Annotation of parser3/src/main/pa_string.C, revision 1.260
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:
1.260 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.259 2016/12/29 15:24:49 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.242 moko 19: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
20:
1.249 moko 21: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
22: T result = 0;
1.242 moko 23: const char *pos = str;
24:
25: while (isspace(*pos)) /* skip leading whitespace */
26: ++pos;
27:
28: if (base == 16 && *pos == '0') { /* handle option prefix */
29: ++pos;
30: if (*pos == 'x' || *pos == 'X') {
31: ++pos;
32: }
33: }
34:
35: if (base == 0) { /* dynamic base */
36: base = 10; /* default is 10 */
37: if (*pos == '0') {
38: ++pos;
1.245 moko 39: if (*pos == 'x' || *pos == 'X'){
1.242 moko 40: ++pos;
41: base=16;
1.245 moko 42: }
1.242 moko 43: }
44: }
45:
46: if (base < 2 || base > 16) { /* illegal base */
47: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
48: }
49:
1.249 moko 50: T cutoff = max / base;
51: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 52:
53: while(true) {
54: int digit;
55:
56: if ((*pos >= '0') && (*pos <= '9')) {
57: digit = (*pos - '0');
58: } else if (*pos >= 'a') {
59: digit = (*pos - 'a' + 10);
60: } else if (*pos >= 'A') {
61: digit = (*pos - 'A' + 10);
62: } else break;
63:
64: if (digit >= base) {
65: break;
66: }
67:
68: ++pos;
69:
70: /* adjust number, with overflow check */
71: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
72: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
73: } else {
74: result = result * base + digit;
75: }
76: }
77:
78: while(char c=*pos++)
79: if(!isspace(c))
80: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
81:
82: return result;
83: }
84:
1.249 moko 85: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
1.254 moko 86: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
1.249 moko 87: }
88:
89: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
1.254 moko 90: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
1.249 moko 91: }
92:
1.193 paf 93: int pa_atoi(const char* str, const String* problem_source) {
94: if(!str)
95: return 0;
96:
1.242 moko 97: while(isspace(*str))
1.193 paf 98: str++;
1.242 moko 99:
1.193 paf 100: if(!*str)
101: return 0;
102:
1.200 paf 103: bool negative=false;
104: if(str[0]=='-') {
105: negative=true;
106: str++;
107: } else if(str[0]=='+') {
108: str++;
109: }
1.193 paf 110:
1.242 moko 111: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 112:
1.242 moko 113: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 114: return -(int)result;
1.242 moko 115:
116: if(result<=INT_MAX)
1.243 moko 117: return (int)result;
1.242 moko 118:
119: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 120: }
121:
122: double pa_atod(const char* str, const String* problem_source) {
123: if(!str)
124: return 0;
125:
1.242 moko 126: while(isspace(*str))
1.193 paf 127: str++;
1.242 moko 128:
1.193 paf 129: if(!*str)
130: return 0;
131:
1.200 paf 132: bool negative=false;
133: if(str[0]=='-') {
134: negative=true;
135: str++;
136: } else if(str[0]=='+') {
137: str++;
138: }
1.242 moko 139:
140: double result;
1.247 moko 141: if(str[0]=='0') {
142: if(str[1]=='x' || str[1]=='X') {
1.242 moko 143: // 0xABC
1.249 moko 144: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 145: return negative ? -result : result;
146: } else {
1.200 paf 147: // skip leading 0000, to disable octal interpretation
1.242 moko 148: do str++; while(*str=='0');
1.200 paf 149: }
1.247 moko 150: }
1.242 moko 151:
152: char *error_pos;
153: result=strtod(str, &error_pos);
1.193 paf 154:
155: while(char c=*error_pos++)
1.199 paf 156: if(!isspace((unsigned char)c))
1.242 moko 157: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 158:
1.242 moko 159: return negative ? -result : result;
1.193 paf 160: }
161:
1.176 paf 162: // cord lib extension
163:
164: #ifndef DOXYGEN
165: typedef struct {
1.254 moko 166: ssize_t countdown;
167: int target; /* Character we're looking for */
1.176 paf 168: } chr_data;
169: #endif
1.248 moko 170:
1.176 paf 171: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
172: {
1.254 moko 173: register chr_data * d = (chr_data *)client_data;
174:
175: if (d -> countdown<=0) return(2);
176: d -> countdown -= size;
177: if (c > d -> target) return(1);
178: return(0);
1.176 paf 179: }
1.248 moko 180:
1.176 paf 181: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
182: {
1.254 moko 183: chr_data d;
1.176 paf 184:
1.254 moko 185: d.countdown = n;
186: d.target = c;
187: 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 188: }
189:
1.187 paf 190: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 191: {
1.254 moko 192: int* result=(int*)client_data;
193: (*result)++;
194: return(0); // 0=continue
1.178 paf 195: }
1.248 moko 196:
1.178 paf 197: size_t CORD_block_count(CORD x)
198: {
199: size_t result=0;
200: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 201: return result;
1.178 paf 202: }
203:
1.174 paf 204: // helpers
1.139 paf 205:
1.174 paf 206: /// String::match uses this as replace & global search table columns
1.139 paf 207:
1.174 paf 208: const int MAX_MATCH_GROUPS=100;
1.139 paf 209:
1.174 paf 210: class String_match_table_template_columns: public ArrayString {
211: public:
212: String_match_table_template_columns() {
213: *this+=new String("prematch");
214: *this+=new String("match");
215: *this+=new String("postmatch");
216: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 217: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 218: }
219: }
220: };
221:
222: Table string_match_table_template(new String_match_table_template_columns);
223:
1.176 paf 224: // String::Body methods
1.140 paf 225:
1.176 paf 226: String::Body String::Body::Format(int value) {
1.174 paf 227: char local[MAX_NUMBER];
228: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 229: return String::Body(pa_strdup(local, length));
1.120 paf 230: }
231:
1.254 moko 232: 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 233: size_t our_length=length();
234: if(!our_length)
235: return *this;
1.229 misha 236:
237: // check if any UTF-8 in chars
238: bool fast=true;
239: if(chars && source_charset && source_charset->isUTF8()){
240: const char* pos=chars;
241: while(unsigned char c=*pos++)
242: if(c>127){
243: fast=false;
244: break;
245: }
246: }
247:
248: size_t start=0;
249: size_t end=our_length;
1.195 paf 250: if(!chars)
251: chars=" \t\n"; // white space
252:
1.229 misha 253: if(fast){
254: // from left...
255: if(kind!=TRIM_END) {
256: CORD_pos pos; set_pos(pos, 0);
257: while(true) {
258: char c=CORD_pos_fetch(pos);
259: if(strchr(chars, c)) {
260: if(++start==our_length)
261: return 0; // all chars are empty, just return empty string
262: } else
263: break;
264:
265: CORD_next(pos);
266: }
267: }
268:
269: // from right..
270: if(kind!=TRIM_START) {
271: CORD_pos pos; set_pos(pos, end-1);
272: while(true) {
273: char c=CORD_pos_fetch(pos);
274: if(strchr(chars, c)) {
275: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
276: return 0; // all chars are empty, just return empty string
277: } else
278: break;
279:
280: CORD_prev(pos);
281: }
282: }
283: } else {
1.234 misha 284: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 285: const XMLByte* src_end=src_begin+our_length;
286:
287: // from left...
288: if(kind!=TRIM_END) {
289: while(src_begin<src_end){
290: uint char_length=1;
291: const XMLByte* ptr=src_begin;
1.231 misha 292: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
293: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 294: char_length++;
295:
296: bool found=false;
297: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
298: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
299: found=true;
300: break;
301: }
302:
303: if(found){
304: start+=char_length;
305: if(start==our_length)
306: return 0; // all chars are empty, just return empty string
307: } else
308: break;
309: }
310: }
1.196 paf 311:
1.229 misha 312: // from right..
313: if(kind!=TRIM_START) {
314: while(src_begin<src_end){
315: uint char_length=1;
1.231 misha 316: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
317: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 318: char_length++;
319:
320: bool found=false;
321: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
322: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
323: found=true;
324: break;
325: }
326:
327: if(found){
328: end-=char_length;
329: if(end==0)
330: return 0; // all chars are empty, just return empty string
331: } else
332: break;
333: }
1.196 paf 334: }
335: }
1.195 paf 336:
337: if(start==0 && end==our_length) // nobody moved a thing
338: return *this;
339:
340: if(out_start)
341: *out_start=start;
342: size_t new_length=end-start;
343: if(out_length)
344: *out_length=new_length;
345:
346: return mid(start, new_length);
347: }
348:
1.174 paf 349: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
350: uint& result=*static_cast<uint*>(client_data);
351: generic_hash_code(result, c);
352: return 0;
353: }
1.248 moko 354:
1.174 paf 355: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
356: uint& result=*static_cast<uint*>(client_data);
357: generic_hash_code(result, s);
358: return 0;
1.248 moko 359: }
360:
1.225 misha 361: uint String::Body::get_hash_code() const {
362: #ifdef HASH_CODE_CACHING
363: if(hash_code)
364: return hash_code;
365: #else
366: uint hash_code=0;
367: #endif
1.220 misha 368: if (body && CORD_IS_STRING(body)){
1.250 moko 369: generic_hash_code(hash_code, (const char *)body);
1.220 misha 370: } else {
371: CORD_iter5(body, 0,
372: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 373: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 374: }
1.225 misha 375: return hash_code;
1.94 parser 376: }
377:
1.241 misha 378: struct CORD_pos_info {
379: const char* chars;
380: size_t left;
381: size_t pos;
382: };
383:
384: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
385: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
386: if(info->pos < info->left){
387: info->pos=STRING_NOT_FOUND;
388: return 1;
389: }
390: if(strchr(info->chars, c))
391: return 1;
392: --(info->pos);
393: return 0;
394: }
395:
396: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
397: if(is_empty() || !chars || !strlen(chars))
398: return STRING_NOT_FOUND;
399: CORD_pos_info info={chars, left, right};
400: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
401: return info.pos;
402: else
403: return STRING_NOT_FOUND;
404: }
405:
406:
407: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
408: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
409: if(info->pos < info->left) {
410: info->pos=STRING_NOT_FOUND;
411: return 1;
412: }
413: if(!strchr(info->chars, c))
414: return 1;
415: --(info->pos);
416: return 0;
417: }
418:
419: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
420: if(is_empty() || !chars || !strlen(chars))
421: return STRING_NOT_FOUND;
422: CORD_pos_info info={chars, left, right};
423: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
424: return info.pos;
425: else
426: return STRING_NOT_FOUND;
427: }
428:
1.174 paf 429: // String methods
430:
431: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
432: if(!known_length)
1.9 paf 433: return *this;
1.122 paf 434:
1.176 paf 435: // first: langs
436: langs.append(body, lang, known_length);
437: // next: letters themselves
1.174 paf 438: body.append_know_length(str, known_length);
1.1 paf 439:
1.174 paf 440: ASSERT_STRING_INVARIANT(*this);
1.1 paf 441: return *this;
442: }
1.248 moko 443:
1.174 paf 444: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
445: if(!str)
446: return *this;
447: size_t known_length=helper_length?helper_length:strlen(str);
448: if(!known_length)
449: return *this;
1.1 paf 450:
1.174 paf 451: return append_know_length(str, known_length, lang);
1.5 paf 452: }
1.248 moko 453:
1.244 moko 454: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 455: char buf[MAX_NUMBER];
456: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
457: }
1.248 moko 458:
1.174 paf 459: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
460: size_t known_length=helper_length?helper_length:strlen(str);
461: if(!known_length)
462: return *this;
1.5 paf 463:
1.176 paf 464: // first: langs
465: langs.append(body, lang, known_length);
466: // next: letters themselves
1.174 paf 467: body.append_strdup_know_length(str, known_length);
1.33 paf 468:
1.174 paf 469: ASSERT_STRING_INVARIANT(*this);
470: return *this;
1.5 paf 471: }
1.46 paf 472:
1.234 misha 473: struct CORD_length_info {
474: size_t len;
475: size_t skip;
476: };
477:
478: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 479: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
480: return 0;
1.221 misha 481: }
482:
1.234 misha 483: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
484: int CORD_batched_len(const char c, CORD_length_info* info){
485: if (info->skip==0){
486: info->len++;
487: info->skip = lengthUTF8Char(c)-1;
488: } else {
489: info->skip--;
490: }
1.220 misha 491: return 0;
492: }
493:
1.210 misha 494: size_t String::length(Charset& charset) const {
495: if(charset.isUTF8()){
1.234 misha 496: CORD_length_info info = {0, 0};
497: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
498: return info.len;
1.210 misha 499: } else
500: return body.length();
501: }
502:
1.174 paf 503: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
504: String& String::mid(size_t substr_begin, size_t substr_end) const {
505: String& result=*new String;
506:
507: size_t self_length=length();
508: substr_begin=min(substr_begin, self_length);
509: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 510: size_t substr_length=substr_end-substr_begin;
511: if(!substr_length)
1.107 parser 512: return result;
1.53 paf 513:
1.176 paf 514: // first: their langs
515: result.langs.append(result.body, langs, substr_begin, substr_length);
516: // next: letters themselves
517: result.body=body.mid(substr_begin, substr_length);
1.174 paf 518:
519: ASSERT_STRING_INVARIANT(result);
1.53 paf 520: return result;
1.54 paf 521: }
522:
1.211 misha 523: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
524: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 525: String& result=*new String;
526:
1.255 moko 527: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 528:
529: if(!self_length)
530: return result;
531:
532: from=min(min(to, from), self_length);
1.210 misha 533: to=min(max(to, from), self_length);
1.211 misha 534:
1.210 misha 535: size_t substr_length=to-from;
1.211 misha 536:
1.210 misha 537: if(!substr_length)
538: return result;
539:
540: if(charset.isUTF8()){
1.234 misha 541: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 542: const XMLByte* src_end=src_begin+body.length();
1.210 misha 543:
1.212 misha 544: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 545: from=getUTF8BytePos(src_begin, src_end, from);
546: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 547: if(!substr_length)
548: return result;
549: }
550:
551: // first: their langs
552: result.langs.append(result.body, langs, from, substr_length);
553: // next: letters themselves
554: result.body=body.mid(from, substr_length);
555:
556: ASSERT_STRING_INVARIANT(result);
557: return result;
558: }
559:
1.176 paf 560: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 561: size_t substr_length=substr.length();
562: while(true) {
563: size_t substr_begin=body.pos(substr, this_offset);
564:
565: if(substr_begin==CORD_NOT_FOUND)
566: return STRING_NOT_FOUND;
1.174 paf 567:
1.183 paf 568: if(langs.check_lang(lang, substr_begin, substr_length))
569: return substr_begin;
570:
571: this_offset=substr_begin+substr_length;
572: }
1.58 paf 573: }
574:
1.254 moko 575: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 576: return pos(substr.body, this_offset, lang);
1.60 paf 577: }
578:
1.254 moko 579: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 580:
581: if(charset.isUTF8()){
1.234 misha 582: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 583: const XMLByte* srcEnd=srcPtr+body.length();
584:
585: // convert 'this_offset' from 'characters' to 'bytes'
586: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
587:
1.229 misha 588: size_t result=pos(substr.body, this_offset, lang);
589: return (result==CORD_NOT_FOUND)
590: ? STRING_NOT_FOUND
591: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 592: } else {
1.229 misha 593: size_t result=pos(substr.body, this_offset, lang);
594: return (result==CORD_NOT_FOUND)
595: ? STRING_NOT_FOUND
596: : result;
1.210 misha 597: }
598: }
599:
1.256 moko 600: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 601: if(is_empty())
602: return;
1.174 paf 603: size_t self_length=length();
1.237 misha 604: if(size_t delim_length=strlen(delim)) {
1.186 paf 605: size_t pos_before;
1.60 paf 606: // while we have 'delim'...
1.256 moko 607: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 608: result+=&mid(pos_after, pos_before);
1.174 paf 609: pos_after=pos_before+delim_length;
1.60 paf 610: }
611: // last piece
1.256 moko 612: if(pos_after<self_length)
1.174 paf 613: result+=&mid(pos_after, self_length);
1.60 paf 614: } else { // empty delim
615: result+=this;
616: }
617: }
618:
1.256 moko 619: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 620: if(is_empty())
621: return;
622: if(!delim.is_empty()) {
1.186 paf 623: size_t pos_before;
1.60 paf 624: // while we have 'delim'...
1.256 moko 625: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 626: result+=&mid(pos_after, pos_before);
1.174 paf 627: pos_after=pos_before+delim.length();
1.60 paf 628: }
629: // last piece
1.256 moko 630: if(pos_after<length())
1.174 paf 631: result+=&mid(pos_after, length());
1.60 paf 632: } else { // empty delim
633: result+=this;
634: }
1.61 paf 635: }
636:
1.254 moko 637: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 638:
1.222 misha 639: // vregex->info(); // I have no idea what does it for?
1.63 paf 640:
1.222 misha 641: bool need_pre_post_match=vregex->is_pre_post_match_needed();
642: bool global=vregex->is_global_search();
1.63 paf 643:
1.174 paf 644: const char* subject=cstr();
1.234 misha 645: size_t subject_length=length();
1.222 misha 646: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
647: int ovector[ovector_size];
1.155 paf 648:
1.173 paf 649: Table::Action_options table_options;
1.174 paf 650: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 651:
1.154 paf 652: int prestart=0;
653: int poststart=0;
1.174 paf 654: int postfinish=length();
1.63 paf 655: while(true) {
1.222 misha 656: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 657:
1.222 misha 658: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
659: break;
1.63 paf 660:
1.154 paf 661: int prefinish=ovector[0];
662: poststart=ovector[1];
1.236 moko 663:
664: if (prestart==poststart && subject[poststart]=='\n'){
665: prestart++;
666: continue;
667: }
668:
1.232 misha 669: ArrayString* row=new ArrayString(3);
1.174 paf 670: if(need_pre_post_match) {
671: *row+=&mid(0, prefinish); // .prematch column value
672: *row+=&mid(prefinish, poststart); // .match
673: *row+=&mid(poststart, postfinish); // .postmatch
674: } else {
1.185 paf 675: *row+=&Empty; // .prematch column value
676: *row+=&Empty; // .match
677: *row+=&Empty; // .postmatch
1.174 paf 678: }
1.63 paf 679:
1.222 misha 680: for(int i=1; i<exec_result; i++) {
1.69 paf 681: // -1:-1 case handled peacefully by mid() itself
1.232 misha 682: *row+=(ovector[i*2+0]>=0 && ovector[i*2+1]>0)?&mid(ovector[i*2+0], ovector[i*2+1]):new String; // .i column value
1.63 paf 683: }
684:
1.209 misha 685: matches_count++;
1.174 paf 686: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 687:
1.222 misha 688: if(!global || prestart==poststart) // last step
689: break;
690:
1.154 paf 691: prestart=poststart;
1.222 misha 692: }
1.63 paf 693:
1.222 misha 694: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
695: return vregex->is_just_count() ? 0 : &table;
1.82 parser 696: }
697:
1.174 paf 698: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
699: String& result=*new String();
700: if(is_empty())
701: return result;
702:
703: char* new_cstr=cstrm();
1.216 misha 704:
1.181 paf 705: if(source_charset.isUTF8()) {
1.216 misha 706: size_t new_cstr_len=length();
1.181 paf 707: switch(kind) {
708: case CC_UPPER:
1.192 paf 709: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 710: break;
711: case CC_LOWER:
1.192 paf 712: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 713: break;
714: default:
715: assert(!"unknown change case kind");
716: break; // never
717: }
718:
719: } else {
720: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 721:
1.181 paf 722: const unsigned char *a;
723: const unsigned char *b;
724: switch(kind) {
725: case CC_UPPER:
726: a=tables+lcc_offset;
727: b=tables+fcc_offset;
728: break;
729: case CC_LOWER:
730: a=tables+lcc_offset;
731: b=0;
732: break;
733: default:
734: assert(!"unknown change case kind");
735: a=b=0; // calm, compiler
736: break; // never
737: }
738:
1.192 paf 739: char *dest=new_cstr;
1.181 paf 740: unsigned char index;
1.190 paf 741: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 742: unsigned char c=a[index];
743: if(b)
744: c=b[c];
745:
746: *dest++=(char)c;
747: }
1.174 paf 748: }
1.176 paf 749: result.langs=langs;
1.174 paf 750: result.body=new_cstr;
1.89 parser 751:
1.101 parser 752: return result;
753: }
754:
1.213 misha 755: const String& String::escape(Charset& source_charset) const {
756: if(is_empty())
757: return *this;
758:
759: return Charset::escape(*this, source_charset);
760: }
761:
1.238 misha 762: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
763: result.langs.append(result.body, langs, langs_offset, length); \
764: result.body.append_strdup_know_length(from_cstr, length);
765:
1.174 paf 766: const String& String::replace(const Dictionary& dict) const {
1.238 misha 767: if(!dict.count() || is_empty())
768: return *this;
769:
1.174 paf 770: String& result=*new String();
771: const char* old_cstr=cstr();
772: const char* prematch_begin=old_cstr;
773:
1.238 misha 774: if(dict.count()==1) {
775: // optimized simple case
776:
777: Dictionary::Subst subst=dict.get(0);
1.239 moko 778: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 779: // prematch
1.238 misha 780: if(size_t prematch_length=p-prematch_begin) {
781: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 782: }
783:
1.174 paf 784: // match
1.238 misha 785: prematch_begin=p+subst.from_length;
1.174 paf 786:
1.184 paf 787: if(const String* b=subst.to) // are there any b?
1.174 paf 788: result<<*b;
1.238 misha 789: }
790:
791: } else {
792:
793: const char* current=old_cstr;
794: while(*current) {
795: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
796: // prematch
797: if(size_t prematch_length=current-prematch_begin) {
798: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
799: }
800:
801: // match
802: // skip 'a' in 'current'; move prematch_begin
803: current+=subst.from_length; prematch_begin=current;
804:
805: if(const String* b=subst.to) // are there any b?
806: result<<*b;
807: } else // simply advance
808: current++;
809: }
810:
1.174 paf 811: }
1.156 paf 812:
1.238 misha 813: if(prematch_begin==old_cstr) // not modified
814: return *this;
815:
1.174 paf 816: // postmatch
1.238 misha 817: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
818: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 819: }
1.156 paf 820:
1.174 paf 821: ASSERT_STRING_INVARIANT(result);
1.82 parser 822: return result;
1.61 paf 823: }
1.113 parser 824:
1.180 paf 825: static int serialize_body_char(char c, char** cur) {
826: *((*cur)++)=c;
827: return 0; // 0=continue
1.248 moko 828: }
829:
1.174 paf 830: static int serialize_body_piece(const char* s, char** cur) {
831: size_t length=strlen(s);
832: memcpy(*cur, s, length); *cur+=length;
1.178 paf 833: return 0; // 0=continue
1.248 moko 834: }
835:
1.178 paf 836: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
837: // lang
1.191 paf 838: **cur=alang; (*cur)++;
839: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 840: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
841:
842: return 0; // 0=continue
843: }
1.248 moko 844:
1.174 paf 845: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 846: size_t fragments_count=langs.count();
1.202 paf 847: size_t body_length=body.length();
1.174 paf 848: size_t buf_length=
1.178 paf 849: prolog_length //1
850: +sizeof(size_t) //2
1.202 paf 851: +body_length //3
852: +1 // 4 for zero terminator used in deserialize
853: +sizeof(size_t) //5
854: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
855:
1.174 paf 856: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
857:
858: // 1: prolog
859: char *cur=result.str+prolog_length;
1.202 paf 860: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
861: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
862: // 3: letters
863: body.for_each(serialize_body_char, serialize_body_piece, &cur);
864: // 4: zero terminator
865: *cur++=0;
866: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 867: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 868: // 6: lang info
1.178 paf 869: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 870:
1.174 paf 871: return result;
1.113 parser 872: }
1.248 moko 873:
1.202 paf 874: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
875: size_t in_buf=buf_size;
876: if(in_buf<=prolog_size)
1.148 paf 877: return false;
1.202 paf 878: in_buf-=prolog_size;
1.135 paf 879:
1.174 paf 880: // 1: prolog
1.202 paf 881: const char* cur=(const char* )buf+prolog_size;
1.113 parser 882:
1.207 paf 883: // 2: chars.count
1.202 paf 884: size_t body_length;
885: if(in_buf<sizeof(body_length)) // body.length don't fit?
886: return false;
887: // [WARNING: not cast, addresses must be %4=0 on sparc]
888: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
889: in_buf-=sizeof(body_length);
890:
891: if(in_buf<body_length+1) // letters+terminator don't fit?
892: return false;
893: // 4: zero terminator
894: if(cur[body_length] != 0) // in place?
895: return false;
896: // 3: letters
1.251 moko 897: body=String::Body(String::C(cur, body_length));
1.202 paf 898: cur+=body_length+1;
899: in_buf-=body_length+1;
900:
901: // 5: langs.count
1.191 paf 902: size_t fragments_count;
1.202 paf 903: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 904: return false;
1.191 paf 905: // [WARNING: not cast, addresses must be %4=0 on sparc]
906: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 907: in_buf-=sizeof(fragments_count);
1.174 paf 908:
909: if(fragments_count) {
1.202 paf 910: // 6: lang info
1.174 paf 911: size_t total_length=0;
912: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 913: char lang;
914: size_t fragment_length;
915: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 916: if(in_buf<piece_length) // lang+length
1.174 paf 917: return false;
918:
1.191 paf 919: // lang
920: lang=*cur++;
921: // length [WARNING: not cast, addresses must be %4=0 on sparc]
922: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
923:
1.206 paf 924: size_t combined_length=total_length+fragment_length;
925: if(combined_length>body_length)
926: return false; // file curruption
1.191 paf 927: // uchar needed to prevent propagating 0x80 bit to upper bytes
928: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 929: total_length=combined_length;
1.202 paf 930: in_buf-=piece_length;
1.174 paf 931: }
1.128 paf 932:
1.202 paf 933: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 934: return false;
1.174 paf 935: }
1.202 paf 936: if(in_buf!=0) // some strange extra bytes
937: return false;
1.113 parser 938:
1.174 paf 939: ASSERT_STRING_INVARIANT(*this);
1.148 paf 940: return true;
1.176 paf 941: }
942:
1.201 paf 943: void String::Body::dump() const {
944: CORD_dump(body);
945: }
946:
1.257 moko 947: const char* String::Languages::visualize() const {
1.177 paf 948: if(opt.is_not_just_lang)
1.233 misha 949: return CORD_to_const_char_star(langs, 0);
1.176 paf 950: else
1.257 moko 951: return 0;
1.176 paf 952: }
1.248 moko 953:
1.201 paf 954: void String::Languages::dump() const {
955: if(opt.is_not_just_lang)
956: CORD_dump(langs);
957: else
958: puts((const char*)&langs);
959: }
1.248 moko 960:
1.257 moko 961: void String::dump() const {
962: body.dump();
963: langs.dump();
964: }
1.176 paf 965:
1.257 moko 966: static char *n_chars(char c, size_t length){
967: char *result=(char *)pa_malloc_atomic(length+1);
968: memset(result, c, length);
969: result[length] = '\0';
970: return result;
1.113 parser 971: }
1.229 misha 972:
1.257 moko 973: char* String::visualize_langs() const {
1.258 moko 974: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 975: }
1.229 misha 976:
977: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 978: if(is_empty())
1.195 paf 979: return *this;
980:
981: size_t substr_begin, substr_length;
1.229 misha 982: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 983: if(new_body==body) // we received unchanged pointer, do likewise
984: return *this;
985: // new_body differs from body, adjust langs along
986:
987: String& result=*new String;
988: if(!new_body) // body.trim produced empty result
989: return result;
990: // body.trim produced nonempty result
991:
992: // first: their langs
993: result.langs.append(result.body, langs, substr_begin, substr_length);
994: // next: letters themselves
995: result.body=new_body;
996:
997: ASSERT_STRING_INVARIANT(result);
998: return result;
1.198 paf 999: }
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