Annotation of parser3/src/main/pa_string.C, revision 1.261
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.260 moko 4: Copyright (c) 2001-2017 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:
1.261 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.260 2017/02/07 22:00:44 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.261 ! moko 19: #define COMPILE_ASSERT(x) extern int assert_checker[(int)(x) ? 1 : -1]
! 20: COMPILE_ASSERT(sizeof(String::Languages) == sizeof(CORD));
! 21:
1.242 moko 22: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
23:
1.249 moko 24: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
25: T result = 0;
1.242 moko 26: const char *pos = str;
27:
28: while (isspace(*pos)) /* skip leading whitespace */
29: ++pos;
30:
31: if (base == 16 && *pos == '0') { /* handle option prefix */
32: ++pos;
33: if (*pos == 'x' || *pos == 'X') {
34: ++pos;
35: }
36: }
37:
38: if (base == 0) { /* dynamic base */
39: base = 10; /* default is 10 */
40: if (*pos == '0') {
41: ++pos;
1.245 moko 42: if (*pos == 'x' || *pos == 'X'){
1.242 moko 43: ++pos;
44: base=16;
1.245 moko 45: }
1.242 moko 46: }
47: }
48:
49: if (base < 2 || base > 16) { /* illegal base */
50: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
51: }
52:
1.249 moko 53: T cutoff = max / base;
54: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 55:
56: while(true) {
57: int digit;
58:
59: if ((*pos >= '0') && (*pos <= '9')) {
60: digit = (*pos - '0');
61: } else if (*pos >= 'a') {
62: digit = (*pos - 'a' + 10);
63: } else if (*pos >= 'A') {
64: digit = (*pos - 'A' + 10);
65: } else break;
66:
67: if (digit >= base) {
68: break;
69: }
70:
71: ++pos;
72:
73: /* adjust number, with overflow check */
74: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
75: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
76: } else {
77: result = result * base + digit;
78: }
79: }
80:
81: while(char c=*pos++)
82: if(!isspace(c))
83: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
84:
85: return result;
86: }
87:
1.249 moko 88: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.254 moko 89: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 90: }
91:
92: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
1.254 moko 93: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
1.249 moko 94: }
95:
1.193 paf 96: int pa_atoi(const char* str, const String* problem_source) {
97: if(!str)
98: return 0;
99:
1.242 moko 100: while(isspace(*str))
1.193 paf 101: str++;
1.242 moko 102:
1.193 paf 103: if(!*str)
104: return 0;
105:
1.200 paf 106: bool negative=false;
107: if(str[0]=='-') {
108: negative=true;
109: str++;
110: } else if(str[0]=='+') {
111: str++;
112: }
1.193 paf 113:
1.242 moko 114: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 115:
1.242 moko 116: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 117: return -(int)result;
1.242 moko 118:
119: if(result<=INT_MAX)
1.243 moko 120: return (int)result;
1.242 moko 121:
122: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 123: }
124:
125: double pa_atod(const char* str, const String* problem_source) {
126: if(!str)
127: return 0;
128:
1.242 moko 129: while(isspace(*str))
1.193 paf 130: str++;
1.242 moko 131:
1.193 paf 132: if(!*str)
133: return 0;
134:
1.200 paf 135: bool negative=false;
136: if(str[0]=='-') {
137: negative=true;
138: str++;
139: } else if(str[0]=='+') {
140: str++;
141: }
1.242 moko 142:
143: double result;
1.247 moko 144: if(str[0]=='0') {
145: if(str[1]=='x' || str[1]=='X') {
1.242 moko 146: // 0xABC
1.249 moko 147: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 148: return negative ? -result : result;
149: } else {
1.200 paf 150: // skip leading 0000, to disable octal interpretation
1.242 moko 151: do str++; while(*str=='0');
1.200 paf 152: }
1.247 moko 153: }
1.242 moko 154:
155: char *error_pos;
156: result=strtod(str, &error_pos);
1.193 paf 157:
158: while(char c=*error_pos++)
1.199 paf 159: if(!isspace((unsigned char)c))
1.242 moko 160: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 161:
1.242 moko 162: return negative ? -result : result;
1.193 paf 163: }
164:
1.176 paf 165: // cord lib extension
166:
167: #ifndef DOXYGEN
168: typedef struct {
1.254 moko 169: ssize_t countdown;
170: int target; /* Character we're looking for */
1.176 paf 171: } chr_data;
172: #endif
1.248 moko 173:
1.176 paf 174: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
175: {
1.254 moko 176: register chr_data * d = (chr_data *)client_data;
177:
178: if (d -> countdown<=0) return(2);
179: d -> countdown -= size;
180: if (c > d -> target) return(1);
181: return(0);
1.176 paf 182: }
1.248 moko 183:
1.176 paf 184: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
185: {
1.254 moko 186: chr_data d;
1.176 paf 187:
1.254 moko 188: d.countdown = n;
189: d.target = c;
190: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
1.176 paf 191: }
192:
1.187 paf 193: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 194: {
1.254 moko 195: int* result=(int*)client_data;
196: (*result)++;
197: return(0); // 0=continue
1.178 paf 198: }
1.248 moko 199:
1.178 paf 200: size_t CORD_block_count(CORD x)
201: {
202: size_t result=0;
203: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 204: return result;
1.178 paf 205: }
206:
1.174 paf 207: // helpers
1.139 paf 208:
1.174 paf 209: /// String::match uses this as replace & global search table columns
1.139 paf 210:
1.174 paf 211: const int MAX_MATCH_GROUPS=100;
1.139 paf 212:
1.174 paf 213: class String_match_table_template_columns: public ArrayString {
214: public:
215: String_match_table_template_columns() {
216: *this+=new String("prematch");
217: *this+=new String("match");
218: *this+=new String("postmatch");
219: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 220: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 221: }
222: }
223: };
224:
225: Table string_match_table_template(new String_match_table_template_columns);
226:
1.176 paf 227: // String::Body methods
1.140 paf 228:
1.176 paf 229: String::Body String::Body::Format(int value) {
1.174 paf 230: char local[MAX_NUMBER];
231: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 232: return String::Body(pa_strdup(local, length));
1.120 paf 233: }
234:
1.254 moko 235: String::Body String::Body::trim(String::Trim_kind kind, const char* chars, size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 236: size_t our_length=length();
237: if(!our_length)
238: return *this;
1.229 misha 239:
240: // check if any UTF-8 in chars
241: bool fast=true;
242: if(chars && source_charset && source_charset->isUTF8()){
243: const char* pos=chars;
244: while(unsigned char c=*pos++)
245: if(c>127){
246: fast=false;
247: break;
248: }
249: }
250:
251: size_t start=0;
252: size_t end=our_length;
1.195 paf 253: if(!chars)
254: chars=" \t\n"; // white space
255:
1.229 misha 256: if(fast){
257: // from left...
258: if(kind!=TRIM_END) {
259: CORD_pos pos; set_pos(pos, 0);
260: while(true) {
261: char c=CORD_pos_fetch(pos);
262: if(strchr(chars, c)) {
263: if(++start==our_length)
264: return 0; // all chars are empty, just return empty string
265: } else
266: break;
267:
268: CORD_next(pos);
269: }
270: }
271:
272: // from right..
273: if(kind!=TRIM_START) {
274: CORD_pos pos; set_pos(pos, end-1);
275: while(true) {
276: char c=CORD_pos_fetch(pos);
277: if(strchr(chars, c)) {
278: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
279: return 0; // all chars are empty, just return empty string
280: } else
281: break;
282:
283: CORD_prev(pos);
284: }
285: }
286: } else {
1.234 misha 287: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 288: const XMLByte* src_end=src_begin+our_length;
289:
290: // from left...
291: if(kind!=TRIM_END) {
292: while(src_begin<src_end){
293: uint char_length=1;
294: const XMLByte* ptr=src_begin;
1.231 misha 295: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
296: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 297: char_length++;
298:
299: bool found=false;
300: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
301: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
302: found=true;
303: break;
304: }
305:
306: if(found){
307: start+=char_length;
308: if(start==our_length)
309: return 0; // all chars are empty, just return empty string
310: } else
311: break;
312: }
313: }
1.196 paf 314:
1.229 misha 315: // from right..
316: if(kind!=TRIM_START) {
317: while(src_begin<src_end){
318: uint char_length=1;
1.231 misha 319: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
320: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 321: char_length++;
322:
323: bool found=false;
324: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
325: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
326: found=true;
327: break;
328: }
329:
330: if(found){
331: end-=char_length;
332: if(end==0)
333: return 0; // all chars are empty, just return empty string
334: } else
335: break;
336: }
1.196 paf 337: }
338: }
1.195 paf 339:
340: if(start==0 && end==our_length) // nobody moved a thing
341: return *this;
342:
343: if(out_start)
344: *out_start=start;
345: size_t new_length=end-start;
346: if(out_length)
347: *out_length=new_length;
348:
349: return mid(start, new_length);
350: }
351:
1.174 paf 352: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
353: uint& result=*static_cast<uint*>(client_data);
354: generic_hash_code(result, c);
355: return 0;
356: }
1.248 moko 357:
1.174 paf 358: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
359: uint& result=*static_cast<uint*>(client_data);
360: generic_hash_code(result, s);
361: return 0;
1.248 moko 362: }
363:
1.225 misha 364: uint String::Body::get_hash_code() const {
365: #ifdef HASH_CODE_CACHING
366: if(hash_code)
367: return hash_code;
368: #else
369: uint hash_code=0;
370: #endif
1.220 misha 371: if (body && CORD_IS_STRING(body)){
1.250 moko 372: generic_hash_code(hash_code, (const char *)body);
1.220 misha 373: } else {
374: CORD_iter5(body, 0,
375: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 376: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 377: }
1.225 misha 378: return hash_code;
1.94 parser 379: }
380:
1.241 misha 381: struct CORD_pos_info {
382: const char* chars;
383: size_t left;
384: size_t pos;
385: };
386:
387: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
388: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
389: if(info->pos < info->left){
390: info->pos=STRING_NOT_FOUND;
391: return 1;
392: }
393: if(strchr(info->chars, c))
394: return 1;
395: --(info->pos);
396: return 0;
397: }
398:
399: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
400: if(is_empty() || !chars || !strlen(chars))
401: return STRING_NOT_FOUND;
402: CORD_pos_info info={chars, left, right};
403: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
404: return info.pos;
405: else
406: return STRING_NOT_FOUND;
407: }
408:
409:
410: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
411: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
412: if(info->pos < info->left) {
413: info->pos=STRING_NOT_FOUND;
414: return 1;
415: }
416: if(!strchr(info->chars, c))
417: return 1;
418: --(info->pos);
419: return 0;
420: }
421:
422: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
423: if(is_empty() || !chars || !strlen(chars))
424: return STRING_NOT_FOUND;
425: CORD_pos_info info={chars, left, right};
426: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
427: return info.pos;
428: else
429: return STRING_NOT_FOUND;
430: }
431:
1.174 paf 432: // String methods
433:
434: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
435: if(!known_length)
1.9 paf 436: return *this;
1.122 paf 437:
1.176 paf 438: // first: langs
439: langs.append(body, lang, known_length);
440: // next: letters themselves
1.174 paf 441: body.append_know_length(str, known_length);
1.1 paf 442:
1.174 paf 443: ASSERT_STRING_INVARIANT(*this);
1.1 paf 444: return *this;
445: }
1.248 moko 446:
1.174 paf 447: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
448: if(!str)
449: return *this;
450: size_t known_length=helper_length?helper_length:strlen(str);
451: if(!known_length)
452: return *this;
1.1 paf 453:
1.174 paf 454: return append_know_length(str, known_length, lang);
1.5 paf 455: }
1.248 moko 456:
1.244 moko 457: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 458: char buf[MAX_NUMBER];
459: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
460: }
1.248 moko 461:
1.174 paf 462: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
463: size_t known_length=helper_length?helper_length:strlen(str);
464: if(!known_length)
465: return *this;
1.5 paf 466:
1.176 paf 467: // first: langs
468: langs.append(body, lang, known_length);
469: // next: letters themselves
1.174 paf 470: body.append_strdup_know_length(str, known_length);
1.33 paf 471:
1.174 paf 472: ASSERT_STRING_INVARIANT(*this);
473: return *this;
1.5 paf 474: }
1.46 paf 475:
1.234 misha 476: struct CORD_length_info {
477: size_t len;
478: size_t skip;
479: };
480:
481: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 482: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
483: return 0;
1.221 misha 484: }
485:
1.234 misha 486: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
487: int CORD_batched_len(const char c, CORD_length_info* info){
488: if (info->skip==0){
489: info->len++;
490: info->skip = lengthUTF8Char(c)-1;
491: } else {
492: info->skip--;
493: }
1.220 misha 494: return 0;
495: }
496:
1.210 misha 497: size_t String::length(Charset& charset) const {
498: if(charset.isUTF8()){
1.234 misha 499: CORD_length_info info = {0, 0};
500: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
501: return info.len;
1.210 misha 502: } else
503: return body.length();
504: }
505:
1.174 paf 506: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
507: String& String::mid(size_t substr_begin, size_t substr_end) const {
508: String& result=*new String;
509:
510: size_t self_length=length();
511: substr_begin=min(substr_begin, self_length);
512: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 513: size_t substr_length=substr_end-substr_begin;
514: if(!substr_length)
1.107 parser 515: return result;
1.53 paf 516:
1.176 paf 517: // first: their langs
518: result.langs.append(result.body, langs, substr_begin, substr_length);
519: // next: letters themselves
520: result.body=body.mid(substr_begin, substr_length);
1.174 paf 521:
522: ASSERT_STRING_INVARIANT(result);
1.53 paf 523: return result;
1.54 paf 524: }
525:
1.211 misha 526: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
527: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 528: String& result=*new String;
529:
1.255 moko 530: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 531:
532: if(!self_length)
533: return result;
534:
535: from=min(min(to, from), self_length);
1.210 misha 536: to=min(max(to, from), self_length);
1.211 misha 537:
1.210 misha 538: size_t substr_length=to-from;
1.211 misha 539:
1.210 misha 540: if(!substr_length)
541: return result;
542:
543: if(charset.isUTF8()){
1.234 misha 544: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 545: const XMLByte* src_end=src_begin+body.length();
1.210 misha 546:
1.212 misha 547: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 548: from=getUTF8BytePos(src_begin, src_end, from);
549: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 550: if(!substr_length)
551: return result;
552: }
553:
554: // first: their langs
555: result.langs.append(result.body, langs, from, substr_length);
556: // next: letters themselves
557: result.body=body.mid(from, substr_length);
558:
559: ASSERT_STRING_INVARIANT(result);
560: return result;
561: }
562:
1.176 paf 563: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 564: size_t substr_length=substr.length();
565: while(true) {
566: size_t substr_begin=body.pos(substr, this_offset);
567:
568: if(substr_begin==CORD_NOT_FOUND)
569: return STRING_NOT_FOUND;
1.174 paf 570:
1.183 paf 571: if(langs.check_lang(lang, substr_begin, substr_length))
572: return substr_begin;
573:
574: this_offset=substr_begin+substr_length;
575: }
1.58 paf 576: }
577:
1.254 moko 578: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 579: return pos(substr.body, this_offset, lang);
1.60 paf 580: }
581:
1.254 moko 582: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 583:
584: if(charset.isUTF8()){
1.234 misha 585: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 586: const XMLByte* srcEnd=srcPtr+body.length();
587:
588: // convert 'this_offset' from 'characters' to 'bytes'
589: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
590:
1.229 misha 591: size_t result=pos(substr.body, this_offset, lang);
592: return (result==CORD_NOT_FOUND)
593: ? STRING_NOT_FOUND
594: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 595: } else {
1.229 misha 596: size_t result=pos(substr.body, this_offset, lang);
597: return (result==CORD_NOT_FOUND)
598: ? STRING_NOT_FOUND
599: : result;
1.210 misha 600: }
601: }
602:
1.256 moko 603: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 604: if(is_empty())
605: return;
1.174 paf 606: size_t self_length=length();
1.237 misha 607: if(size_t delim_length=strlen(delim)) {
1.186 paf 608: size_t pos_before;
1.60 paf 609: // while we have 'delim'...
1.256 moko 610: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 611: result+=&mid(pos_after, pos_before);
1.174 paf 612: pos_after=pos_before+delim_length;
1.60 paf 613: }
614: // last piece
1.256 moko 615: if(pos_after<self_length)
1.174 paf 616: result+=&mid(pos_after, self_length);
1.60 paf 617: } else { // empty delim
618: result+=this;
619: }
620: }
621:
1.256 moko 622: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 623: if(is_empty())
624: return;
625: if(!delim.is_empty()) {
1.186 paf 626: size_t pos_before;
1.60 paf 627: // while we have 'delim'...
1.256 moko 628: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 629: result+=&mid(pos_after, pos_before);
1.174 paf 630: pos_after=pos_before+delim.length();
1.60 paf 631: }
632: // last piece
1.256 moko 633: if(pos_after<length())
1.174 paf 634: result+=&mid(pos_after, length());
1.60 paf 635: } else { // empty delim
636: result+=this;
637: }
1.61 paf 638: }
639:
1.254 moko 640: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 641:
1.222 misha 642: // vregex->info(); // I have no idea what does it for?
1.63 paf 643:
1.222 misha 644: bool need_pre_post_match=vregex->is_pre_post_match_needed();
645: bool global=vregex->is_global_search();
1.63 paf 646:
1.174 paf 647: const char* subject=cstr();
1.234 misha 648: size_t subject_length=length();
1.222 misha 649: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
650: int ovector[ovector_size];
1.155 paf 651:
1.173 paf 652: Table::Action_options table_options;
1.174 paf 653: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 654:
1.154 paf 655: int prestart=0;
656: int poststart=0;
1.174 paf 657: int postfinish=length();
1.63 paf 658: while(true) {
1.222 misha 659: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 660:
1.222 misha 661: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
662: break;
1.63 paf 663:
1.154 paf 664: int prefinish=ovector[0];
665: poststart=ovector[1];
1.236 moko 666:
667: if (prestart==poststart && subject[poststart]=='\n'){
668: prestart++;
669: continue;
670: }
671:
1.232 misha 672: ArrayString* row=new ArrayString(3);
1.174 paf 673: if(need_pre_post_match) {
674: *row+=&mid(0, prefinish); // .prematch column value
675: *row+=&mid(prefinish, poststart); // .match
676: *row+=&mid(poststart, postfinish); // .postmatch
677: } else {
1.185 paf 678: *row+=&Empty; // .prematch column value
679: *row+=&Empty; // .match
680: *row+=&Empty; // .postmatch
1.174 paf 681: }
1.63 paf 682:
1.222 misha 683: for(int i=1; i<exec_result; i++) {
1.69 paf 684: // -1:-1 case handled peacefully by mid() itself
1.232 misha 685: *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 686: }
687:
1.209 misha 688: matches_count++;
1.174 paf 689: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 690:
1.222 misha 691: if(!global || prestart==poststart) // last step
692: break;
693:
1.154 paf 694: prestart=poststart;
1.222 misha 695: }
1.63 paf 696:
1.222 misha 697: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
698: return vregex->is_just_count() ? 0 : &table;
1.82 parser 699: }
700:
1.174 paf 701: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
702: String& result=*new String();
703: if(is_empty())
704: return result;
705:
706: char* new_cstr=cstrm();
1.216 misha 707:
1.181 paf 708: if(source_charset.isUTF8()) {
1.216 misha 709: size_t new_cstr_len=length();
1.181 paf 710: switch(kind) {
711: case CC_UPPER:
1.192 paf 712: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 713: break;
714: case CC_LOWER:
1.192 paf 715: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 716: break;
717: default:
718: assert(!"unknown change case kind");
719: break; // never
720: }
721:
722: } else {
723: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 724:
1.181 paf 725: const unsigned char *a;
726: const unsigned char *b;
727: switch(kind) {
728: case CC_UPPER:
729: a=tables+lcc_offset;
730: b=tables+fcc_offset;
731: break;
732: case CC_LOWER:
733: a=tables+lcc_offset;
734: b=0;
735: break;
736: default:
737: assert(!"unknown change case kind");
738: a=b=0; // calm, compiler
739: break; // never
740: }
741:
1.192 paf 742: char *dest=new_cstr;
1.181 paf 743: unsigned char index;
1.190 paf 744: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 745: unsigned char c=a[index];
746: if(b)
747: c=b[c];
748:
749: *dest++=(char)c;
750: }
1.174 paf 751: }
1.176 paf 752: result.langs=langs;
1.174 paf 753: result.body=new_cstr;
1.89 parser 754:
1.101 parser 755: return result;
756: }
757:
1.213 misha 758: const String& String::escape(Charset& source_charset) const {
759: if(is_empty())
760: return *this;
761:
762: return Charset::escape(*this, source_charset);
763: }
764:
1.238 misha 765: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
766: result.langs.append(result.body, langs, langs_offset, length); \
767: result.body.append_strdup_know_length(from_cstr, length);
768:
1.174 paf 769: const String& String::replace(const Dictionary& dict) const {
1.238 misha 770: if(!dict.count() || is_empty())
771: return *this;
772:
1.174 paf 773: String& result=*new String();
774: const char* old_cstr=cstr();
775: const char* prematch_begin=old_cstr;
776:
1.238 misha 777: if(dict.count()==1) {
778: // optimized simple case
779:
780: Dictionary::Subst subst=dict.get(0);
1.239 moko 781: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 782: // prematch
1.238 misha 783: if(size_t prematch_length=p-prematch_begin) {
784: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 785: }
786:
1.174 paf 787: // match
1.238 misha 788: prematch_begin=p+subst.from_length;
1.174 paf 789:
1.184 paf 790: if(const String* b=subst.to) // are there any b?
1.174 paf 791: result<<*b;
1.238 misha 792: }
793:
794: } else {
795:
796: const char* current=old_cstr;
797: while(*current) {
798: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
799: // prematch
800: if(size_t prematch_length=current-prematch_begin) {
801: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
802: }
803:
804: // match
805: // skip 'a' in 'current'; move prematch_begin
806: current+=subst.from_length; prematch_begin=current;
807:
808: if(const String* b=subst.to) // are there any b?
809: result<<*b;
810: } else // simply advance
811: current++;
812: }
813:
1.174 paf 814: }
1.156 paf 815:
1.238 misha 816: if(prematch_begin==old_cstr) // not modified
817: return *this;
818:
1.174 paf 819: // postmatch
1.238 misha 820: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
821: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 822: }
1.156 paf 823:
1.174 paf 824: ASSERT_STRING_INVARIANT(result);
1.82 parser 825: return result;
1.61 paf 826: }
1.113 parser 827:
1.180 paf 828: static int serialize_body_char(char c, char** cur) {
829: *((*cur)++)=c;
830: return 0; // 0=continue
1.248 moko 831: }
832:
1.174 paf 833: static int serialize_body_piece(const char* s, char** cur) {
834: size_t length=strlen(s);
835: memcpy(*cur, s, length); *cur+=length;
1.178 paf 836: return 0; // 0=continue
1.248 moko 837: }
838:
1.178 paf 839: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
840: // lang
1.191 paf 841: **cur=alang; (*cur)++;
842: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 843: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
844:
845: return 0; // 0=continue
846: }
1.248 moko 847:
1.174 paf 848: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 849: size_t fragments_count=langs.count();
1.202 paf 850: size_t body_length=body.length();
1.174 paf 851: size_t buf_length=
1.178 paf 852: prolog_length //1
853: +sizeof(size_t) //2
1.202 paf 854: +body_length //3
855: +1 // 4 for zero terminator used in deserialize
856: +sizeof(size_t) //5
857: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
858:
1.174 paf 859: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
860:
861: // 1: prolog
862: char *cur=result.str+prolog_length;
1.202 paf 863: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
864: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
865: // 3: letters
866: body.for_each(serialize_body_char, serialize_body_piece, &cur);
867: // 4: zero terminator
868: *cur++=0;
869: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 870: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 871: // 6: lang info
1.178 paf 872: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 873:
1.174 paf 874: return result;
1.113 parser 875: }
1.248 moko 876:
1.202 paf 877: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
878: size_t in_buf=buf_size;
879: if(in_buf<=prolog_size)
1.148 paf 880: return false;
1.202 paf 881: in_buf-=prolog_size;
1.135 paf 882:
1.174 paf 883: // 1: prolog
1.202 paf 884: const char* cur=(const char* )buf+prolog_size;
1.113 parser 885:
1.207 paf 886: // 2: chars.count
1.202 paf 887: size_t body_length;
888: if(in_buf<sizeof(body_length)) // body.length don't fit?
889: return false;
890: // [WARNING: not cast, addresses must be %4=0 on sparc]
891: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
892: in_buf-=sizeof(body_length);
893:
894: if(in_buf<body_length+1) // letters+terminator don't fit?
895: return false;
896: // 4: zero terminator
897: if(cur[body_length] != 0) // in place?
898: return false;
899: // 3: letters
1.251 moko 900: body=String::Body(String::C(cur, body_length));
1.202 paf 901: cur+=body_length+1;
902: in_buf-=body_length+1;
903:
904: // 5: langs.count
1.191 paf 905: size_t fragments_count;
1.202 paf 906: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 907: return false;
1.191 paf 908: // [WARNING: not cast, addresses must be %4=0 on sparc]
909: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 910: in_buf-=sizeof(fragments_count);
1.174 paf 911:
912: if(fragments_count) {
1.202 paf 913: // 6: lang info
1.174 paf 914: size_t total_length=0;
915: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 916: char lang;
917: size_t fragment_length;
918: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 919: if(in_buf<piece_length) // lang+length
1.174 paf 920: return false;
921:
1.191 paf 922: // lang
923: lang=*cur++;
924: // length [WARNING: not cast, addresses must be %4=0 on sparc]
925: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
926:
1.206 paf 927: size_t combined_length=total_length+fragment_length;
928: if(combined_length>body_length)
929: return false; // file curruption
1.191 paf 930: // uchar needed to prevent propagating 0x80 bit to upper bytes
931: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 932: total_length=combined_length;
1.202 paf 933: in_buf-=piece_length;
1.174 paf 934: }
1.128 paf 935:
1.202 paf 936: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 937: return false;
1.174 paf 938: }
1.202 paf 939: if(in_buf!=0) // some strange extra bytes
940: return false;
1.113 parser 941:
1.174 paf 942: ASSERT_STRING_INVARIANT(*this);
1.148 paf 943: return true;
1.176 paf 944: }
945:
1.201 paf 946: void String::Body::dump() const {
947: CORD_dump(body);
948: }
949:
1.257 moko 950: const char* String::Languages::visualize() const {
1.177 paf 951: if(opt.is_not_just_lang)
1.233 misha 952: return CORD_to_const_char_star(langs, 0);
1.176 paf 953: else
1.257 moko 954: return 0;
1.176 paf 955: }
1.248 moko 956:
1.201 paf 957: void String::Languages::dump() const {
958: if(opt.is_not_just_lang)
959: CORD_dump(langs);
960: else
961: puts((const char*)&langs);
962: }
1.248 moko 963:
1.257 moko 964: void String::dump() const {
965: body.dump();
966: langs.dump();
967: }
1.176 paf 968:
1.257 moko 969: static char *n_chars(char c, size_t length){
970: char *result=(char *)pa_malloc_atomic(length+1);
971: memset(result, c, length);
972: result[length] = '\0';
973: return result;
1.113 parser 974: }
1.229 misha 975:
1.257 moko 976: char* String::visualize_langs() const {
1.258 moko 977: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 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