Annotation of parser3/src/main/pa_string.C, revision 1.245
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:
1.245 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.244 2013/07/16 15:21:06 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;
1.193 paf 134: if(str[0]=='0')
1.242 moko 135: if(str[1]=='x' || str[1]=='X'){
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: }
1.242 moko 143:
144: char *error_pos;
145: result=strtod(str, &error_pos);
1.193 paf 146:
147: while(char c=*error_pos++)
1.199 paf 148: if(!isspace((unsigned char)c))
1.242 moko 149: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 150:
1.242 moko 151: return negative ? -result : result;
1.193 paf 152: }
153:
1.176 paf 154: // cord lib extension
155:
156: #ifndef DOXYGEN
157: typedef struct {
158: ssize_t countdown;
1.204 paf 159: int target; /* Character we're looking for */
1.176 paf 160: } chr_data;
161: #endif
162: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
163: {
164: register chr_data * d = (chr_data *)client_data;
165:
166: if (d -> countdown<=0) return(2);
167: d -> countdown -= size;
168: if (c > d -> target) return(1);
169: return(0);
170: }
171: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
172: {
173: chr_data d;
174:
175: d.countdown = n;
176: d.target = c;
177: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
178: }
179:
1.187 paf 180: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 181: {
182: int* result=(int*)client_data;
183: (*result)++;
184: return(0); // 0=continue
185: }
186: size_t CORD_block_count(CORD x)
187: {
188: size_t result=0;
189: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
190: return result;
191: }
192:
1.174 paf 193: // helpers
1.139 paf 194:
1.174 paf 195: /// String::match uses this as replace & global search table columns
1.139 paf 196:
1.174 paf 197: const int MAX_MATCH_GROUPS=100;
1.139 paf 198:
1.174 paf 199: class String_match_table_template_columns: public ArrayString {
200: public:
201: String_match_table_template_columns() {
202: *this+=new String("prematch");
203: *this+=new String("match");
204: *this+=new String("postmatch");
205: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 206: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 207: }
208: }
209: };
210:
211: Table string_match_table_template(new String_match_table_template_columns);
212:
1.176 paf 213: // String::Body methods
1.140 paf 214:
1.176 paf 215: String::Body String::Body::Format(int value) {
1.174 paf 216: char local[MAX_NUMBER];
217: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 218: return String::Body(pa_strdup(local, length));
1.120 paf 219: }
220:
1.195 paf 221: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 222: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 223: size_t our_length=length();
224: if(!our_length)
225: return *this;
1.229 misha 226:
227: // check if any UTF-8 in chars
228: bool fast=true;
229: if(chars && source_charset && source_charset->isUTF8()){
230: const char* pos=chars;
231: while(unsigned char c=*pos++)
232: if(c>127){
233: fast=false;
234: break;
235: }
236: }
237:
238: size_t start=0;
239: size_t end=our_length;
1.195 paf 240: if(!chars)
241: chars=" \t\n"; // white space
242:
1.229 misha 243: if(fast){
244: // from left...
245: if(kind!=TRIM_END) {
246: CORD_pos pos; set_pos(pos, 0);
247: while(true) {
248: char c=CORD_pos_fetch(pos);
249: if(strchr(chars, c)) {
250: if(++start==our_length)
251: return 0; // all chars are empty, just return empty string
252: } else
253: break;
254:
255: CORD_next(pos);
256: }
257: }
258:
259: // from right..
260: if(kind!=TRIM_START) {
261: CORD_pos pos; set_pos(pos, end-1);
262: while(true) {
263: char c=CORD_pos_fetch(pos);
264: if(strchr(chars, c)) {
265: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
266: return 0; // all chars are empty, just return empty string
267: } else
268: break;
269:
270: CORD_prev(pos);
271: }
272: }
273: } else {
1.234 misha 274: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 275: const XMLByte* src_end=src_begin+our_length;
276:
277: // from left...
278: if(kind!=TRIM_END) {
279: while(src_begin<src_end){
280: uint char_length=1;
281: const XMLByte* ptr=src_begin;
1.231 misha 282: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
283: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 284: char_length++;
285:
286: bool found=false;
287: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
288: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
289: found=true;
290: break;
291: }
292:
293: if(found){
294: start+=char_length;
295: if(start==our_length)
296: return 0; // all chars are empty, just return empty string
297: } else
298: break;
299: }
300: }
1.196 paf 301:
1.229 misha 302: // from right..
303: if(kind!=TRIM_START) {
304: while(src_begin<src_end){
305: uint char_length=1;
1.231 misha 306: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
307: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 308: char_length++;
309:
310: bool found=false;
311: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
312: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
313: found=true;
314: break;
315: }
316:
317: if(found){
318: end-=char_length;
319: if(end==0)
320: return 0; // all chars are empty, just return empty string
321: } else
322: break;
323: }
1.196 paf 324: }
325: }
1.195 paf 326:
327: if(start==0 && end==our_length) // nobody moved a thing
328: return *this;
329:
330: if(out_start)
331: *out_start=start;
332: size_t new_length=end-start;
333: if(out_length)
334: *out_length=new_length;
335:
336: return mid(start, new_length);
337: }
338:
1.174 paf 339: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
340: uint& result=*static_cast<uint*>(client_data);
341: generic_hash_code(result, c);
342: return 0;
343: }
344: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
345: uint& result=*static_cast<uint*>(client_data);
346: generic_hash_code(result, s);
347: return 0;
348: };
1.225 misha 349: uint String::Body::get_hash_code() const {
350: #ifdef HASH_CODE_CACHING
351: if(hash_code)
352: return hash_code;
353: #else
354: uint hash_code=0;
355: #endif
1.220 misha 356: if (body && CORD_IS_STRING(body)){
1.225 misha 357: generic_hash_code(hash_code, body);
1.220 misha 358: } else {
359: CORD_iter5(body, 0,
360: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 361: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 362: }
1.225 misha 363: return hash_code;
1.94 parser 364: }
365:
1.241 misha 366: struct CORD_pos_info {
367: const char* chars;
368: size_t left;
369: size_t pos;
370: };
371:
372: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
373: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
374: if(info->pos < info->left){
375: info->pos=STRING_NOT_FOUND;
376: return 1;
377: }
378: if(strchr(info->chars, c))
379: return 1;
380: --(info->pos);
381: return 0;
382: }
383:
384: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
385: if(is_empty() || !chars || !strlen(chars))
386: return STRING_NOT_FOUND;
387: CORD_pos_info info={chars, left, right};
388: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
389: return info.pos;
390: else
391: return STRING_NOT_FOUND;
392: }
393:
394:
395: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
396: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
397: if(info->pos < info->left) {
398: info->pos=STRING_NOT_FOUND;
399: return 1;
400: }
401: if(!strchr(info->chars, c))
402: return 1;
403: --(info->pos);
404: return 0;
405: }
406:
407: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
408: if(is_empty() || !chars || !strlen(chars))
409: return STRING_NOT_FOUND;
410: CORD_pos_info info={chars, left, right};
411: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
412: return info.pos;
413: else
414: return STRING_NOT_FOUND;
415: }
416:
1.174 paf 417: // String methods
418:
419: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
420: if(!known_length)
1.9 paf 421: return *this;
1.122 paf 422:
1.176 paf 423: // first: langs
424: langs.append(body, lang, known_length);
425: // next: letters themselves
1.174 paf 426: body.append_know_length(str, known_length);
1.1 paf 427:
1.174 paf 428: ASSERT_STRING_INVARIANT(*this);
1.1 paf 429: return *this;
430: }
1.174 paf 431: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
432: if(!str)
433: return *this;
434: size_t known_length=helper_length?helper_length:strlen(str);
435: if(!known_length)
436: return *this;
1.1 paf 437:
1.174 paf 438: return append_know_length(str, known_length, lang);
1.5 paf 439: }
1.244 moko 440: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 441: char buf[MAX_NUMBER];
442: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
443: }
1.174 paf 444: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
445: size_t known_length=helper_length?helper_length:strlen(str);
446: if(!known_length)
447: return *this;
1.5 paf 448:
1.176 paf 449: // first: langs
450: langs.append(body, lang, known_length);
451: // next: letters themselves
1.174 paf 452: body.append_strdup_know_length(str, known_length);
1.33 paf 453:
1.174 paf 454: ASSERT_STRING_INVARIANT(*this);
455: return *this;
1.5 paf 456: }
1.46 paf 457:
1.234 misha 458: struct CORD_length_info {
459: size_t len;
460: size_t skip;
461: };
462:
463: int CORD_batched_len(const char* s, CORD_length_info* info){
464: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 465: }
466:
1.234 misha 467: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
468: int CORD_batched_len(const char c, CORD_length_info* info){
469: if (info->skip==0){
470: info->len++;
471: info->skip = lengthUTF8Char(c)-1;
472: } else {
473: info->skip--;
474: }
1.220 misha 475: return 0;
476: }
477:
1.210 misha 478: size_t String::length(Charset& charset) const {
479: if(charset.isUTF8()){
1.234 misha 480: CORD_length_info info = {0, 0};
481: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
482: return info.len;
1.210 misha 483: } else
484: return body.length();
485: }
486:
1.174 paf 487: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
488: String& String::mid(size_t substr_begin, size_t substr_end) const {
489: String& result=*new String;
490:
491: size_t self_length=length();
492: substr_begin=min(substr_begin, self_length);
493: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 494: size_t substr_length=substr_end-substr_begin;
495: if(!substr_length)
1.107 parser 496: return result;
1.53 paf 497:
1.176 paf 498: // first: their langs
499: result.langs.append(result.body, langs, substr_begin, substr_length);
500: // next: letters themselves
501: result.body=body.mid(substr_begin, substr_length);
1.174 paf 502:
503: ASSERT_STRING_INVARIANT(result);
1.53 paf 504: return result;
1.54 paf 505: }
506:
1.211 misha 507: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
508: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 509: String& result=*new String;
510:
1.211 misha 511: size_t self_length=(helper_length)?helper_length:length(charset);
512:
513: if(!self_length)
514: return result;
515:
516: from=min(min(to, from), self_length);
1.210 misha 517: to=min(max(to, from), self_length);
1.211 misha 518:
1.210 misha 519: size_t substr_length=to-from;
1.211 misha 520:
1.210 misha 521: if(!substr_length)
522: return result;
523:
524: if(charset.isUTF8()){
1.234 misha 525: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 526: const XMLByte* src_end=src_begin+body.length();
1.210 misha 527:
1.212 misha 528: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 529: from=getUTF8BytePos(src_begin, src_end, from);
530: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 531: if(!substr_length)
532: return result;
533: }
534:
535: // first: their langs
536: result.langs.append(result.body, langs, from, substr_length);
537: // next: letters themselves
538: result.body=body.mid(from, substr_length);
539:
540: ASSERT_STRING_INVARIANT(result);
541: return result;
542: }
543:
1.176 paf 544: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 545: size_t substr_length=substr.length();
546: while(true) {
547: size_t substr_begin=body.pos(substr, this_offset);
548:
549: if(substr_begin==CORD_NOT_FOUND)
550: return STRING_NOT_FOUND;
1.174 paf 551:
1.183 paf 552: if(langs.check_lang(lang, substr_begin, substr_length))
553: return substr_begin;
554:
555: this_offset=substr_begin+substr_length;
556: }
1.58 paf 557: }
558:
1.174 paf 559: size_t String::pos(const String& substr,
560: size_t this_offset, Language lang) const {
561: return pos(substr.body, this_offset, lang);
1.60 paf 562: }
563:
1.210 misha 564: size_t String::pos(Charset& charset, const String& substr,
565: size_t this_offset, Language lang) const {
566:
567: if(charset.isUTF8()){
1.234 misha 568: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 569: const XMLByte* srcEnd=srcPtr+body.length();
570:
571: // convert 'this_offset' from 'characters' to 'bytes'
572: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
573:
1.229 misha 574: size_t result=pos(substr.body, this_offset, lang);
575: return (result==CORD_NOT_FOUND)
576: ? STRING_NOT_FOUND
577: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 578: } else {
1.229 misha 579: size_t result=pos(substr.body, this_offset, lang);
580: return (result==CORD_NOT_FOUND)
581: ? STRING_NOT_FOUND
582: : result;
1.210 misha 583: }
584: }
585:
1.174 paf 586: void String::split(ArrayString& result,
587: size_t& pos_after,
588: const char* delim,
589: Language lang, int limit) const {
1.237 misha 590: if(is_empty())
591: return;
1.174 paf 592: size_t self_length=length();
1.237 misha 593: if(size_t delim_length=strlen(delim)) {
1.186 paf 594: size_t pos_before;
1.60 paf 595: // while we have 'delim'...
1.174 paf 596: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 597: result+=&mid(pos_after, pos_before);
1.174 paf 598: pos_after=pos_before+delim_length;
1.60 paf 599: }
600: // last piece
1.174 paf 601: if(pos_after<self_length && limit) {
602: result+=&mid(pos_after, self_length);
603: pos_after=self_length;
1.60 paf 604: }
605: } else { // empty delim
606: result+=this;
1.174 paf 607: pos_after+=self_length;
1.60 paf 608: }
609: }
610:
1.174 paf 611: void String::split(ArrayString& result,
612: size_t& pos_after,
613: const String& delim, Language lang,
614: int limit) const {
1.237 misha 615: if(is_empty())
616: return;
617: if(!delim.is_empty()) {
1.186 paf 618: size_t pos_before;
1.60 paf 619: // while we have 'delim'...
1.174 paf 620: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 621: result+=&mid(pos_after, pos_before);
1.174 paf 622: pos_after=pos_before+delim.length();
1.60 paf 623: }
624: // last piece
1.174 paf 625: if(pos_after<length() && limit) {
626: result+=&mid(pos_after, length());
627: pos_after=length();
1.60 paf 628: }
629: } else { // empty delim
630: result+=this;
1.174 paf 631: pos_after+=length();
1.60 paf 632: }
1.61 paf 633: }
634:
1.222 misha 635: Table* String::match(VRegex* vregex,
636: Row_action row_action, void *info,
637: int& matches_count) const {
1.209 misha 638:
1.222 misha 639: // vregex->info(); // I have no idea what does it for?
1.63 paf 640:
1.222 misha 641: bool need_pre_post_match=vregex->is_pre_post_match_needed();
642: bool global=vregex->is_global_search();
1.63 paf 643:
1.174 paf 644: const char* subject=cstr();
1.234 misha 645: size_t subject_length=length();
1.222 misha 646: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
647: int ovector[ovector_size];
1.155 paf 648:
1.173 paf 649: Table::Action_options table_options;
1.174 paf 650: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 651:
1.154 paf 652: int prestart=0;
653: int poststart=0;
1.174 paf 654: int postfinish=length();
1.63 paf 655: while(true) {
1.222 misha 656: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 657:
1.222 misha 658: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
659: break;
1.63 paf 660:
1.154 paf 661: int prefinish=ovector[0];
662: poststart=ovector[1];
1.236 moko 663:
664: if (prestart==poststart && subject[poststart]=='\n'){
665: prestart++;
666: continue;
667: }
668:
1.232 misha 669: ArrayString* row=new ArrayString(3);
1.174 paf 670: if(need_pre_post_match) {
671: *row+=&mid(0, prefinish); // .prematch column value
672: *row+=&mid(prefinish, poststart); // .match
673: *row+=&mid(poststart, postfinish); // .postmatch
674: } else {
1.185 paf 675: *row+=&Empty; // .prematch column value
676: *row+=&Empty; // .match
677: *row+=&Empty; // .postmatch
1.174 paf 678: }
1.63 paf 679:
1.222 misha 680: for(int i=1; i<exec_result; i++) {
1.69 paf 681: // -1:-1 case handled peacefully by mid() itself
1.232 misha 682: *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 683: }
684:
1.209 misha 685: matches_count++;
1.174 paf 686: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 687:
1.222 misha 688: if(!global || prestart==poststart) // last step
689: break;
690:
1.154 paf 691: prestart=poststart;
1.222 misha 692: }
1.63 paf 693:
1.222 misha 694: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
695: return vregex->is_just_count() ? 0 : &table;
1.82 parser 696: }
697:
1.174 paf 698: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
699: String& result=*new String();
700: if(is_empty())
701: return result;
702:
703: char* new_cstr=cstrm();
1.216 misha 704:
1.181 paf 705: if(source_charset.isUTF8()) {
1.216 misha 706: size_t new_cstr_len=length();
1.181 paf 707: switch(kind) {
708: case CC_UPPER:
1.192 paf 709: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 710: break;
711: case CC_LOWER:
1.192 paf 712: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 713: break;
714: default:
715: assert(!"unknown change case kind");
716: break; // never
717: }
718:
719: } else {
720: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 721:
1.181 paf 722: const unsigned char *a;
723: const unsigned char *b;
724: switch(kind) {
725: case CC_UPPER:
726: a=tables+lcc_offset;
727: b=tables+fcc_offset;
728: break;
729: case CC_LOWER:
730: a=tables+lcc_offset;
731: b=0;
732: break;
733: default:
734: assert(!"unknown change case kind");
735: a=b=0; // calm, compiler
736: break; // never
737: }
738:
1.192 paf 739: char *dest=new_cstr;
1.181 paf 740: unsigned char index;
1.190 paf 741: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 742: unsigned char c=a[index];
743: if(b)
744: c=b[c];
745:
746: *dest++=(char)c;
747: }
1.174 paf 748: }
1.176 paf 749: result.langs=langs;
1.174 paf 750: result.body=new_cstr;
1.89 parser 751:
1.101 parser 752: return result;
753: }
754:
1.213 misha 755: const String& String::escape(Charset& source_charset) const {
756: if(is_empty())
757: return *this;
758:
759: return Charset::escape(*this, source_charset);
760: }
761:
1.238 misha 762: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
763: result.langs.append(result.body, langs, langs_offset, length); \
764: result.body.append_strdup_know_length(from_cstr, length);
765:
1.174 paf 766: const String& String::replace(const Dictionary& dict) const {
1.238 misha 767: if(!dict.count() || is_empty())
768: return *this;
769:
1.174 paf 770: String& result=*new String();
771: const char* old_cstr=cstr();
772: const char* prematch_begin=old_cstr;
773:
1.238 misha 774: if(dict.count()==1) {
775: // optimized simple case
776:
777: Dictionary::Subst subst=dict.get(0);
1.239 moko 778: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 779: // prematch
1.238 misha 780: if(size_t prematch_length=p-prematch_begin) {
781: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 782: }
783:
1.174 paf 784: // match
1.238 misha 785: prematch_begin=p+subst.from_length;
1.174 paf 786:
1.184 paf 787: if(const String* b=subst.to) // are there any b?
1.174 paf 788: result<<*b;
1.238 misha 789: }
790:
791: } else {
792:
793: const char* current=old_cstr;
794: while(*current) {
795: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
796: // prematch
797: if(size_t prematch_length=current-prematch_begin) {
798: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
799: }
800:
801: // match
802: // skip 'a' in 'current'; move prematch_begin
803: current+=subst.from_length; prematch_begin=current;
804:
805: if(const String* b=subst.to) // are there any b?
806: result<<*b;
807: } else // simply advance
808: current++;
809: }
810:
1.174 paf 811: }
1.156 paf 812:
1.238 misha 813: if(prematch_begin==old_cstr) // not modified
814: return *this;
815:
1.174 paf 816: // postmatch
1.238 misha 817: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
818: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 819: }
1.156 paf 820:
1.174 paf 821: ASSERT_STRING_INVARIANT(result);
1.82 parser 822: return result;
1.61 paf 823: }
1.113 parser 824:
1.180 paf 825: static int serialize_body_char(char c, char** cur) {
826: *((*cur)++)=c;
827: return 0; // 0=continue
828: };
1.174 paf 829: static int serialize_body_piece(const char* s, char** cur) {
830: size_t length=strlen(s);
831: memcpy(*cur, s, length); *cur+=length;
1.178 paf 832: return 0; // 0=continue
1.174 paf 833: };
1.178 paf 834: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
835: // lang
1.191 paf 836: **cur=alang; (*cur)++;
837: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 838: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
839:
840: return 0; // 0=continue
841: }
1.174 paf 842: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 843: size_t fragments_count=langs.count();
1.202 paf 844: size_t body_length=body.length();
1.174 paf 845: size_t buf_length=
1.178 paf 846: prolog_length //1
847: +sizeof(size_t) //2
1.202 paf 848: +body_length //3
849: +1 // 4 for zero terminator used in deserialize
850: +sizeof(size_t) //5
851: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
852:
1.174 paf 853: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
854:
855: // 1: prolog
856: char *cur=result.str+prolog_length;
1.202 paf 857: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
858: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
859: // 3: letters
860: body.for_each(serialize_body_char, serialize_body_piece, &cur);
861: // 4: zero terminator
862: *cur++=0;
863: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 864: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 865: // 6: lang info
1.178 paf 866: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 867:
1.174 paf 868: return result;
1.113 parser 869: }
1.202 paf 870: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
871: size_t in_buf=buf_size;
872: if(in_buf<=prolog_size)
1.148 paf 873: return false;
1.202 paf 874: in_buf-=prolog_size;
1.135 paf 875:
1.174 paf 876: // 1: prolog
1.202 paf 877: const char* cur=(const char* )buf+prolog_size;
1.113 parser 878:
1.207 paf 879: // 2: chars.count
1.202 paf 880: size_t body_length;
881: if(in_buf<sizeof(body_length)) // body.length don't fit?
882: return false;
883: // [WARNING: not cast, addresses must be %4=0 on sparc]
884: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
885: in_buf-=sizeof(body_length);
886:
887: if(in_buf<body_length+1) // letters+terminator don't fit?
888: return false;
889: // 4: zero terminator
890: if(cur[body_length] != 0) // in place?
891: return false;
892: // 3: letters
1.227 misha 893: body=String::Body(*cur?cur:0);
1.229 misha 894: #ifdef STRING_LENGTH_CACHING
895: body.set_length(body_length);
896: #endif
1.202 paf 897: cur+=body_length+1;
898: in_buf-=body_length+1;
899:
900: // 5: langs.count
1.191 paf 901: size_t fragments_count;
1.202 paf 902: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 903: return false;
1.191 paf 904: // [WARNING: not cast, addresses must be %4=0 on sparc]
905: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 906: in_buf-=sizeof(fragments_count);
1.174 paf 907:
908: if(fragments_count) {
1.202 paf 909: // 6: lang info
1.174 paf 910: size_t total_length=0;
911: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 912: char lang;
913: size_t fragment_length;
914: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 915: if(in_buf<piece_length) // lang+length
1.174 paf 916: return false;
917:
1.191 paf 918: // lang
919: lang=*cur++;
920: // length [WARNING: not cast, addresses must be %4=0 on sparc]
921: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
922:
1.206 paf 923: size_t combined_length=total_length+fragment_length;
924: if(combined_length>body_length)
925: return false; // file curruption
1.191 paf 926: // uchar needed to prevent propagating 0x80 bit to upper bytes
927: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 928: total_length=combined_length;
1.202 paf 929: in_buf-=piece_length;
1.174 paf 930: }
1.128 paf 931:
1.202 paf 932: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 933: return false;
1.174 paf 934: }
1.202 paf 935: if(in_buf!=0) // some strange extra bytes
936: return false;
1.113 parser 937:
1.174 paf 938: ASSERT_STRING_INVARIANT(*this);
1.148 paf 939: return true;
1.176 paf 940: }
941:
942: const char* String::Body::v() const {
1.233 misha 943: return CORD_to_const_char_star(body, length());
1.176 paf 944: }
1.201 paf 945: void String::Body::dump() const {
946: CORD_dump(body);
947: }
948:
1.176 paf 949: const char* String::Languages::v() const {
1.177 paf 950: if(opt.is_not_just_lang)
1.233 misha 951: return CORD_to_const_char_star(langs, 0);
1.176 paf 952: else
953: return (const char*)&langs;
954: }
1.201 paf 955: void String::Languages::dump() const {
956: if(opt.is_not_just_lang)
957: CORD_dump(langs);
958: else
959: puts((const char*)&langs);
960: }
1.176 paf 961: const char* String::v() const {
1.198 paf 962: const uint LIMIT_VIEW=20;
1.176 paf 963: char* buf=(char*)malloc(MAX_STRING);
964: const char*body_view=body.v();
965: const char*langs_view=langs.v();
966: snprintf(buf, MAX_STRING,
1.178 paf 967: "%d:%.*s%s} "
1.176 paf 968: "{%d:%s",
1.178 paf 969: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 970: strlen(body_view), body_view
971: );
972:
973: return buf;
1.113 parser 974: }
1.229 misha 975:
1.201 paf 976: void String::dump() const {
977: body.dump();
978: langs.dump();
979: }
1.229 misha 980:
981: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 982: if(is_empty())
1.195 paf 983: return *this;
984:
985: size_t substr_begin, substr_length;
1.229 misha 986: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 987: if(new_body==body) // we received unchanged pointer, do likewise
988: return *this;
989: // new_body differs from body, adjust langs along
990:
991: String& result=*new String;
992: if(!new_body) // body.trim produced empty result
993: return result;
994: // body.trim produced nonempty result
995:
996: // first: their langs
997: result.langs.append(result.body, langs, substr_begin, substr_length);
998: // next: letters themselves
999: result.body=new_body;
1000:
1001: ASSERT_STRING_INVARIANT(result);
1002: return result;
1.198 paf 1003: }
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