Annotation of parser3/src/main/pa_string.C, revision 1.247
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.240 moko 4: Copyright (c) 2001-2012 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:
1.247 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.246 2015/04/06 22:27:26 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.242 moko 19:
20: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
21:
22: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
23: unsigned int result = 0;
24: const char *pos = str;
25:
26: while (isspace(*pos)) /* skip leading whitespace */
27: ++pos;
28:
29: if (base == 16 && *pos == '0') { /* handle option prefix */
30: ++pos;
31: if (*pos == 'x' || *pos == 'X') {
32: ++pos;
33: }
34: }
35:
36: if (base == 0) { /* dynamic base */
37: base = 10; /* default is 10 */
38: if (*pos == '0') {
39: ++pos;
1.245 moko 40: if (*pos == 'x' || *pos == 'X'){
1.242 moko 41: ++pos;
42: base=16;
1.245 moko 43: }
1.242 moko 44: }
45: }
46:
47: if (base < 2 || base > 16) { /* illegal base */
48: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
49: }
50:
51: unsigned int cutoff = UINT_MAX / base;
52: int cutoff_digit = UINT_MAX - cutoff * base;
53:
54: while(true) {
55: int digit;
56:
57: if ((*pos >= '0') && (*pos <= '9')) {
58: digit = (*pos - '0');
59: } else if (*pos >= 'a') {
60: digit = (*pos - 'a' + 10);
61: } else if (*pos >= 'A') {
62: digit = (*pos - 'A' + 10);
63: } else break;
64:
65: if (digit >= base) {
66: break;
67: }
68:
69: ++pos;
70:
71: /* adjust number, with overflow check */
72: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
73: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
74: } else {
75: result = result * base + digit;
76: }
77: }
78:
79: while(char c=*pos++)
80: if(!isspace(c))
81: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
82:
83: return result;
84: }
85:
1.193 paf 86: int pa_atoi(const char* str, const String* problem_source) {
87: if(!str)
88: return 0;
89:
1.242 moko 90: while(isspace(*str))
1.193 paf 91: str++;
1.242 moko 92:
1.193 paf 93: if(!*str)
94: return 0;
95:
1.200 paf 96: bool negative=false;
97: if(str[0]=='-') {
98: negative=true;
99: str++;
100: } else if(str[0]=='+') {
101: str++;
102: }
1.193 paf 103:
1.242 moko 104: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 105:
1.242 moko 106: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 107: return -(int)result;
1.242 moko 108:
109: if(result<=INT_MAX)
1.243 moko 110: return (int)result;
1.242 moko 111:
112: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 113: }
114:
115: double pa_atod(const char* str, const String* problem_source) {
116: if(!str)
117: return 0;
118:
1.242 moko 119: while(isspace(*str))
1.193 paf 120: str++;
1.242 moko 121:
1.193 paf 122: if(!*str)
123: return 0;
124:
1.200 paf 125: bool negative=false;
126: if(str[0]=='-') {
127: negative=true;
128: str++;
129: } else if(str[0]=='+') {
130: str++;
131: }
1.242 moko 132:
133: double result;
1.247 ! moko 134: if(str[0]=='0') {
! 135: if(str[1]=='x' || str[1]=='X') {
1.242 moko 136: // 0xABC
137: result=(double)pa_atoui(str, 0, problem_source);
138: return negative ? -result : result;
139: } else {
1.200 paf 140: // skip leading 0000, to disable octal interpretation
1.242 moko 141: do str++; while(*str=='0');
1.200 paf 142: }
1.247 ! moko 143: }
1.242 moko 144:
145: char *error_pos;
146: result=strtod(str, &error_pos);
1.193 paf 147:
148: while(char c=*error_pos++)
1.199 paf 149: if(!isspace((unsigned char)c))
1.242 moko 150: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 151:
1.242 moko 152: return negative ? -result : result;
1.193 paf 153: }
154:
1.176 paf 155: // cord lib extension
156:
157: #ifndef DOXYGEN
158: typedef struct {
159: ssize_t countdown;
1.204 paf 160: int target; /* Character we're looking for */
1.176 paf 161: } chr_data;
162: #endif
163: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
164: {
165: register chr_data * d = (chr_data *)client_data;
166:
167: if (d -> countdown<=0) return(2);
168: d -> countdown -= size;
169: if (c > d -> target) return(1);
170: return(0);
171: }
172: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
173: {
174: chr_data d;
175:
176: d.countdown = n;
177: d.target = c;
178: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
179: }
180:
1.187 paf 181: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 182: {
183: int* result=(int*)client_data;
184: (*result)++;
185: return(0); // 0=continue
186: }
187: size_t CORD_block_count(CORD x)
188: {
189: size_t result=0;
190: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
191: return result;
192: }
193:
1.174 paf 194: // helpers
1.139 paf 195:
1.174 paf 196: /// String::match uses this as replace & global search table columns
1.139 paf 197:
1.174 paf 198: const int MAX_MATCH_GROUPS=100;
1.139 paf 199:
1.174 paf 200: class String_match_table_template_columns: public ArrayString {
201: public:
202: String_match_table_template_columns() {
203: *this+=new String("prematch");
204: *this+=new String("match");
205: *this+=new String("postmatch");
206: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 207: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 208: }
209: }
210: };
211:
212: Table string_match_table_template(new String_match_table_template_columns);
213:
1.176 paf 214: // String::Body methods
1.140 paf 215:
1.176 paf 216: String::Body String::Body::Format(int value) {
1.174 paf 217: char local[MAX_NUMBER];
218: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 219: return String::Body(pa_strdup(local, length));
1.120 paf 220: }
221:
1.195 paf 222: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 223: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 224: size_t our_length=length();
225: if(!our_length)
226: return *this;
1.229 misha 227:
228: // check if any UTF-8 in chars
229: bool fast=true;
230: if(chars && source_charset && source_charset->isUTF8()){
231: const char* pos=chars;
232: while(unsigned char c=*pos++)
233: if(c>127){
234: fast=false;
235: break;
236: }
237: }
238:
239: size_t start=0;
240: size_t end=our_length;
1.195 paf 241: if(!chars)
242: chars=" \t\n"; // white space
243:
1.229 misha 244: if(fast){
245: // from left...
246: if(kind!=TRIM_END) {
247: CORD_pos pos; set_pos(pos, 0);
248: while(true) {
249: char c=CORD_pos_fetch(pos);
250: if(strchr(chars, c)) {
251: if(++start==our_length)
252: return 0; // all chars are empty, just return empty string
253: } else
254: break;
255:
256: CORD_next(pos);
257: }
258: }
259:
260: // from right..
261: if(kind!=TRIM_START) {
262: CORD_pos pos; set_pos(pos, end-1);
263: while(true) {
264: char c=CORD_pos_fetch(pos);
265: if(strchr(chars, c)) {
266: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
267: return 0; // all chars are empty, just return empty string
268: } else
269: break;
270:
271: CORD_prev(pos);
272: }
273: }
274: } else {
1.234 misha 275: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 276: const XMLByte* src_end=src_begin+our_length;
277:
278: // from left...
279: if(kind!=TRIM_END) {
280: while(src_begin<src_end){
281: uint char_length=1;
282: const XMLByte* ptr=src_begin;
1.231 misha 283: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
284: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 285: char_length++;
286:
287: bool found=false;
288: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
289: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
290: found=true;
291: break;
292: }
293:
294: if(found){
295: start+=char_length;
296: if(start==our_length)
297: return 0; // all chars are empty, just return empty string
298: } else
299: break;
300: }
301: }
1.196 paf 302:
1.229 misha 303: // from right..
304: if(kind!=TRIM_START) {
305: while(src_begin<src_end){
306: uint char_length=1;
1.231 misha 307: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
308: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 309: char_length++;
310:
311: bool found=false;
312: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
313: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
314: found=true;
315: break;
316: }
317:
318: if(found){
319: end-=char_length;
320: if(end==0)
321: return 0; // all chars are empty, just return empty string
322: } else
323: break;
324: }
1.196 paf 325: }
326: }
1.195 paf 327:
328: if(start==0 && end==our_length) // nobody moved a thing
329: return *this;
330:
331: if(out_start)
332: *out_start=start;
333: size_t new_length=end-start;
334: if(out_length)
335: *out_length=new_length;
336:
337: return mid(start, new_length);
338: }
339:
1.174 paf 340: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
341: uint& result=*static_cast<uint*>(client_data);
342: generic_hash_code(result, c);
343: return 0;
344: }
345: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
346: uint& result=*static_cast<uint*>(client_data);
347: generic_hash_code(result, s);
348: return 0;
349: };
1.225 misha 350: uint String::Body::get_hash_code() const {
351: #ifdef HASH_CODE_CACHING
352: if(hash_code)
353: return hash_code;
354: #else
355: uint hash_code=0;
356: #endif
1.220 misha 357: if (body && CORD_IS_STRING(body)){
1.225 misha 358: generic_hash_code(hash_code, body);
1.220 misha 359: } else {
360: CORD_iter5(body, 0,
361: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 362: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 363: }
1.225 misha 364: return hash_code;
1.94 parser 365: }
366:
1.241 misha 367: struct CORD_pos_info {
368: const char* chars;
369: size_t left;
370: size_t pos;
371: };
372:
373: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
374: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
375: if(info->pos < info->left){
376: info->pos=STRING_NOT_FOUND;
377: return 1;
378: }
379: if(strchr(info->chars, c))
380: return 1;
381: --(info->pos);
382: return 0;
383: }
384:
385: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
386: if(is_empty() || !chars || !strlen(chars))
387: return STRING_NOT_FOUND;
388: CORD_pos_info info={chars, left, right};
389: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
390: return info.pos;
391: else
392: return STRING_NOT_FOUND;
393: }
394:
395:
396: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
397: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
398: if(info->pos < info->left) {
399: info->pos=STRING_NOT_FOUND;
400: return 1;
401: }
402: if(!strchr(info->chars, c))
403: return 1;
404: --(info->pos);
405: return 0;
406: }
407:
408: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
409: if(is_empty() || !chars || !strlen(chars))
410: return STRING_NOT_FOUND;
411: CORD_pos_info info={chars, left, right};
412: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
413: return info.pos;
414: else
415: return STRING_NOT_FOUND;
416: }
417:
1.174 paf 418: // String methods
419:
420: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
421: if(!known_length)
1.9 paf 422: return *this;
1.122 paf 423:
1.176 paf 424: // first: langs
425: langs.append(body, lang, known_length);
426: // next: letters themselves
1.174 paf 427: body.append_know_length(str, known_length);
1.1 paf 428:
1.174 paf 429: ASSERT_STRING_INVARIANT(*this);
1.1 paf 430: return *this;
431: }
1.174 paf 432: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
433: if(!str)
434: return *this;
435: size_t known_length=helper_length?helper_length:strlen(str);
436: if(!known_length)
437: return *this;
1.1 paf 438:
1.174 paf 439: return append_know_length(str, known_length, lang);
1.5 paf 440: }
1.244 moko 441: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 442: char buf[MAX_NUMBER];
443: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
444: }
1.174 paf 445: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
446: size_t known_length=helper_length?helper_length:strlen(str);
447: if(!known_length)
448: return *this;
1.5 paf 449:
1.176 paf 450: // first: langs
451: langs.append(body, lang, known_length);
452: // next: letters themselves
1.174 paf 453: body.append_strdup_know_length(str, known_length);
1.33 paf 454:
1.174 paf 455: ASSERT_STRING_INVARIANT(*this);
456: return *this;
1.5 paf 457: }
1.46 paf 458:
1.234 misha 459: struct CORD_length_info {
460: size_t len;
461: size_t skip;
462: };
463:
464: int CORD_batched_len(const char* s, CORD_length_info* info){
465: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 466: }
467:
1.234 misha 468: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
469: int CORD_batched_len(const char c, CORD_length_info* info){
470: if (info->skip==0){
471: info->len++;
472: info->skip = lengthUTF8Char(c)-1;
473: } else {
474: info->skip--;
475: }
1.220 misha 476: return 0;
477: }
478:
1.210 misha 479: size_t String::length(Charset& charset) const {
480: if(charset.isUTF8()){
1.234 misha 481: CORD_length_info info = {0, 0};
482: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
483: return info.len;
1.210 misha 484: } else
485: return body.length();
486: }
487:
1.174 paf 488: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
489: String& String::mid(size_t substr_begin, size_t substr_end) const {
490: String& result=*new String;
491:
492: size_t self_length=length();
493: substr_begin=min(substr_begin, self_length);
494: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 495: size_t substr_length=substr_end-substr_begin;
496: if(!substr_length)
1.107 parser 497: return result;
1.53 paf 498:
1.176 paf 499: // first: their langs
500: result.langs.append(result.body, langs, substr_begin, substr_length);
501: // next: letters themselves
502: result.body=body.mid(substr_begin, substr_length);
1.174 paf 503:
504: ASSERT_STRING_INVARIANT(result);
1.53 paf 505: return result;
1.54 paf 506: }
507:
1.211 misha 508: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
509: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 510: String& result=*new String;
511:
1.211 misha 512: size_t self_length=(helper_length)?helper_length:length(charset);
513:
514: if(!self_length)
515: return result;
516:
517: from=min(min(to, from), self_length);
1.210 misha 518: to=min(max(to, from), self_length);
1.211 misha 519:
1.210 misha 520: size_t substr_length=to-from;
1.211 misha 521:
1.210 misha 522: if(!substr_length)
523: return result;
524:
525: if(charset.isUTF8()){
1.234 misha 526: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 527: const XMLByte* src_end=src_begin+body.length();
1.210 misha 528:
1.212 misha 529: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 530: from=getUTF8BytePos(src_begin, src_end, from);
531: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 532: if(!substr_length)
533: return result;
534: }
535:
536: // first: their langs
537: result.langs.append(result.body, langs, from, substr_length);
538: // next: letters themselves
539: result.body=body.mid(from, substr_length);
540:
541: ASSERT_STRING_INVARIANT(result);
542: return result;
543: }
544:
1.176 paf 545: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 546: size_t substr_length=substr.length();
547: while(true) {
548: size_t substr_begin=body.pos(substr, this_offset);
549:
550: if(substr_begin==CORD_NOT_FOUND)
551: return STRING_NOT_FOUND;
1.174 paf 552:
1.183 paf 553: if(langs.check_lang(lang, substr_begin, substr_length))
554: return substr_begin;
555:
556: this_offset=substr_begin+substr_length;
557: }
1.58 paf 558: }
559:
1.174 paf 560: size_t String::pos(const String& substr,
561: size_t this_offset, Language lang) const {
562: return pos(substr.body, this_offset, lang);
1.60 paf 563: }
564:
1.210 misha 565: size_t String::pos(Charset& charset, const String& substr,
566: size_t this_offset, Language lang) const {
567:
568: if(charset.isUTF8()){
1.234 misha 569: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 570: const XMLByte* srcEnd=srcPtr+body.length();
571:
572: // convert 'this_offset' from 'characters' to 'bytes'
573: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
574:
1.229 misha 575: size_t result=pos(substr.body, this_offset, lang);
576: return (result==CORD_NOT_FOUND)
577: ? STRING_NOT_FOUND
578: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 579: } else {
1.229 misha 580: size_t result=pos(substr.body, this_offset, lang);
581: return (result==CORD_NOT_FOUND)
582: ? STRING_NOT_FOUND
583: : result;
1.210 misha 584: }
585: }
586:
1.174 paf 587: void String::split(ArrayString& result,
588: size_t& pos_after,
589: const char* delim,
590: Language lang, int limit) const {
1.237 misha 591: if(is_empty())
592: return;
1.174 paf 593: size_t self_length=length();
1.237 misha 594: if(size_t delim_length=strlen(delim)) {
1.186 paf 595: size_t pos_before;
1.60 paf 596: // while we have 'delim'...
1.174 paf 597: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 598: result+=&mid(pos_after, pos_before);
1.174 paf 599: pos_after=pos_before+delim_length;
1.60 paf 600: }
601: // last piece
1.174 paf 602: if(pos_after<self_length && limit) {
603: result+=&mid(pos_after, self_length);
604: pos_after=self_length;
1.60 paf 605: }
606: } else { // empty delim
607: result+=this;
1.174 paf 608: pos_after+=self_length;
1.60 paf 609: }
610: }
611:
1.174 paf 612: void String::split(ArrayString& result,
613: size_t& pos_after,
614: const String& delim, Language lang,
615: int limit) const {
1.237 misha 616: if(is_empty())
617: return;
618: if(!delim.is_empty()) {
1.186 paf 619: size_t pos_before;
1.60 paf 620: // while we have 'delim'...
1.174 paf 621: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 622: result+=&mid(pos_after, pos_before);
1.174 paf 623: pos_after=pos_before+delim.length();
1.60 paf 624: }
625: // last piece
1.174 paf 626: if(pos_after<length() && limit) {
627: result+=&mid(pos_after, length());
628: pos_after=length();
1.60 paf 629: }
630: } else { // empty delim
631: result+=this;
1.174 paf 632: pos_after+=length();
1.60 paf 633: }
1.61 paf 634: }
635:
1.222 misha 636: Table* String::match(VRegex* vregex,
637: Row_action row_action, void *info,
638: int& matches_count) const {
1.209 misha 639:
1.222 misha 640: // vregex->info(); // I have no idea what does it for?
1.63 paf 641:
1.222 misha 642: bool need_pre_post_match=vregex->is_pre_post_match_needed();
643: bool global=vregex->is_global_search();
1.63 paf 644:
1.174 paf 645: const char* subject=cstr();
1.234 misha 646: size_t subject_length=length();
1.222 misha 647: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
648: int ovector[ovector_size];
1.155 paf 649:
1.173 paf 650: Table::Action_options table_options;
1.174 paf 651: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 652:
1.154 paf 653: int prestart=0;
654: int poststart=0;
1.174 paf 655: int postfinish=length();
1.63 paf 656: while(true) {
1.222 misha 657: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 658:
1.222 misha 659: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
660: break;
1.63 paf 661:
1.154 paf 662: int prefinish=ovector[0];
663: poststart=ovector[1];
1.236 moko 664:
665: if (prestart==poststart && subject[poststart]=='\n'){
666: prestart++;
667: continue;
668: }
669:
1.232 misha 670: ArrayString* row=new ArrayString(3);
1.174 paf 671: if(need_pre_post_match) {
672: *row+=&mid(0, prefinish); // .prematch column value
673: *row+=&mid(prefinish, poststart); // .match
674: *row+=&mid(poststart, postfinish); // .postmatch
675: } else {
1.185 paf 676: *row+=&Empty; // .prematch column value
677: *row+=&Empty; // .match
678: *row+=&Empty; // .postmatch
1.174 paf 679: }
1.63 paf 680:
1.222 misha 681: for(int i=1; i<exec_result; i++) {
1.69 paf 682: // -1:-1 case handled peacefully by mid() itself
1.232 misha 683: *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 684: }
685:
1.209 misha 686: matches_count++;
1.174 paf 687: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 688:
1.222 misha 689: if(!global || prestart==poststart) // last step
690: break;
691:
1.154 paf 692: prestart=poststart;
1.222 misha 693: }
1.63 paf 694:
1.222 misha 695: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
696: return vregex->is_just_count() ? 0 : &table;
1.82 parser 697: }
698:
1.174 paf 699: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
700: String& result=*new String();
701: if(is_empty())
702: return result;
703:
704: char* new_cstr=cstrm();
1.216 misha 705:
1.181 paf 706: if(source_charset.isUTF8()) {
1.216 misha 707: size_t new_cstr_len=length();
1.181 paf 708: switch(kind) {
709: case CC_UPPER:
1.192 paf 710: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 711: break;
712: case CC_LOWER:
1.192 paf 713: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 714: break;
715: default:
716: assert(!"unknown change case kind");
717: break; // never
718: }
719:
720: } else {
721: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 722:
1.181 paf 723: const unsigned char *a;
724: const unsigned char *b;
725: switch(kind) {
726: case CC_UPPER:
727: a=tables+lcc_offset;
728: b=tables+fcc_offset;
729: break;
730: case CC_LOWER:
731: a=tables+lcc_offset;
732: b=0;
733: break;
734: default:
735: assert(!"unknown change case kind");
736: a=b=0; // calm, compiler
737: break; // never
738: }
739:
1.192 paf 740: char *dest=new_cstr;
1.181 paf 741: unsigned char index;
1.190 paf 742: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 743: unsigned char c=a[index];
744: if(b)
745: c=b[c];
746:
747: *dest++=(char)c;
748: }
1.174 paf 749: }
1.176 paf 750: result.langs=langs;
1.174 paf 751: result.body=new_cstr;
1.89 parser 752:
1.101 parser 753: return result;
754: }
755:
1.213 misha 756: const String& String::escape(Charset& source_charset) const {
757: if(is_empty())
758: return *this;
759:
760: return Charset::escape(*this, source_charset);
761: }
762:
1.238 misha 763: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
764: result.langs.append(result.body, langs, langs_offset, length); \
765: result.body.append_strdup_know_length(from_cstr, length);
766:
1.174 paf 767: const String& String::replace(const Dictionary& dict) const {
1.238 misha 768: if(!dict.count() || is_empty())
769: return *this;
770:
1.174 paf 771: String& result=*new String();
772: const char* old_cstr=cstr();
773: const char* prematch_begin=old_cstr;
774:
1.238 misha 775: if(dict.count()==1) {
776: // optimized simple case
777:
778: Dictionary::Subst subst=dict.get(0);
1.239 moko 779: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 780: // prematch
1.238 misha 781: if(size_t prematch_length=p-prematch_begin) {
782: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 783: }
784:
1.174 paf 785: // match
1.238 misha 786: prematch_begin=p+subst.from_length;
1.174 paf 787:
1.184 paf 788: if(const String* b=subst.to) // are there any b?
1.174 paf 789: result<<*b;
1.238 misha 790: }
791:
792: } else {
793:
794: const char* current=old_cstr;
795: while(*current) {
796: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
797: // prematch
798: if(size_t prematch_length=current-prematch_begin) {
799: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
800: }
801:
802: // match
803: // skip 'a' in 'current'; move prematch_begin
804: current+=subst.from_length; prematch_begin=current;
805:
806: if(const String* b=subst.to) // are there any b?
807: result<<*b;
808: } else // simply advance
809: current++;
810: }
811:
1.174 paf 812: }
1.156 paf 813:
1.238 misha 814: if(prematch_begin==old_cstr) // not modified
815: return *this;
816:
1.174 paf 817: // postmatch
1.238 misha 818: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
819: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 820: }
1.156 paf 821:
1.174 paf 822: ASSERT_STRING_INVARIANT(result);
1.82 parser 823: return result;
1.61 paf 824: }
1.113 parser 825:
1.180 paf 826: static int serialize_body_char(char c, char** cur) {
827: *((*cur)++)=c;
828: return 0; // 0=continue
829: };
1.174 paf 830: static int serialize_body_piece(const char* s, char** cur) {
831: size_t length=strlen(s);
832: memcpy(*cur, s, length); *cur+=length;
1.178 paf 833: return 0; // 0=continue
1.174 paf 834: };
1.178 paf 835: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
836: // lang
1.191 paf 837: **cur=alang; (*cur)++;
838: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 839: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
840:
841: return 0; // 0=continue
842: }
1.174 paf 843: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 844: size_t fragments_count=langs.count();
1.202 paf 845: size_t body_length=body.length();
1.174 paf 846: size_t buf_length=
1.178 paf 847: prolog_length //1
848: +sizeof(size_t) //2
1.202 paf 849: +body_length //3
850: +1 // 4 for zero terminator used in deserialize
851: +sizeof(size_t) //5
852: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
853:
1.174 paf 854: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
855:
856: // 1: prolog
857: char *cur=result.str+prolog_length;
1.202 paf 858: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
859: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
860: // 3: letters
861: body.for_each(serialize_body_char, serialize_body_piece, &cur);
862: // 4: zero terminator
863: *cur++=0;
864: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 865: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 866: // 6: lang info
1.178 paf 867: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 868:
1.174 paf 869: return result;
1.113 parser 870: }
1.202 paf 871: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
872: size_t in_buf=buf_size;
873: if(in_buf<=prolog_size)
1.148 paf 874: return false;
1.202 paf 875: in_buf-=prolog_size;
1.135 paf 876:
1.174 paf 877: // 1: prolog
1.202 paf 878: const char* cur=(const char* )buf+prolog_size;
1.113 parser 879:
1.207 paf 880: // 2: chars.count
1.202 paf 881: size_t body_length;
882: if(in_buf<sizeof(body_length)) // body.length don't fit?
883: return false;
884: // [WARNING: not cast, addresses must be %4=0 on sparc]
885: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
886: in_buf-=sizeof(body_length);
887:
888: if(in_buf<body_length+1) // letters+terminator don't fit?
889: return false;
890: // 4: zero terminator
891: if(cur[body_length] != 0) // in place?
892: return false;
893: // 3: letters
1.227 misha 894: body=String::Body(*cur?cur:0);
1.229 misha 895: #ifdef STRING_LENGTH_CACHING
896: body.set_length(body_length);
897: #endif
1.202 paf 898: cur+=body_length+1;
899: in_buf-=body_length+1;
900:
901: // 5: langs.count
1.191 paf 902: size_t fragments_count;
1.202 paf 903: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 904: return false;
1.191 paf 905: // [WARNING: not cast, addresses must be %4=0 on sparc]
906: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 907: in_buf-=sizeof(fragments_count);
1.174 paf 908:
909: if(fragments_count) {
1.202 paf 910: // 6: lang info
1.174 paf 911: size_t total_length=0;
912: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 913: char lang;
914: size_t fragment_length;
915: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 916: if(in_buf<piece_length) // lang+length
1.174 paf 917: return false;
918:
1.191 paf 919: // lang
920: lang=*cur++;
921: // length [WARNING: not cast, addresses must be %4=0 on sparc]
922: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
923:
1.206 paf 924: size_t combined_length=total_length+fragment_length;
925: if(combined_length>body_length)
926: return false; // file curruption
1.191 paf 927: // uchar needed to prevent propagating 0x80 bit to upper bytes
928: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 929: total_length=combined_length;
1.202 paf 930: in_buf-=piece_length;
1.174 paf 931: }
1.128 paf 932:
1.202 paf 933: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 934: return false;
1.174 paf 935: }
1.202 paf 936: if(in_buf!=0) // some strange extra bytes
937: return false;
1.113 parser 938:
1.174 paf 939: ASSERT_STRING_INVARIANT(*this);
1.148 paf 940: return true;
1.176 paf 941: }
942:
943: const char* String::Body::v() const {
1.233 misha 944: return CORD_to_const_char_star(body, length());
1.176 paf 945: }
1.201 paf 946: void String::Body::dump() const {
947: CORD_dump(body);
948: }
949:
1.176 paf 950: const char* String::Languages::v() const {
1.177 paf 951: if(opt.is_not_just_lang)
1.233 misha 952: return CORD_to_const_char_star(langs, 0);
1.176 paf 953: else
954: return (const char*)&langs;
955: }
1.201 paf 956: void String::Languages::dump() const {
957: if(opt.is_not_just_lang)
958: CORD_dump(langs);
959: else
960: puts((const char*)&langs);
961: }
1.176 paf 962: const char* String::v() const {
1.198 paf 963: const uint LIMIT_VIEW=20;
1.246 moko 964: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 965: const char*body_view=body.v();
966: const char*langs_view=langs.v();
967: snprintf(buf, MAX_STRING,
1.178 paf 968: "%d:%.*s%s} "
1.176 paf 969: "{%d:%s",
1.178 paf 970: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 971: strlen(body_view), body_view
972: );
973:
974: return buf;
1.113 parser 975: }
1.229 misha 976:
1.201 paf 977: void String::dump() const {
978: body.dump();
979: langs.dump();
980: }
1.229 misha 981:
982: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 983: if(is_empty())
1.195 paf 984: return *this;
985:
986: size_t substr_begin, substr_length;
1.229 misha 987: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 988: if(new_body==body) // we received unchanged pointer, do likewise
989: return *this;
990: // new_body differs from body, adjust langs along
991:
992: String& result=*new String;
993: if(!new_body) // body.trim produced empty result
994: return result;
995: // body.trim produced nonempty result
996:
997: // first: their langs
998: result.langs.append(result.body, langs, substr_begin, substr_length);
999: // next: letters themselves
1000: result.body=new_body;
1001:
1002: ASSERT_STRING_INVARIANT(result);
1003: return result;
1.198 paf 1004: }
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