Annotation of parser3/src/main/pa_string.C, revision 1.243
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:
1.243 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.242 2012-05-28 10:33:18 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;
40: if (*pos == 'x' || *pos == 'X')
41: ++pos;
42: base=16;
43: }
44: }
45:
46: if (base < 2 || base > 16) { /* illegal base */
47: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
48: }
49:
50: unsigned int cutoff = UINT_MAX / base;
51: int cutoff_digit = UINT_MAX - cutoff * base;
52:
53: while(true) {
54: int digit;
55:
56: if ((*pos >= '0') && (*pos <= '9')) {
57: digit = (*pos - '0');
58: } else if (*pos >= 'a') {
59: digit = (*pos - 'a' + 10);
60: } else if (*pos >= 'A') {
61: digit = (*pos - 'A' + 10);
62: } else break;
63:
64: if (digit >= base) {
65: break;
66: }
67:
68: ++pos;
69:
70: /* adjust number, with overflow check */
71: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
72: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
73: } else {
74: result = result * base + digit;
75: }
76: }
77:
78: while(char c=*pos++)
79: if(!isspace(c))
80: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
81:
82: return result;
83: }
84:
1.193 paf 85: int pa_atoi(const char* str, const String* problem_source) {
86: if(!str)
87: return 0;
88:
1.242 moko 89: while(isspace(*str))
1.193 paf 90: str++;
1.242 moko 91:
1.193 paf 92: if(!*str)
93: return 0;
94:
1.200 paf 95: bool negative=false;
96: if(str[0]=='-') {
97: negative=true;
98: str++;
99: } else if(str[0]=='+') {
100: str++;
101: }
1.193 paf 102:
1.242 moko 103: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 104:
1.242 moko 105: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 ! moko 106: return -(int)result;
1.242 moko 107:
108: if(result<=INT_MAX)
1.243 ! moko 109: return (int)result;
1.242 moko 110:
111: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 112: }
113:
114: double pa_atod(const char* str, const String* problem_source) {
115: if(!str)
116: return 0;
117:
1.242 moko 118: while(isspace(*str))
1.193 paf 119: str++;
1.242 moko 120:
1.193 paf 121: if(!*str)
122: return 0;
123:
1.200 paf 124: bool negative=false;
125: if(str[0]=='-') {
126: negative=true;
127: str++;
128: } else if(str[0]=='+') {
129: str++;
130: }
1.242 moko 131:
132: double result;
1.193 paf 133: if(str[0]=='0')
1.242 moko 134: if(str[1]=='x' || str[1]=='X'){
135: // 0xABC
136: result=(double)pa_atoui(str, 0, problem_source);
137: return negative ? -result : result;
138: } else {
1.200 paf 139: // skip leading 0000, to disable octal interpretation
1.242 moko 140: do str++; while(*str=='0');
1.200 paf 141: }
1.242 moko 142:
143: char *error_pos;
144: result=strtod(str, &error_pos);
1.193 paf 145:
146: while(char c=*error_pos++)
1.199 paf 147: if(!isspace((unsigned char)c))
1.242 moko 148: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 149:
1.242 moko 150: return negative ? -result : result;
1.193 paf 151: }
152:
1.176 paf 153: // cord lib extension
154:
155: #ifndef DOXYGEN
156: typedef struct {
157: ssize_t countdown;
1.204 paf 158: int target; /* Character we're looking for */
1.176 paf 159: } chr_data;
160: #endif
161: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
162: {
163: register chr_data * d = (chr_data *)client_data;
164:
165: if (d -> countdown<=0) return(2);
166: d -> countdown -= size;
167: if (c > d -> target) return(1);
168: return(0);
169: }
170: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
171: {
172: chr_data d;
173:
174: d.countdown = n;
175: d.target = c;
176: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
177: }
178:
1.187 paf 179: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 180: {
181: int* result=(int*)client_data;
182: (*result)++;
183: return(0); // 0=continue
184: }
185: size_t CORD_block_count(CORD x)
186: {
187: size_t result=0;
188: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
189: return result;
190: }
191:
1.174 paf 192: // helpers
1.139 paf 193:
1.174 paf 194: /// String::match uses this as replace & global search table columns
1.139 paf 195:
1.174 paf 196: const int MAX_MATCH_GROUPS=100;
1.139 paf 197:
1.174 paf 198: class String_match_table_template_columns: public ArrayString {
199: public:
200: String_match_table_template_columns() {
201: *this+=new String("prematch");
202: *this+=new String("match");
203: *this+=new String("postmatch");
204: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 205: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 206: }
207: }
208: };
209:
210: Table string_match_table_template(new String_match_table_template_columns);
211:
1.176 paf 212: // String::Body methods
1.140 paf 213:
1.176 paf 214: String::Body String::Body::Format(int value) {
1.174 paf 215: char local[MAX_NUMBER];
216: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 217: return String::Body(pa_strdup(local, length));
1.120 paf 218: }
219:
1.195 paf 220: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 221: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 222: size_t our_length=length();
223: if(!our_length)
224: return *this;
1.229 misha 225:
226: // check if any UTF-8 in chars
227: bool fast=true;
228: if(chars && source_charset && source_charset->isUTF8()){
229: const char* pos=chars;
230: while(unsigned char c=*pos++)
231: if(c>127){
232: fast=false;
233: break;
234: }
235: }
236:
237: size_t start=0;
238: size_t end=our_length;
1.195 paf 239: if(!chars)
240: chars=" \t\n"; // white space
241:
1.229 misha 242: if(fast){
243: // from left...
244: if(kind!=TRIM_END) {
245: CORD_pos pos; set_pos(pos, 0);
246: while(true) {
247: char c=CORD_pos_fetch(pos);
248: if(strchr(chars, c)) {
249: if(++start==our_length)
250: return 0; // all chars are empty, just return empty string
251: } else
252: break;
253:
254: CORD_next(pos);
255: }
256: }
257:
258: // from right..
259: if(kind!=TRIM_START) {
260: CORD_pos pos; set_pos(pos, end-1);
261: while(true) {
262: char c=CORD_pos_fetch(pos);
263: if(strchr(chars, c)) {
264: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
265: return 0; // all chars are empty, just return empty string
266: } else
267: break;
268:
269: CORD_prev(pos);
270: }
271: }
272: } else {
1.234 misha 273: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 274: const XMLByte* src_end=src_begin+our_length;
275:
276: // from left...
277: if(kind!=TRIM_END) {
278: while(src_begin<src_end){
279: uint char_length=1;
280: const XMLByte* ptr=src_begin;
1.231 misha 281: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
282: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 283: char_length++;
284:
285: bool found=false;
286: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
287: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
288: found=true;
289: break;
290: }
291:
292: if(found){
293: start+=char_length;
294: if(start==our_length)
295: return 0; // all chars are empty, just return empty string
296: } else
297: break;
298: }
299: }
1.196 paf 300:
1.229 misha 301: // from right..
302: if(kind!=TRIM_START) {
303: while(src_begin<src_end){
304: uint char_length=1;
1.231 misha 305: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
306: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 307: char_length++;
308:
309: bool found=false;
310: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
311: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
312: found=true;
313: break;
314: }
315:
316: if(found){
317: end-=char_length;
318: if(end==0)
319: return 0; // all chars are empty, just return empty string
320: } else
321: break;
322: }
1.196 paf 323: }
324: }
1.195 paf 325:
326: if(start==0 && end==our_length) // nobody moved a thing
327: return *this;
328:
329: if(out_start)
330: *out_start=start;
331: size_t new_length=end-start;
332: if(out_length)
333: *out_length=new_length;
334:
335: return mid(start, new_length);
336: }
337:
1.174 paf 338: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
339: uint& result=*static_cast<uint*>(client_data);
340: generic_hash_code(result, c);
341: return 0;
342: }
343: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
344: uint& result=*static_cast<uint*>(client_data);
345: generic_hash_code(result, s);
346: return 0;
347: };
1.225 misha 348: uint String::Body::get_hash_code() const {
349: #ifdef HASH_CODE_CACHING
350: if(hash_code)
351: return hash_code;
352: #else
353: uint hash_code=0;
354: #endif
1.220 misha 355: if (body && CORD_IS_STRING(body)){
1.225 misha 356: generic_hash_code(hash_code, body);
1.220 misha 357: } else {
358: CORD_iter5(body, 0,
359: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 360: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 361: }
1.225 misha 362: return hash_code;
1.94 parser 363: }
364:
1.241 misha 365: struct CORD_pos_info {
366: const char* chars;
367: size_t left;
368: size_t pos;
369: };
370:
371: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
372: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
373: if(info->pos < info->left){
374: info->pos=STRING_NOT_FOUND;
375: return 1;
376: }
377: if(strchr(info->chars, c))
378: return 1;
379: --(info->pos);
380: return 0;
381: }
382:
383: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
384: if(is_empty() || !chars || !strlen(chars))
385: return STRING_NOT_FOUND;
386: CORD_pos_info info={chars, left, right};
387: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
388: return info.pos;
389: else
390: return STRING_NOT_FOUND;
391: }
392:
393:
394: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
395: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
396: if(info->pos < info->left) {
397: info->pos=STRING_NOT_FOUND;
398: return 1;
399: }
400: if(!strchr(info->chars, c))
401: return 1;
402: --(info->pos);
403: return 0;
404: }
405:
406: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
407: if(is_empty() || !chars || !strlen(chars))
408: return STRING_NOT_FOUND;
409: CORD_pos_info info={chars, left, right};
410: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
411: return info.pos;
412: else
413: return STRING_NOT_FOUND;
414: }
415:
1.174 paf 416: // String methods
417:
418: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
419: if(!known_length)
1.9 paf 420: return *this;
1.122 paf 421:
1.176 paf 422: // first: langs
423: langs.append(body, lang, known_length);
424: // next: letters themselves
1.174 paf 425: body.append_know_length(str, known_length);
1.1 paf 426:
1.174 paf 427: ASSERT_STRING_INVARIANT(*this);
1.1 paf 428: return *this;
429: }
1.174 paf 430: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
431: if(!str)
432: return *this;
433: size_t known_length=helper_length?helper_length:strlen(str);
434: if(!known_length)
435: return *this;
1.1 paf 436:
1.174 paf 437: return append_know_length(str, known_length, lang);
1.5 paf 438: }
1.226 misha 439: String::String(int value, char *format) : langs(L_CLEAN){
440: char buf[MAX_NUMBER];
441: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
442: }
1.174 paf 443: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
444: size_t known_length=helper_length?helper_length:strlen(str);
445: if(!known_length)
446: return *this;
1.5 paf 447:
1.176 paf 448: // first: langs
449: langs.append(body, lang, known_length);
450: // next: letters themselves
1.174 paf 451: body.append_strdup_know_length(str, known_length);
1.33 paf 452:
1.174 paf 453: ASSERT_STRING_INVARIANT(*this);
454: return *this;
1.5 paf 455: }
1.46 paf 456:
1.234 misha 457: struct CORD_length_info {
458: size_t len;
459: size_t skip;
460: };
461:
462: int CORD_batched_len(const char* s, CORD_length_info* info){
463: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 464: }
465:
1.234 misha 466: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
467: int CORD_batched_len(const char c, CORD_length_info* info){
468: if (info->skip==0){
469: info->len++;
470: info->skip = lengthUTF8Char(c)-1;
471: } else {
472: info->skip--;
473: }
1.220 misha 474: return 0;
475: }
476:
1.210 misha 477: size_t String::length(Charset& charset) const {
478: if(charset.isUTF8()){
1.234 misha 479: CORD_length_info info = {0, 0};
480: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
481: return info.len;
1.210 misha 482: } else
483: return body.length();
484: }
485:
1.174 paf 486: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
487: String& String::mid(size_t substr_begin, size_t substr_end) const {
488: String& result=*new String;
489:
490: size_t self_length=length();
491: substr_begin=min(substr_begin, self_length);
492: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 493: size_t substr_length=substr_end-substr_begin;
494: if(!substr_length)
1.107 parser 495: return result;
1.53 paf 496:
1.176 paf 497: // first: their langs
498: result.langs.append(result.body, langs, substr_begin, substr_length);
499: // next: letters themselves
500: result.body=body.mid(substr_begin, substr_length);
1.174 paf 501:
502: ASSERT_STRING_INVARIANT(result);
1.53 paf 503: return result;
1.54 paf 504: }
505:
1.211 misha 506: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
507: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 508: String& result=*new String;
509:
1.211 misha 510: size_t self_length=(helper_length)?helper_length:length(charset);
511:
512: if(!self_length)
513: return result;
514:
515: from=min(min(to, from), self_length);
1.210 misha 516: to=min(max(to, from), self_length);
1.211 misha 517:
1.210 misha 518: size_t substr_length=to-from;
1.211 misha 519:
1.210 misha 520: if(!substr_length)
521: return result;
522:
523: if(charset.isUTF8()){
1.234 misha 524: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 525: const XMLByte* src_end=src_begin+body.length();
1.210 misha 526:
1.212 misha 527: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 528: from=getUTF8BytePos(src_begin, src_end, from);
529: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 530: if(!substr_length)
531: return result;
532: }
533:
534: // first: their langs
535: result.langs.append(result.body, langs, from, substr_length);
536: // next: letters themselves
537: result.body=body.mid(from, substr_length);
538:
539: ASSERT_STRING_INVARIANT(result);
540: return result;
541: }
542:
1.176 paf 543: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 544: size_t substr_length=substr.length();
545: while(true) {
546: size_t substr_begin=body.pos(substr, this_offset);
547:
548: if(substr_begin==CORD_NOT_FOUND)
549: return STRING_NOT_FOUND;
1.174 paf 550:
1.183 paf 551: if(langs.check_lang(lang, substr_begin, substr_length))
552: return substr_begin;
553:
554: this_offset=substr_begin+substr_length;
555: }
1.58 paf 556: }
557:
1.174 paf 558: size_t String::pos(const String& substr,
559: size_t this_offset, Language lang) const {
560: return pos(substr.body, this_offset, lang);
1.60 paf 561: }
562:
1.210 misha 563: size_t String::pos(Charset& charset, const String& substr,
564: size_t this_offset, Language lang) const {
565:
566: if(charset.isUTF8()){
1.234 misha 567: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 568: const XMLByte* srcEnd=srcPtr+body.length();
569:
570: // convert 'this_offset' from 'characters' to 'bytes'
571: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
572:
1.229 misha 573: size_t result=pos(substr.body, this_offset, lang);
574: return (result==CORD_NOT_FOUND)
575: ? STRING_NOT_FOUND
576: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 577: } else {
1.229 misha 578: size_t result=pos(substr.body, this_offset, lang);
579: return (result==CORD_NOT_FOUND)
580: ? STRING_NOT_FOUND
581: : result;
1.210 misha 582: }
583: }
584:
1.174 paf 585: void String::split(ArrayString& result,
586: size_t& pos_after,
587: const char* delim,
588: Language lang, int limit) const {
1.237 misha 589: if(is_empty())
590: return;
1.174 paf 591: size_t self_length=length();
1.237 misha 592: if(size_t delim_length=strlen(delim)) {
1.186 paf 593: size_t pos_before;
1.60 paf 594: // while we have 'delim'...
1.174 paf 595: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 596: result+=&mid(pos_after, pos_before);
1.174 paf 597: pos_after=pos_before+delim_length;
1.60 paf 598: }
599: // last piece
1.174 paf 600: if(pos_after<self_length && limit) {
601: result+=&mid(pos_after, self_length);
602: pos_after=self_length;
1.60 paf 603: }
604: } else { // empty delim
605: result+=this;
1.174 paf 606: pos_after+=self_length;
1.60 paf 607: }
608: }
609:
1.174 paf 610: void String::split(ArrayString& result,
611: size_t& pos_after,
612: const String& delim, Language lang,
613: int limit) const {
1.237 misha 614: if(is_empty())
615: return;
616: if(!delim.is_empty()) {
1.186 paf 617: size_t pos_before;
1.60 paf 618: // while we have 'delim'...
1.174 paf 619: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 620: result+=&mid(pos_after, pos_before);
1.174 paf 621: pos_after=pos_before+delim.length();
1.60 paf 622: }
623: // last piece
1.174 paf 624: if(pos_after<length() && limit) {
625: result+=&mid(pos_after, length());
626: pos_after=length();
1.60 paf 627: }
628: } else { // empty delim
629: result+=this;
1.174 paf 630: pos_after+=length();
1.60 paf 631: }
1.61 paf 632: }
633:
1.222 misha 634: Table* String::match(VRegex* vregex,
635: Row_action row_action, void *info,
636: int& matches_count) const {
1.209 misha 637:
1.222 misha 638: // vregex->info(); // I have no idea what does it for?
1.63 paf 639:
1.222 misha 640: bool need_pre_post_match=vregex->is_pre_post_match_needed();
641: bool global=vregex->is_global_search();
1.63 paf 642:
1.174 paf 643: const char* subject=cstr();
1.234 misha 644: size_t subject_length=length();
1.222 misha 645: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
646: int ovector[ovector_size];
1.155 paf 647:
1.173 paf 648: Table::Action_options table_options;
1.174 paf 649: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 650:
1.154 paf 651: int prestart=0;
652: int poststart=0;
1.174 paf 653: int postfinish=length();
1.63 paf 654: while(true) {
1.222 misha 655: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 656:
1.222 misha 657: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
658: break;
1.63 paf 659:
1.154 paf 660: int prefinish=ovector[0];
661: poststart=ovector[1];
1.236 moko 662:
663: if (prestart==poststart && subject[poststart]=='\n'){
664: prestart++;
665: continue;
666: }
667:
1.232 misha 668: ArrayString* row=new ArrayString(3);
1.174 paf 669: if(need_pre_post_match) {
670: *row+=&mid(0, prefinish); // .prematch column value
671: *row+=&mid(prefinish, poststart); // .match
672: *row+=&mid(poststart, postfinish); // .postmatch
673: } else {
1.185 paf 674: *row+=&Empty; // .prematch column value
675: *row+=&Empty; // .match
676: *row+=&Empty; // .postmatch
1.174 paf 677: }
1.63 paf 678:
1.222 misha 679: for(int i=1; i<exec_result; i++) {
1.69 paf 680: // -1:-1 case handled peacefully by mid() itself
1.232 misha 681: *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 682: }
683:
1.209 misha 684: matches_count++;
1.174 paf 685: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 686:
1.222 misha 687: if(!global || prestart==poststart) // last step
688: break;
689:
1.154 paf 690: prestart=poststart;
1.222 misha 691: }
1.63 paf 692:
1.222 misha 693: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
694: return vregex->is_just_count() ? 0 : &table;
1.82 parser 695: }
696:
1.174 paf 697: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
698: String& result=*new String();
699: if(is_empty())
700: return result;
701:
702: char* new_cstr=cstrm();
1.216 misha 703:
1.181 paf 704: if(source_charset.isUTF8()) {
1.216 misha 705: size_t new_cstr_len=length();
1.181 paf 706: switch(kind) {
707: case CC_UPPER:
1.192 paf 708: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 709: break;
710: case CC_LOWER:
1.192 paf 711: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 712: break;
713: default:
714: assert(!"unknown change case kind");
715: break; // never
716: }
717:
718: } else {
719: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 720:
1.181 paf 721: const unsigned char *a;
722: const unsigned char *b;
723: switch(kind) {
724: case CC_UPPER:
725: a=tables+lcc_offset;
726: b=tables+fcc_offset;
727: break;
728: case CC_LOWER:
729: a=tables+lcc_offset;
730: b=0;
731: break;
732: default:
733: assert(!"unknown change case kind");
734: a=b=0; // calm, compiler
735: break; // never
736: }
737:
1.192 paf 738: char *dest=new_cstr;
1.181 paf 739: unsigned char index;
1.190 paf 740: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 741: unsigned char c=a[index];
742: if(b)
743: c=b[c];
744:
745: *dest++=(char)c;
746: }
1.174 paf 747: }
1.176 paf 748: result.langs=langs;
1.174 paf 749: result.body=new_cstr;
1.89 parser 750:
1.101 parser 751: return result;
752: }
753:
1.213 misha 754: const String& String::escape(Charset& source_charset) const {
755: if(is_empty())
756: return *this;
757:
758: return Charset::escape(*this, source_charset);
759: }
760:
1.238 misha 761: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
762: result.langs.append(result.body, langs, langs_offset, length); \
763: result.body.append_strdup_know_length(from_cstr, length);
764:
1.174 paf 765: const String& String::replace(const Dictionary& dict) const {
1.238 misha 766: if(!dict.count() || is_empty())
767: return *this;
768:
1.174 paf 769: String& result=*new String();
770: const char* old_cstr=cstr();
771: const char* prematch_begin=old_cstr;
772:
1.238 misha 773: if(dict.count()==1) {
774: // optimized simple case
775:
776: Dictionary::Subst subst=dict.get(0);
1.239 moko 777: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 778: // prematch
1.238 misha 779: if(size_t prematch_length=p-prematch_begin) {
780: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 781: }
782:
1.174 paf 783: // match
1.238 misha 784: prematch_begin=p+subst.from_length;
1.174 paf 785:
1.184 paf 786: if(const String* b=subst.to) // are there any b?
1.174 paf 787: result<<*b;
1.238 misha 788: }
789:
790: } else {
791:
792: const char* current=old_cstr;
793: while(*current) {
794: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
795: // prematch
796: if(size_t prematch_length=current-prematch_begin) {
797: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
798: }
799:
800: // match
801: // skip 'a' in 'current'; move prematch_begin
802: current+=subst.from_length; prematch_begin=current;
803:
804: if(const String* b=subst.to) // are there any b?
805: result<<*b;
806: } else // simply advance
807: current++;
808: }
809:
1.174 paf 810: }
1.156 paf 811:
1.238 misha 812: if(prematch_begin==old_cstr) // not modified
813: return *this;
814:
1.174 paf 815: // postmatch
1.238 misha 816: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
817: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 818: }
1.156 paf 819:
1.174 paf 820: ASSERT_STRING_INVARIANT(result);
1.82 parser 821: return result;
1.61 paf 822: }
1.113 parser 823:
1.180 paf 824: static int serialize_body_char(char c, char** cur) {
825: *((*cur)++)=c;
826: return 0; // 0=continue
827: };
1.174 paf 828: static int serialize_body_piece(const char* s, char** cur) {
829: size_t length=strlen(s);
830: memcpy(*cur, s, length); *cur+=length;
1.178 paf 831: return 0; // 0=continue
1.174 paf 832: };
1.178 paf 833: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
834: // lang
1.191 paf 835: **cur=alang; (*cur)++;
836: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 837: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
838:
839: return 0; // 0=continue
840: }
1.174 paf 841: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 842: size_t fragments_count=langs.count();
1.202 paf 843: size_t body_length=body.length();
1.174 paf 844: size_t buf_length=
1.178 paf 845: prolog_length //1
846: +sizeof(size_t) //2
1.202 paf 847: +body_length //3
848: +1 // 4 for zero terminator used in deserialize
849: +sizeof(size_t) //5
850: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
851:
1.174 paf 852: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
853:
854: // 1: prolog
855: char *cur=result.str+prolog_length;
1.202 paf 856: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
857: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
858: // 3: letters
859: body.for_each(serialize_body_char, serialize_body_piece, &cur);
860: // 4: zero terminator
861: *cur++=0;
862: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 863: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 864: // 6: lang info
1.178 paf 865: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 866:
1.174 paf 867: return result;
1.113 parser 868: }
1.202 paf 869: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
870: size_t in_buf=buf_size;
871: if(in_buf<=prolog_size)
1.148 paf 872: return false;
1.202 paf 873: in_buf-=prolog_size;
1.135 paf 874:
1.174 paf 875: // 1: prolog
1.202 paf 876: const char* cur=(const char* )buf+prolog_size;
1.113 parser 877:
1.207 paf 878: // 2: chars.count
1.202 paf 879: size_t body_length;
880: if(in_buf<sizeof(body_length)) // body.length don't fit?
881: return false;
882: // [WARNING: not cast, addresses must be %4=0 on sparc]
883: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
884: in_buf-=sizeof(body_length);
885:
886: if(in_buf<body_length+1) // letters+terminator don't fit?
887: return false;
888: // 4: zero terminator
889: if(cur[body_length] != 0) // in place?
890: return false;
891: // 3: letters
1.227 misha 892: body=String::Body(*cur?cur:0);
1.229 misha 893: #ifdef STRING_LENGTH_CACHING
894: body.set_length(body_length);
895: #endif
1.202 paf 896: cur+=body_length+1;
897: in_buf-=body_length+1;
898:
899: // 5: langs.count
1.191 paf 900: size_t fragments_count;
1.202 paf 901: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 902: return false;
1.191 paf 903: // [WARNING: not cast, addresses must be %4=0 on sparc]
904: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 905: in_buf-=sizeof(fragments_count);
1.174 paf 906:
907: if(fragments_count) {
1.202 paf 908: // 6: lang info
1.174 paf 909: size_t total_length=0;
910: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 911: char lang;
912: size_t fragment_length;
913: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 914: if(in_buf<piece_length) // lang+length
1.174 paf 915: return false;
916:
1.191 paf 917: // lang
918: lang=*cur++;
919: // length [WARNING: not cast, addresses must be %4=0 on sparc]
920: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
921:
1.206 paf 922: size_t combined_length=total_length+fragment_length;
923: if(combined_length>body_length)
924: return false; // file curruption
1.191 paf 925: // uchar needed to prevent propagating 0x80 bit to upper bytes
926: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 927: total_length=combined_length;
1.202 paf 928: in_buf-=piece_length;
1.174 paf 929: }
1.128 paf 930:
1.202 paf 931: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 932: return false;
1.174 paf 933: }
1.202 paf 934: if(in_buf!=0) // some strange extra bytes
935: return false;
1.113 parser 936:
1.174 paf 937: ASSERT_STRING_INVARIANT(*this);
1.148 paf 938: return true;
1.176 paf 939: }
940:
941: const char* String::Body::v() const {
1.233 misha 942: return CORD_to_const_char_star(body, length());
1.176 paf 943: }
1.201 paf 944: void String::Body::dump() const {
945: CORD_dump(body);
946: }
947:
1.176 paf 948: const char* String::Languages::v() const {
1.177 paf 949: if(opt.is_not_just_lang)
1.233 misha 950: return CORD_to_const_char_star(langs, 0);
1.176 paf 951: else
952: return (const char*)&langs;
953: }
1.201 paf 954: void String::Languages::dump() const {
955: if(opt.is_not_just_lang)
956: CORD_dump(langs);
957: else
958: puts((const char*)&langs);
959: }
1.176 paf 960: const char* String::v() const {
1.198 paf 961: const uint LIMIT_VIEW=20;
1.176 paf 962: char* buf=(char*)malloc(MAX_STRING);
963: const char*body_view=body.v();
964: const char*langs_view=langs.v();
965: snprintf(buf, MAX_STRING,
1.178 paf 966: "%d:%.*s%s} "
1.176 paf 967: "{%d:%s",
1.178 paf 968: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 969: strlen(body_view), body_view
970: );
971:
972: return buf;
1.113 parser 973: }
1.229 misha 974:
1.201 paf 975: void String::dump() const {
976: body.dump();
977: langs.dump();
978: }
1.229 misha 979:
980: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 981: if(is_empty())
1.195 paf 982: return *this;
983:
984: size_t substr_begin, substr_length;
1.229 misha 985: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 986: if(new_body==body) // we received unchanged pointer, do likewise
987: return *this;
988: // new_body differs from body, adjust langs along
989:
990: String& result=*new String;
991: if(!new_body) // body.trim produced empty result
992: return result;
993: // body.trim produced nonempty result
994:
995: // first: their langs
996: result.langs.append(result.body, langs, substr_begin, substr_length);
997: // next: letters themselves
998: result.body=new_body;
999:
1000: ASSERT_STRING_INVARIANT(result);
1001: return result;
1.198 paf 1002: }
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