Annotation of parser3/src/main/pa_string.C, revision 1.283
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.277 moko 4: Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com)
1.269 moko 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:
1.283 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.282 2025/05/25 17:44:27 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:
1.274 moko 49: if (base < 2 || base > 16) /* illegal base */
1.242 moko 50: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
1.274 moko 51: if (*pos == '-')
52: throw Exception("number.format", problem_source, problem_source ? "out of range (negative)" : "'%s' is out if range (negative)", str);
1.242 moko 53:
1.249 moko 54: T cutoff = max / base;
55: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 56:
57: while(true) {
58: int digit;
59:
60: if ((*pos >= '0') && (*pos <= '9')) {
61: digit = (*pos - '0');
62: } else if (*pos >= 'a') {
63: digit = (*pos - 'a' + 10);
64: } else if (*pos >= 'A') {
65: digit = (*pos - 'A' + 10);
66: } else break;
67:
68: if (digit >= base) {
69: break;
70: }
71:
72: ++pos;
73:
74: /* adjust number, with overflow check */
75: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
76: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
77: } else {
78: result = result * base + digit;
79: }
80: }
81:
82: while(char c=*pos++)
83: if(!isspace(c))
1.274 moko 84: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str);
1.242 moko 85:
86: return result;
87: }
88:
1.249 moko 89: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.263 moko 90: if(!str)
91: return 0;
92:
1.254 moko 93: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 94: }
95:
1.265 moko 96: uint64_t pa_atoul(const char *str, int base, const String* problem_source){
1.263 moko 97: if(!str)
98: return 0;
99:
1.265 moko 100: return pa_ato_any<uint64_t>(str, base, problem_source, ULLONG_MAX);
1.249 moko 101: }
102:
1.264 moko 103: int pa_atoi(const char* str, int base, const String* problem_source) {
1.193 paf 104: if(!str)
105: return 0;
106:
1.242 moko 107: while(isspace(*str))
1.193 paf 108: str++;
1.242 moko 109:
1.193 paf 110: if(!*str)
111: return 0;
112:
1.270 moko 113: const char *str_copy=str;
1.200 paf 114: bool negative=false;
115: if(str[0]=='-') {
116: negative=true;
117: str++;
1.270 moko 118: if(!*str || isspace(*str))
1.276 moko 119: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str_copy);
1.200 paf 120: } else if(str[0]=='+') {
121: str++;
1.270 moko 122: if(!*str || isspace(*str))
1.276 moko 123: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str_copy);
1.200 paf 124: }
1.193 paf 125:
1.264 moko 126: unsigned int result=pa_atoui(str, base, problem_source);
1.193 paf 127:
1.242 moko 128: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 129: return -(int)result;
1.242 moko 130:
131: if(result<=INT_MAX)
1.243 moko 132: return (int)result;
1.270 moko 133:
134: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str_copy);
1.193 paf 135: }
136:
1.270 moko 137: double pa_atod(const char* str, const String* problem_source /* never null */) {
1.193 paf 138: if(!str)
139: return 0;
140:
1.242 moko 141: while(isspace(*str))
1.193 paf 142: str++;
1.242 moko 143:
1.193 paf 144: if(!*str)
145: return 0;
146:
1.200 paf 147: bool negative=false;
148: if(str[0]=='-') {
149: negative=true;
150: str++;
1.270 moko 151: if(!*str || isspace(*str))
152: throw Exception("number.format", problem_source, "invalid number (double)");
1.200 paf 153: } else if(str[0]=='+') {
154: str++;
1.270 moko 155: if(!*str || isspace(*str))
156: throw Exception("number.format", problem_source, "invalid number (double)");
1.200 paf 157: }
1.242 moko 158:
159: double result;
1.247 moko 160: if(str[0]=='0') {
161: if(str[1]=='x' || str[1]=='X') {
1.242 moko 162: // 0xABC
1.249 moko 163: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 164: return negative ? -result : result;
165: } else {
1.200 paf 166: // skip leading 0000, to disable octal interpretation
1.242 moko 167: do str++; while(*str=='0');
1.200 paf 168: }
1.247 moko 169: }
1.242 moko 170:
171: char *error_pos;
172: result=strtod(str, &error_pos);
1.193 paf 173:
1.270 moko 174: while(const char c=*error_pos++)
175: if(!isspace(c))
176: throw Exception("number.format", problem_source, "invalid number (double)");
1.193 paf 177:
1.242 moko 178: return negative ? -result : result;
1.193 paf 179: }
180:
1.176 paf 181: // cord lib extension
182:
183: #ifndef DOXYGEN
184: typedef struct {
1.254 moko 185: ssize_t countdown;
186: int target; /* Character we're looking for */
1.176 paf 187: } chr_data;
188: #endif
1.248 moko 189:
1.176 paf 190: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
191: {
1.272 moko 192: chr_data * d = (chr_data *)client_data;
1.254 moko 193:
194: if (d -> countdown<=0) return(2);
195: d -> countdown -= size;
196: if (c > d -> target) return(1);
197: return(0);
1.176 paf 198: }
1.248 moko 199:
1.176 paf 200: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
201: {
1.254 moko 202: chr_data d;
1.176 paf 203:
1.254 moko 204: d.countdown = n;
205: d.target = c;
206: 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 207: }
208:
1.187 paf 209: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 210: {
1.254 moko 211: int* result=(int*)client_data;
212: (*result)++;
213: return(0); // 0=continue
1.178 paf 214: }
1.248 moko 215:
1.178 paf 216: size_t CORD_block_count(CORD x)
217: {
218: size_t result=0;
219: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 220: return result;
1.178 paf 221: }
222:
1.174 paf 223: // helpers
1.139 paf 224:
1.174 paf 225: /// String::match uses this as replace & global search table columns
1.139 paf 226:
1.174 paf 227: const int MAX_MATCH_GROUPS=100;
1.139 paf 228:
1.174 paf 229: class String_match_table_template_columns: public ArrayString {
230: public:
231: String_match_table_template_columns() {
232: *this+=new String("prematch");
233: *this+=new String("match");
234: *this+=new String("postmatch");
235: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.282 moko 236: *this+=new String(String::Body::uitoa(1+i), String::L_CLEAN);
1.174 paf 237: }
238: }
239: };
240:
1.278 moko 241: static Table &string_match_table_template(){
242: static Table *singleton=NULL;
1.280 moko 243: if(!singleton)
1.278 moko 244: singleton=new Table(new String_match_table_template_columns);
245: return *singleton;
246: }
1.174 paf 247:
1.176 paf 248: // String::Body methods
1.140 paf 249:
1.282 moko 250: String::Body String::Body::uitoa(size_t aindex) {
251: static const size_t CACHE_SIZE=256;
252: static String::Body cache[CACHE_SIZE];
253: if (aindex < CACHE_SIZE)
254: return !cache[aindex] ? (cache[aindex]=Body(pa_uitoa(aindex))) : cache[aindex];
255: else
256: return Body(pa_uitoa(aindex));
257: }
258:
1.254 moko 259: 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 260: size_t our_length=length();
261: if(!our_length)
262: return *this;
1.229 misha 263:
264: // check if any UTF-8 in chars
265: bool fast=true;
266: if(chars && source_charset && source_charset->isUTF8()){
267: const char* pos=chars;
268: while(unsigned char c=*pos++)
269: if(c>127){
270: fast=false;
271: break;
272: }
273: }
274:
275: size_t start=0;
276: size_t end=our_length;
1.195 paf 277: if(!chars)
1.281 moko 278: chars=" \t\r\n"; // white space
1.195 paf 279:
1.229 misha 280: if(fast){
281: // from left...
282: if(kind!=TRIM_END) {
283: CORD_pos pos; set_pos(pos, 0);
284: while(true) {
285: char c=CORD_pos_fetch(pos);
286: if(strchr(chars, c)) {
287: if(++start==our_length)
1.283 ! moko 288: return String::Body(); // all chars are empty, just return empty string
1.229 misha 289: } else
1.281 moko 290: break;
1.229 misha 291:
292: CORD_next(pos);
293: }
294: }
295:
296: // from right..
297: if(kind!=TRIM_START) {
298: CORD_pos pos; set_pos(pos, end-1);
299: while(true) {
300: char c=CORD_pos_fetch(pos);
301: if(strchr(chars, c)) {
302: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
1.283 ! moko 303: return String::Body(); // all chars are empty, just return empty string
1.229 misha 304: } else
1.281 moko 305: break;
1.229 misha 306:
307: CORD_prev(pos);
308: }
309: }
310: } else {
1.234 misha 311: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 312: const XMLByte* src_end=src_begin+our_length;
313:
314: // from left...
315: if(kind!=TRIM_END) {
316: while(src_begin<src_end){
317: uint char_length=1;
318: const XMLByte* ptr=src_begin;
1.231 misha 319: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
320: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 321: char_length++;
322:
323: bool found=false;
324: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
325: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
326: found=true;
327: break;
328: }
329:
330: if(found){
331: start+=char_length;
332: if(start==our_length)
1.283 ! moko 333: return String::Body(); // all chars are empty, just return empty string
1.229 misha 334: } else
335: break;
336: }
337: }
1.196 paf 338:
1.229 misha 339: // from right..
340: if(kind!=TRIM_START) {
341: while(src_begin<src_end){
342: uint char_length=1;
1.231 misha 343: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
344: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 345: char_length++;
346:
347: bool found=false;
348: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
349: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
350: found=true;
351: break;
352: }
353:
354: if(found){
355: end-=char_length;
356: if(end==0)
1.283 ! moko 357: return String::Body(); // all chars are empty, just return empty string
1.229 misha 358: } else
359: break;
360: }
1.196 paf 361: }
362: }
1.195 paf 363:
364: if(start==0 && end==our_length) // nobody moved a thing
365: return *this;
366:
367: if(out_start)
368: *out_start=start;
369: size_t new_length=end-start;
370: if(out_length)
371: *out_length=new_length;
372:
373: return mid(start, new_length);
374: }
375:
1.174 paf 376: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
377: uint& result=*static_cast<uint*>(client_data);
378: generic_hash_code(result, c);
379: return 0;
380: }
1.248 moko 381:
1.174 paf 382: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
383: uint& result=*static_cast<uint*>(client_data);
384: generic_hash_code(result, s);
385: return 0;
1.248 moko 386: }
387:
1.225 misha 388: uint String::Body::get_hash_code() const {
389: #ifdef HASH_CODE_CACHING
390: if(hash_code)
391: return hash_code;
392: #else
393: uint hash_code=0;
394: #endif
1.220 misha 395: if (body && CORD_IS_STRING(body)){
1.250 moko 396: generic_hash_code(hash_code, (const char *)body);
1.220 misha 397: } else {
398: CORD_iter5(body, 0,
399: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 400: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 401: }
1.225 misha 402: return hash_code;
1.94 parser 403: }
404:
1.241 misha 405: struct CORD_pos_info {
406: const char* chars;
407: size_t left;
408: size_t pos;
409: };
410:
411: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
412: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
413: if(info->pos < info->left){
414: info->pos=STRING_NOT_FOUND;
415: return 1;
416: }
417: if(strchr(info->chars, c))
418: return 1;
419: --(info->pos);
420: return 0;
421: }
422:
423: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
424: if(is_empty() || !chars || !strlen(chars))
425: return STRING_NOT_FOUND;
426: CORD_pos_info info={chars, left, right};
427: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
428: return info.pos;
429: else
430: return STRING_NOT_FOUND;
431: }
432:
433:
434: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
435: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
436: if(info->pos < info->left) {
437: info->pos=STRING_NOT_FOUND;
438: return 1;
439: }
440: if(!strchr(info->chars, c))
441: return 1;
442: --(info->pos);
443: return 0;
444: }
445:
446: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
447: if(is_empty() || !chars || !strlen(chars))
448: return STRING_NOT_FOUND;
449: CORD_pos_info info={chars, left, right};
450: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
451: return info.pos;
452: else
453: return STRING_NOT_FOUND;
454: }
455:
1.174 paf 456: // String methods
457:
458: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
459: if(!known_length)
1.9 paf 460: return *this;
1.122 paf 461:
1.176 paf 462: // first: langs
463: langs.append(body, lang, known_length);
464: // next: letters themselves
1.174 paf 465: body.append_know_length(str, known_length);
1.1 paf 466:
1.174 paf 467: ASSERT_STRING_INVARIANT(*this);
1.1 paf 468: return *this;
469: }
1.248 moko 470:
1.174 paf 471: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
472: if(!str)
473: return *this;
474: size_t known_length=helper_length?helper_length:strlen(str);
475: if(!known_length)
476: return *this;
1.1 paf 477:
1.174 paf 478: return append_know_length(str, known_length, lang);
1.5 paf 479: }
1.248 moko 480:
1.244 moko 481: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 482: char buf[MAX_NUMBER];
483: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
484: }
1.248 moko 485:
1.174 paf 486: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
487: size_t known_length=helper_length?helper_length:strlen(str);
488: if(!known_length)
489: return *this;
1.5 paf 490:
1.176 paf 491: // first: langs
492: langs.append(body, lang, known_length);
493: // next: letters themselves
1.174 paf 494: body.append_strdup_know_length(str, known_length);
1.33 paf 495:
1.174 paf 496: ASSERT_STRING_INVARIANT(*this);
497: return *this;
1.5 paf 498: }
1.46 paf 499:
1.234 misha 500: struct CORD_length_info {
501: size_t len;
502: size_t skip;
503: };
504:
505: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 506: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
507: return 0;
1.221 misha 508: }
509:
1.234 misha 510: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
511: int CORD_batched_len(const char c, CORD_length_info* info){
512: if (info->skip==0){
513: info->len++;
514: info->skip = lengthUTF8Char(c)-1;
515: } else {
516: info->skip--;
517: }
1.220 misha 518: return 0;
519: }
520:
1.210 misha 521: size_t String::length(Charset& charset) const {
522: if(charset.isUTF8()){
1.234 misha 523: CORD_length_info info = {0, 0};
524: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
525: return info.len;
1.210 misha 526: } else
527: return body.length();
528: }
529:
1.174 paf 530: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
531: String& String::mid(size_t substr_begin, size_t substr_end) const {
532: String& result=*new String;
533:
534: size_t self_length=length();
535: substr_begin=min(substr_begin, self_length);
536: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 537: size_t substr_length=substr_end-substr_begin;
538: if(!substr_length)
1.107 parser 539: return result;
1.53 paf 540:
1.176 paf 541: // first: their langs
542: result.langs.append(result.body, langs, substr_begin, substr_length);
543: // next: letters themselves
544: result.body=body.mid(substr_begin, substr_length);
1.174 paf 545:
546: ASSERT_STRING_INVARIANT(result);
1.53 paf 547: return result;
1.54 paf 548: }
549:
1.211 misha 550: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
551: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 552: String& result=*new String;
553:
1.255 moko 554: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 555:
556: if(!self_length)
557: return result;
558:
559: from=min(min(to, from), self_length);
1.210 misha 560: to=min(max(to, from), self_length);
1.211 misha 561:
1.210 misha 562: size_t substr_length=to-from;
1.211 misha 563:
1.210 misha 564: if(!substr_length)
565: return result;
566:
567: if(charset.isUTF8()){
1.234 misha 568: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 569: const XMLByte* src_end=src_begin+body.length();
1.210 misha 570:
1.212 misha 571: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 572: from=getUTF8BytePos(src_begin, src_end, from);
573: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 574: if(!substr_length)
575: return result;
576: }
577:
578: // first: their langs
579: result.langs.append(result.body, langs, from, substr_length);
580: // next: letters themselves
581: result.body=body.mid(from, substr_length);
582:
583: ASSERT_STRING_INVARIANT(result);
584: return result;
585: }
586:
1.176 paf 587: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 588: size_t substr_length=substr.length();
589: while(true) {
590: size_t substr_begin=body.pos(substr, this_offset);
591:
592: if(substr_begin==CORD_NOT_FOUND)
593: return STRING_NOT_FOUND;
1.174 paf 594:
1.183 paf 595: if(langs.check_lang(lang, substr_begin, substr_length))
596: return substr_begin;
597:
598: this_offset=substr_begin+substr_length;
599: }
1.58 paf 600: }
601:
1.254 moko 602: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 603: return pos(substr.body, this_offset, lang);
1.60 paf 604: }
605:
1.254 moko 606: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 607:
608: if(charset.isUTF8()){
1.234 misha 609: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 610: const XMLByte* srcEnd=srcPtr+body.length();
611:
612: // convert 'this_offset' from 'characters' to 'bytes'
613: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
614:
1.229 misha 615: size_t result=pos(substr.body, this_offset, lang);
616: return (result==CORD_NOT_FOUND)
617: ? STRING_NOT_FOUND
618: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 619: } else {
1.229 misha 620: size_t result=pos(substr.body, this_offset, lang);
621: return (result==CORD_NOT_FOUND)
622: ? STRING_NOT_FOUND
623: : result;
1.210 misha 624: }
625: }
626:
1.256 moko 627: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 628: if(is_empty())
629: return;
1.174 paf 630: size_t self_length=length();
1.237 misha 631: if(size_t delim_length=strlen(delim)) {
1.186 paf 632: size_t pos_before;
1.60 paf 633: // while we have 'delim'...
1.283 ! moko 634: while((pos_before=pos(String::Body(delim), pos_after, lang)) != STRING_NOT_FOUND) {
1.69 paf 635: result+=&mid(pos_after, pos_before);
1.174 paf 636: pos_after=pos_before+delim_length;
1.60 paf 637: }
638: // last piece
1.256 moko 639: if(pos_after<self_length)
1.174 paf 640: result+=&mid(pos_after, self_length);
1.60 paf 641: } else { // empty delim
642: result+=this;
643: }
644: }
645:
1.256 moko 646: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 647: if(is_empty())
648: return;
649: if(!delim.is_empty()) {
1.186 paf 650: size_t pos_before;
1.60 paf 651: // while we have 'delim'...
1.256 moko 652: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 653: result+=&mid(pos_after, pos_before);
1.174 paf 654: pos_after=pos_before+delim.length();
1.60 paf 655: }
656: // last piece
1.256 moko 657: if(pos_after<length())
1.174 paf 658: result+=&mid(pos_after, length());
1.60 paf 659: } else { // empty delim
660: result+=this;
661: }
1.61 paf 662: }
663:
1.254 moko 664: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 665:
1.222 misha 666: // vregex->info(); // I have no idea what does it for?
1.63 paf 667:
1.222 misha 668: bool need_pre_post_match=vregex->is_pre_post_match_needed();
669: bool global=vregex->is_global_search();
1.63 paf 670:
1.174 paf 671: const char* subject=cstr();
1.234 misha 672: size_t subject_length=length();
1.271 moko 673: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */
1.222 misha 674: int ovector[ovector_size];
1.155 paf 675:
1.173 paf 676: Table::Action_options table_options;
1.278 moko 677: Table& table=*new Table(string_match_table_template(), table_options);
1.63 paf 678:
1.154 paf 679: int prestart=0;
680: int poststart=0;
1.174 paf 681: int postfinish=length();
1.267 moko 682: int action_was_executed=-1;
1.63 paf 683: while(true) {
1.222 misha 684: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 685:
1.222 misha 686: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
687: break;
1.63 paf 688:
1.154 paf 689: int prefinish=ovector[0];
690: poststart=ovector[1];
1.236 moko 691:
1.267 moko 692: if (prestart==poststart && action_was_executed==1){
1.236 moko 693: prestart++;
1.267 moko 694: action_was_executed=0;
1.236 moko 695: continue;
696: }
697:
1.232 misha 698: ArrayString* row=new ArrayString(3);
1.174 paf 699: if(need_pre_post_match) {
700: *row+=&mid(0, prefinish); // .prematch column value
701: *row+=&mid(prefinish, poststart); // .match
702: *row+=&mid(poststart, postfinish); // .postmatch
703: } else {
1.185 paf 704: *row+=&Empty; // .prematch column value
705: *row+=&Empty; // .match
706: *row+=&Empty; // .postmatch
1.174 paf 707: }
1.63 paf 708:
1.222 misha 709: for(int i=1; i<exec_result; i++) {
1.69 paf 710: // -1:-1 case handled peacefully by mid() itself
1.232 misha 711: *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 712: }
713:
1.209 misha 714: matches_count++;
1.267 moko 715: row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info);
1.63 paf 716:
1.275 moko 717: if(!global || (size_t)poststart>=subject_length) // last step, avoid prestart++ after last char
1.222 misha 718: break;
719:
1.154 paf 720: prestart=poststart;
1.267 moko 721: action_was_executed=1;
1.222 misha 722: }
1.63 paf 723:
1.222 misha 724: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
725: return vregex->is_just_count() ? 0 : &table;
1.82 parser 726: }
727:
1.174 paf 728: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
729: String& result=*new String();
730: if(is_empty())
731: return result;
732:
733: char* new_cstr=cstrm();
1.216 misha 734:
1.181 paf 735: if(source_charset.isUTF8()) {
1.216 misha 736: size_t new_cstr_len=length();
1.181 paf 737: switch(kind) {
738: case CC_UPPER:
1.192 paf 739: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 740: break;
741: case CC_LOWER:
1.192 paf 742: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 743: break;
744: default:
745: assert(!"unknown change case kind");
746: break; // never
747: }
748:
749: } else {
750: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 751:
1.181 paf 752: const unsigned char *a;
753: const unsigned char *b;
754: switch(kind) {
755: case CC_UPPER:
756: a=tables+lcc_offset;
757: b=tables+fcc_offset;
758: break;
759: case CC_LOWER:
760: a=tables+lcc_offset;
761: b=0;
762: break;
763: default:
764: assert(!"unknown change case kind");
765: a=b=0; // calm, compiler
766: break; // never
767: }
768:
1.192 paf 769: char *dest=new_cstr;
1.181 paf 770: unsigned char index;
1.190 paf 771: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 772: unsigned char c=a[index];
773: if(b)
774: c=b[c];
775:
776: *dest++=(char)c;
777: }
1.174 paf 778: }
1.176 paf 779: result.langs=langs;
1.283 ! moko 780: result.body=String::Body(new_cstr);
1.89 parser 781:
1.101 parser 782: return result;
783: }
784:
1.213 misha 785: const String& String::escape(Charset& source_charset) const {
786: if(is_empty())
787: return *this;
788:
789: return Charset::escape(*this, source_charset);
790: }
791:
1.238 misha 792: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
793: result.langs.append(result.body, langs, langs_offset, length); \
794: result.body.append_strdup_know_length(from_cstr, length);
795:
1.174 paf 796: const String& String::replace(const Dictionary& dict) const {
1.238 misha 797: if(!dict.count() || is_empty())
798: return *this;
799:
1.174 paf 800: String& result=*new String();
801: const char* old_cstr=cstr();
802: const char* prematch_begin=old_cstr;
803:
1.238 misha 804: if(dict.count()==1) {
805: // optimized simple case
806:
807: Dictionary::Subst subst=dict.get(0);
1.239 moko 808: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 809: // prematch
1.238 misha 810: if(size_t prematch_length=p-prematch_begin) {
811: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 812: }
813:
1.174 paf 814: // match
1.238 misha 815: prematch_begin=p+subst.from_length;
1.174 paf 816:
1.184 paf 817: if(const String* b=subst.to) // are there any b?
1.174 paf 818: result<<*b;
1.238 misha 819: }
820:
821: } else {
822:
823: const char* current=old_cstr;
824: while(*current) {
825: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
826: // prematch
827: if(size_t prematch_length=current-prematch_begin) {
828: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
829: }
830:
831: // match
832: // skip 'a' in 'current'; move prematch_begin
833: current+=subst.from_length; prematch_begin=current;
834:
835: if(const String* b=subst.to) // are there any b?
836: result<<*b;
837: } else // simply advance
838: current++;
839: }
840:
1.174 paf 841: }
1.156 paf 842:
1.238 misha 843: if(prematch_begin==old_cstr) // not modified
844: return *this;
845:
1.174 paf 846: // postmatch
1.238 misha 847: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
848: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 849: }
1.156 paf 850:
1.174 paf 851: ASSERT_STRING_INVARIANT(result);
1.82 parser 852: return result;
1.61 paf 853: }
1.113 parser 854:
1.180 paf 855: static int serialize_body_char(char c, char** cur) {
856: *((*cur)++)=c;
857: return 0; // 0=continue
1.248 moko 858: }
859:
1.174 paf 860: static int serialize_body_piece(const char* s, char** cur) {
861: size_t length=strlen(s);
862: memcpy(*cur, s, length); *cur+=length;
1.178 paf 863: return 0; // 0=continue
1.248 moko 864: }
865:
1.178 paf 866: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
867: // lang
1.191 paf 868: **cur=alang; (*cur)++;
869: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 870: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
871:
872: return 0; // 0=continue
873: }
1.248 moko 874:
1.174 paf 875: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 876: size_t fragments_count=langs.count();
1.202 paf 877: size_t body_length=body.length();
1.174 paf 878: size_t buf_length=
1.178 paf 879: prolog_length //1
880: +sizeof(size_t) //2
1.202 paf 881: +body_length //3
882: +1 // 4 for zero terminator used in deserialize
883: +sizeof(size_t) //5
884: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
885:
1.174 paf 886: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
887:
888: // 1: prolog
889: char *cur=result.str+prolog_length;
1.202 paf 890: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
891: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
892: // 3: letters
893: body.for_each(serialize_body_char, serialize_body_piece, &cur);
894: // 4: zero terminator
895: *cur++=0;
896: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 897: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 898: // 6: lang info
1.178 paf 899: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 900:
1.174 paf 901: return result;
1.113 parser 902: }
1.248 moko 903:
1.202 paf 904: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
905: size_t in_buf=buf_size;
906: if(in_buf<=prolog_size)
1.148 paf 907: return false;
1.202 paf 908: in_buf-=prolog_size;
1.135 paf 909:
1.174 paf 910: // 1: prolog
1.202 paf 911: const char* cur=(const char* )buf+prolog_size;
1.113 parser 912:
1.207 paf 913: // 2: chars.count
1.202 paf 914: size_t body_length;
915: if(in_buf<sizeof(body_length)) // body.length don't fit?
916: return false;
917: // [WARNING: not cast, addresses must be %4=0 on sparc]
918: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
919: in_buf-=sizeof(body_length);
920:
921: if(in_buf<body_length+1) // letters+terminator don't fit?
922: return false;
923: // 4: zero terminator
924: if(cur[body_length] != 0) // in place?
925: return false;
926: // 3: letters
1.251 moko 927: body=String::Body(String::C(cur, body_length));
1.202 paf 928: cur+=body_length+1;
929: in_buf-=body_length+1;
930:
931: // 5: langs.count
1.191 paf 932: size_t fragments_count;
1.202 paf 933: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 934: return false;
1.191 paf 935: // [WARNING: not cast, addresses must be %4=0 on sparc]
936: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 937: in_buf-=sizeof(fragments_count);
1.174 paf 938:
939: if(fragments_count) {
1.202 paf 940: // 6: lang info
1.174 paf 941: size_t total_length=0;
942: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 943: char lang;
944: size_t fragment_length;
945: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 946: if(in_buf<piece_length) // lang+length
1.174 paf 947: return false;
948:
1.191 paf 949: // lang
950: lang=*cur++;
951: // length [WARNING: not cast, addresses must be %4=0 on sparc]
952: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
953:
1.206 paf 954: size_t combined_length=total_length+fragment_length;
955: if(combined_length>body_length)
956: return false; // file curruption
1.191 paf 957: // uchar needed to prevent propagating 0x80 bit to upper bytes
958: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 959: total_length=combined_length;
1.202 paf 960: in_buf-=piece_length;
1.174 paf 961: }
1.128 paf 962:
1.202 paf 963: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 964: return false;
1.174 paf 965: }
1.202 paf 966: if(in_buf!=0) // some strange extra bytes
967: return false;
1.113 parser 968:
1.174 paf 969: ASSERT_STRING_INVARIANT(*this);
1.148 paf 970: return true;
1.176 paf 971: }
972:
1.201 paf 973: void String::Body::dump() const {
974: CORD_dump(body);
975: }
976:
1.257 moko 977: const char* String::Languages::visualize() const {
1.177 paf 978: if(opt.is_not_just_lang)
1.233 misha 979: return CORD_to_const_char_star(langs, 0);
1.176 paf 980: else
1.257 moko 981: return 0;
1.176 paf 982: }
1.248 moko 983:
1.201 paf 984: void String::Languages::dump() const {
985: if(opt.is_not_just_lang)
986: CORD_dump(langs);
987: else
988: puts((const char*)&langs);
989: }
1.248 moko 990:
1.257 moko 991: void String::dump() const {
992: body.dump();
993: langs.dump();
994: }
1.176 paf 995:
1.257 moko 996: static char *n_chars(char c, size_t length){
997: char *result=(char *)pa_malloc_atomic(length+1);
998: memset(result, c, length);
999: result[length] = '\0';
1000: return result;
1.113 parser 1001: }
1.229 misha 1002:
1.257 moko 1003: char* String::visualize_langs() const {
1.258 moko 1004: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 1005: }
1.229 misha 1006:
1007: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 1008: if(is_empty())
1.195 paf 1009: return *this;
1010:
1011: size_t substr_begin, substr_length;
1.229 misha 1012: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1013: if(new_body==body) // we received unchanged pointer, do likewise
1014: return *this;
1015: // new_body differs from body, adjust langs along
1016:
1017: String& result=*new String;
1018: if(!new_body) // body.trim produced empty result
1019: return result;
1020: // body.trim produced nonempty result
1021:
1022: // first: their langs
1023: result.langs.append(result.body, langs, substr_begin, substr_length);
1024: // next: letters themselves
1025: result.body=new_body;
1026:
1027: ASSERT_STRING_INVARIANT(result);
1028: return result;
1.198 paf 1029: }
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