Annotation of parser3/src/main/pa_string.C, revision 1.252
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:
1.252 ! moko 15: #ifndef ULLONG_MAX
! 16: #define ULLONG_MAX 18446744073709551615ULL
! 17: #endif
! 18:
! 19: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.251 2015/10/09 11:42:39 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){
90: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
91: }
92:
93: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
94: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
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 {
170: ssize_t countdown;
1.204 paf 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: {
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);
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: {
187: chr_data d;
188:
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'*/);
192: }
193:
1.187 paf 194: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 195: {
196: int* result=(int*)client_data;
197: (*result)++;
198: return(0); // 0=continue
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);
205: return result;
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:
1.195 paf 236: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 237: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 238: size_t our_length=length();
239: if(!our_length)
240: return *this;
1.229 misha 241:
242: // check if any UTF-8 in chars
243: bool fast=true;
244: if(chars && source_charset && source_charset->isUTF8()){
245: const char* pos=chars;
246: while(unsigned char c=*pos++)
247: if(c>127){
248: fast=false;
249: break;
250: }
251: }
252:
253: size_t start=0;
254: size_t end=our_length;
1.195 paf 255: if(!chars)
256: chars=" \t\n"; // white space
257:
1.229 misha 258: if(fast){
259: // from left...
260: if(kind!=TRIM_END) {
261: CORD_pos pos; set_pos(pos, 0);
262: while(true) {
263: char c=CORD_pos_fetch(pos);
264: if(strchr(chars, c)) {
265: if(++start==our_length)
266: return 0; // all chars are empty, just return empty string
267: } else
268: break;
269:
270: CORD_next(pos);
271: }
272: }
273:
274: // from right..
275: if(kind!=TRIM_START) {
276: CORD_pos pos; set_pos(pos, end-1);
277: while(true) {
278: char c=CORD_pos_fetch(pos);
279: if(strchr(chars, c)) {
280: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
281: return 0; // all chars are empty, just return empty string
282: } else
283: break;
284:
285: CORD_prev(pos);
286: }
287: }
288: } else {
1.234 misha 289: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 290: const XMLByte* src_end=src_begin+our_length;
291:
292: // from left...
293: if(kind!=TRIM_END) {
294: while(src_begin<src_end){
295: uint char_length=1;
296: const XMLByte* ptr=src_begin;
1.231 misha 297: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
298: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 299: char_length++;
300:
301: bool found=false;
302: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
303: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
304: found=true;
305: break;
306: }
307:
308: if(found){
309: start+=char_length;
310: if(start==our_length)
311: return 0; // all chars are empty, just return empty string
312: } else
313: break;
314: }
315: }
1.196 paf 316:
1.229 misha 317: // from right..
318: if(kind!=TRIM_START) {
319: while(src_begin<src_end){
320: uint char_length=1;
1.231 misha 321: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
322: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 323: char_length++;
324:
325: bool found=false;
326: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
327: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
328: found=true;
329: break;
330: }
331:
332: if(found){
333: end-=char_length;
334: if(end==0)
335: return 0; // all chars are empty, just return empty string
336: } else
337: break;
338: }
1.196 paf 339: }
340: }
1.195 paf 341:
342: if(start==0 && end==our_length) // nobody moved a thing
343: return *this;
344:
345: if(out_start)
346: *out_start=start;
347: size_t new_length=end-start;
348: if(out_length)
349: *out_length=new_length;
350:
351: return mid(start, new_length);
352: }
353:
1.174 paf 354: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
355: uint& result=*static_cast<uint*>(client_data);
356: generic_hash_code(result, c);
357: return 0;
358: }
1.248 moko 359:
1.174 paf 360: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
361: uint& result=*static_cast<uint*>(client_data);
362: generic_hash_code(result, s);
363: return 0;
1.248 moko 364: }
365:
1.225 misha 366: uint String::Body::get_hash_code() const {
367: #ifdef HASH_CODE_CACHING
368: if(hash_code)
369: return hash_code;
370: #else
371: uint hash_code=0;
372: #endif
1.220 misha 373: if (body && CORD_IS_STRING(body)){
1.250 moko 374: generic_hash_code(hash_code, (const char *)body);
1.220 misha 375: } else {
376: CORD_iter5(body, 0,
377: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 378: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 379: }
1.225 misha 380: return hash_code;
1.94 parser 381: }
382:
1.241 misha 383: struct CORD_pos_info {
384: const char* chars;
385: size_t left;
386: size_t pos;
387: };
388:
389: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
390: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
391: if(info->pos < info->left){
392: info->pos=STRING_NOT_FOUND;
393: return 1;
394: }
395: if(strchr(info->chars, c))
396: return 1;
397: --(info->pos);
398: return 0;
399: }
400:
401: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
402: if(is_empty() || !chars || !strlen(chars))
403: return STRING_NOT_FOUND;
404: CORD_pos_info info={chars, left, right};
405: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
406: return info.pos;
407: else
408: return STRING_NOT_FOUND;
409: }
410:
411:
412: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
413: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
414: if(info->pos < info->left) {
415: info->pos=STRING_NOT_FOUND;
416: return 1;
417: }
418: if(!strchr(info->chars, c))
419: return 1;
420: --(info->pos);
421: return 0;
422: }
423:
424: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
425: if(is_empty() || !chars || !strlen(chars))
426: return STRING_NOT_FOUND;
427: CORD_pos_info info={chars, left, right};
428: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
429: return info.pos;
430: else
431: return STRING_NOT_FOUND;
432: }
433:
1.174 paf 434: // String methods
435:
436: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
437: if(!known_length)
1.9 paf 438: return *this;
1.122 paf 439:
1.176 paf 440: // first: langs
441: langs.append(body, lang, known_length);
442: // next: letters themselves
1.174 paf 443: body.append_know_length(str, known_length);
1.1 paf 444:
1.174 paf 445: ASSERT_STRING_INVARIANT(*this);
1.1 paf 446: return *this;
447: }
1.248 moko 448:
1.174 paf 449: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
450: if(!str)
451: return *this;
452: size_t known_length=helper_length?helper_length:strlen(str);
453: if(!known_length)
454: return *this;
1.1 paf 455:
1.174 paf 456: return append_know_length(str, known_length, lang);
1.5 paf 457: }
1.248 moko 458:
1.244 moko 459: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 460: char buf[MAX_NUMBER];
461: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
462: }
1.248 moko 463:
1.174 paf 464: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
465: size_t known_length=helper_length?helper_length:strlen(str);
466: if(!known_length)
467: return *this;
1.5 paf 468:
1.176 paf 469: // first: langs
470: langs.append(body, lang, known_length);
471: // next: letters themselves
1.174 paf 472: body.append_strdup_know_length(str, known_length);
1.33 paf 473:
1.174 paf 474: ASSERT_STRING_INVARIANT(*this);
475: return *this;
1.5 paf 476: }
1.46 paf 477:
1.234 misha 478: struct CORD_length_info {
479: size_t len;
480: size_t skip;
481: };
482:
483: int CORD_batched_len(const char* s, CORD_length_info* info){
484: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 485: }
486:
1.234 misha 487: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
488: int CORD_batched_len(const char c, CORD_length_info* info){
489: if (info->skip==0){
490: info->len++;
491: info->skip = lengthUTF8Char(c)-1;
492: } else {
493: info->skip--;
494: }
1.220 misha 495: return 0;
496: }
497:
1.210 misha 498: size_t String::length(Charset& charset) const {
499: if(charset.isUTF8()){
1.234 misha 500: CORD_length_info info = {0, 0};
501: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
502: return info.len;
1.210 misha 503: } else
504: return body.length();
505: }
506:
1.174 paf 507: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
508: String& String::mid(size_t substr_begin, size_t substr_end) const {
509: String& result=*new String;
510:
511: size_t self_length=length();
512: substr_begin=min(substr_begin, self_length);
513: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 514: size_t substr_length=substr_end-substr_begin;
515: if(!substr_length)
1.107 parser 516: return result;
1.53 paf 517:
1.176 paf 518: // first: their langs
519: result.langs.append(result.body, langs, substr_begin, substr_length);
520: // next: letters themselves
521: result.body=body.mid(substr_begin, substr_length);
1.174 paf 522:
523: ASSERT_STRING_INVARIANT(result);
1.53 paf 524: return result;
1.54 paf 525: }
526:
1.211 misha 527: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
528: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 529: String& result=*new String;
530:
1.211 misha 531: size_t self_length=(helper_length)?helper_length:length(charset);
532:
533: if(!self_length)
534: return result;
535:
536: from=min(min(to, from), self_length);
1.210 misha 537: to=min(max(to, from), self_length);
1.211 misha 538:
1.210 misha 539: size_t substr_length=to-from;
1.211 misha 540:
1.210 misha 541: if(!substr_length)
542: return result;
543:
544: if(charset.isUTF8()){
1.234 misha 545: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 546: const XMLByte* src_end=src_begin+body.length();
1.210 misha 547:
1.212 misha 548: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 549: from=getUTF8BytePos(src_begin, src_end, from);
550: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 551: if(!substr_length)
552: return result;
553: }
554:
555: // first: their langs
556: result.langs.append(result.body, langs, from, substr_length);
557: // next: letters themselves
558: result.body=body.mid(from, substr_length);
559:
560: ASSERT_STRING_INVARIANT(result);
561: return result;
562: }
563:
1.176 paf 564: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 565: size_t substr_length=substr.length();
566: while(true) {
567: size_t substr_begin=body.pos(substr, this_offset);
568:
569: if(substr_begin==CORD_NOT_FOUND)
570: return STRING_NOT_FOUND;
1.174 paf 571:
1.183 paf 572: if(langs.check_lang(lang, substr_begin, substr_length))
573: return substr_begin;
574:
575: this_offset=substr_begin+substr_length;
576: }
1.58 paf 577: }
578:
1.174 paf 579: size_t String::pos(const String& substr,
580: size_t this_offset, Language lang) const {
581: return pos(substr.body, this_offset, lang);
1.60 paf 582: }
583:
1.210 misha 584: size_t String::pos(Charset& charset, const String& substr,
585: size_t this_offset, Language lang) const {
586:
587: if(charset.isUTF8()){
1.234 misha 588: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 589: const XMLByte* srcEnd=srcPtr+body.length();
590:
591: // convert 'this_offset' from 'characters' to 'bytes'
592: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
593:
1.229 misha 594: size_t result=pos(substr.body, this_offset, lang);
595: return (result==CORD_NOT_FOUND)
596: ? STRING_NOT_FOUND
597: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 598: } else {
1.229 misha 599: size_t result=pos(substr.body, this_offset, lang);
600: return (result==CORD_NOT_FOUND)
601: ? STRING_NOT_FOUND
602: : result;
1.210 misha 603: }
604: }
605:
1.174 paf 606: void String::split(ArrayString& result,
607: size_t& pos_after,
608: const char* delim,
609: Language lang, int limit) const {
1.237 misha 610: if(is_empty())
611: return;
1.174 paf 612: size_t self_length=length();
1.237 misha 613: if(size_t delim_length=strlen(delim)) {
1.186 paf 614: size_t pos_before;
1.60 paf 615: // while we have 'delim'...
1.174 paf 616: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 617: result+=&mid(pos_after, pos_before);
1.174 paf 618: pos_after=pos_before+delim_length;
1.60 paf 619: }
620: // last piece
1.174 paf 621: if(pos_after<self_length && limit) {
622: result+=&mid(pos_after, self_length);
623: pos_after=self_length;
1.60 paf 624: }
625: } else { // empty delim
626: result+=this;
1.174 paf 627: pos_after+=self_length;
1.60 paf 628: }
629: }
630:
1.174 paf 631: void String::split(ArrayString& result,
632: size_t& pos_after,
633: const String& delim, Language lang,
634: int limit) const {
1.237 misha 635: if(is_empty())
636: return;
637: if(!delim.is_empty()) {
1.186 paf 638: size_t pos_before;
1.60 paf 639: // while we have 'delim'...
1.174 paf 640: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 641: result+=&mid(pos_after, pos_before);
1.174 paf 642: pos_after=pos_before+delim.length();
1.60 paf 643: }
644: // last piece
1.174 paf 645: if(pos_after<length() && limit) {
646: result+=&mid(pos_after, length());
647: pos_after=length();
1.60 paf 648: }
649: } else { // empty delim
650: result+=this;
1.174 paf 651: pos_after+=length();
1.60 paf 652: }
1.61 paf 653: }
654:
1.222 misha 655: Table* String::match(VRegex* vregex,
656: Row_action row_action, void *info,
657: int& matches_count) const {
1.209 misha 658:
1.222 misha 659: // vregex->info(); // I have no idea what does it for?
1.63 paf 660:
1.222 misha 661: bool need_pre_post_match=vregex->is_pre_post_match_needed();
662: bool global=vregex->is_global_search();
1.63 paf 663:
1.174 paf 664: const char* subject=cstr();
1.234 misha 665: size_t subject_length=length();
1.222 misha 666: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
667: int ovector[ovector_size];
1.155 paf 668:
1.173 paf 669: Table::Action_options table_options;
1.174 paf 670: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 671:
1.154 paf 672: int prestart=0;
673: int poststart=0;
1.174 paf 674: int postfinish=length();
1.63 paf 675: while(true) {
1.222 misha 676: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 677:
1.222 misha 678: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
679: break;
1.63 paf 680:
1.154 paf 681: int prefinish=ovector[0];
682: poststart=ovector[1];
1.236 moko 683:
684: if (prestart==poststart && subject[poststart]=='\n'){
685: prestart++;
686: continue;
687: }
688:
1.232 misha 689: ArrayString* row=new ArrayString(3);
1.174 paf 690: if(need_pre_post_match) {
691: *row+=&mid(0, prefinish); // .prematch column value
692: *row+=&mid(prefinish, poststart); // .match
693: *row+=&mid(poststart, postfinish); // .postmatch
694: } else {
1.185 paf 695: *row+=&Empty; // .prematch column value
696: *row+=&Empty; // .match
697: *row+=&Empty; // .postmatch
1.174 paf 698: }
1.63 paf 699:
1.222 misha 700: for(int i=1; i<exec_result; i++) {
1.69 paf 701: // -1:-1 case handled peacefully by mid() itself
1.232 misha 702: *row+=(ovector[i*2+0]>=0 && ovector[i*2+1]>0)?&mid(ovector[i*2+0], ovector[i*2+1]):new String; // .i column value
1.63 paf 703: }
704:
1.209 misha 705: matches_count++;
1.174 paf 706: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 707:
1.222 misha 708: if(!global || prestart==poststart) // last step
709: break;
710:
1.154 paf 711: prestart=poststart;
1.222 misha 712: }
1.63 paf 713:
1.222 misha 714: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
715: return vregex->is_just_count() ? 0 : &table;
1.82 parser 716: }
717:
1.174 paf 718: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
719: String& result=*new String();
720: if(is_empty())
721: return result;
722:
723: char* new_cstr=cstrm();
1.216 misha 724:
1.181 paf 725: if(source_charset.isUTF8()) {
1.216 misha 726: size_t new_cstr_len=length();
1.181 paf 727: switch(kind) {
728: case CC_UPPER:
1.192 paf 729: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 730: break;
731: case CC_LOWER:
1.192 paf 732: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 733: break;
734: default:
735: assert(!"unknown change case kind");
736: break; // never
737: }
738:
739: } else {
740: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 741:
1.181 paf 742: const unsigned char *a;
743: const unsigned char *b;
744: switch(kind) {
745: case CC_UPPER:
746: a=tables+lcc_offset;
747: b=tables+fcc_offset;
748: break;
749: case CC_LOWER:
750: a=tables+lcc_offset;
751: b=0;
752: break;
753: default:
754: assert(!"unknown change case kind");
755: a=b=0; // calm, compiler
756: break; // never
757: }
758:
1.192 paf 759: char *dest=new_cstr;
1.181 paf 760: unsigned char index;
1.190 paf 761: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 762: unsigned char c=a[index];
763: if(b)
764: c=b[c];
765:
766: *dest++=(char)c;
767: }
1.174 paf 768: }
1.176 paf 769: result.langs=langs;
1.174 paf 770: result.body=new_cstr;
1.89 parser 771:
1.101 parser 772: return result;
773: }
774:
1.213 misha 775: const String& String::escape(Charset& source_charset) const {
776: if(is_empty())
777: return *this;
778:
779: return Charset::escape(*this, source_charset);
780: }
781:
1.238 misha 782: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
783: result.langs.append(result.body, langs, langs_offset, length); \
784: result.body.append_strdup_know_length(from_cstr, length);
785:
1.174 paf 786: const String& String::replace(const Dictionary& dict) const {
1.238 misha 787: if(!dict.count() || is_empty())
788: return *this;
789:
1.174 paf 790: String& result=*new String();
791: const char* old_cstr=cstr();
792: const char* prematch_begin=old_cstr;
793:
1.238 misha 794: if(dict.count()==1) {
795: // optimized simple case
796:
797: Dictionary::Subst subst=dict.get(0);
1.239 moko 798: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 799: // prematch
1.238 misha 800: if(size_t prematch_length=p-prematch_begin) {
801: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 802: }
803:
1.174 paf 804: // match
1.238 misha 805: prematch_begin=p+subst.from_length;
1.174 paf 806:
1.184 paf 807: if(const String* b=subst.to) // are there any b?
1.174 paf 808: result<<*b;
1.238 misha 809: }
810:
811: } else {
812:
813: const char* current=old_cstr;
814: while(*current) {
815: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
816: // prematch
817: if(size_t prematch_length=current-prematch_begin) {
818: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
819: }
820:
821: // match
822: // skip 'a' in 'current'; move prematch_begin
823: current+=subst.from_length; prematch_begin=current;
824:
825: if(const String* b=subst.to) // are there any b?
826: result<<*b;
827: } else // simply advance
828: current++;
829: }
830:
1.174 paf 831: }
1.156 paf 832:
1.238 misha 833: if(prematch_begin==old_cstr) // not modified
834: return *this;
835:
1.174 paf 836: // postmatch
1.238 misha 837: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
838: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 839: }
1.156 paf 840:
1.174 paf 841: ASSERT_STRING_INVARIANT(result);
1.82 parser 842: return result;
1.61 paf 843: }
1.113 parser 844:
1.180 paf 845: static int serialize_body_char(char c, char** cur) {
846: *((*cur)++)=c;
847: return 0; // 0=continue
1.248 moko 848: }
849:
1.174 paf 850: static int serialize_body_piece(const char* s, char** cur) {
851: size_t length=strlen(s);
852: memcpy(*cur, s, length); *cur+=length;
1.178 paf 853: return 0; // 0=continue
1.248 moko 854: }
855:
1.178 paf 856: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
857: // lang
1.191 paf 858: **cur=alang; (*cur)++;
859: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 860: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
861:
862: return 0; // 0=continue
863: }
1.248 moko 864:
1.174 paf 865: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 866: size_t fragments_count=langs.count();
1.202 paf 867: size_t body_length=body.length();
1.174 paf 868: size_t buf_length=
1.178 paf 869: prolog_length //1
870: +sizeof(size_t) //2
1.202 paf 871: +body_length //3
872: +1 // 4 for zero terminator used in deserialize
873: +sizeof(size_t) //5
874: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
875:
1.174 paf 876: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
877:
878: // 1: prolog
879: char *cur=result.str+prolog_length;
1.202 paf 880: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
881: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
882: // 3: letters
883: body.for_each(serialize_body_char, serialize_body_piece, &cur);
884: // 4: zero terminator
885: *cur++=0;
886: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 887: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 888: // 6: lang info
1.178 paf 889: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 890:
1.174 paf 891: return result;
1.113 parser 892: }
1.248 moko 893:
1.202 paf 894: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
895: size_t in_buf=buf_size;
896: if(in_buf<=prolog_size)
1.148 paf 897: return false;
1.202 paf 898: in_buf-=prolog_size;
1.135 paf 899:
1.174 paf 900: // 1: prolog
1.202 paf 901: const char* cur=(const char* )buf+prolog_size;
1.113 parser 902:
1.207 paf 903: // 2: chars.count
1.202 paf 904: size_t body_length;
905: if(in_buf<sizeof(body_length)) // body.length don't fit?
906: return false;
907: // [WARNING: not cast, addresses must be %4=0 on sparc]
908: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
909: in_buf-=sizeof(body_length);
910:
911: if(in_buf<body_length+1) // letters+terminator don't fit?
912: return false;
913: // 4: zero terminator
914: if(cur[body_length] != 0) // in place?
915: return false;
916: // 3: letters
1.251 moko 917: body=String::Body(String::C(cur, body_length));
1.202 paf 918: cur+=body_length+1;
919: in_buf-=body_length+1;
920:
921: // 5: langs.count
1.191 paf 922: size_t fragments_count;
1.202 paf 923: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 924: return false;
1.191 paf 925: // [WARNING: not cast, addresses must be %4=0 on sparc]
926: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 927: in_buf-=sizeof(fragments_count);
1.174 paf 928:
929: if(fragments_count) {
1.202 paf 930: // 6: lang info
1.174 paf 931: size_t total_length=0;
932: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 933: char lang;
934: size_t fragment_length;
935: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 936: if(in_buf<piece_length) // lang+length
1.174 paf 937: return false;
938:
1.191 paf 939: // lang
940: lang=*cur++;
941: // length [WARNING: not cast, addresses must be %4=0 on sparc]
942: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
943:
1.206 paf 944: size_t combined_length=total_length+fragment_length;
945: if(combined_length>body_length)
946: return false; // file curruption
1.191 paf 947: // uchar needed to prevent propagating 0x80 bit to upper bytes
948: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 949: total_length=combined_length;
1.202 paf 950: in_buf-=piece_length;
1.174 paf 951: }
1.128 paf 952:
1.202 paf 953: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 954: return false;
1.174 paf 955: }
1.202 paf 956: if(in_buf!=0) // some strange extra bytes
957: return false;
1.113 parser 958:
1.174 paf 959: ASSERT_STRING_INVARIANT(*this);
1.148 paf 960: return true;
1.176 paf 961: }
962:
963: const char* String::Body::v() const {
1.233 misha 964: return CORD_to_const_char_star(body, length());
1.176 paf 965: }
1.248 moko 966:
1.201 paf 967: void String::Body::dump() const {
968: CORD_dump(body);
969: }
970:
1.176 paf 971: const char* String::Languages::v() const {
1.177 paf 972: if(opt.is_not_just_lang)
1.233 misha 973: return CORD_to_const_char_star(langs, 0);
1.176 paf 974: else
975: return (const char*)&langs;
976: }
1.248 moko 977:
1.201 paf 978: void String::Languages::dump() const {
979: if(opt.is_not_just_lang)
980: CORD_dump(langs);
981: else
982: puts((const char*)&langs);
983: }
1.248 moko 984:
1.176 paf 985: const char* String::v() const {
1.198 paf 986: const uint LIMIT_VIEW=20;
1.246 moko 987: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 988: const char*body_view=body.v();
989: const char*langs_view=langs.v();
990: snprintf(buf, MAX_STRING,
1.178 paf 991: "%d:%.*s%s} "
1.176 paf 992: "{%d:%s",
1.178 paf 993: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 994: strlen(body_view), body_view
995: );
996:
997: return buf;
1.113 parser 998: }
1.229 misha 999:
1.201 paf 1000: void String::dump() const {
1001: body.dump();
1002: langs.dump();
1003: }
1.229 misha 1004:
1005: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 1006: if(is_empty())
1.195 paf 1007: return *this;
1008:
1009: size_t substr_begin, substr_length;
1.229 misha 1010: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1011: if(new_body==body) // we received unchanged pointer, do likewise
1012: return *this;
1013: // new_body differs from body, adjust langs along
1014:
1015: String& result=*new String;
1016: if(!new_body) // body.trim produced empty result
1017: return result;
1018: // body.trim produced nonempty result
1019:
1020: // first: their langs
1021: result.langs.append(result.body, langs, substr_begin, substr_length);
1022: // next: letters themselves
1023: result.body=new_body;
1024:
1025: ASSERT_STRING_INVARIANT(result);
1026: return result;
1.198 paf 1027: }
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