Annotation of parser3/src/main/pa_string.C, revision 1.255
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.253 moko 4: Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
1.138 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.164 paf 6: */
1.46 paf 7:
1.12 paf 8: #include "pa_string.h"
1.22 paf 9: #include "pa_exception.h"
1.61 paf 10: #include "pa_table.h"
1.101 parser 11: #include "pa_dictionary.h"
1.132 paf 12: #include "pa_charset.h"
1.222 misha 13: #include "pa_vregex.h"
1.215 misha 14:
1.252 moko 15: #ifndef ULLONG_MAX
16: #define ULLONG_MAX 18446744073709551615ULL
17: #endif
18:
1.255 ! moko 19: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.254 2016/09/05 21:59:22 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 20:
1.185 paf 21: const String String::Empty;
22:
1.242 moko 23: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
24:
1.249 moko 25: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
26: T result = 0;
1.242 moko 27: const char *pos = str;
28:
29: while (isspace(*pos)) /* skip leading whitespace */
30: ++pos;
31:
32: if (base == 16 && *pos == '0') { /* handle option prefix */
33: ++pos;
34: if (*pos == 'x' || *pos == 'X') {
35: ++pos;
36: }
37: }
38:
39: if (base == 0) { /* dynamic base */
40: base = 10; /* default is 10 */
41: if (*pos == '0') {
42: ++pos;
1.245 moko 43: if (*pos == 'x' || *pos == 'X'){
1.242 moko 44: ++pos;
45: base=16;
1.245 moko 46: }
1.242 moko 47: }
48: }
49:
50: if (base < 2 || base > 16) { /* illegal base */
51: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
52: }
53:
1.249 moko 54: T cutoff = max / base;
55: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 56:
57: while(true) {
58: int digit;
59:
60: if ((*pos >= '0') && (*pos <= '9')) {
61: digit = (*pos - '0');
62: } else if (*pos >= 'a') {
63: digit = (*pos - 'a' + 10);
64: } else if (*pos >= 'A') {
65: digit = (*pos - 'A' + 10);
66: } else break;
67:
68: if (digit >= base) {
69: break;
70: }
71:
72: ++pos;
73:
74: /* adjust number, with overflow check */
75: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
76: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
77: } else {
78: result = result * base + digit;
79: }
80: }
81:
82: while(char c=*pos++)
83: if(!isspace(c))
84: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
85:
86: return result;
87: }
88:
1.249 moko 89: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.254 moko 90: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 91: }
92:
93: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
1.254 moko 94: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
1.249 moko 95: }
96:
1.193 paf 97: int pa_atoi(const char* str, const String* problem_source) {
98: if(!str)
99: return 0;
100:
1.242 moko 101: while(isspace(*str))
1.193 paf 102: str++;
1.242 moko 103:
1.193 paf 104: if(!*str)
105: return 0;
106:
1.200 paf 107: bool negative=false;
108: if(str[0]=='-') {
109: negative=true;
110: str++;
111: } else if(str[0]=='+') {
112: str++;
113: }
1.193 paf 114:
1.242 moko 115: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 116:
1.242 moko 117: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 118: return -(int)result;
1.242 moko 119:
120: if(result<=INT_MAX)
1.243 moko 121: return (int)result;
1.242 moko 122:
123: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 124: }
125:
126: double pa_atod(const char* str, const String* problem_source) {
127: if(!str)
128: return 0;
129:
1.242 moko 130: while(isspace(*str))
1.193 paf 131: str++;
1.242 moko 132:
1.193 paf 133: if(!*str)
134: return 0;
135:
1.200 paf 136: bool negative=false;
137: if(str[0]=='-') {
138: negative=true;
139: str++;
140: } else if(str[0]=='+') {
141: str++;
142: }
1.242 moko 143:
144: double result;
1.247 moko 145: if(str[0]=='0') {
146: if(str[1]=='x' || str[1]=='X') {
1.242 moko 147: // 0xABC
1.249 moko 148: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 149: return negative ? -result : result;
150: } else {
1.200 paf 151: // skip leading 0000, to disable octal interpretation
1.242 moko 152: do str++; while(*str=='0');
1.200 paf 153: }
1.247 moko 154: }
1.242 moko 155:
156: char *error_pos;
157: result=strtod(str, &error_pos);
1.193 paf 158:
159: while(char c=*error_pos++)
1.199 paf 160: if(!isspace((unsigned char)c))
1.242 moko 161: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 162:
1.242 moko 163: return negative ? -result : result;
1.193 paf 164: }
165:
1.176 paf 166: // cord lib extension
167:
168: #ifndef DOXYGEN
169: typedef struct {
1.254 moko 170: ssize_t countdown;
171: int target; /* Character we're looking for */
1.176 paf 172: } chr_data;
173: #endif
1.248 moko 174:
1.176 paf 175: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
176: {
1.254 moko 177: register chr_data * d = (chr_data *)client_data;
178:
179: if (d -> countdown<=0) return(2);
180: d -> countdown -= size;
181: if (c > d -> target) return(1);
182: return(0);
1.176 paf 183: }
1.248 moko 184:
1.176 paf 185: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
186: {
1.254 moko 187: chr_data d;
1.176 paf 188:
1.254 moko 189: d.countdown = n;
190: d.target = c;
191: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
1.176 paf 192: }
193:
1.187 paf 194: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 195: {
1.254 moko 196: int* result=(int*)client_data;
197: (*result)++;
198: return(0); // 0=continue
1.178 paf 199: }
1.248 moko 200:
1.178 paf 201: size_t CORD_block_count(CORD x)
202: {
203: size_t result=0;
204: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 205: return result;
1.178 paf 206: }
207:
1.174 paf 208: // helpers
1.139 paf 209:
1.174 paf 210: /// String::match uses this as replace & global search table columns
1.139 paf 211:
1.174 paf 212: const int MAX_MATCH_GROUPS=100;
1.139 paf 213:
1.174 paf 214: class String_match_table_template_columns: public ArrayString {
215: public:
216: String_match_table_template_columns() {
217: *this+=new String("prematch");
218: *this+=new String("match");
219: *this+=new String("postmatch");
220: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 221: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 222: }
223: }
224: };
225:
226: Table string_match_table_template(new String_match_table_template_columns);
227:
1.176 paf 228: // String::Body methods
1.140 paf 229:
1.176 paf 230: String::Body String::Body::Format(int value) {
1.174 paf 231: char local[MAX_NUMBER];
232: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 233: return String::Body(pa_strdup(local, length));
1.120 paf 234: }
235:
1.254 moko 236: 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 237: size_t our_length=length();
238: if(!our_length)
239: return *this;
1.229 misha 240:
241: // check if any UTF-8 in chars
242: bool fast=true;
243: if(chars && source_charset && source_charset->isUTF8()){
244: const char* pos=chars;
245: while(unsigned char c=*pos++)
246: if(c>127){
247: fast=false;
248: break;
249: }
250: }
251:
252: size_t start=0;
253: size_t end=our_length;
1.195 paf 254: if(!chars)
255: chars=" \t\n"; // white space
256:
1.229 misha 257: if(fast){
258: // from left...
259: if(kind!=TRIM_END) {
260: CORD_pos pos; set_pos(pos, 0);
261: while(true) {
262: char c=CORD_pos_fetch(pos);
263: if(strchr(chars, c)) {
264: if(++start==our_length)
265: return 0; // all chars are empty, just return empty string
266: } else
267: break;
268:
269: CORD_next(pos);
270: }
271: }
272:
273: // from right..
274: if(kind!=TRIM_START) {
275: CORD_pos pos; set_pos(pos, end-1);
276: while(true) {
277: char c=CORD_pos_fetch(pos);
278: if(strchr(chars, c)) {
279: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
280: return 0; // all chars are empty, just return empty string
281: } else
282: break;
283:
284: CORD_prev(pos);
285: }
286: }
287: } else {
1.234 misha 288: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 289: const XMLByte* src_end=src_begin+our_length;
290:
291: // from left...
292: if(kind!=TRIM_END) {
293: while(src_begin<src_end){
294: uint char_length=1;
295: const XMLByte* ptr=src_begin;
1.231 misha 296: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
297: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 298: char_length++;
299:
300: bool found=false;
301: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
302: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
303: found=true;
304: break;
305: }
306:
307: if(found){
308: start+=char_length;
309: if(start==our_length)
310: return 0; // all chars are empty, just return empty string
311: } else
312: break;
313: }
314: }
1.196 paf 315:
1.229 misha 316: // from right..
317: if(kind!=TRIM_START) {
318: while(src_begin<src_end){
319: uint char_length=1;
1.231 misha 320: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
321: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 322: char_length++;
323:
324: bool found=false;
325: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
326: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
327: found=true;
328: break;
329: }
330:
331: if(found){
332: end-=char_length;
333: if(end==0)
334: return 0; // all chars are empty, just return empty string
335: } else
336: break;
337: }
1.196 paf 338: }
339: }
1.195 paf 340:
341: if(start==0 && end==our_length) // nobody moved a thing
342: return *this;
343:
344: if(out_start)
345: *out_start=start;
346: size_t new_length=end-start;
347: if(out_length)
348: *out_length=new_length;
349:
350: return mid(start, new_length);
351: }
352:
1.174 paf 353: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
354: uint& result=*static_cast<uint*>(client_data);
355: generic_hash_code(result, c);
356: return 0;
357: }
1.248 moko 358:
1.174 paf 359: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
360: uint& result=*static_cast<uint*>(client_data);
361: generic_hash_code(result, s);
362: return 0;
1.248 moko 363: }
364:
1.225 misha 365: uint String::Body::get_hash_code() const {
366: #ifdef HASH_CODE_CACHING
367: if(hash_code)
368: return hash_code;
369: #else
370: uint hash_code=0;
371: #endif
1.220 misha 372: if (body && CORD_IS_STRING(body)){
1.250 moko 373: generic_hash_code(hash_code, (const char *)body);
1.220 misha 374: } else {
375: CORD_iter5(body, 0,
376: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 377: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 378: }
1.225 misha 379: return hash_code;
1.94 parser 380: }
381:
1.241 misha 382: struct CORD_pos_info {
383: const char* chars;
384: size_t left;
385: size_t pos;
386: };
387:
388: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
389: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
390: if(info->pos < info->left){
391: info->pos=STRING_NOT_FOUND;
392: return 1;
393: }
394: if(strchr(info->chars, c))
395: return 1;
396: --(info->pos);
397: return 0;
398: }
399:
400: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
401: if(is_empty() || !chars || !strlen(chars))
402: return STRING_NOT_FOUND;
403: CORD_pos_info info={chars, left, right};
404: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
405: return info.pos;
406: else
407: return STRING_NOT_FOUND;
408: }
409:
410:
411: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
412: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
413: if(info->pos < info->left) {
414: info->pos=STRING_NOT_FOUND;
415: return 1;
416: }
417: if(!strchr(info->chars, c))
418: return 1;
419: --(info->pos);
420: return 0;
421: }
422:
423: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
424: if(is_empty() || !chars || !strlen(chars))
425: return STRING_NOT_FOUND;
426: CORD_pos_info info={chars, left, right};
427: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
428: return info.pos;
429: else
430: return STRING_NOT_FOUND;
431: }
432:
1.174 paf 433: // String methods
434:
435: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
436: if(!known_length)
1.9 paf 437: return *this;
1.122 paf 438:
1.176 paf 439: // first: langs
440: langs.append(body, lang, known_length);
441: // next: letters themselves
1.174 paf 442: body.append_know_length(str, known_length);
1.1 paf 443:
1.174 paf 444: ASSERT_STRING_INVARIANT(*this);
1.1 paf 445: return *this;
446: }
1.248 moko 447:
1.174 paf 448: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
449: if(!str)
450: return *this;
451: size_t known_length=helper_length?helper_length:strlen(str);
452: if(!known_length)
453: return *this;
1.1 paf 454:
1.174 paf 455: return append_know_length(str, known_length, lang);
1.5 paf 456: }
1.248 moko 457:
1.244 moko 458: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 459: char buf[MAX_NUMBER];
460: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
461: }
1.248 moko 462:
1.174 paf 463: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
464: size_t known_length=helper_length?helper_length:strlen(str);
465: if(!known_length)
466: return *this;
1.5 paf 467:
1.176 paf 468: // first: langs
469: langs.append(body, lang, known_length);
470: // next: letters themselves
1.174 paf 471: body.append_strdup_know_length(str, known_length);
1.33 paf 472:
1.174 paf 473: ASSERT_STRING_INVARIANT(*this);
474: return *this;
1.5 paf 475: }
1.46 paf 476:
1.234 misha 477: struct CORD_length_info {
478: size_t len;
479: size_t skip;
480: };
481:
482: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 ! moko 483: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
! 484: return 0;
1.221 misha 485: }
486:
1.234 misha 487: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
488: int CORD_batched_len(const char c, CORD_length_info* info){
489: if (info->skip==0){
490: info->len++;
491: info->skip = lengthUTF8Char(c)-1;
492: } else {
493: info->skip--;
494: }
1.220 misha 495: return 0;
496: }
497:
1.210 misha 498: size_t String::length(Charset& charset) const {
499: if(charset.isUTF8()){
1.234 misha 500: CORD_length_info info = {0, 0};
501: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
502: return info.len;
1.210 misha 503: } else
504: return body.length();
505: }
506:
1.174 paf 507: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
508: String& String::mid(size_t substr_begin, size_t substr_end) const {
509: String& result=*new String;
510:
511: size_t self_length=length();
512: substr_begin=min(substr_begin, self_length);
513: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 514: size_t substr_length=substr_end-substr_begin;
515: if(!substr_length)
1.107 parser 516: return result;
1.53 paf 517:
1.176 paf 518: // first: their langs
519: result.langs.append(result.body, langs, substr_begin, substr_length);
520: // next: letters themselves
521: result.body=body.mid(substr_begin, substr_length);
1.174 paf 522:
523: ASSERT_STRING_INVARIANT(result);
1.53 paf 524: return result;
1.54 paf 525: }
526:
1.211 misha 527: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
528: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 529: String& result=*new String;
530:
1.255 ! moko 531: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 532:
533: if(!self_length)
534: return result;
535:
536: from=min(min(to, from), self_length);
1.210 misha 537: to=min(max(to, from), self_length);
1.211 misha 538:
1.210 misha 539: size_t substr_length=to-from;
1.211 misha 540:
1.210 misha 541: if(!substr_length)
542: return result;
543:
544: if(charset.isUTF8()){
1.234 misha 545: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 546: const XMLByte* src_end=src_begin+body.length();
1.210 misha 547:
1.212 misha 548: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 549: from=getUTF8BytePos(src_begin, src_end, from);
550: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 551: if(!substr_length)
552: return result;
553: }
554:
555: // first: their langs
556: result.langs.append(result.body, langs, from, substr_length);
557: // next: letters themselves
558: result.body=body.mid(from, substr_length);
559:
560: ASSERT_STRING_INVARIANT(result);
561: return result;
562: }
563:
1.176 paf 564: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 565: size_t substr_length=substr.length();
566: while(true) {
567: size_t substr_begin=body.pos(substr, this_offset);
568:
569: if(substr_begin==CORD_NOT_FOUND)
570: return STRING_NOT_FOUND;
1.174 paf 571:
1.183 paf 572: if(langs.check_lang(lang, substr_begin, substr_length))
573: return substr_begin;
574:
575: this_offset=substr_begin+substr_length;
576: }
1.58 paf 577: }
578:
1.254 moko 579: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 580: return pos(substr.body, this_offset, lang);
1.60 paf 581: }
582:
1.254 moko 583: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 584:
585: if(charset.isUTF8()){
1.234 misha 586: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 587: const XMLByte* srcEnd=srcPtr+body.length();
588:
589: // convert 'this_offset' from 'characters' to 'bytes'
590: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
591:
1.229 misha 592: size_t result=pos(substr.body, this_offset, lang);
593: return (result==CORD_NOT_FOUND)
594: ? STRING_NOT_FOUND
595: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 596: } else {
1.229 misha 597: size_t result=pos(substr.body, this_offset, lang);
598: return (result==CORD_NOT_FOUND)
599: ? STRING_NOT_FOUND
600: : result;
1.210 misha 601: }
602: }
603:
1.254 moko 604: void String::split(ArrayString& result, size_t& pos_after, const char* delim, Language lang, int limit) const {
1.237 misha 605: if(is_empty())
606: return;
1.174 paf 607: size_t self_length=length();
1.237 misha 608: if(size_t delim_length=strlen(delim)) {
1.186 paf 609: size_t pos_before;
1.60 paf 610: // while we have 'delim'...
1.174 paf 611: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 612: result+=&mid(pos_after, pos_before);
1.174 paf 613: pos_after=pos_before+delim_length;
1.60 paf 614: }
615: // last piece
1.174 paf 616: if(pos_after<self_length && limit) {
617: result+=&mid(pos_after, self_length);
618: pos_after=self_length;
1.60 paf 619: }
620: } else { // empty delim
621: result+=this;
1.174 paf 622: pos_after+=self_length;
1.60 paf 623: }
624: }
625:
1.254 moko 626: void String::split(ArrayString& result, size_t& pos_after, const String& delim, Language lang, int limit) const {
1.237 misha 627: if(is_empty())
628: return;
629: if(!delim.is_empty()) {
1.186 paf 630: size_t pos_before;
1.60 paf 631: // while we have 'delim'...
1.174 paf 632: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 633: result+=&mid(pos_after, pos_before);
1.174 paf 634: pos_after=pos_before+delim.length();
1.60 paf 635: }
636: // last piece
1.174 paf 637: if(pos_after<length() && limit) {
638: result+=&mid(pos_after, length());
639: pos_after=length();
1.60 paf 640: }
641: } else { // empty delim
642: result+=this;
1.174 paf 643: pos_after+=length();
1.60 paf 644: }
1.61 paf 645: }
646:
1.254 moko 647: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 648:
1.222 misha 649: // vregex->info(); // I have no idea what does it for?
1.63 paf 650:
1.222 misha 651: bool need_pre_post_match=vregex->is_pre_post_match_needed();
652: bool global=vregex->is_global_search();
1.63 paf 653:
1.174 paf 654: const char* subject=cstr();
1.234 misha 655: size_t subject_length=length();
1.222 misha 656: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
657: int ovector[ovector_size];
1.155 paf 658:
1.173 paf 659: Table::Action_options table_options;
1.174 paf 660: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 661:
1.154 paf 662: int prestart=0;
663: int poststart=0;
1.174 paf 664: int postfinish=length();
1.63 paf 665: while(true) {
1.222 misha 666: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 667:
1.222 misha 668: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
669: break;
1.63 paf 670:
1.154 paf 671: int prefinish=ovector[0];
672: poststart=ovector[1];
1.236 moko 673:
674: if (prestart==poststart && subject[poststart]=='\n'){
675: prestart++;
676: continue;
677: }
678:
1.232 misha 679: ArrayString* row=new ArrayString(3);
1.174 paf 680: if(need_pre_post_match) {
681: *row+=&mid(0, prefinish); // .prematch column value
682: *row+=&mid(prefinish, poststart); // .match
683: *row+=&mid(poststart, postfinish); // .postmatch
684: } else {
1.185 paf 685: *row+=&Empty; // .prematch column value
686: *row+=&Empty; // .match
687: *row+=&Empty; // .postmatch
1.174 paf 688: }
1.63 paf 689:
1.222 misha 690: for(int i=1; i<exec_result; i++) {
1.69 paf 691: // -1:-1 case handled peacefully by mid() itself
1.232 misha 692: *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 693: }
694:
1.209 misha 695: matches_count++;
1.174 paf 696: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 697:
1.222 misha 698: if(!global || prestart==poststart) // last step
699: break;
700:
1.154 paf 701: prestart=poststart;
1.222 misha 702: }
1.63 paf 703:
1.222 misha 704: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
705: return vregex->is_just_count() ? 0 : &table;
1.82 parser 706: }
707:
1.174 paf 708: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
709: String& result=*new String();
710: if(is_empty())
711: return result;
712:
713: char* new_cstr=cstrm();
1.216 misha 714:
1.181 paf 715: if(source_charset.isUTF8()) {
1.216 misha 716: size_t new_cstr_len=length();
1.181 paf 717: switch(kind) {
718: case CC_UPPER:
1.192 paf 719: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 720: break;
721: case CC_LOWER:
1.192 paf 722: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 723: break;
724: default:
725: assert(!"unknown change case kind");
726: break; // never
727: }
728:
729: } else {
730: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 731:
1.181 paf 732: const unsigned char *a;
733: const unsigned char *b;
734: switch(kind) {
735: case CC_UPPER:
736: a=tables+lcc_offset;
737: b=tables+fcc_offset;
738: break;
739: case CC_LOWER:
740: a=tables+lcc_offset;
741: b=0;
742: break;
743: default:
744: assert(!"unknown change case kind");
745: a=b=0; // calm, compiler
746: break; // never
747: }
748:
1.192 paf 749: char *dest=new_cstr;
1.181 paf 750: unsigned char index;
1.190 paf 751: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 752: unsigned char c=a[index];
753: if(b)
754: c=b[c];
755:
756: *dest++=(char)c;
757: }
1.174 paf 758: }
1.176 paf 759: result.langs=langs;
1.174 paf 760: result.body=new_cstr;
1.89 parser 761:
1.101 parser 762: return result;
763: }
764:
1.213 misha 765: const String& String::escape(Charset& source_charset) const {
766: if(is_empty())
767: return *this;
768:
769: return Charset::escape(*this, source_charset);
770: }
771:
1.238 misha 772: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
773: result.langs.append(result.body, langs, langs_offset, length); \
774: result.body.append_strdup_know_length(from_cstr, length);
775:
1.174 paf 776: const String& String::replace(const Dictionary& dict) const {
1.238 misha 777: if(!dict.count() || is_empty())
778: return *this;
779:
1.174 paf 780: String& result=*new String();
781: const char* old_cstr=cstr();
782: const char* prematch_begin=old_cstr;
783:
1.238 misha 784: if(dict.count()==1) {
785: // optimized simple case
786:
787: Dictionary::Subst subst=dict.get(0);
1.239 moko 788: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 789: // prematch
1.238 misha 790: if(size_t prematch_length=p-prematch_begin) {
791: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 792: }
793:
1.174 paf 794: // match
1.238 misha 795: prematch_begin=p+subst.from_length;
1.174 paf 796:
1.184 paf 797: if(const String* b=subst.to) // are there any b?
1.174 paf 798: result<<*b;
1.238 misha 799: }
800:
801: } else {
802:
803: const char* current=old_cstr;
804: while(*current) {
805: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
806: // prematch
807: if(size_t prematch_length=current-prematch_begin) {
808: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
809: }
810:
811: // match
812: // skip 'a' in 'current'; move prematch_begin
813: current+=subst.from_length; prematch_begin=current;
814:
815: if(const String* b=subst.to) // are there any b?
816: result<<*b;
817: } else // simply advance
818: current++;
819: }
820:
1.174 paf 821: }
1.156 paf 822:
1.238 misha 823: if(prematch_begin==old_cstr) // not modified
824: return *this;
825:
1.174 paf 826: // postmatch
1.238 misha 827: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
828: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 829: }
1.156 paf 830:
1.174 paf 831: ASSERT_STRING_INVARIANT(result);
1.82 parser 832: return result;
1.61 paf 833: }
1.113 parser 834:
1.180 paf 835: static int serialize_body_char(char c, char** cur) {
836: *((*cur)++)=c;
837: return 0; // 0=continue
1.248 moko 838: }
839:
1.174 paf 840: static int serialize_body_piece(const char* s, char** cur) {
841: size_t length=strlen(s);
842: memcpy(*cur, s, length); *cur+=length;
1.178 paf 843: return 0; // 0=continue
1.248 moko 844: }
845:
1.178 paf 846: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
847: // lang
1.191 paf 848: **cur=alang; (*cur)++;
849: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 850: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
851:
852: return 0; // 0=continue
853: }
1.248 moko 854:
1.174 paf 855: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 856: size_t fragments_count=langs.count();
1.202 paf 857: size_t body_length=body.length();
1.174 paf 858: size_t buf_length=
1.178 paf 859: prolog_length //1
860: +sizeof(size_t) //2
1.202 paf 861: +body_length //3
862: +1 // 4 for zero terminator used in deserialize
863: +sizeof(size_t) //5
864: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
865:
1.174 paf 866: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
867:
868: // 1: prolog
869: char *cur=result.str+prolog_length;
1.202 paf 870: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
871: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
872: // 3: letters
873: body.for_each(serialize_body_char, serialize_body_piece, &cur);
874: // 4: zero terminator
875: *cur++=0;
876: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 877: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 878: // 6: lang info
1.178 paf 879: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 880:
1.174 paf 881: return result;
1.113 parser 882: }
1.248 moko 883:
1.202 paf 884: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
885: size_t in_buf=buf_size;
886: if(in_buf<=prolog_size)
1.148 paf 887: return false;
1.202 paf 888: in_buf-=prolog_size;
1.135 paf 889:
1.174 paf 890: // 1: prolog
1.202 paf 891: const char* cur=(const char* )buf+prolog_size;
1.113 parser 892:
1.207 paf 893: // 2: chars.count
1.202 paf 894: size_t body_length;
895: if(in_buf<sizeof(body_length)) // body.length don't fit?
896: return false;
897: // [WARNING: not cast, addresses must be %4=0 on sparc]
898: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
899: in_buf-=sizeof(body_length);
900:
901: if(in_buf<body_length+1) // letters+terminator don't fit?
902: return false;
903: // 4: zero terminator
904: if(cur[body_length] != 0) // in place?
905: return false;
906: // 3: letters
1.251 moko 907: body=String::Body(String::C(cur, body_length));
1.202 paf 908: cur+=body_length+1;
909: in_buf-=body_length+1;
910:
911: // 5: langs.count
1.191 paf 912: size_t fragments_count;
1.202 paf 913: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 914: return false;
1.191 paf 915: // [WARNING: not cast, addresses must be %4=0 on sparc]
916: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 917: in_buf-=sizeof(fragments_count);
1.174 paf 918:
919: if(fragments_count) {
1.202 paf 920: // 6: lang info
1.174 paf 921: size_t total_length=0;
922: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 923: char lang;
924: size_t fragment_length;
925: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 926: if(in_buf<piece_length) // lang+length
1.174 paf 927: return false;
928:
1.191 paf 929: // lang
930: lang=*cur++;
931: // length [WARNING: not cast, addresses must be %4=0 on sparc]
932: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
933:
1.206 paf 934: size_t combined_length=total_length+fragment_length;
935: if(combined_length>body_length)
936: return false; // file curruption
1.191 paf 937: // uchar needed to prevent propagating 0x80 bit to upper bytes
938: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 939: total_length=combined_length;
1.202 paf 940: in_buf-=piece_length;
1.174 paf 941: }
1.128 paf 942:
1.202 paf 943: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 944: return false;
1.174 paf 945: }
1.202 paf 946: if(in_buf!=0) // some strange extra bytes
947: return false;
1.113 parser 948:
1.174 paf 949: ASSERT_STRING_INVARIANT(*this);
1.148 paf 950: return true;
1.176 paf 951: }
952:
953: const char* String::Body::v() const {
1.233 misha 954: return CORD_to_const_char_star(body, length());
1.176 paf 955: }
1.248 moko 956:
1.201 paf 957: void String::Body::dump() const {
958: CORD_dump(body);
959: }
960:
1.176 paf 961: const char* String::Languages::v() const {
1.177 paf 962: if(opt.is_not_just_lang)
1.233 misha 963: return CORD_to_const_char_star(langs, 0);
1.176 paf 964: else
965: return (const char*)&langs;
966: }
1.248 moko 967:
1.201 paf 968: void String::Languages::dump() const {
969: if(opt.is_not_just_lang)
970: CORD_dump(langs);
971: else
972: puts((const char*)&langs);
973: }
1.248 moko 974:
1.176 paf 975: const char* String::v() const {
1.198 paf 976: const uint LIMIT_VIEW=20;
1.246 moko 977: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 978: const char*body_view=body.v();
979: const char*langs_view=langs.v();
980: snprintf(buf, MAX_STRING,
1.178 paf 981: "%d:%.*s%s} "
1.176 paf 982: "{%d:%s",
1.178 paf 983: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 984: strlen(body_view), body_view
985: );
986:
987: return buf;
1.113 parser 988: }
1.229 misha 989:
1.201 paf 990: void String::dump() const {
991: body.dump();
992: langs.dump();
993: }
1.229 misha 994:
995: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 996: if(is_empty())
1.195 paf 997: return *this;
998:
999: size_t substr_begin, substr_length;
1.229 misha 1000: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1001: if(new_body==body) // we received unchanged pointer, do likewise
1002: return *this;
1003: // new_body differs from body, adjust langs along
1004:
1005: String& result=*new String;
1006: if(!new_body) // body.trim produced empty result
1007: return result;
1008: // body.trim produced nonempty result
1009:
1010: // first: their langs
1011: result.langs.append(result.body, langs, substr_begin, substr_length);
1012: // next: letters themselves
1013: result.body=new_body;
1014:
1015: ASSERT_STRING_INVARIANT(result);
1016: return result;
1.198 paf 1017: }
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