Annotation of parser3/src/main/pa_string.C, revision 1.290
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.285 moko 4: Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com)
1.269 moko 5: Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <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:
1.290 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.289 2026/11/03 12:25:52 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:
1.176 paf 22: // cord lib extension
23:
24: #ifndef DOXYGEN
25: typedef struct {
1.254 moko 26: ssize_t countdown;
27: int target; /* Character we're looking for */
1.176 paf 28: } chr_data;
29: #endif
1.248 moko 30:
1.176 paf 31: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
32: {
1.272 moko 33: chr_data * d = (chr_data *)client_data;
1.254 moko 34:
35: if (d -> countdown<=0) return(2);
36: d -> countdown -= size;
37: if (c > d -> target) return(1);
38: return(0);
1.176 paf 39: }
1.248 moko 40:
1.176 paf 41: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
42: {
1.254 moko 43: chr_data d;
1.176 paf 44:
1.254 moko 45: d.countdown = n;
46: d.target = c;
47: 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 48: }
49:
1.187 paf 50: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 51: {
1.254 moko 52: int* result=(int*)client_data;
53: (*result)++;
54: return(0); // 0=continue
1.178 paf 55: }
1.248 moko 56:
1.178 paf 57: size_t CORD_block_count(CORD x)
58: {
59: size_t result=0;
60: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
1.254 moko 61: return result;
1.178 paf 62: }
63:
1.174 paf 64: // helpers
1.139 paf 65:
1.174 paf 66: /// String::match uses this as replace & global search table columns
1.139 paf 67:
1.174 paf 68: const int MAX_MATCH_GROUPS=100;
1.139 paf 69:
1.174 paf 70: class String_match_table_template_columns: public ArrayString {
71: public:
72: String_match_table_template_columns() {
73: *this+=new String("prematch");
74: *this+=new String("match");
75: *this+=new String("postmatch");
76: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.282 moko 77: *this+=new String(String::Body::uitoa(1+i), String::L_CLEAN);
1.174 paf 78: }
79: }
80: };
81:
1.278 moko 82: static Table &string_match_table_template(){
83: static Table *singleton=NULL;
1.280 moko 84: if(!singleton)
1.278 moko 85: singleton=new Table(new String_match_table_template_columns);
86: return *singleton;
87: }
1.174 paf 88:
1.176 paf 89: // String::Body methods
1.140 paf 90:
1.282 moko 91: String::Body String::Body::uitoa(size_t aindex) {
92: static const size_t CACHE_SIZE=256;
93: static String::Body cache[CACHE_SIZE];
94: if (aindex < CACHE_SIZE)
95: return !cache[aindex] ? (cache[aindex]=Body(pa_uitoa(aindex))) : cache[aindex];
96: else
97: return Body(pa_uitoa(aindex));
98: }
99:
1.254 moko 100: 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 101: size_t our_length=length();
102: if(!our_length)
103: return *this;
1.229 misha 104:
105: // check if any UTF-8 in chars
106: bool fast=true;
107: if(chars && source_charset && source_charset->isUTF8()){
108: const char* pos=chars;
109: while(unsigned char c=*pos++)
110: if(c>127){
111: fast=false;
112: break;
113: }
114: }
115:
116: size_t start=0;
117: size_t end=our_length;
1.195 paf 118: if(!chars)
1.281 moko 119: chars=" \t\r\n"; // white space
1.195 paf 120:
1.229 misha 121: if(fast){
122: // from left...
123: if(kind!=TRIM_END) {
124: CORD_pos pos; set_pos(pos, 0);
125: while(true) {
126: char c=CORD_pos_fetch(pos);
127: if(strchr(chars, c)) {
128: if(++start==our_length)
1.283 moko 129: return String::Body(); // all chars are empty, just return empty string
1.229 misha 130: } else
1.281 moko 131: break;
1.229 misha 132:
133: CORD_next(pos);
134: }
135: }
136:
137: // from right..
138: if(kind!=TRIM_START) {
139: CORD_pos pos; set_pos(pos, end-1);
140: while(true) {
141: char c=CORD_pos_fetch(pos);
142: if(strchr(chars, c)) {
143: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
1.283 moko 144: return String::Body(); // all chars are empty, just return empty string
1.229 misha 145: } else
1.281 moko 146: break;
1.229 misha 147:
148: CORD_prev(pos);
149: }
150: }
151: } else {
1.234 misha 152: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 153: const XMLByte* src_end=src_begin+our_length;
154:
155: // from left...
156: if(kind!=TRIM_END) {
157: while(src_begin<src_end){
158: uint char_length=1;
159: const XMLByte* ptr=src_begin;
1.231 misha 160: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
161: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 162: char_length++;
163:
164: bool found=false;
165: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
166: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
167: found=true;
168: break;
169: }
170:
171: if(found){
172: start+=char_length;
173: if(start==our_length)
1.283 moko 174: return String::Body(); // all chars are empty, just return empty string
1.229 misha 175: } else
176: break;
177: }
178: }
1.196 paf 179:
1.229 misha 180: // from right..
181: if(kind!=TRIM_START) {
182: while(src_begin<src_end){
183: uint char_length=1;
1.231 misha 184: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
185: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 186: char_length++;
187:
188: bool found=false;
189: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
190: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
191: found=true;
192: break;
193: }
194:
195: if(found){
196: end-=char_length;
197: if(end==0)
1.283 moko 198: return String::Body(); // all chars are empty, just return empty string
1.229 misha 199: } else
200: break;
201: }
1.196 paf 202: }
203: }
1.195 paf 204:
205: if(start==0 && end==our_length) // nobody moved a thing
206: return *this;
207:
208: if(out_start)
209: *out_start=start;
210: size_t new_length=end-start;
211: if(out_length)
212: *out_length=new_length;
213:
214: return mid(start, new_length);
215: }
216:
1.174 paf 217: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
218: uint& result=*static_cast<uint*>(client_data);
219: generic_hash_code(result, c);
220: return 0;
221: }
1.248 moko 222:
1.174 paf 223: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
224: uint& result=*static_cast<uint*>(client_data);
225: generic_hash_code(result, s);
226: return 0;
1.248 moko 227: }
228:
1.225 misha 229: uint String::Body::get_hash_code() const {
230: #ifdef HASH_CODE_CACHING
231: if(hash_code)
232: return hash_code;
233: #else
234: uint hash_code=0;
235: #endif
1.220 misha 236: if (body && CORD_IS_STRING(body)){
1.250 moko 237: generic_hash_code(hash_code, (const char *)body);
1.220 misha 238: } else {
239: CORD_iter5(body, 0,
240: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 241: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 242: }
1.225 misha 243: return hash_code;
1.94 parser 244: }
245:
1.241 misha 246: struct CORD_pos_info {
247: const char* chars;
248: size_t left;
249: size_t pos;
250: };
251:
252: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
253: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
254: if(info->pos < info->left){
255: info->pos=STRING_NOT_FOUND;
256: return 1;
257: }
258: if(strchr(info->chars, c))
259: return 1;
260: --(info->pos);
261: return 0;
262: }
263:
264: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
265: if(is_empty() || !chars || !strlen(chars))
266: return STRING_NOT_FOUND;
267: CORD_pos_info info={chars, left, right};
268: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
269: return info.pos;
270: else
271: return STRING_NOT_FOUND;
272: }
273:
1.288 moko 274: size_t String::Body::rpos(const Body substr, size_t offset) const {
275: size_t substr_length=substr.length();
276: if(!substr_length || substr_length>length())
277: return STRING_NOT_FOUND; // empty substr is not found, as in pos
278:
279: const char* self_str=cstr();
280: const char* substr_str=substr.cstr();
281:
282: size_t last=length()-substr_length; // the last position substr fits at
283: for(size_t i=(offset<last ? offset : last)+1; i--; )
284: if(self_str[i]==*substr_str && memcmp(self_str+i, substr_str, substr_length)==0)
285: return i;
286:
287: return STRING_NOT_FOUND;
288: }
289:
1.241 misha 290:
291: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
292: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
293: if(info->pos < info->left) {
294: info->pos=STRING_NOT_FOUND;
295: return 1;
296: }
297: if(!strchr(info->chars, c))
298: return 1;
299: --(info->pos);
300: return 0;
301: }
302:
303: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
304: if(is_empty() || !chars || !strlen(chars))
305: return STRING_NOT_FOUND;
306: CORD_pos_info info={chars, left, right};
307: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
308: return info.pos;
309: else
310: return STRING_NOT_FOUND;
311: }
312:
1.174 paf 313: // String methods
314:
315: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
316: if(!known_length)
1.9 paf 317: return *this;
1.122 paf 318:
1.176 paf 319: // first: langs
320: langs.append(body, lang, known_length);
321: // next: letters themselves
1.174 paf 322: body.append_know_length(str, known_length);
1.1 paf 323:
1.174 paf 324: ASSERT_STRING_INVARIANT(*this);
1.1 paf 325: return *this;
326: }
1.248 moko 327:
1.174 paf 328: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
329: if(!str)
330: return *this;
331: size_t known_length=helper_length?helper_length:strlen(str);
332: if(!known_length)
333: return *this;
1.1 paf 334:
1.174 paf 335: return append_know_length(str, known_length, lang);
1.5 paf 336: }
1.248 moko 337:
1.244 moko 338: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 339: char buf[MAX_NUMBER];
1.287 moko 340: body.append_strdup_know_length(buf, pa_snprintf(buf, MAX_NUMBER, format, value));
1.226 misha 341: }
1.248 moko 342:
1.174 paf 343: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
344: size_t known_length=helper_length?helper_length:strlen(str);
345: if(!known_length)
346: return *this;
1.5 paf 347:
1.176 paf 348: // first: langs
349: langs.append(body, lang, known_length);
350: // next: letters themselves
1.174 paf 351: body.append_strdup_know_length(str, known_length);
1.33 paf 352:
1.174 paf 353: ASSERT_STRING_INVARIANT(*this);
354: return *this;
1.5 paf 355: }
1.46 paf 356:
1.234 misha 357: struct CORD_length_info {
358: size_t len;
359: size_t skip;
360: };
361:
362: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 363: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
364: return 0;
1.221 misha 365: }
366:
1.234 misha 367: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
368: int CORD_batched_len(const char c, CORD_length_info* info){
369: if (info->skip==0){
370: info->len++;
371: info->skip = lengthUTF8Char(c)-1;
372: } else {
373: info->skip--;
374: }
1.220 misha 375: return 0;
376: }
377:
1.210 misha 378: size_t String::length(Charset& charset) const {
379: if(charset.isUTF8()){
1.234 misha 380: CORD_length_info info = {0, 0};
381: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
382: return info.len;
1.210 misha 383: } else
384: return body.length();
385: }
386:
1.174 paf 387: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
388: String& String::mid(size_t substr_begin, size_t substr_end) const {
389: String& result=*new String;
390:
391: size_t self_length=length();
392: substr_begin=min(substr_begin, self_length);
393: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 394: size_t substr_length=substr_end-substr_begin;
395: if(!substr_length)
1.107 parser 396: return result;
1.53 paf 397:
1.176 paf 398: // first: their langs
399: result.langs.append(result.body, langs, substr_begin, substr_length);
400: // next: letters themselves
401: result.body=body.mid(substr_begin, substr_length);
1.174 paf 402:
403: ASSERT_STRING_INVARIANT(result);
1.53 paf 404: return result;
1.54 paf 405: }
406:
1.289 moko 407: // from and to in characters, not in bytes (it's important for utf-8)
408: String& String::mid(Charset& charset, size_t from, size_t to) const {
1.210 misha 409: String& result=*new String;
410:
1.289 moko 411: // utf-8 length is not needed: converting to bytes stops at the end of the string by itself
412: size_t self_length=charset.isUTF8() ? to : length();
1.211 misha 413:
414: if(!self_length)
415: return result;
416:
417: from=min(min(to, from), self_length);
1.210 misha 418: to=min(max(to, from), self_length);
1.211 misha 419:
1.210 misha 420: size_t substr_length=to-from;
1.211 misha 421:
1.210 misha 422: if(!substr_length)
423: return result;
424:
425: if(charset.isUTF8()){
1.234 misha 426: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 427: const XMLByte* src_end=src_begin+body.length();
1.210 misha 428:
1.212 misha 429: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 430: from=getUTF8BytePos(src_begin, src_end, from);
431: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 432: if(!substr_length)
433: return result;
434: }
435:
436: // first: their langs
437: result.langs.append(result.body, langs, from, substr_length);
438: // next: letters themselves
439: result.body=body.mid(from, substr_length);
440:
441: ASSERT_STRING_INVARIANT(result);
442: return result;
443: }
444:
1.289 moko 445: // n in characters, not in bytes (it's important for utf-8)
446: const String& String::right(Charset& charset, size_t n) const {
447: size_t self_length=body.length();
448: size_t from;
449:
450: if(charset.isUTF8()){
451: const XMLByte* src_begin=(const XMLByte*)cstr();
1.290 ! moko 452: const XMLByte* src_end=src_begin+self_length;
! 453: const XMLByte* ptr=src_end;
1.289 moko 454:
1.290 ! moko 455: // step back by n characters: continuation bytes (80-BF) belong to the leading byte before them,
! 456: // but only if it expects exactly that many, otherwise the byte is a character of its own (as length() counts it);
! 457: // the only exception is the last character cut off by the end of the string, length() counts it as a whole
! 458: while(n-- && ptr>src_begin) {
! 459: const XMLByte* lead=ptr-1;
! 460: while(lead>src_begin && (*lead & 0xC0)==0x80 && ptr-lead<6) // 6 bytes is the longest character
! 461: lead--;
! 462: size_t expected=lengthUTF8Char(*lead);
! 463: size_t present=ptr-lead;
! 464: ptr=(expected==present || (expected>present && ptr==src_end)) ? lead : ptr-1;
! 465: }
1.289 moko 466: from=ptr-src_begin;
467: } else
468: from=self_length>n ? self_length-n : 0;
469:
470: return from ? mid(from, self_length) : *this;
471: }
472:
1.176 paf 473: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 474: size_t substr_length=substr.length();
475: while(true) {
476: size_t substr_begin=body.pos(substr, this_offset);
477:
478: if(substr_begin==CORD_NOT_FOUND)
479: return STRING_NOT_FOUND;
1.174 paf 480:
1.183 paf 481: if(langs.check_lang(lang, substr_begin, substr_length))
482: return substr_begin;
483:
484: this_offset=substr_begin+substr_length;
485: }
1.58 paf 486: }
487:
1.254 moko 488: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 489: return pos(substr.body, this_offset, lang);
1.60 paf 490: }
491:
1.254 moko 492: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 493:
494: if(charset.isUTF8()){
1.234 misha 495: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 496: const XMLByte* srcEnd=srcPtr+body.length();
497:
498: // convert 'this_offset' from 'characters' to 'bytes'
499: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
500:
1.229 misha 501: size_t result=pos(substr.body, this_offset, lang);
502: return (result==CORD_NOT_FOUND)
503: ? STRING_NOT_FOUND
504: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 505: } else {
1.229 misha 506: size_t result=pos(substr.body, this_offset, lang);
507: return (result==CORD_NOT_FOUND)
508: ? STRING_NOT_FOUND
509: : result;
1.210 misha 510: }
511: }
512:
1.288 moko 513: size_t String::rpos(Charset& charset, const String& substr, size_t this_offset) const {
514: if(charset.isUTF8()){
515: const XMLByte* srcPtr=(const XMLByte*)cstr();
516: const XMLByte* srcEnd=srcPtr+body.length();
517:
518: // convert 'this_offset' from 'characters' to 'bytes'
519: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
520:
521: size_t result=body.rpos(substr.body, this_offset);
522: return (result==STRING_NOT_FOUND)
523: ? STRING_NOT_FOUND
524: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
525: } else
526: return body.rpos(substr.body, this_offset);
527: }
528:
1.256 moko 529: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 530: if(is_empty())
531: return;
1.174 paf 532: size_t self_length=length();
1.237 misha 533: if(size_t delim_length=strlen(delim)) {
1.186 paf 534: size_t pos_before;
1.60 paf 535: // while we have 'delim'...
1.283 moko 536: while((pos_before=pos(String::Body(delim), pos_after, lang)) != STRING_NOT_FOUND) {
1.69 paf 537: result+=&mid(pos_after, pos_before);
1.174 paf 538: pos_after=pos_before+delim_length;
1.60 paf 539: }
540: // last piece
1.256 moko 541: if(pos_after<self_length)
1.174 paf 542: result+=&mid(pos_after, self_length);
1.60 paf 543: } else { // empty delim
544: result+=this;
545: }
546: }
547:
1.256 moko 548: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 549: if(is_empty())
550: return;
551: if(!delim.is_empty()) {
1.186 paf 552: size_t pos_before;
1.60 paf 553: // while we have 'delim'...
1.256 moko 554: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 555: result+=&mid(pos_after, pos_before);
1.174 paf 556: pos_after=pos_before+delim.length();
1.60 paf 557: }
558: // last piece
1.256 moko 559: if(pos_after<length())
1.174 paf 560: result+=&mid(pos_after, length());
1.60 paf 561: } else { // empty delim
562: result+=this;
563: }
1.61 paf 564: }
565:
1.254 moko 566: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 567:
1.222 misha 568: // vregex->info(); // I have no idea what does it for?
1.63 paf 569:
1.222 misha 570: bool need_pre_post_match=vregex->is_pre_post_match_needed();
571: bool global=vregex->is_global_search();
1.63 paf 572:
1.174 paf 573: const char* subject=cstr();
1.234 misha 574: size_t subject_length=length();
1.271 moko 575: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */
1.222 misha 576: int ovector[ovector_size];
1.155 paf 577:
1.173 paf 578: Table::Action_options table_options;
1.278 moko 579: Table& table=*new Table(string_match_table_template(), table_options);
1.63 paf 580:
1.154 paf 581: int prestart=0;
582: int poststart=0;
1.174 paf 583: int postfinish=length();
1.267 moko 584: int action_was_executed=-1;
1.63 paf 585: while(true) {
1.222 misha 586: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 587:
1.222 misha 588: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
589: break;
1.63 paf 590:
1.154 paf 591: int prefinish=ovector[0];
592: poststart=ovector[1];
1.236 moko 593:
1.267 moko 594: if (prestart==poststart && action_was_executed==1){
1.236 moko 595: prestart++;
1.267 moko 596: action_was_executed=0;
1.236 moko 597: continue;
598: }
599:
1.232 misha 600: ArrayString* row=new ArrayString(3);
1.174 paf 601: if(need_pre_post_match) {
602: *row+=&mid(0, prefinish); // .prematch column value
603: *row+=&mid(prefinish, poststart); // .match
604: *row+=&mid(poststart, postfinish); // .postmatch
605: } else {
1.185 paf 606: *row+=&Empty; // .prematch column value
607: *row+=&Empty; // .match
608: *row+=&Empty; // .postmatch
1.174 paf 609: }
1.63 paf 610:
1.222 misha 611: for(int i=1; i<exec_result; i++) {
1.69 paf 612: // -1:-1 case handled peacefully by mid() itself
1.232 misha 613: *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 614: }
615:
1.209 misha 616: matches_count++;
1.267 moko 617: row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info);
1.63 paf 618:
1.275 moko 619: if(!global || (size_t)poststart>=subject_length) // last step, avoid prestart++ after last char
1.222 misha 620: break;
621:
1.154 paf 622: prestart=poststart;
1.267 moko 623: action_was_executed=1;
1.222 misha 624: }
1.63 paf 625:
1.222 misha 626: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
627: return vregex->is_just_count() ? 0 : &table;
1.82 parser 628: }
629:
1.174 paf 630: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
631: String& result=*new String();
632: if(is_empty())
633: return result;
634:
635: char* new_cstr=cstrm();
1.216 misha 636:
1.181 paf 637: if(source_charset.isUTF8()) {
1.216 misha 638: size_t new_cstr_len=length();
1.181 paf 639: switch(kind) {
640: case CC_UPPER:
1.192 paf 641: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 642: break;
643: case CC_LOWER:
1.192 paf 644: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 645: break;
646: default:
647: assert(!"unknown change case kind");
648: break; // never
649: }
650:
651: } else {
652: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 653:
1.181 paf 654: const unsigned char *a;
655: const unsigned char *b;
656: switch(kind) {
657: case CC_UPPER:
658: a=tables+lcc_offset;
659: b=tables+fcc_offset;
660: break;
661: case CC_LOWER:
662: a=tables+lcc_offset;
663: b=0;
664: break;
665: default:
666: assert(!"unknown change case kind");
667: a=b=0; // calm, compiler
668: break; // never
669: }
670:
1.192 paf 671: char *dest=new_cstr;
1.181 paf 672: unsigned char index;
1.190 paf 673: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 674: unsigned char c=a[index];
675: if(b)
676: c=b[c];
677:
678: *dest++=(char)c;
679: }
1.174 paf 680: }
1.176 paf 681: result.langs=langs;
1.283 moko 682: result.body=String::Body(new_cstr);
1.89 parser 683:
1.101 parser 684: return result;
685: }
686:
1.286 moko 687: String& String::change_char(char from, char to) const {
688: String& result=*new String();
689: if(is_empty())
690: return result;
691:
692: char* new_cstr=cstrm();
693: for(char* c=new_cstr; *c; c++)
694: if(*c==from)
695: *c=to;
696:
697: result.langs=langs;
698: result.body=String::Body(new_cstr);
699:
700: return result;
701: }
702:
1.213 misha 703: const String& String::escape(Charset& source_charset) const {
704: if(is_empty())
705: return *this;
706:
707: return Charset::escape(*this, source_charset);
708: }
709:
1.238 misha 710: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
711: result.langs.append(result.body, langs, langs_offset, length); \
712: result.body.append_strdup_know_length(from_cstr, length);
713:
1.174 paf 714: const String& String::replace(const Dictionary& dict) const {
1.238 misha 715: if(!dict.count() || is_empty())
716: return *this;
717:
1.174 paf 718: String& result=*new String();
719: const char* old_cstr=cstr();
720: const char* prematch_begin=old_cstr;
721:
1.238 misha 722: if(dict.count()==1) {
723: // optimized simple case
724:
725: Dictionary::Subst subst=dict.get(0);
1.239 moko 726: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 727: // prematch
1.238 misha 728: if(size_t prematch_length=p-prematch_begin) {
729: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 730: }
731:
1.174 paf 732: // match
1.238 misha 733: prematch_begin=p+subst.from_length;
1.174 paf 734:
1.184 paf 735: if(const String* b=subst.to) // are there any b?
1.174 paf 736: result<<*b;
1.238 misha 737: }
738:
739: } else {
740:
741: const char* current=old_cstr;
742: while(*current) {
743: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
744: // prematch
745: if(size_t prematch_length=current-prematch_begin) {
746: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
747: }
748:
749: // match
750: // skip 'a' in 'current'; move prematch_begin
751: current+=subst.from_length; prematch_begin=current;
752:
753: if(const String* b=subst.to) // are there any b?
754: result<<*b;
755: } else // simply advance
756: current++;
757: }
758:
1.174 paf 759: }
1.156 paf 760:
1.238 misha 761: if(prematch_begin==old_cstr) // not modified
762: return *this;
763:
1.174 paf 764: // postmatch
1.238 misha 765: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
766: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 767: }
1.156 paf 768:
1.174 paf 769: ASSERT_STRING_INVARIANT(result);
1.82 parser 770: return result;
1.61 paf 771: }
1.113 parser 772:
1.180 paf 773: static int serialize_body_char(char c, char** cur) {
774: *((*cur)++)=c;
775: return 0; // 0=continue
1.248 moko 776: }
777:
1.174 paf 778: static int serialize_body_piece(const char* s, char** cur) {
779: size_t length=strlen(s);
780: memcpy(*cur, s, length); *cur+=length;
1.178 paf 781: return 0; // 0=continue
1.248 moko 782: }
783:
1.178 paf 784: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
785: // lang
1.191 paf 786: **cur=alang; (*cur)++;
787: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 788: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
789:
790: return 0; // 0=continue
791: }
1.248 moko 792:
1.174 paf 793: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 794: size_t fragments_count=langs.count();
1.202 paf 795: size_t body_length=body.length();
1.174 paf 796: size_t buf_length=
1.178 paf 797: prolog_length //1
798: +sizeof(size_t) //2
1.202 paf 799: +body_length //3
800: +1 // 4 for zero terminator used in deserialize
801: +sizeof(size_t) //5
802: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
803:
1.174 paf 804: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
805:
806: // 1: prolog
807: char *cur=result.str+prolog_length;
1.202 paf 808: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
809: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
810: // 3: letters
811: body.for_each(serialize_body_char, serialize_body_piece, &cur);
812: // 4: zero terminator
813: *cur++=0;
814: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 815: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 816: // 6: lang info
1.178 paf 817: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 818:
1.174 paf 819: return result;
1.113 parser 820: }
1.248 moko 821:
1.202 paf 822: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
823: size_t in_buf=buf_size;
824: if(in_buf<=prolog_size)
1.148 paf 825: return false;
1.202 paf 826: in_buf-=prolog_size;
1.135 paf 827:
1.174 paf 828: // 1: prolog
1.202 paf 829: const char* cur=(const char* )buf+prolog_size;
1.113 parser 830:
1.207 paf 831: // 2: chars.count
1.202 paf 832: size_t body_length;
833: if(in_buf<sizeof(body_length)) // body.length don't fit?
834: return false;
835: // [WARNING: not cast, addresses must be %4=0 on sparc]
836: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
837: in_buf-=sizeof(body_length);
838:
839: if(in_buf<body_length+1) // letters+terminator don't fit?
840: return false;
841: // 4: zero terminator
842: if(cur[body_length] != 0) // in place?
843: return false;
844: // 3: letters
1.251 moko 845: body=String::Body(String::C(cur, body_length));
1.202 paf 846: cur+=body_length+1;
847: in_buf-=body_length+1;
848:
849: // 5: langs.count
1.191 paf 850: size_t fragments_count;
1.202 paf 851: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 852: return false;
1.191 paf 853: // [WARNING: not cast, addresses must be %4=0 on sparc]
854: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 855: in_buf-=sizeof(fragments_count);
1.174 paf 856:
857: if(fragments_count) {
1.202 paf 858: // 6: lang info
1.174 paf 859: size_t total_length=0;
860: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 861: char lang;
862: size_t fragment_length;
863: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 864: if(in_buf<piece_length) // lang+length
1.174 paf 865: return false;
866:
1.191 paf 867: // lang
868: lang=*cur++;
869: // length [WARNING: not cast, addresses must be %4=0 on sparc]
870: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
871:
1.206 paf 872: size_t combined_length=total_length+fragment_length;
873: if(combined_length>body_length)
874: return false; // file curruption
1.191 paf 875: // uchar needed to prevent propagating 0x80 bit to upper bytes
876: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 877: total_length=combined_length;
1.202 paf 878: in_buf-=piece_length;
1.174 paf 879: }
1.128 paf 880:
1.202 paf 881: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 882: return false;
1.174 paf 883: }
1.202 paf 884: if(in_buf!=0) // some strange extra bytes
885: return false;
1.113 parser 886:
1.174 paf 887: ASSERT_STRING_INVARIANT(*this);
1.148 paf 888: return true;
1.176 paf 889: }
890:
1.201 paf 891: void String::Body::dump() const {
892: CORD_dump(body);
893: }
894:
1.257 moko 895: const char* String::Languages::visualize() const {
1.177 paf 896: if(opt.is_not_just_lang)
1.233 misha 897: return CORD_to_const_char_star(langs, 0);
1.176 paf 898: else
1.257 moko 899: return 0;
1.176 paf 900: }
1.248 moko 901:
1.201 paf 902: void String::Languages::dump() const {
903: if(opt.is_not_just_lang)
904: CORD_dump(langs);
905: else
906: puts((const char*)&langs);
907: }
1.248 moko 908:
1.257 moko 909: void String::dump() const {
910: body.dump();
911: langs.dump();
912: }
1.176 paf 913:
1.257 moko 914: static char *n_chars(char c, size_t length){
915: char *result=(char *)pa_malloc_atomic(length+1);
916: memset(result, c, length);
917: result[length] = '\0';
918: return result;
1.113 parser 919: }
1.229 misha 920:
1.257 moko 921: char* String::visualize_langs() const {
1.258 moko 922: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 923: }
1.229 misha 924:
925: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 926: if(is_empty())
1.195 paf 927: return *this;
928:
929: size_t substr_begin, substr_length;
1.229 misha 930: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 931: if(new_body==body) // we received unchanged pointer, do likewise
932: return *this;
933: // new_body differs from body, adjust langs along
934:
935: String& result=*new String;
936: if(!new_body) // body.trim produced empty result
937: return result;
938: // body.trim produced nonempty result
939:
940: // first: their langs
941: result.langs.append(result.body, langs, substr_begin, substr_length);
942: // next: letters themselves
943: result.body=new_body;
944:
945: ASSERT_STRING_INVARIANT(result);
946: return result;
1.198 paf 947: }
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