Annotation of parser3/src/main/pa_string.C, revision 1.266
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.266 ! moko 4: Copyright (c) 2001-2020 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:
1.266 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.265 2020/12/02 15:25:41 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.262 moko 19: #define COMPILE_ASSERT(x) extern int assert_checker[(x) ? 1 : -1]
1.261 moko 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){
1.263 moko 89: if(!str)
90: return 0;
91:
1.254 moko 92: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 93: }
94:
1.265 moko 95: uint64_t pa_atoul(const char *str, int base, const String* problem_source){
1.263 moko 96: if(!str)
97: return 0;
98:
1.265 moko 99: return pa_ato_any<uint64_t>(str, base, problem_source, ULLONG_MAX);
1.249 moko 100: }
101:
1.264 moko 102: int pa_atoi(const char* str, int base, const String* problem_source) {
1.193 paf 103: if(!str)
104: return 0;
105:
1.242 moko 106: while(isspace(*str))
1.193 paf 107: str++;
1.242 moko 108:
1.193 paf 109: if(!*str)
110: return 0;
111:
1.200 paf 112: bool negative=false;
113: if(str[0]=='-') {
114: negative=true;
115: str++;
116: } else if(str[0]=='+') {
117: str++;
118: }
1.193 paf 119:
1.264 moko 120: unsigned int result=pa_atoui(str, base, problem_source);
1.193 paf 121:
1.242 moko 122: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 123: return -(int)result;
1.242 moko 124:
125: if(result<=INT_MAX)
1.243 moko 126: return (int)result;
1.242 moko 127:
128: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 129: }
130:
131: double pa_atod(const char* str, const String* problem_source) {
132: if(!str)
133: return 0;
134:
1.242 moko 135: while(isspace(*str))
1.193 paf 136: str++;
1.242 moko 137:
1.193 paf 138: if(!*str)
139: return 0;
140:
1.200 paf 141: bool negative=false;
142: if(str[0]=='-') {
143: negative=true;
144: str++;
145: } else if(str[0]=='+') {
146: str++;
147: }
1.242 moko 148:
149: double result;
1.247 moko 150: if(str[0]=='0') {
151: if(str[1]=='x' || str[1]=='X') {
1.242 moko 152: // 0xABC
1.249 moko 153: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 154: return negative ? -result : result;
155: } else {
1.200 paf 156: // skip leading 0000, to disable octal interpretation
1.242 moko 157: do str++; while(*str=='0');
1.200 paf 158: }
1.247 moko 159: }
1.242 moko 160:
161: char *error_pos;
162: result=strtod(str, &error_pos);
1.193 paf 163:
164: while(char c=*error_pos++)
1.199 paf 165: if(!isspace((unsigned char)c))
1.242 moko 166: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 167:
1.242 moko 168: return negative ? -result : result;
1.193 paf 169: }
170:
1.176 paf 171: // cord lib extension
172:
173: #ifndef DOXYGEN
174: typedef struct {
1.254 moko 175: ssize_t countdown;
176: int target; /* Character we're looking for */
1.176 paf 177: } chr_data;
178: #endif
1.248 moko 179:
1.176 paf 180: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
181: {
1.254 moko 182: register chr_data * d = (chr_data *)client_data;
183:
184: if (d -> countdown<=0) return(2);
185: d -> countdown -= size;
186: if (c > d -> target) return(1);
187: return(0);
1.176 paf 188: }
1.248 moko 189:
1.176 paf 190: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
191: {
1.254 moko 192: chr_data d;
1.176 paf 193:
1.254 moko 194: d.countdown = n;
195: d.target = c;
196: 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 197: }
198:
1.187 paf 199: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 200: {
1.254 moko 201: int* result=(int*)client_data;
202: (*result)++;
203: return(0); // 0=continue
1.178 paf 204: }
1.248 moko 205:
1.178 paf 206: size_t CORD_block_count(CORD x)
207: {
208: size_t result=0;
209: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 210: return result;
1.178 paf 211: }
212:
1.174 paf 213: // helpers
1.139 paf 214:
1.174 paf 215: /// String::match uses this as replace & global search table columns
1.139 paf 216:
1.174 paf 217: const int MAX_MATCH_GROUPS=100;
1.139 paf 218:
1.174 paf 219: class String_match_table_template_columns: public ArrayString {
220: public:
221: String_match_table_template_columns() {
222: *this+=new String("prematch");
223: *this+=new String("match");
224: *this+=new String("postmatch");
225: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 226: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 227: }
228: }
229: };
230:
231: Table string_match_table_template(new String_match_table_template_columns);
232:
1.176 paf 233: // String::Body methods
1.140 paf 234:
1.176 paf 235: String::Body String::Body::Format(int value) {
1.174 paf 236: char local[MAX_NUMBER];
237: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 238: return String::Body(pa_strdup(local, length));
1.120 paf 239: }
240:
1.254 moko 241: 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 242: size_t our_length=length();
243: if(!our_length)
244: return *this;
1.229 misha 245:
246: // check if any UTF-8 in chars
247: bool fast=true;
248: if(chars && source_charset && source_charset->isUTF8()){
249: const char* pos=chars;
250: while(unsigned char c=*pos++)
251: if(c>127){
252: fast=false;
253: break;
254: }
255: }
256:
257: size_t start=0;
258: size_t end=our_length;
1.195 paf 259: if(!chars)
260: chars=" \t\n"; // white space
261:
1.229 misha 262: if(fast){
263: // from left...
264: if(kind!=TRIM_END) {
265: CORD_pos pos; set_pos(pos, 0);
266: while(true) {
267: char c=CORD_pos_fetch(pos);
268: if(strchr(chars, c)) {
269: if(++start==our_length)
270: return 0; // all chars are empty, just return empty string
271: } else
272: break;
273:
274: CORD_next(pos);
275: }
276: }
277:
278: // from right..
279: if(kind!=TRIM_START) {
280: CORD_pos pos; set_pos(pos, end-1);
281: while(true) {
282: char c=CORD_pos_fetch(pos);
283: if(strchr(chars, c)) {
284: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
285: return 0; // all chars are empty, just return empty string
286: } else
287: break;
288:
289: CORD_prev(pos);
290: }
291: }
292: } else {
1.234 misha 293: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 294: const XMLByte* src_end=src_begin+our_length;
295:
296: // from left...
297: if(kind!=TRIM_END) {
298: while(src_begin<src_end){
299: uint char_length=1;
300: const XMLByte* ptr=src_begin;
1.231 misha 301: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
302: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 303: char_length++;
304:
305: bool found=false;
306: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
307: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
308: found=true;
309: break;
310: }
311:
312: if(found){
313: start+=char_length;
314: if(start==our_length)
315: return 0; // all chars are empty, just return empty string
316: } else
317: break;
318: }
319: }
1.196 paf 320:
1.229 misha 321: // from right..
322: if(kind!=TRIM_START) {
323: while(src_begin<src_end){
324: uint char_length=1;
1.231 misha 325: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
326: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 327: char_length++;
328:
329: bool found=false;
330: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
331: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
332: found=true;
333: break;
334: }
335:
336: if(found){
337: end-=char_length;
338: if(end==0)
339: return 0; // all chars are empty, just return empty string
340: } else
341: break;
342: }
1.196 paf 343: }
344: }
1.195 paf 345:
346: if(start==0 && end==our_length) // nobody moved a thing
347: return *this;
348:
349: if(out_start)
350: *out_start=start;
351: size_t new_length=end-start;
352: if(out_length)
353: *out_length=new_length;
354:
355: return mid(start, new_length);
356: }
357:
1.174 paf 358: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
359: uint& result=*static_cast<uint*>(client_data);
360: generic_hash_code(result, c);
361: return 0;
362: }
1.248 moko 363:
1.174 paf 364: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
365: uint& result=*static_cast<uint*>(client_data);
366: generic_hash_code(result, s);
367: return 0;
1.248 moko 368: }
369:
1.225 misha 370: uint String::Body::get_hash_code() const {
371: #ifdef HASH_CODE_CACHING
372: if(hash_code)
373: return hash_code;
374: #else
375: uint hash_code=0;
376: #endif
1.220 misha 377: if (body && CORD_IS_STRING(body)){
1.250 moko 378: generic_hash_code(hash_code, (const char *)body);
1.220 misha 379: } else {
380: CORD_iter5(body, 0,
381: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 382: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 383: }
1.225 misha 384: return hash_code;
1.94 parser 385: }
386:
1.241 misha 387: struct CORD_pos_info {
388: const char* chars;
389: size_t left;
390: size_t pos;
391: };
392:
393: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
394: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
395: if(info->pos < info->left){
396: info->pos=STRING_NOT_FOUND;
397: return 1;
398: }
399: if(strchr(info->chars, c))
400: return 1;
401: --(info->pos);
402: return 0;
403: }
404:
405: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
406: if(is_empty() || !chars || !strlen(chars))
407: return STRING_NOT_FOUND;
408: CORD_pos_info info={chars, left, right};
409: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
410: return info.pos;
411: else
412: return STRING_NOT_FOUND;
413: }
414:
415:
416: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
417: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
418: if(info->pos < info->left) {
419: info->pos=STRING_NOT_FOUND;
420: return 1;
421: }
422: if(!strchr(info->chars, c))
423: return 1;
424: --(info->pos);
425: return 0;
426: }
427:
428: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
429: if(is_empty() || !chars || !strlen(chars))
430: return STRING_NOT_FOUND;
431: CORD_pos_info info={chars, left, right};
432: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
433: return info.pos;
434: else
435: return STRING_NOT_FOUND;
436: }
437:
1.174 paf 438: // String methods
439:
440: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
441: if(!known_length)
1.9 paf 442: return *this;
1.122 paf 443:
1.176 paf 444: // first: langs
445: langs.append(body, lang, known_length);
446: // next: letters themselves
1.174 paf 447: body.append_know_length(str, known_length);
1.1 paf 448:
1.174 paf 449: ASSERT_STRING_INVARIANT(*this);
1.1 paf 450: return *this;
451: }
1.248 moko 452:
1.174 paf 453: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
454: if(!str)
455: return *this;
456: size_t known_length=helper_length?helper_length:strlen(str);
457: if(!known_length)
458: return *this;
1.1 paf 459:
1.174 paf 460: return append_know_length(str, known_length, lang);
1.5 paf 461: }
1.248 moko 462:
1.244 moko 463: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 464: char buf[MAX_NUMBER];
465: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
466: }
1.248 moko 467:
1.174 paf 468: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
469: size_t known_length=helper_length?helper_length:strlen(str);
470: if(!known_length)
471: return *this;
1.5 paf 472:
1.176 paf 473: // first: langs
474: langs.append(body, lang, known_length);
475: // next: letters themselves
1.174 paf 476: body.append_strdup_know_length(str, known_length);
1.33 paf 477:
1.174 paf 478: ASSERT_STRING_INVARIANT(*this);
479: return *this;
1.5 paf 480: }
1.46 paf 481:
1.234 misha 482: struct CORD_length_info {
483: size_t len;
484: size_t skip;
485: };
486:
487: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 488: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
489: return 0;
1.221 misha 490: }
491:
1.234 misha 492: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
493: int CORD_batched_len(const char c, CORD_length_info* info){
494: if (info->skip==0){
495: info->len++;
496: info->skip = lengthUTF8Char(c)-1;
497: } else {
498: info->skip--;
499: }
1.220 misha 500: return 0;
501: }
502:
1.210 misha 503: size_t String::length(Charset& charset) const {
504: if(charset.isUTF8()){
1.234 misha 505: CORD_length_info info = {0, 0};
506: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
507: return info.len;
1.210 misha 508: } else
509: return body.length();
510: }
511:
1.174 paf 512: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
513: String& String::mid(size_t substr_begin, size_t substr_end) const {
514: String& result=*new String;
515:
516: size_t self_length=length();
517: substr_begin=min(substr_begin, self_length);
518: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 519: size_t substr_length=substr_end-substr_begin;
520: if(!substr_length)
1.107 parser 521: return result;
1.53 paf 522:
1.176 paf 523: // first: their langs
524: result.langs.append(result.body, langs, substr_begin, substr_length);
525: // next: letters themselves
526: result.body=body.mid(substr_begin, substr_length);
1.174 paf 527:
528: ASSERT_STRING_INVARIANT(result);
1.53 paf 529: return result;
1.54 paf 530: }
531:
1.211 misha 532: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
533: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 534: String& result=*new String;
535:
1.255 moko 536: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 537:
538: if(!self_length)
539: return result;
540:
541: from=min(min(to, from), self_length);
1.210 misha 542: to=min(max(to, from), self_length);
1.211 misha 543:
1.210 misha 544: size_t substr_length=to-from;
1.211 misha 545:
1.210 misha 546: if(!substr_length)
547: return result;
548:
549: if(charset.isUTF8()){
1.234 misha 550: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 551: const XMLByte* src_end=src_begin+body.length();
1.210 misha 552:
1.212 misha 553: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 554: from=getUTF8BytePos(src_begin, src_end, from);
555: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 556: if(!substr_length)
557: return result;
558: }
559:
560: // first: their langs
561: result.langs.append(result.body, langs, from, substr_length);
562: // next: letters themselves
563: result.body=body.mid(from, substr_length);
564:
565: ASSERT_STRING_INVARIANT(result);
566: return result;
567: }
568:
1.176 paf 569: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 570: size_t substr_length=substr.length();
571: while(true) {
572: size_t substr_begin=body.pos(substr, this_offset);
573:
574: if(substr_begin==CORD_NOT_FOUND)
575: return STRING_NOT_FOUND;
1.174 paf 576:
1.183 paf 577: if(langs.check_lang(lang, substr_begin, substr_length))
578: return substr_begin;
579:
580: this_offset=substr_begin+substr_length;
581: }
1.58 paf 582: }
583:
1.254 moko 584: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 585: return pos(substr.body, this_offset, lang);
1.60 paf 586: }
587:
1.254 moko 588: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 589:
590: if(charset.isUTF8()){
1.234 misha 591: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 592: const XMLByte* srcEnd=srcPtr+body.length();
593:
594: // convert 'this_offset' from 'characters' to 'bytes'
595: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
596:
1.229 misha 597: size_t result=pos(substr.body, this_offset, lang);
598: return (result==CORD_NOT_FOUND)
599: ? STRING_NOT_FOUND
600: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 601: } else {
1.229 misha 602: size_t result=pos(substr.body, this_offset, lang);
603: return (result==CORD_NOT_FOUND)
604: ? STRING_NOT_FOUND
605: : result;
1.210 misha 606: }
607: }
608:
1.256 moko 609: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 610: if(is_empty())
611: return;
1.174 paf 612: size_t self_length=length();
1.237 misha 613: if(size_t delim_length=strlen(delim)) {
1.186 paf 614: size_t pos_before;
1.60 paf 615: // while we have 'delim'...
1.256 moko 616: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 617: result+=&mid(pos_after, pos_before);
1.174 paf 618: pos_after=pos_before+delim_length;
1.60 paf 619: }
620: // last piece
1.256 moko 621: if(pos_after<self_length)
1.174 paf 622: result+=&mid(pos_after, self_length);
1.60 paf 623: } else { // empty delim
624: result+=this;
625: }
626: }
627:
1.256 moko 628: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 629: if(is_empty())
630: return;
631: if(!delim.is_empty()) {
1.186 paf 632: size_t pos_before;
1.60 paf 633: // while we have 'delim'...
1.256 moko 634: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 635: result+=&mid(pos_after, pos_before);
1.174 paf 636: pos_after=pos_before+delim.length();
1.60 paf 637: }
638: // last piece
1.256 moko 639: if(pos_after<length())
1.174 paf 640: result+=&mid(pos_after, length());
1.60 paf 641: } else { // empty delim
642: result+=this;
643: }
1.61 paf 644: }
645:
1.254 moko 646: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 647:
1.222 misha 648: // vregex->info(); // I have no idea what does it for?
1.63 paf 649:
1.222 misha 650: bool need_pre_post_match=vregex->is_pre_post_match_needed();
651: bool global=vregex->is_global_search();
1.63 paf 652:
1.174 paf 653: const char* subject=cstr();
1.234 misha 654: size_t subject_length=length();
1.222 misha 655: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
656: int ovector[ovector_size];
1.155 paf 657:
1.173 paf 658: Table::Action_options table_options;
1.174 paf 659: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 660:
1.154 paf 661: int prestart=0;
662: int poststart=0;
1.174 paf 663: int postfinish=length();
1.63 paf 664: while(true) {
1.222 misha 665: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 666:
1.222 misha 667: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
668: break;
1.63 paf 669:
1.154 paf 670: int prefinish=ovector[0];
671: poststart=ovector[1];
1.236 moko 672:
673: if (prestart==poststart && subject[poststart]=='\n'){
674: prestart++;
675: continue;
676: }
677:
1.232 misha 678: ArrayString* row=new ArrayString(3);
1.174 paf 679: if(need_pre_post_match) {
680: *row+=&mid(0, prefinish); // .prematch column value
681: *row+=&mid(prefinish, poststart); // .match
682: *row+=&mid(poststart, postfinish); // .postmatch
683: } else {
1.185 paf 684: *row+=&Empty; // .prematch column value
685: *row+=&Empty; // .match
686: *row+=&Empty; // .postmatch
1.174 paf 687: }
1.63 paf 688:
1.222 misha 689: for(int i=1; i<exec_result; i++) {
1.69 paf 690: // -1:-1 case handled peacefully by mid() itself
1.232 misha 691: *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 692: }
693:
1.209 misha 694: matches_count++;
1.174 paf 695: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 696:
1.222 misha 697: if(!global || prestart==poststart) // last step
698: break;
699:
1.154 paf 700: prestart=poststart;
1.222 misha 701: }
1.63 paf 702:
1.222 misha 703: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
704: return vregex->is_just_count() ? 0 : &table;
1.82 parser 705: }
706:
1.174 paf 707: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
708: String& result=*new String();
709: if(is_empty())
710: return result;
711:
712: char* new_cstr=cstrm();
1.216 misha 713:
1.181 paf 714: if(source_charset.isUTF8()) {
1.216 misha 715: size_t new_cstr_len=length();
1.181 paf 716: switch(kind) {
717: case CC_UPPER:
1.192 paf 718: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 719: break;
720: case CC_LOWER:
1.192 paf 721: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 722: break;
723: default:
724: assert(!"unknown change case kind");
725: break; // never
726: }
727:
728: } else {
729: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 730:
1.181 paf 731: const unsigned char *a;
732: const unsigned char *b;
733: switch(kind) {
734: case CC_UPPER:
735: a=tables+lcc_offset;
736: b=tables+fcc_offset;
737: break;
738: case CC_LOWER:
739: a=tables+lcc_offset;
740: b=0;
741: break;
742: default:
743: assert(!"unknown change case kind");
744: a=b=0; // calm, compiler
745: break; // never
746: }
747:
1.192 paf 748: char *dest=new_cstr;
1.181 paf 749: unsigned char index;
1.190 paf 750: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 751: unsigned char c=a[index];
752: if(b)
753: c=b[c];
754:
755: *dest++=(char)c;
756: }
1.174 paf 757: }
1.176 paf 758: result.langs=langs;
1.174 paf 759: result.body=new_cstr;
1.89 parser 760:
1.101 parser 761: return result;
762: }
763:
1.213 misha 764: const String& String::escape(Charset& source_charset) const {
765: if(is_empty())
766: return *this;
767:
768: return Charset::escape(*this, source_charset);
769: }
770:
1.238 misha 771: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
772: result.langs.append(result.body, langs, langs_offset, length); \
773: result.body.append_strdup_know_length(from_cstr, length);
774:
1.174 paf 775: const String& String::replace(const Dictionary& dict) const {
1.238 misha 776: if(!dict.count() || is_empty())
777: return *this;
778:
1.174 paf 779: String& result=*new String();
780: const char* old_cstr=cstr();
781: const char* prematch_begin=old_cstr;
782:
1.238 misha 783: if(dict.count()==1) {
784: // optimized simple case
785:
786: Dictionary::Subst subst=dict.get(0);
1.239 moko 787: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 788: // prematch
1.238 misha 789: if(size_t prematch_length=p-prematch_begin) {
790: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 791: }
792:
1.174 paf 793: // match
1.238 misha 794: prematch_begin=p+subst.from_length;
1.174 paf 795:
1.184 paf 796: if(const String* b=subst.to) // are there any b?
1.174 paf 797: result<<*b;
1.238 misha 798: }
799:
800: } else {
801:
802: const char* current=old_cstr;
803: while(*current) {
804: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
805: // prematch
806: if(size_t prematch_length=current-prematch_begin) {
807: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
808: }
809:
810: // match
811: // skip 'a' in 'current'; move prematch_begin
812: current+=subst.from_length; prematch_begin=current;
813:
814: if(const String* b=subst.to) // are there any b?
815: result<<*b;
816: } else // simply advance
817: current++;
818: }
819:
1.174 paf 820: }
1.156 paf 821:
1.238 misha 822: if(prematch_begin==old_cstr) // not modified
823: return *this;
824:
1.174 paf 825: // postmatch
1.238 misha 826: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
827: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 828: }
1.156 paf 829:
1.174 paf 830: ASSERT_STRING_INVARIANT(result);
1.82 parser 831: return result;
1.61 paf 832: }
1.113 parser 833:
1.180 paf 834: static int serialize_body_char(char c, char** cur) {
835: *((*cur)++)=c;
836: return 0; // 0=continue
1.248 moko 837: }
838:
1.174 paf 839: static int serialize_body_piece(const char* s, char** cur) {
840: size_t length=strlen(s);
841: memcpy(*cur, s, length); *cur+=length;
1.178 paf 842: return 0; // 0=continue
1.248 moko 843: }
844:
1.178 paf 845: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
846: // lang
1.191 paf 847: **cur=alang; (*cur)++;
848: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 849: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
850:
851: return 0; // 0=continue
852: }
1.248 moko 853:
1.174 paf 854: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 855: size_t fragments_count=langs.count();
1.202 paf 856: size_t body_length=body.length();
1.174 paf 857: size_t buf_length=
1.178 paf 858: prolog_length //1
859: +sizeof(size_t) //2
1.202 paf 860: +body_length //3
861: +1 // 4 for zero terminator used in deserialize
862: +sizeof(size_t) //5
863: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
864:
1.174 paf 865: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
866:
867: // 1: prolog
868: char *cur=result.str+prolog_length;
1.202 paf 869: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
870: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
871: // 3: letters
872: body.for_each(serialize_body_char, serialize_body_piece, &cur);
873: // 4: zero terminator
874: *cur++=0;
875: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 876: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 877: // 6: lang info
1.178 paf 878: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 879:
1.174 paf 880: return result;
1.113 parser 881: }
1.248 moko 882:
1.202 paf 883: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
884: size_t in_buf=buf_size;
885: if(in_buf<=prolog_size)
1.148 paf 886: return false;
1.202 paf 887: in_buf-=prolog_size;
1.135 paf 888:
1.174 paf 889: // 1: prolog
1.202 paf 890: const char* cur=(const char* )buf+prolog_size;
1.113 parser 891:
1.207 paf 892: // 2: chars.count
1.202 paf 893: size_t body_length;
894: if(in_buf<sizeof(body_length)) // body.length don't fit?
895: return false;
896: // [WARNING: not cast, addresses must be %4=0 on sparc]
897: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
898: in_buf-=sizeof(body_length);
899:
900: if(in_buf<body_length+1) // letters+terminator don't fit?
901: return false;
902: // 4: zero terminator
903: if(cur[body_length] != 0) // in place?
904: return false;
905: // 3: letters
1.251 moko 906: body=String::Body(String::C(cur, body_length));
1.202 paf 907: cur+=body_length+1;
908: in_buf-=body_length+1;
909:
910: // 5: langs.count
1.191 paf 911: size_t fragments_count;
1.202 paf 912: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 913: return false;
1.191 paf 914: // [WARNING: not cast, addresses must be %4=0 on sparc]
915: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 916: in_buf-=sizeof(fragments_count);
1.174 paf 917:
918: if(fragments_count) {
1.202 paf 919: // 6: lang info
1.174 paf 920: size_t total_length=0;
921: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 922: char lang;
923: size_t fragment_length;
924: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 925: if(in_buf<piece_length) // lang+length
1.174 paf 926: return false;
927:
1.191 paf 928: // lang
929: lang=*cur++;
930: // length [WARNING: not cast, addresses must be %4=0 on sparc]
931: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
932:
1.206 paf 933: size_t combined_length=total_length+fragment_length;
934: if(combined_length>body_length)
935: return false; // file curruption
1.191 paf 936: // uchar needed to prevent propagating 0x80 bit to upper bytes
937: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 938: total_length=combined_length;
1.202 paf 939: in_buf-=piece_length;
1.174 paf 940: }
1.128 paf 941:
1.202 paf 942: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 943: return false;
1.174 paf 944: }
1.202 paf 945: if(in_buf!=0) // some strange extra bytes
946: return false;
1.113 parser 947:
1.174 paf 948: ASSERT_STRING_INVARIANT(*this);
1.148 paf 949: return true;
1.176 paf 950: }
951:
1.201 paf 952: void String::Body::dump() const {
953: CORD_dump(body);
954: }
955:
1.257 moko 956: const char* String::Languages::visualize() const {
1.177 paf 957: if(opt.is_not_just_lang)
1.233 misha 958: return CORD_to_const_char_star(langs, 0);
1.176 paf 959: else
1.257 moko 960: return 0;
1.176 paf 961: }
1.248 moko 962:
1.201 paf 963: void String::Languages::dump() const {
964: if(opt.is_not_just_lang)
965: CORD_dump(langs);
966: else
967: puts((const char*)&langs);
968: }
1.248 moko 969:
1.257 moko 970: void String::dump() const {
971: body.dump();
972: langs.dump();
973: }
1.176 paf 974:
1.257 moko 975: static char *n_chars(char c, size_t length){
976: char *result=(char *)pa_malloc_atomic(length+1);
977: memset(result, c, length);
978: result[length] = '\0';
979: return result;
1.113 parser 980: }
1.229 misha 981:
1.257 moko 982: char* String::visualize_langs() const {
1.258 moko 983: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 984: }
1.229 misha 985:
986: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 987: if(is_empty())
1.195 paf 988: return *this;
989:
990: size_t substr_begin, substr_length;
1.229 misha 991: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 992: if(new_body==body) // we received unchanged pointer, do likewise
993: return *this;
994: // new_body differs from body, adjust langs along
995:
996: String& result=*new String;
997: if(!new_body) // body.trim produced empty result
998: return result;
999: // body.trim produced nonempty result
1000:
1001: // first: their langs
1002: result.langs.append(result.body, langs, substr_begin, substr_length);
1003: // next: letters themselves
1004: result.body=new_body;
1005:
1006: ASSERT_STRING_INVARIANT(result);
1007: return result;
1.198 paf 1008: }
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