Annotation of parser3/src/main/pa_string.C, revision 1.284
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.277 moko 4: Copyright (c) 2001-2024 Art. Lebedev Studio (http://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:
1.284 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.283 2025/05/26 00:52:15 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:
274:
275: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
276: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
277: if(info->pos < info->left) {
278: info->pos=STRING_NOT_FOUND;
279: return 1;
280: }
281: if(!strchr(info->chars, c))
282: return 1;
283: --(info->pos);
284: return 0;
285: }
286:
287: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
288: if(is_empty() || !chars || !strlen(chars))
289: return STRING_NOT_FOUND;
290: CORD_pos_info info={chars, left, right};
291: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
292: return info.pos;
293: else
294: return STRING_NOT_FOUND;
295: }
296:
1.174 paf 297: // String methods
298:
299: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
300: if(!known_length)
1.9 paf 301: return *this;
1.122 paf 302:
1.176 paf 303: // first: langs
304: langs.append(body, lang, known_length);
305: // next: letters themselves
1.174 paf 306: body.append_know_length(str, known_length);
1.1 paf 307:
1.174 paf 308: ASSERT_STRING_INVARIANT(*this);
1.1 paf 309: return *this;
310: }
1.248 moko 311:
1.174 paf 312: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
313: if(!str)
314: return *this;
315: size_t known_length=helper_length?helper_length:strlen(str);
316: if(!known_length)
317: return *this;
1.1 paf 318:
1.174 paf 319: return append_know_length(str, known_length, lang);
1.5 paf 320: }
1.248 moko 321:
1.244 moko 322: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 323: char buf[MAX_NUMBER];
324: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
325: }
1.248 moko 326:
1.174 paf 327: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
328: size_t known_length=helper_length?helper_length:strlen(str);
329: if(!known_length)
330: return *this;
1.5 paf 331:
1.176 paf 332: // first: langs
333: langs.append(body, lang, known_length);
334: // next: letters themselves
1.174 paf 335: body.append_strdup_know_length(str, known_length);
1.33 paf 336:
1.174 paf 337: ASSERT_STRING_INVARIANT(*this);
338: return *this;
1.5 paf 339: }
1.46 paf 340:
1.234 misha 341: struct CORD_length_info {
342: size_t len;
343: size_t skip;
344: };
345:
346: int CORD_batched_len(const char* s, CORD_length_info* info){
1.255 moko 347: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
348: return 0;
1.221 misha 349: }
350:
1.234 misha 351: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
352: int CORD_batched_len(const char c, CORD_length_info* info){
353: if (info->skip==0){
354: info->len++;
355: info->skip = lengthUTF8Char(c)-1;
356: } else {
357: info->skip--;
358: }
1.220 misha 359: return 0;
360: }
361:
1.210 misha 362: size_t String::length(Charset& charset) const {
363: if(charset.isUTF8()){
1.234 misha 364: CORD_length_info info = {0, 0};
365: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
366: return info.len;
1.210 misha 367: } else
368: return body.length();
369: }
370:
1.174 paf 371: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
372: String& String::mid(size_t substr_begin, size_t substr_end) const {
373: String& result=*new String;
374:
375: size_t self_length=length();
376: substr_begin=min(substr_begin, self_length);
377: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 378: size_t substr_length=substr_end-substr_begin;
379: if(!substr_length)
1.107 parser 380: return result;
1.53 paf 381:
1.176 paf 382: // first: their langs
383: result.langs.append(result.body, langs, substr_begin, substr_length);
384: // next: letters themselves
385: result.body=body.mid(substr_begin, substr_length);
1.174 paf 386:
387: ASSERT_STRING_INVARIANT(result);
1.53 paf 388: return result;
1.54 paf 389: }
390:
1.211 misha 391: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
392: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 393: String& result=*new String;
394:
1.255 moko 395: size_t self_length=helper_length ? helper_length : length(charset);
1.211 misha 396:
397: if(!self_length)
398: return result;
399:
400: from=min(min(to, from), self_length);
1.210 misha 401: to=min(max(to, from), self_length);
1.211 misha 402:
1.210 misha 403: size_t substr_length=to-from;
1.211 misha 404:
1.210 misha 405: if(!substr_length)
406: return result;
407:
408: if(charset.isUTF8()){
1.234 misha 409: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 410: const XMLByte* src_end=src_begin+body.length();
1.210 misha 411:
1.212 misha 412: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 413: from=getUTF8BytePos(src_begin, src_end, from);
414: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 415: if(!substr_length)
416: return result;
417: }
418:
419: // first: their langs
420: result.langs.append(result.body, langs, from, substr_length);
421: // next: letters themselves
422: result.body=body.mid(from, substr_length);
423:
424: ASSERT_STRING_INVARIANT(result);
425: return result;
426: }
427:
1.176 paf 428: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 429: size_t substr_length=substr.length();
430: while(true) {
431: size_t substr_begin=body.pos(substr, this_offset);
432:
433: if(substr_begin==CORD_NOT_FOUND)
434: return STRING_NOT_FOUND;
1.174 paf 435:
1.183 paf 436: if(langs.check_lang(lang, substr_begin, substr_length))
437: return substr_begin;
438:
439: this_offset=substr_begin+substr_length;
440: }
1.58 paf 441: }
442:
1.254 moko 443: size_t String::pos(const String& substr, size_t this_offset, Language lang) const {
1.174 paf 444: return pos(substr.body, this_offset, lang);
1.60 paf 445: }
446:
1.254 moko 447: size_t String::pos(Charset& charset, const String& substr, size_t this_offset, Language lang) const {
1.210 misha 448:
449: if(charset.isUTF8()){
1.234 misha 450: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 451: const XMLByte* srcEnd=srcPtr+body.length();
452:
453: // convert 'this_offset' from 'characters' to 'bytes'
454: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
455:
1.229 misha 456: size_t result=pos(substr.body, this_offset, lang);
457: return (result==CORD_NOT_FOUND)
458: ? STRING_NOT_FOUND
459: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 460: } else {
1.229 misha 461: size_t result=pos(substr.body, this_offset, lang);
462: return (result==CORD_NOT_FOUND)
463: ? STRING_NOT_FOUND
464: : result;
1.210 misha 465: }
466: }
467:
1.256 moko 468: void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
1.237 misha 469: if(is_empty())
470: return;
1.174 paf 471: size_t self_length=length();
1.237 misha 472: if(size_t delim_length=strlen(delim)) {
1.186 paf 473: size_t pos_before;
1.60 paf 474: // while we have 'delim'...
1.283 moko 475: while((pos_before=pos(String::Body(delim), pos_after, lang)) != STRING_NOT_FOUND) {
1.69 paf 476: result+=&mid(pos_after, pos_before);
1.174 paf 477: pos_after=pos_before+delim_length;
1.60 paf 478: }
479: // last piece
1.256 moko 480: if(pos_after<self_length)
1.174 paf 481: result+=&mid(pos_after, self_length);
1.60 paf 482: } else { // empty delim
483: result+=this;
484: }
485: }
486:
1.256 moko 487: void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
1.237 misha 488: if(is_empty())
489: return;
490: if(!delim.is_empty()) {
1.186 paf 491: size_t pos_before;
1.60 paf 492: // while we have 'delim'...
1.256 moko 493: while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
1.69 paf 494: result+=&mid(pos_after, pos_before);
1.174 paf 495: pos_after=pos_before+delim.length();
1.60 paf 496: }
497: // last piece
1.256 moko 498: if(pos_after<length())
1.174 paf 499: result+=&mid(pos_after, length());
1.60 paf 500: } else { // empty delim
501: result+=this;
502: }
1.61 paf 503: }
504:
1.254 moko 505: Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const {
1.209 misha 506:
1.222 misha 507: // vregex->info(); // I have no idea what does it for?
1.63 paf 508:
1.222 misha 509: bool need_pre_post_match=vregex->is_pre_post_match_needed();
510: bool global=vregex->is_global_search();
1.63 paf 511:
1.174 paf 512: const char* subject=cstr();
1.234 misha 513: size_t subject_length=length();
1.271 moko 514: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */
1.222 misha 515: int ovector[ovector_size];
1.155 paf 516:
1.173 paf 517: Table::Action_options table_options;
1.278 moko 518: Table& table=*new Table(string_match_table_template(), table_options);
1.63 paf 519:
1.154 paf 520: int prestart=0;
521: int poststart=0;
1.174 paf 522: int postfinish=length();
1.267 moko 523: int action_was_executed=-1;
1.63 paf 524: while(true) {
1.222 misha 525: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 526:
1.222 misha 527: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
528: break;
1.63 paf 529:
1.154 paf 530: int prefinish=ovector[0];
531: poststart=ovector[1];
1.236 moko 532:
1.267 moko 533: if (prestart==poststart && action_was_executed==1){
1.236 moko 534: prestart++;
1.267 moko 535: action_was_executed=0;
1.236 moko 536: continue;
537: }
538:
1.232 misha 539: ArrayString* row=new ArrayString(3);
1.174 paf 540: if(need_pre_post_match) {
541: *row+=&mid(0, prefinish); // .prematch column value
542: *row+=&mid(prefinish, poststart); // .match
543: *row+=&mid(poststart, postfinish); // .postmatch
544: } else {
1.185 paf 545: *row+=&Empty; // .prematch column value
546: *row+=&Empty; // .match
547: *row+=&Empty; // .postmatch
1.174 paf 548: }
1.63 paf 549:
1.222 misha 550: for(int i=1; i<exec_result; i++) {
1.69 paf 551: // -1:-1 case handled peacefully by mid() itself
1.232 misha 552: *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 553: }
554:
1.209 misha 555: matches_count++;
1.267 moko 556: row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info);
1.63 paf 557:
1.275 moko 558: if(!global || (size_t)poststart>=subject_length) // last step, avoid prestart++ after last char
1.222 misha 559: break;
560:
1.154 paf 561: prestart=poststart;
1.267 moko 562: action_was_executed=1;
1.222 misha 563: }
1.63 paf 564:
1.222 misha 565: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
566: return vregex->is_just_count() ? 0 : &table;
1.82 parser 567: }
568:
1.174 paf 569: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
570: String& result=*new String();
571: if(is_empty())
572: return result;
573:
574: char* new_cstr=cstrm();
1.216 misha 575:
1.181 paf 576: if(source_charset.isUTF8()) {
1.216 misha 577: size_t new_cstr_len=length();
1.181 paf 578: switch(kind) {
579: case CC_UPPER:
1.192 paf 580: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 581: break;
582: case CC_LOWER:
1.192 paf 583: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 584: break;
585: default:
586: assert(!"unknown change case kind");
587: break; // never
588: }
589:
590: } else {
591: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 592:
1.181 paf 593: const unsigned char *a;
594: const unsigned char *b;
595: switch(kind) {
596: case CC_UPPER:
597: a=tables+lcc_offset;
598: b=tables+fcc_offset;
599: break;
600: case CC_LOWER:
601: a=tables+lcc_offset;
602: b=0;
603: break;
604: default:
605: assert(!"unknown change case kind");
606: a=b=0; // calm, compiler
607: break; // never
608: }
609:
1.192 paf 610: char *dest=new_cstr;
1.181 paf 611: unsigned char index;
1.190 paf 612: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 613: unsigned char c=a[index];
614: if(b)
615: c=b[c];
616:
617: *dest++=(char)c;
618: }
1.174 paf 619: }
1.176 paf 620: result.langs=langs;
1.283 moko 621: result.body=String::Body(new_cstr);
1.89 parser 622:
1.101 parser 623: return result;
624: }
625:
1.213 misha 626: const String& String::escape(Charset& source_charset) const {
627: if(is_empty())
628: return *this;
629:
630: return Charset::escape(*this, source_charset);
631: }
632:
1.238 misha 633: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
634: result.langs.append(result.body, langs, langs_offset, length); \
635: result.body.append_strdup_know_length(from_cstr, length);
636:
1.174 paf 637: const String& String::replace(const Dictionary& dict) const {
1.238 misha 638: if(!dict.count() || is_empty())
639: return *this;
640:
1.174 paf 641: String& result=*new String();
642: const char* old_cstr=cstr();
643: const char* prematch_begin=old_cstr;
644:
1.238 misha 645: if(dict.count()==1) {
646: // optimized simple case
647:
648: Dictionary::Subst subst=dict.get(0);
1.239 moko 649: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 650: // prematch
1.238 misha 651: if(size_t prematch_length=p-prematch_begin) {
652: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 653: }
654:
1.174 paf 655: // match
1.238 misha 656: prematch_begin=p+subst.from_length;
1.174 paf 657:
1.184 paf 658: if(const String* b=subst.to) // are there any b?
1.174 paf 659: result<<*b;
1.238 misha 660: }
661:
662: } else {
663:
664: const char* current=old_cstr;
665: while(*current) {
666: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
667: // prematch
668: if(size_t prematch_length=current-prematch_begin) {
669: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
670: }
671:
672: // match
673: // skip 'a' in 'current'; move prematch_begin
674: current+=subst.from_length; prematch_begin=current;
675:
676: if(const String* b=subst.to) // are there any b?
677: result<<*b;
678: } else // simply advance
679: current++;
680: }
681:
1.174 paf 682: }
1.156 paf 683:
1.238 misha 684: if(prematch_begin==old_cstr) // not modified
685: return *this;
686:
1.174 paf 687: // postmatch
1.238 misha 688: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
689: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 690: }
1.156 paf 691:
1.174 paf 692: ASSERT_STRING_INVARIANT(result);
1.82 parser 693: return result;
1.61 paf 694: }
1.113 parser 695:
1.180 paf 696: static int serialize_body_char(char c, char** cur) {
697: *((*cur)++)=c;
698: return 0; // 0=continue
1.248 moko 699: }
700:
1.174 paf 701: static int serialize_body_piece(const char* s, char** cur) {
702: size_t length=strlen(s);
703: memcpy(*cur, s, length); *cur+=length;
1.178 paf 704: return 0; // 0=continue
1.248 moko 705: }
706:
1.178 paf 707: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
708: // lang
1.191 paf 709: **cur=alang; (*cur)++;
710: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 711: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
712:
713: return 0; // 0=continue
714: }
1.248 moko 715:
1.174 paf 716: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 717: size_t fragments_count=langs.count();
1.202 paf 718: size_t body_length=body.length();
1.174 paf 719: size_t buf_length=
1.178 paf 720: prolog_length //1
721: +sizeof(size_t) //2
1.202 paf 722: +body_length //3
723: +1 // 4 for zero terminator used in deserialize
724: +sizeof(size_t) //5
725: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
726:
1.174 paf 727: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
728:
729: // 1: prolog
730: char *cur=result.str+prolog_length;
1.202 paf 731: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
732: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
733: // 3: letters
734: body.for_each(serialize_body_char, serialize_body_piece, &cur);
735: // 4: zero terminator
736: *cur++=0;
737: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 738: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 739: // 6: lang info
1.178 paf 740: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 741:
1.174 paf 742: return result;
1.113 parser 743: }
1.248 moko 744:
1.202 paf 745: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
746: size_t in_buf=buf_size;
747: if(in_buf<=prolog_size)
1.148 paf 748: return false;
1.202 paf 749: in_buf-=prolog_size;
1.135 paf 750:
1.174 paf 751: // 1: prolog
1.202 paf 752: const char* cur=(const char* )buf+prolog_size;
1.113 parser 753:
1.207 paf 754: // 2: chars.count
1.202 paf 755: size_t body_length;
756: if(in_buf<sizeof(body_length)) // body.length don't fit?
757: return false;
758: // [WARNING: not cast, addresses must be %4=0 on sparc]
759: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
760: in_buf-=sizeof(body_length);
761:
762: if(in_buf<body_length+1) // letters+terminator don't fit?
763: return false;
764: // 4: zero terminator
765: if(cur[body_length] != 0) // in place?
766: return false;
767: // 3: letters
1.251 moko 768: body=String::Body(String::C(cur, body_length));
1.202 paf 769: cur+=body_length+1;
770: in_buf-=body_length+1;
771:
772: // 5: langs.count
1.191 paf 773: size_t fragments_count;
1.202 paf 774: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 775: return false;
1.191 paf 776: // [WARNING: not cast, addresses must be %4=0 on sparc]
777: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 778: in_buf-=sizeof(fragments_count);
1.174 paf 779:
780: if(fragments_count) {
1.202 paf 781: // 6: lang info
1.174 paf 782: size_t total_length=0;
783: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 784: char lang;
785: size_t fragment_length;
786: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 787: if(in_buf<piece_length) // lang+length
1.174 paf 788: return false;
789:
1.191 paf 790: // lang
791: lang=*cur++;
792: // length [WARNING: not cast, addresses must be %4=0 on sparc]
793: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
794:
1.206 paf 795: size_t combined_length=total_length+fragment_length;
796: if(combined_length>body_length)
797: return false; // file curruption
1.191 paf 798: // uchar needed to prevent propagating 0x80 bit to upper bytes
799: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 800: total_length=combined_length;
1.202 paf 801: in_buf-=piece_length;
1.174 paf 802: }
1.128 paf 803:
1.202 paf 804: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 805: return false;
1.174 paf 806: }
1.202 paf 807: if(in_buf!=0) // some strange extra bytes
808: return false;
1.113 parser 809:
1.174 paf 810: ASSERT_STRING_INVARIANT(*this);
1.148 paf 811: return true;
1.176 paf 812: }
813:
1.201 paf 814: void String::Body::dump() const {
815: CORD_dump(body);
816: }
817:
1.257 moko 818: const char* String::Languages::visualize() const {
1.177 paf 819: if(opt.is_not_just_lang)
1.233 misha 820: return CORD_to_const_char_star(langs, 0);
1.176 paf 821: else
1.257 moko 822: return 0;
1.176 paf 823: }
1.248 moko 824:
1.201 paf 825: void String::Languages::dump() const {
826: if(opt.is_not_just_lang)
827: CORD_dump(langs);
828: else
829: puts((const char*)&langs);
830: }
1.248 moko 831:
1.257 moko 832: void String::dump() const {
833: body.dump();
834: langs.dump();
835: }
1.176 paf 836:
1.257 moko 837: static char *n_chars(char c, size_t length){
838: char *result=(char *)pa_malloc_atomic(length+1);
839: memset(result, c, length);
840: result[length] = '\0';
841: return result;
1.113 parser 842: }
1.229 misha 843:
1.257 moko 844: char* String::visualize_langs() const {
1.258 moko 845: return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
1.201 paf 846: }
1.229 misha 847:
848: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 849: if(is_empty())
1.195 paf 850: return *this;
851:
852: size_t substr_begin, substr_length;
1.229 misha 853: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 854: if(new_body==body) // we received unchanged pointer, do likewise
855: return *this;
856: // new_body differs from body, adjust langs along
857:
858: String& result=*new String;
859: if(!new_body) // body.trim produced empty result
860: return result;
861: // body.trim produced nonempty result
862:
863: // first: their langs
864: result.langs.append(result.body, langs, substr_begin, substr_length);
865: // next: letters themselves
866: result.body=new_body;
867:
868: ASSERT_STRING_INVARIANT(result);
869: return result;
1.198 paf 870: }
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