.180 paf 846: static int serialize_body_char(char c, char** cur) {
847: *((*cur)++)=c;
848: return 0; // 0=continue
1.248 moko 849: }
850:
1.174 paf 851: static int serialize_body_piece(const char* s, char** cur) {
852: size_t length=strlen(s);
853: memcpy(*cur, s, length); *cur+=length;
1.178 paf 854: return 0; // 0=continue
1.248 moko 855: }
856:
1.178 paf 857: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
858: // lang
1.191 paf 859: **cur=alang; (*cur)++;
860: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 861: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
862:
863: return 0; // 0=continue
864: }
1.248 moko 865:
1.174 paf 866: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 867: size_t fragments_count=langs.count();
1.202 paf 868: size_t body_length=body.length();
1.174 paf 869: size_t buf_length=
1.178 paf 870: prolog_length //1
871: +sizeof(size_t) //2
1.202 paf 872: +body_length //3
873: +1 // 4 for zero terminator used in deserialize
874: +sizeof(size_t) //5
875: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
876:
1.174 paf 877: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
878:
879: // 1: prolog
880: char *cur=result.str+prolog_length;
1.202 paf 881: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
882: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
883: // 3: letters
884: body.for_each(serialize_body_char, serialize_body_piece, &cur);
885: // 4: zero terminator
886: *cur++=0;
887: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 888: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 889: // 6: lang info
1.178 paf 890: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 891:
1.174 paf 892: return result;
1.113 parser 893: }
1.248 moko 894:
1.202 paf 895: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
896: size_t in_buf=buf_size;
897: if(in_buf<=prolog_size)
1.148 paf 898: return false;
1.202 paf 899: in_buf-=prolog_size;
1.135 paf 900:
1.174 paf 901: // 1: prolog
1.202 paf 902: const char* cur=(const char* )buf+prolog_size;
1.113 parser 903:
1.207 paf 904: // 2: chars.count
1.202 paf 905: size_t body_length;
906: if(in_buf<sizeof(body_length)) // body.length don't fit?
907: return false;
908: // [WARNING: not cast, addresses must be %4=0 on sparc]
909: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
910: in_buf-=sizeof(body_length);
911:
912: if(in_buf<body_length+1) // letters+terminator don't fit?
913: return false;
914: // 4: zero terminator
915: if(cur[body_length] != 0) // in place?
916: return false;
917: // 3: letters
1.251 moko 918: body=String::Body(String::C(cur, body_length));
1.202 paf 919: cur+=body_length+1;
920: in_buf-=body_length+1;
921:
922: // 5: langs.count
1.191 paf 923: size_t fragments_count;
1.202 paf 924: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 925: return false;
1.191 paf 926: // [WARNING: not cast, addresses must be %4=0 on sparc]
927: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 928: in_buf-=sizeof(fragments_count);
1.174 paf 929:
930: if(fragments_count) {
1.202 paf 931: // 6: lang info
1.174 paf 932: size_t total_length=0;
933: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 934: char lang;
935: size_t fragment_length;
936: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 937: if(in_buf<piece_length) // lang+length
1.174 paf 938: return false;
939:
1.191 paf 940: // lang
941: lang=*cur++;
942: // length [WARNING: not cast, addresses must be %4=0 on sparc]
943: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
944:
1.206 paf 945: size_t combined_length=total_length+fragment_length;
946: if(combined_length>body_length)
947: return false; // file curruption
1.191 paf 948: // uchar needed to prevent propagating 0x80 bit to upper bytes
949: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 950: total_length=combined_length;
1.202 paf 951: in_buf-=piece_length;
1.174 paf 952: }
1.128 paf 953:
1.202 paf 954: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 955: return false;
1.174 paf 956: }
1.202 paf 957: if(in_buf!=0) // some strange extra bytes
958: return false;
1.113 parser 959:
1.174 paf 960: ASSERT_STRING_INVARIANT(*this);
1.148 paf 961: return true;
1.176 paf 962: }
963:
1.201 paf 964: void String::Body::dump() const {
965: CORD_dump(body);
966: }
967:
1.257 moko 968: const char* String::Languages::visualize() const {
1.177 paf 969: if(opt.is_not_just_lang)
1.233 misha 970: return CORD_to_const_char_star(langs, 0);
1.176 paf 971: else
1.257 moko 972: return 0;
1.176 paf 973: }
1.248 moko 974:
1.201 paf 975: void String::Languages::dump() const {
976: if(opt.is_not_just_lang)
977: CORD_dump(langs);
978: else
979: puts((const char*)&langs);
980: }
1.248 moko 981:
1.257 moko 982: void String::dump() const {
983: body.dump();
984: langs.dump();
985: }
1.176 paf 986:
1.257 moko 987: static char *n_chars(char c, size_t length){
988: char *result=(char *)pa_malloc_atomic(length+1);
989: memset(result, c, length);
990: result[length] = '\0';
991: return result;
1.113 parser 992: }
1.229 misha 993:
1.257 moko 994: char* String::visualize_langs() const {
1.258 moko 995: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 996: }
1.229 misha 997:
998: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 999: if(is_empty())
1.195 paf 1000: return *this;
1001:
1002: size_t substr_begin, substr_length;
1.229 misha 1003: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1004: if(new_body==body) // we received unchanged pointer, do likewise
1005: return *this;
1006: // new_body differs from body, adjust langs along
1007:
1008: String& result=*new String;
1009: if(!new_body) // body.trim produced empty result
1010: return result;
1011: // body.trim produced nonempty result
1012:
1013: // first: their langs
1014: result.langs.append(result.body, langs, substr_begin, substr_length);
1015: // next: letters themselves
1016: result.body=new_body;
1017:
1018: ASSERT_STRING_INVARIANT(result);
1019: return result;
1.198 paf 1020: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_string.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main