Annotation of parser3/src/main/pa_string.C, revision 1.254
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.253 moko 4: Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
1.138 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://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:
1.252 moko 15: #ifndef ULLONG_MAX
16: #define ULLONG_MAX 18446744073709551615ULL
17: #endif
18:
1.254 ! moko 19: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.253 2015/10/26 01:21:59 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 20:
1.185 paf 21: const String String::Empty;
22:
1.242 moko 23: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
24:
1.249 moko 25: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
26: T result = 0;
1.242 moko 27: const char *pos = str;
28:
29: while (isspace(*pos)) /* skip leading whitespace */
30: ++pos;
31:
32: if (base == 16 && *pos == '0') { /* handle option prefix */
33: ++pos;
34: if (*pos == 'x' || *pos == 'X') {
35: ++pos;
36: }
37: }
38:
39: if (base == 0) { /* dynamic base */
40: base = 10; /* default is 10 */
41: if (*pos == '0') {
42: ++pos;
1.245 moko 43: if (*pos == 'x' || *pos == 'X'){
1.242 moko 44: ++pos;
45: base=16;
1.245 moko 46: }
1.242 moko 47: }
48: }
49:
50: if (base < 2 || base > 16) { /* illegal base */
51: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
52: }
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))
84: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
85:
86: return result;
87: }
88:
1.249 moko 89: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.254 ! moko 90: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 91: }
92:
93: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
1.254 ! moko 94: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
1.249 moko 95: }
96:
1.193 paf 97: int pa_atoi(const char* str, const String* problem_source) {
98: if(!str)
99: return 0;
100:
1.242 moko 101: while(isspace(*str))
1.193 paf 102: str++;
1.242 moko 103:
1.193 paf 104: if(!*str)
105: return 0;
106:
1.200 paf 107: bool negative=false;
108: if(str[0]=='-') {
109: negative=true;
110: str++;
111: } else if(str[0]=='+') {
112: str++;
113: }
1.193 paf 114:
1.242 moko 115: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 116:
1.242 moko 117: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 118: return -(int)result;
1.242 moko 119:
120: if(result<=INT_MAX)
1.243 moko 121: return (int)result;
1.242 moko 122:
123: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 124: }
125:
126: double pa_atod(const char* str, const String* problem_source) {
127: if(!str)
128: return 0;
129:
1.242 moko 130: while(isspace(*str))
1.193 paf 131: str++;
1.242 moko 132:
1.193 paf 133: if(!*str)
134: return 0;
135:
1.200 paf 136: bool negative=false;
137: if(str[0]=='-') {
138: negative=true;
139: str++;
140: } else if(str[0]=='+') {
141: str++;
142: }
1.242 moko 143:
144: double result;
1.247 moko 145: if(str[0]=='0') {
146: if(str[1]=='x' || str[1]=='X') {
1.242 moko 147: // 0xABC
1.249 moko 148: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 149: return negative ? -result : result;
150: } else {
1.200 paf 151: // skip leading 0000, to disable octal interpretation
1.242 moko 152: do str++; while(*str=='0');
1.200 paf 153: }
1.247 moko 154: }
1.242 moko 155:
156: char *error_pos;
157: result=strtod(str, &error_pos);
1.193 paf 158:
159: while(char c=*error_pos++)
1.199 paf 160: if(!isspace((unsigned char)c))
1.242 moko 161: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 162:
1.242 moko 163: return negative ? -result : result;
1.193 paf 164: }
165:
1.176 paf 166: // cord lib extension
167:
168: #ifndef DOXYGEN
169: typedef struct {
1.254 ! moko 170: ssize_t countdown;
! 171: int target; /* Character we're looking for */
1.176 paf 172: } chr_data;
173: #endif
1.248 moko 174:
1.176 paf 175: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
176: {
1.254 ! moko 177: register chr_data * d = (chr_data *)client_data;
! 178:
! 179: if (d -> countdown<=0) return(2);
! 180: d -> countdown -= size;
! 181: if (c > d -> target) return(1);
! 182: return(0);
1.176 paf 183: }
1.248 moko 184:
1.176 paf 185: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
186: {
1.254 ! moko 187: chr_data d;
1.176 paf 188:
1.254 ! moko 189: d.countdown = n;
! 190: d.target = c;
! 191: 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 192: }
193:
1.187 paf 194: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 195: {
1.254 ! moko 196: int* result=(int*)client_data;
! 197: (*result)++;
! 198: return(0); // 0=continue
1.178 paf 199: }
1.248 moko 200:
1.178 paf 201: size_t CORD_block_count(CORD x)
202: {
203: size_t result=0;
204: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 ! moko 205: return result;
1.178 paf 206: }
207:
1.174 paf 208: // helpers
1.139 paf 209:
1.174 paf 210: /// String::match uses this as replace & global search table columns
1.139 paf 211:
1.174 paf 212: const int MAX_MATCH_GROUPS=100;
1.139 paf 213:
1.174 paf 214: class String_match_table_template_columns: public ArrayString {
215: public:
216: String_match_table_template_columns() {
217: *this+=new String("prematch");
218: *this+=new String("match");
219: *this+=new String("postmatch");
220: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 221: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 222: }
223: }
224: };
225:
226: Table string_match_table_template(new String_match_table_template_columns);
227:
1.176 paf 228: // String::Body methods
1.140 paf 229:
1.176 paf 230: String::Body String::Body::Format(int value) {
1.174 paf 231: char local[MAX_NUMBER];
232: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 233: return String::Body(pa_strdup(local, length));
1.120 paf 234: }
235:
1.254 ! moko 236: 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 237: size_t our_length=length();
238: if(!our_length)
239: return *this;
1.229 misha 240:
241: // check if any UTF-8 in chars
242: bool fast=true;
243: if(chars && source_charset && source_charset->isUTF8()){
244: const char* pos=chars;
245: while(unsigned char c=*pos++)
246: if(c>127){
247: fast=false;
248: break;
249: }
250: }
251:
252: size_t start=0;
253: size_t end=our_length;
1.195 paf 254: if(!chars)
255: chars=" \t\n"; // white space
256:
1.229 misha 257: if(fast){
258: // from left...
259: if(kind!=TRIM_END) {
260: CORD_pos pos; set_pos(pos, 0);
261: while(true) {
262: char c=CORD_pos_fetch(pos);
263: if(strchr(chars, c)) {
264: if(++start==our_length)
265: return 0; // all chars are empty, just return empty string
266: } else
267: break;
268:
269: CORD_next(pos);
270: }
271: }
272:
273: // from right..
274: if(kind!=TRIM_START) {
275: CORD_pos pos; set_pos(pos, end-1);
276: while(true) {
277: char c=CORD_pos_fetch(pos);
278: if(strchr(chars, c)) {
279: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
280: return 0; // all chars are empty, just return empty string
281: } else
282: break;
283:
284: CORD_prev(pos);
285: }
286: }
287: } else {
1.234 misha 288: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 289: const XMLByte* src_end=src_begin+our_length;
290:
291: // from left...
292: if(kind!=TRIM_END) {
293: while(src_begin<src_end){
294: uint char_length=1;
295: const XMLByte* ptr=src_begin;
1.231 misha 296: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
297: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 298: char_length++;
299:
300: bool found=false;
301: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
302: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
303: found=true;
304: break;
305: }
306:
307: if(found){
308: start+=char_length;
309: if(start==our_length)
310: return 0; // all chars are empty, just return empty string
311: } else
312: break;
313: }
314: }
1.196 paf 315:
1.229 misha 316: // from right..
317: if(kind!=TRIM_START) {
318: while(src_begin<src_end){
319: uint char_length=1;
1.231 misha 320: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
321: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 322: char_length++;
323:
324: bool found=false;
325: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
326: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
327: found=true;
328: break;
329: }
330:
331: if(found){
332: end-=char_length;
333: if(end==0)
334: return 0; // all chars are empty, just return empty string
335: } else
336: break;
337: }
1.196 paf 338: }
339: }
1.195 paf 340:
341: if(start==0 && end==our_length) // nobody moved a thing
342: return *this;
343:
344: if(out_start)
345: *out_start=start;
346: size_t new_length=end-start;
347: if(out_length)
348: *out_length=new_length;
349:
350: return mid(start, new_length);
351: }
352:
1.174 paf 353: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
354: uint& result=*static_cast<uint*>(client_data);
355: generic_hash_code(result, c);
356: return 0;
357: }
1.248 moko 358:
1.174 paf 359: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
360: uint& result=*static_cast<uint*>(client_data);
361: generic_hash_code(result, s);
362: return 0;
1.248 moko 363: }
364:
1.225 misha 365: uint String::Body::get_hash_code() const {
366: #ifdef HASH_CODE_CACHING
367: if(hash_code)
368: return hash_code;
369: #else
370: uint hash_code=0;
371: #endif
1.220 misha 372: if (body && CORD_IS_STRING(body)){
1.250 moko 373: generic_hash_code(hash_code, (const char *)body);
1.220 misha 374: } else {
375: CORD_iter5(body, 0,
376: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 377: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 378: }
1.225 misha 379: return hash_code;
1.94 parser 380: }
381:
1.241 misha 382: struct CORD_pos_info {
383: const char* chars;
384: size_t left;
385: size_t pos;
386: };
387:
388: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
389: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
390: if(info->pos < info->left){
391: info->pos=STRING_NOT_FOUND;
392: return 1;
393: }
394: if(strchr(info->chars, c))
395: return 1;
396: --(info->pos);
397: return 0;
398: }
399:
400: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
401: if(is_empty() || !chars || !strlen(chars))
402: return STRING_NOT_FOUND;
403: CORD_pos_info info={chars, left, right};
404: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
405: return info.pos;
406: else
407: return STRING_NOT_FOUND;
408: }
409:
410:
411: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
412: static int CORD_iter_fn_rskip(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::rskipchars(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_rskip, &info))
428: return info.pos;
429: else
430: return STRING_NOT_FOUND;
431: }
432:
1.174 paf 433: // String methods
434:
435: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
436: if(!known_length)
1.9 paf 437: return *this;
1.122 paf 438:
1.176 paf 439: // first: langs
440: langs.append(body, lang, known_length);
441: // next: letters themselves
1.174 paf 442: body.append_know_length(str, known_length);
1.1 paf 443:
1.174 paf 444: ASSERT_STRING_INVARIANT(*this);
1.1 paf 445: return *this;
446: }
1.248 moko 447:
1.174 paf 448: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
449: if(!str)
450: return *this;
451: size_t known_length=helper_length?helper_length:strlen(str);
452: if(!known_length)
453: return *this;
1.1 paf 454:
1.174 paf 455: return append_know_length(str, known_length, lang);
1.5 paf 456: }
1.248 moko 457:
1.244 moko 458: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 459: char buf[MAX_NUMBER];
460: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
461: }
1.248 moko 462:
1.174 paf 463: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
464: size_t known_length=helper_length?helper_length:strlen(str);
465: if(!known_length)
466: return *this;
1.5 paf 467:
1.176 paf 468: // first: langs
469: langs.append(body, lang, known_length);
470: // next: letters themselves
1.174 paf 471: body.append_strdup_know_length(str, known_length);
1.33 paf 472:
1.174 paf 473: ASSERT_STRING_INVARIANT(*this);
474: return *this;
1.5 paf 475: }
1.46 paf 476:
1.234 misha 477: struct CORD_length_info {
478: size_t len;
479: size_t skip;
480: };
481:
482: int CORD_batched_len(const char* s, CORD_length_info* info){
483: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 484: }
485:
1.234 misha 486: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
487: int CORD_batched_len(const char c, CORD_length_info* info){
488: if (info->skip==0){
489: info->len++;
490: info->skip = lengthUTF8Char(c)-1;
491: } else {
492: info->skip--;
493: }
1.220 misha 494: return 0;
495: }
496:
1.210 misha 497: size_t String::length(Charset& charset) const {
498: if(charset.isUTF8()){
1.234 misha 499: CORD_length_info info = {0, 0};
500: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
501: return info.len;
1.210 misha 502: } else
503: return body.length();
504: }
505:
1.174 paf 506: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
507: String& String::mid(size_t substr_begin, size_t substr_end) const {
508: String& result=*new String;
509:
510: size_t self_length=length();
511: substr_begin=min(substr_begin, self_length);
512: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 513: size_t substr_length=substr_end-substr_begin;
514: if(!substr_length)
1.107 parser 515: return result;
1.53 paf 516:
1.176 paf 517: // first: their langs
518: result.langs.append(result.body, langs, substr_begin, substr_length);
519: // next: letters themselves
520: result.body=body.mid(substr_begin, substr_length);
1.174 paf 521:
522: ASSERT_STRING_INVARIANT(result);
1.53 paf 523: return result;
1.54 paf 524: }
525:
1.211 misha 526: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
527: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 528: String& result=*new String;
529:
1.211 misha 530: size_t self_length=(helper_length)?helper_length:length(charset);
531:
532: if(!self_length)
533: return result;
534:
535: from=min(min(to, from), self_length);
1.210 misha 536: to=min(max(to, from), self_length);
1.211 misha 537:
1.210 misha 538: size_t substr_length=to-from;
1.211 misha 539:
1.210 misha 540: if(!substr_length)
541: return result;
542:
543: if(charset.isUTF8()){
1.234 misha 544: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 545: const XMLByte* src_end=src_begin+body.length();
1.210 misha 546:
1.212 misha 547: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 548: from=getUTF8BytePos(src_begin, src_end, from);
549: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 550: if(!substr_length)
551: return result;
552: }
553:
554: // first: their langs
555: result.langs.append(result.body, langs, from, substr_length);
556: // next: letters themselves
557: result.body=body.mid(from, substr_length);
558:
559: ASSERT_STRING_INVARIANT(result);
560: return result;
561: }
562:
1.176 paf 563: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 564: size_t substr_length=substr.length();
565: while(true) {
566: size_t substr_begin=body.pos(substr, this_offset);
567:
568: if(substr_begin==CORD_NOT_FOUND)
569: return STRING_NOT_FOUND;
1.174 paf 570:
1.183 paf 571: if(langs.check_lang(lang, substr_begin, substr_length))
572: return substr_begin;
573:
574: this_offset=substr_begin+substr_length;
575: }
1.58 paf 576: }
577:
1.254 ! moko 578: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 579: return pos(substr.body, this_offset, lang);
1.60 paf 580: }
581:
1.254 ! moko 582: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 583:
584: if(charset.isUTF8()){
1.234 misha 585: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 586: const XMLByte* srcEnd=srcPtr+body.length();
587:
588: // convert 'this_offset' from 'characters' to 'bytes'
589: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
590:
1.229 misha 591: size_t result=pos(substr.body, this_offset, lang);
592: return (result==CORD_NOT_FOUND)
593: ? STRING_NOT_FOUND
594: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 595: } else {
1.229 misha 596: size_t result=pos(substr.body, this_offset, lang);
597: return (result==CORD_NOT_FOUND)
598: ? STRING_NOT_FOUND
599: : result;
1.210 misha 600: }
601: }
602:
1.254 ! moko 603: void String::split(ArrayString& result, size_t& pos_after, const char* delim, Language lang, int limit) const {
1.237 misha 604: if(is_empty())
605: return;
1.174 paf 606: size_t self_length=length();
1.237 misha 607: if(size_t delim_length=strlen(delim)) {
1.186 paf 608: size_t pos_before;
1.60 paf 609: // while we have 'delim'...
1.174 paf 610: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 611: result+=&mid(pos_after, pos_before);
1.174 paf 612: pos_after=pos_before+delim_length;
1.60 paf 613: }
614: // last piece
1.174 paf 615: if(pos_after<self_length && limit) {
616: result+=&mid(pos_after, self_length);
617: pos_after=self_length;
1.60 paf 618: }
619: } else { // empty delim
620: result+=this;
1.174 paf 621: pos_after+=self_length;
1.60 paf 622: }
623: }
624:
1.254 ! moko 625: void String::split(ArrayString& result, size_t& pos_after, const String& delim, Language lang, int limit) const {
1.237 misha 626: if(is_empty())
627: return;
628: if(!delim.is_empty()) {
1.186 paf 629: size_t pos_before;
1.60 paf 630: // while we have 'delim'...
1.174 paf 631: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 632: result+=&mid(pos_after, pos_before);
1.174 paf 633: pos_after=pos_before+delim.length();
1.60 paf 634: }
635: // last piece
1.174 paf 636: if(pos_after<length() && limit) {
637: result+=&mid(pos_after, length());
638: pos_after=length();
1.60 paf 639: }
640: } else { // empty delim
641: result+=this;
1.174 paf 642: pos_after+=length();
1.60 paf 643: }
1.61 paf 644: }
645:
1.254 ! moko 646: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 647:
1.222 misha 648: // vregex->info(); // I have no idea what does it for?
1.63 paf 649:
1.222 misha 650: bool need_pre_post_match=vregex->is_pre_post_match_needed();
651: bool global=vregex->is_global_search();
1.63 paf 652:
1.174 paf 653: const char* subject=cstr();
1.234 misha 654: size_t subject_length=length();
1.222 misha 655: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
656: int ovector[ovector_size];
1.155 paf 657:
1.173 paf 658: Table::Action_options table_options;
1.174 paf 659: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 660:
1.154 paf 661: int prestart=0;
662: int poststart=0;
1.174 paf 663: int postfinish=length();
1.63 paf 664: while(true) {
1.222 misha 665: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 666:
1.222 misha 667: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
668: break;
1.63 paf 669:
1.154 paf 670: int prefinish=ovector[0];
671: poststart=ovector[1];
1.236 moko 672:
673: if (prestart==poststart && subject[poststart]=='\n'){
674: prestart++;
675: continue;
676: }
677:
1.232 misha 678: ArrayString* row=new ArrayString(3);
1.174 paf 679: if(need_pre_post_match) {
680: *row+=&mid(0, prefinish); // .prematch column value
681: *row+=&mid(prefinish, poststart); // .match
682: *row+=&mid(poststart, postfinish); // .postmatch
683: } else {
1.185 paf 684: *row+=&Empty; // .prematch column value
685: *row+=&Empty; // .match
686: *row+=&Empty; // .postmatch
1.174 paf 687: }
1.63 paf 688:
1.222 misha 689: for(int i=1; i<exec_result; i++) {
1.69 paf 690: // -1:-1 case handled peacefully by mid() itself
1.232 misha 691: *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 692: }
693:
1.209 misha 694: matches_count++;
1.174 paf 695: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 696:
1.222 misha 697: if(!global || prestart==poststart) // last step
698: break;
699:
1.154 paf 700: prestart=poststart;
1.222 misha 701: }
1.63 paf 702:
1.222 misha 703: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
704: return vregex->is_just_count() ? 0 : &table;
1.82 parser 705: }
706:
1.174 paf 707: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
708: String& result=*new String();
709: if(is_empty())
710: return result;
711:
712: char* new_cstr=cstrm();
1.216 misha 713:
1.181 paf 714: if(source_charset.isUTF8()) {
1.216 misha 715: size_t new_cstr_len=length();
1.181 paf 716: switch(kind) {
717: case CC_UPPER:
1.192 paf 718: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 719: break;
720: case CC_LOWER:
1.192 paf 721: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 722: break;
723: default:
724: assert(!"unknown change case kind");
725: break; // never
726: }
727:
728: } else {
729: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 730:
1.181 paf 731: const unsigned char *a;
732: const unsigned char *b;
733: switch(kind) {
734: case CC_UPPER:
735: a=tables+lcc_offset;
736: b=tables+fcc_offset;
737: break;
738: case CC_LOWER:
739: a=tables+lcc_offset;
740: b=0;
741: break;
742: default:
743: assert(!"unknown change case kind");
744: a=b=0; // calm, compiler
745: break; // never
746: }
747:
1.192 paf 748: char *dest=new_cstr;
1.181 paf 749: unsigned char index;
1.190 paf 750: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 751: unsigned char c=a[index];
752: if(b)
753: c=b[c];
754:
755: *dest++=(char)c;
756: }
1.174 paf 757: }
1.176 paf 758: result.langs=langs;
1.174 paf 759: result.body=new_cstr;
1.89 parser 760:
1.101 parser 761: return result;
762: }
763:
1.213 misha 764: const String& String::escape(Charset& source_charset) const {
765: if(is_empty())
766: return *this;
767:
768: return Charset::escape(*this, source_charset);
769: }
770:
1.238 misha 771: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
772: result.langs.append(result.body, langs, langs_offset, length); \
773: result.body.append_strdup_know_length(from_cstr, length);
774:
1.174 paf 775: const String& String::replace(const Dictionary& dict) const {
1.238 misha 776: if(!dict.count() || is_empty())
777: return *this;
778:
1.174 paf 779: String& result=*new String();
780: const char* old_cstr=cstr();
781: const char* prematch_begin=old_cstr;
782:
1.238 misha 783: if(dict.count()==1) {
784: // optimized simple case
785:
786: Dictionary::Subst subst=dict.get(0);
1.239 moko 787: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 788: // prematch
1.238 misha 789: if(size_t prematch_length=p-prematch_begin) {
790: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 791: }
792:
1.174 paf 793: // match
1.238 misha 794: prematch_begin=p+subst.from_length;
1.174 paf 795:
1.184 paf 796: if(const String* b=subst.to) // are there any b?
1.174 paf 797: result<<*b;
1.238 misha 798: }
799:
800: } else {
801:
802: const char* current=old_cstr;
803: while(*current) {
804: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
805: // prematch
806: if(size_t prematch_length=current-prematch_begin) {
807: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
808: }
809:
810: // match
811: // skip 'a' in 'current'; move prematch_begin
812: current+=subst.from_length; prematch_begin=current;
813:
814: if(const String* b=subst.to) // are there any b?
815: result<<*b;
816: } else // simply advance
817: current++;
818: }
819:
1.174 paf 820: }
1.156 paf 821:
1.238 misha 822: if(prematch_begin==old_cstr) // not modified
823: return *this;
824:
1.174 paf 825: // postmatch
1.238 misha 826: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
827: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 828: }
1.156 paf 829:
1.174 paf 830: ASSERT_STRING_INVARIANT(result);
1.82 parser 831: return result;
1.61 paf 832: }
1.113 parser 833:
1.180 paf 834: static int serialize_body_char(char c, char** cur) {
835: *((*cur)++)=c;
836: return 0; // 0=continue
1.248 moko 837: }
838:
1.174 paf 839: static int serialize_body_piece(const char* s, char** cur) {
840: size_t length=strlen(s);
841: memcpy(*cur, s, length); *cur+=length;
1.178 paf 842: return 0; // 0=continue
1.248 moko 843: }
844:
1.178 paf 845: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
846: // lang
1.191 paf 847: **cur=alang; (*cur)++;
848: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 849: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
850:
851: return 0; // 0=continue
852: }
1.248 moko 853:
1.174 paf 854: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 855: size_t fragments_count=langs.count();
1.202 paf 856: size_t body_length=body.length();
1.174 paf 857: size_t buf_length=
1.178 paf 858: prolog_length //1
859: +sizeof(size_t) //2
1.202 paf 860: +body_length //3
861: +1 // 4 for zero terminator used in deserialize
862: +sizeof(size_t) //5
863: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
864:
1.174 paf 865: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
866:
867: // 1: prolog
868: char *cur=result.str+prolog_length;
1.202 paf 869: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
870: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
871: // 3: letters
872: body.for_each(serialize_body_char, serialize_body_piece, &cur);
873: // 4: zero terminator
874: *cur++=0;
875: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 876: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 877: // 6: lang info
1.178 paf 878: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 879:
1.174 paf 880: return result;
1.113 parser 881: }
1.248 moko 882:
1.202 paf 883: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
884: size_t in_buf=buf_size;
885: if(in_buf<=prolog_size)
1.148 paf 886: return false;
1.202 paf 887: in_buf-=prolog_size;
1.135 paf 888:
1.174 paf 889: // 1: prolog
1.202 paf 890: const char* cur=(const char* )buf+prolog_size;
1.113 parser 891:
1.207 paf 892: // 2: chars.count
1.202 paf 893: size_t body_length;
894: if(in_buf<sizeof(body_length)) // body.length don't fit?
895: return false;
896: // [WARNING: not cast, addresses must be %4=0 on sparc]
897: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
898: in_buf-=sizeof(body_length);
899:
900: if(in_buf<body_length+1) // letters+terminator don't fit?
901: return false;
902: // 4: zero terminator
903: if(cur[body_length] != 0) // in place?
904: return false;
905: // 3: letters
1.251 moko 906: body=String::Body(String::C(cur, body_length));
1.202 paf 907: cur+=body_length+1;
908: in_buf-=body_length+1;
909:
910: // 5: langs.count
1.191 paf 911: size_t fragments_count;
1.202 paf 912: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 913: return false;
1.191 paf 914: // [WARNING: not cast, addresses must be %4=0 on sparc]
915: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 916: in_buf-=sizeof(fragments_count);
1.174 paf 917:
918: if(fragments_count) {
1.202 paf 919: // 6: lang info
1.174 paf 920: size_t total_length=0;
921: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 922: char lang;
923: size_t fragment_length;
924: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 925: if(in_buf<piece_length) // lang+length
1.174 paf 926: return false;
927:
1.191 paf 928: // lang
929: lang=*cur++;
930: // length [WARNING: not cast, addresses must be %4=0 on sparc]
931: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
932:
1.206 paf 933: size_t combined_length=total_length+fragment_length;
934: if(combined_length>body_length)
935: return false; // file curruption
1.191 paf 936: // uchar needed to prevent propagating 0x80 bit to upper bytes
937: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 938: total_length=combined_length;
1.202 paf 939: in_buf-=piece_length;
1.174 paf 940: }
1.128 paf 941:
1.202 paf 942: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 943: return false;
1.174 paf 944: }
1.202 paf 945: if(in_buf!=0) // some strange extra bytes
946: return false;
1.113 parser 947:
1.174 paf 948: ASSERT_STRING_INVARIANT(*this);
1.148 paf 949: return true;
1.176 paf 950: }
951:
952: const char* String::Body::v() const {
1.233 misha 953: return CORD_to_const_char_star(body, length());
1.176 paf 954: }
1.248 moko 955:
1.201 paf 956: void String::Body::dump() const {
957: CORD_dump(body);
958: }
959:
1.176 paf 960: const char* String::Languages::v() const {
1.177 paf 961: if(opt.is_not_just_lang)
1.233 misha 962: return CORD_to_const_char_star(langs, 0);
1.176 paf 963: else
964: return (const char*)&langs;
965: }
1.248 moko 966:
1.201 paf 967: void String::Languages::dump() const {
968: if(opt.is_not_just_lang)
969: CORD_dump(langs);
970: else
971: puts((const char*)&langs);
972: }
1.248 moko 973:
1.176 paf 974: const char* String::v() const {
1.198 paf 975: const uint LIMIT_VIEW=20;
1.246 moko 976: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 977: const char*body_view=body.v();
978: const char*langs_view=langs.v();
979: snprintf(buf, MAX_STRING,
1.178 paf 980: "%d:%.*s%s} "
1.176 paf 981: "{%d:%s",
1.178 paf 982: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 983: strlen(body_view), body_view
984: );
985:
986: return buf;
1.113 parser 987: }
1.229 misha 988:
1.201 paf 989: void String::dump() const {
990: body.dump();
991: langs.dump();
992: }
1.229 misha 993:
994: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 995: if(is_empty())
1.195 paf 996: return *this;
997:
998: size_t substr_begin, substr_length;
1.229 misha 999: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1000: if(new_body==body) // we received unchanged pointer, do likewise
1001: return *this;
1002: // new_body differs from body, adjust langs along
1003:
1004: String& result=*new String;
1005: if(!new_body) // body.trim produced empty result
1006: return result;
1007: // body.trim produced nonempty result
1008:
1009: // first: their langs
1010: result.langs.append(result.body, langs, substr_begin, substr_length);
1011: // next: letters themselves
1012: result.body=new_body;
1013:
1014: ASSERT_STRING_INVARIANT(result);
1015: return result;
1.198 paf 1016: }
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