Annotation of parser3/src/main/pa_string.C, revision 1.238
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.219 misha 4: Copyright (c) 2001-2009 ArtLebedev Group (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:
1.238 ! misha 8: static const char * const IDENT_STRING_C="$Date: 2011-05-05 23:45:45 $";
1.70 paf 9:
1.12 paf 10: #include "pa_string.h"
1.22 paf 11: #include "pa_exception.h"
1.61 paf 12: #include "pa_table.h"
1.101 parser 13: #include "pa_dictionary.h"
1.132 paf 14: #include "pa_charset.h"
1.222 misha 15: #include "pa_vregex.h"
1.215 misha 16:
1.185 paf 17: const String String::Empty;
18:
1.193 paf 19: int pa_atoi(const char* str, const String* problem_source) {
20: if(!str)
21: return 0;
22:
1.199 paf 23: while(*str && isspace((unsigned char)*str))
1.193 paf 24: str++;
25: if(!*str)
26: return 0;
27:
28: int result;
29: char *error_pos;
1.200 paf 30: bool negative=false;
31: if(str[0]=='-') {
32: negative=true;
33: str++;
34: } else if(str[0]=='+') {
35: str++;
36: }
1.193 paf 37: // 0xABC
38: if(str[0]=='0')
39: if(str[1]=='x' || str[1]=='X')
40: result=(int)(unsigned long)strtol(str, &error_pos, 0);
1.197 paf 41: else {
42: // skip leading 0000, to disable octal interpretation
43: do str++; while(*str=='0');
44: result=(int)strtol(str, &error_pos, 0);
45: }
1.193 paf 46: else
47: result=(int)strtol(str, &error_pos, 0);
1.200 paf 48: if(negative)
49: result=-result;
1.193 paf 50:
51: while(char c=*error_pos++)
1.199 paf 52: if(!isspace((unsigned char)c))
1.193 paf 53: throw Exception("number.format",
54: problem_source,
55: problem_source?"invalid number (int)": "'%s' is invalid number (int)", str);
56:
57: return result;
58: }
59:
60: double pa_atod(const char* str, const String* problem_source) {
61: if(!str)
62: return 0;
63:
1.199 paf 64: while(*str && isspace((unsigned char)*str))
1.193 paf 65: str++;
66: if(!*str)
67: return 0;
68:
69: double result;
70: char *error_pos;
1.200 paf 71: bool negative=false;
72: if(str[0]=='-') {
73: negative=true;
74: str++;
75: } else if(str[0]=='+') {
76: str++;
77: }
1.193 paf 78: // 0xABC
79: if(str[0]=='0')
80: if(str[1]=='x' || str[1]=='X')
81: result=(double)(unsigned long)strtol(str, &error_pos, 0);
1.200 paf 82: else {
83: // skip leading 0000, to disable octal interpretation
84: do str++; while(*str=='0');
85: result=(double)strtod(str, &error_pos);
86: }
1.193 paf 87: else
88: result=(double)strtod(str, &error_pos);
1.200 paf 89: if(negative)
90: result=-result;
1.193 paf 91:
92: while(char c=*error_pos++)
1.199 paf 93: if(!isspace((unsigned char)c))
1.193 paf 94: throw Exception("number.format",
95: problem_source,
96: problem_source?"invalid number (double)": "'%s' is invalid number (double)", str);
97:
98: return result;
99: }
100:
1.176 paf 101: // cord lib extension
102:
103: #ifndef DOXYGEN
104: typedef struct {
105: ssize_t countdown;
1.204 paf 106: int target; /* Character we're looking for */
1.176 paf 107: } chr_data;
108: #endif
109: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
110: {
111: register chr_data * d = (chr_data *)client_data;
112:
113: if (d -> countdown<=0) return(2);
114: d -> countdown -= size;
115: if (c > d -> target) return(1);
116: return(0);
117: }
118: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
119: {
120: chr_data d;
121:
122: d.countdown = n;
123: d.target = c;
124: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
125: }
126:
1.187 paf 127: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 128: {
129: int* result=(int*)client_data;
130: (*result)++;
131: return(0); // 0=continue
132: }
133: size_t CORD_block_count(CORD x)
134: {
135: size_t result=0;
136: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
137: return result;
138: }
139:
1.174 paf 140: // helpers
1.139 paf 141:
1.174 paf 142: /// String::match uses this as replace & global search table columns
1.139 paf 143:
1.174 paf 144: const int MAX_MATCH_GROUPS=100;
1.139 paf 145:
1.174 paf 146: class String_match_table_template_columns: public ArrayString {
147: public:
148: String_match_table_template_columns() {
149: *this+=new String("prematch");
150: *this+=new String("match");
151: *this+=new String("postmatch");
152: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 153: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 154: }
155: }
156: };
157:
158: Table string_match_table_template(new String_match_table_template_columns);
159:
1.176 paf 160: // String::Body methods
1.140 paf 161:
1.176 paf 162: String::Body String::Body::Format(int value) {
1.174 paf 163: char local[MAX_NUMBER];
164: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 165: return String::Body(pa_strdup(local, length));
1.120 paf 166: }
167:
1.195 paf 168: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 169: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 170: size_t our_length=length();
171: if(!our_length)
172: return *this;
1.229 misha 173:
174: // check if any UTF-8 in chars
175: bool fast=true;
176: if(chars && source_charset && source_charset->isUTF8()){
177: const char* pos=chars;
178: while(unsigned char c=*pos++)
179: if(c>127){
180: fast=false;
181: break;
182: }
183: }
184:
185: size_t start=0;
186: size_t end=our_length;
1.195 paf 187: if(!chars)
188: chars=" \t\n"; // white space
189:
1.229 misha 190: if(fast){
191: // from left...
192: if(kind!=TRIM_END) {
193: CORD_pos pos; set_pos(pos, 0);
194: while(true) {
195: char c=CORD_pos_fetch(pos);
196: if(strchr(chars, c)) {
197: if(++start==our_length)
198: return 0; // all chars are empty, just return empty string
199: } else
200: break;
201:
202: CORD_next(pos);
203: }
204: }
205:
206: // from right..
207: if(kind!=TRIM_START) {
208: CORD_pos pos; set_pos(pos, end-1);
209: while(true) {
210: char c=CORD_pos_fetch(pos);
211: if(strchr(chars, c)) {
212: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
213: return 0; // all chars are empty, just return empty string
214: } else
215: break;
216:
217: CORD_prev(pos);
218: }
219: }
220: } else {
1.234 misha 221: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 222: const XMLByte* src_end=src_begin+our_length;
223:
224: // from left...
225: if(kind!=TRIM_END) {
226: while(src_begin<src_end){
227: uint char_length=1;
228: const XMLByte* ptr=src_begin;
1.231 misha 229: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
230: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 231: char_length++;
232:
233: bool found=false;
234: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
235: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
236: found=true;
237: break;
238: }
239:
240: if(found){
241: start+=char_length;
242: if(start==our_length)
243: return 0; // all chars are empty, just return empty string
244: } else
245: break;
246: }
247: }
1.196 paf 248:
1.229 misha 249: // from right..
250: if(kind!=TRIM_START) {
251: while(src_begin<src_end){
252: uint char_length=1;
1.231 misha 253: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
254: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 255: char_length++;
256:
257: bool found=false;
258: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
259: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
260: found=true;
261: break;
262: }
263:
264: if(found){
265: end-=char_length;
266: if(end==0)
267: return 0; // all chars are empty, just return empty string
268: } else
269: break;
270: }
1.196 paf 271: }
272: }
1.195 paf 273:
274: if(start==0 && end==our_length) // nobody moved a thing
275: return *this;
276:
277: if(out_start)
278: *out_start=start;
279: size_t new_length=end-start;
280: if(out_length)
281: *out_length=new_length;
282:
283: return mid(start, new_length);
284: }
285:
1.174 paf 286: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
287: uint& result=*static_cast<uint*>(client_data);
288: generic_hash_code(result, c);
289: return 0;
290: }
291: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
292: uint& result=*static_cast<uint*>(client_data);
293: generic_hash_code(result, s);
294: return 0;
295: };
1.225 misha 296: uint String::Body::get_hash_code() const {
297: #ifdef HASH_CODE_CACHING
298: if(hash_code)
299: return hash_code;
300: #else
301: uint hash_code=0;
302: #endif
1.220 misha 303: if (body && CORD_IS_STRING(body)){
1.225 misha 304: generic_hash_code(hash_code, body);
1.220 misha 305: } else {
306: CORD_iter5(body, 0,
307: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 308: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 309: }
1.225 misha 310: return hash_code;
1.94 parser 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: }
1.174 paf 327: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
328: if(!str)
329: return *this;
330: size_t known_length=helper_length?helper_length:strlen(str);
331: if(!known_length)
332: return *this;
1.1 paf 333:
1.174 paf 334: return append_know_length(str, known_length, lang);
1.5 paf 335: }
1.226 misha 336: String::String(int value, char *format) : langs(L_CLEAN){
337: char buf[MAX_NUMBER];
338: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
339: }
1.174 paf 340: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
341: size_t known_length=helper_length?helper_length:strlen(str);
342: if(!known_length)
343: return *this;
1.5 paf 344:
1.176 paf 345: // first: langs
346: langs.append(body, lang, known_length);
347: // next: letters themselves
1.174 paf 348: body.append_strdup_know_length(str, known_length);
1.33 paf 349:
1.174 paf 350: ASSERT_STRING_INVARIANT(*this);
351: return *this;
1.5 paf 352: }
1.46 paf 353:
1.234 misha 354: struct CORD_length_info {
355: size_t len;
356: size_t skip;
357: };
358:
359: int CORD_batched_len(const char* s, CORD_length_info* info){
360: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 361: }
362:
1.234 misha 363: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
364: int CORD_batched_len(const char c, CORD_length_info* info){
365: if (info->skip==0){
366: info->len++;
367: info->skip = lengthUTF8Char(c)-1;
368: } else {
369: info->skip--;
370: }
1.220 misha 371: return 0;
372: }
373:
1.210 misha 374: size_t String::length(Charset& charset) const {
375: if(charset.isUTF8()){
1.234 misha 376: CORD_length_info info = {0, 0};
377: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
378: return info.len;
1.210 misha 379: } else
380: return body.length();
381: }
382:
1.174 paf 383: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
384: String& String::mid(size_t substr_begin, size_t substr_end) const {
385: String& result=*new String;
386:
387: size_t self_length=length();
388: substr_begin=min(substr_begin, self_length);
389: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 390: size_t substr_length=substr_end-substr_begin;
391: if(!substr_length)
1.107 parser 392: return result;
1.53 paf 393:
1.176 paf 394: // first: their langs
395: result.langs.append(result.body, langs, substr_begin, substr_length);
396: // next: letters themselves
397: result.body=body.mid(substr_begin, substr_length);
1.174 paf 398:
399: ASSERT_STRING_INVARIANT(result);
1.53 paf 400: return result;
1.54 paf 401: }
402:
1.211 misha 403: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
404: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 405: String& result=*new String;
406:
1.211 misha 407: size_t self_length=(helper_length)?helper_length:length(charset);
408:
409: if(!self_length)
410: return result;
411:
412: from=min(min(to, from), self_length);
1.210 misha 413: to=min(max(to, from), self_length);
1.211 misha 414:
1.210 misha 415: size_t substr_length=to-from;
1.211 misha 416:
1.210 misha 417: if(!substr_length)
418: return result;
419:
420: if(charset.isUTF8()){
1.234 misha 421: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 422: const XMLByte* src_end=src_begin+body.length();
1.210 misha 423:
1.212 misha 424: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 425: from=getUTF8BytePos(src_begin, src_end, from);
426: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 427: if(!substr_length)
428: return result;
429: }
430:
431: // first: their langs
432: result.langs.append(result.body, langs, from, substr_length);
433: // next: letters themselves
434: result.body=body.mid(from, substr_length);
435:
436: ASSERT_STRING_INVARIANT(result);
437: return result;
438: }
439:
1.176 paf 440: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 441: size_t substr_length=substr.length();
442: while(true) {
443: size_t substr_begin=body.pos(substr, this_offset);
444:
445: if(substr_begin==CORD_NOT_FOUND)
446: return STRING_NOT_FOUND;
1.174 paf 447:
1.183 paf 448: if(langs.check_lang(lang, substr_begin, substr_length))
449: return substr_begin;
450:
451: this_offset=substr_begin+substr_length;
452: }
1.58 paf 453: }
454:
1.174 paf 455: size_t String::pos(const String& substr,
456: size_t this_offset, Language lang) const {
457: return pos(substr.body, this_offset, lang);
1.60 paf 458: }
459:
1.210 misha 460: size_t String::pos(Charset& charset, const String& substr,
461: size_t this_offset, Language lang) const {
462:
463: if(charset.isUTF8()){
1.234 misha 464: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 465: const XMLByte* srcEnd=srcPtr+body.length();
466:
467: // convert 'this_offset' from 'characters' to 'bytes'
468: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
469:
1.229 misha 470: size_t result=pos(substr.body, this_offset, lang);
471: return (result==CORD_NOT_FOUND)
472: ? STRING_NOT_FOUND
473: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 474: } else {
1.229 misha 475: size_t result=pos(substr.body, this_offset, lang);
476: return (result==CORD_NOT_FOUND)
477: ? STRING_NOT_FOUND
478: : result;
1.210 misha 479: }
480: }
481:
1.174 paf 482: void String::split(ArrayString& result,
483: size_t& pos_after,
484: const char* delim,
485: Language lang, int limit) const {
1.237 misha 486: if(is_empty())
487: return;
1.174 paf 488: size_t self_length=length();
1.237 misha 489: if(size_t delim_length=strlen(delim)) {
1.186 paf 490: size_t pos_before;
1.60 paf 491: // while we have 'delim'...
1.174 paf 492: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 493: result+=&mid(pos_after, pos_before);
1.174 paf 494: pos_after=pos_before+delim_length;
1.60 paf 495: }
496: // last piece
1.174 paf 497: if(pos_after<self_length && limit) {
498: result+=&mid(pos_after, self_length);
499: pos_after=self_length;
1.60 paf 500: }
501: } else { // empty delim
502: result+=this;
1.174 paf 503: pos_after+=self_length;
1.60 paf 504: }
505: }
506:
1.174 paf 507: void String::split(ArrayString& result,
508: size_t& pos_after,
509: const String& delim, Language lang,
510: int limit) const {
1.237 misha 511: if(is_empty())
512: return;
513: if(!delim.is_empty()) {
1.186 paf 514: size_t pos_before;
1.60 paf 515: // while we have 'delim'...
1.174 paf 516: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 517: result+=&mid(pos_after, pos_before);
1.174 paf 518: pos_after=pos_before+delim.length();
1.60 paf 519: }
520: // last piece
1.174 paf 521: if(pos_after<length() && limit) {
522: result+=&mid(pos_after, length());
523: pos_after=length();
1.60 paf 524: }
525: } else { // empty delim
526: result+=this;
1.174 paf 527: pos_after+=length();
1.60 paf 528: }
1.61 paf 529: }
530:
1.222 misha 531: Table* String::match(VRegex* vregex,
532: Row_action row_action, void *info,
533: int& matches_count) const {
1.209 misha 534:
1.222 misha 535: // vregex->info(); // I have no idea what does it for?
1.63 paf 536:
1.222 misha 537: bool need_pre_post_match=vregex->is_pre_post_match_needed();
538: bool global=vregex->is_global_search();
1.63 paf 539:
1.174 paf 540: const char* subject=cstr();
1.234 misha 541: size_t subject_length=length();
1.222 misha 542: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
543: int ovector[ovector_size];
1.155 paf 544:
1.173 paf 545: Table::Action_options table_options;
1.174 paf 546: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 547:
1.154 paf 548: int prestart=0;
549: int poststart=0;
1.174 paf 550: int postfinish=length();
1.63 paf 551: while(true) {
1.222 misha 552: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 553:
1.222 misha 554: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
555: break;
1.63 paf 556:
1.154 paf 557: int prefinish=ovector[0];
558: poststart=ovector[1];
1.236 moko 559:
560: if (prestart==poststart && subject[poststart]=='\n'){
561: prestart++;
562: continue;
563: }
564:
1.232 misha 565: ArrayString* row=new ArrayString(3);
1.174 paf 566: if(need_pre_post_match) {
567: *row+=&mid(0, prefinish); // .prematch column value
568: *row+=&mid(prefinish, poststart); // .match
569: *row+=&mid(poststart, postfinish); // .postmatch
570: } else {
1.185 paf 571: *row+=&Empty; // .prematch column value
572: *row+=&Empty; // .match
573: *row+=&Empty; // .postmatch
1.174 paf 574: }
1.63 paf 575:
1.222 misha 576: for(int i=1; i<exec_result; i++) {
1.69 paf 577: // -1:-1 case handled peacefully by mid() itself
1.232 misha 578: *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 579: }
580:
1.209 misha 581: matches_count++;
1.174 paf 582: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 583:
1.222 misha 584: if(!global || prestart==poststart) // last step
585: break;
586:
1.154 paf 587: prestart=poststart;
1.222 misha 588: }
1.63 paf 589:
1.222 misha 590: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
591: return vregex->is_just_count() ? 0 : &table;
1.82 parser 592: }
593:
1.174 paf 594: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
595: String& result=*new String();
596: if(is_empty())
597: return result;
598:
599: char* new_cstr=cstrm();
1.216 misha 600:
1.181 paf 601: if(source_charset.isUTF8()) {
1.216 misha 602: size_t new_cstr_len=length();
1.181 paf 603: switch(kind) {
604: case CC_UPPER:
1.192 paf 605: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 606: break;
607: case CC_LOWER:
1.192 paf 608: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 609: break;
610: default:
611: assert(!"unknown change case kind");
612: break; // never
613: }
614:
615: } else {
616: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 617:
1.181 paf 618: const unsigned char *a;
619: const unsigned char *b;
620: switch(kind) {
621: case CC_UPPER:
622: a=tables+lcc_offset;
623: b=tables+fcc_offset;
624: break;
625: case CC_LOWER:
626: a=tables+lcc_offset;
627: b=0;
628: break;
629: default:
630: assert(!"unknown change case kind");
631: a=b=0; // calm, compiler
632: break; // never
633: }
634:
1.192 paf 635: char *dest=new_cstr;
1.181 paf 636: unsigned char index;
1.190 paf 637: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 638: unsigned char c=a[index];
639: if(b)
640: c=b[c];
641:
642: *dest++=(char)c;
643: }
1.174 paf 644: }
1.176 paf 645: result.langs=langs;
1.174 paf 646: result.body=new_cstr;
1.89 parser 647:
1.101 parser 648: return result;
649: }
650:
1.213 misha 651: const String& String::escape(Charset& source_charset) const {
652: if(is_empty())
653: return *this;
654:
655: return Charset::escape(*this, source_charset);
656: }
657:
1.238 ! misha 658: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
! 659: result.langs.append(result.body, langs, langs_offset, length); \
! 660: result.body.append_strdup_know_length(from_cstr, length);
! 661:
1.174 paf 662: const String& String::replace(const Dictionary& dict) const {
1.238 ! misha 663: if(!dict.count() || is_empty())
! 664: return *this;
! 665:
1.174 paf 666: String& result=*new String();
667: const char* old_cstr=cstr();
668: const char* prematch_begin=old_cstr;
669:
1.238 ! misha 670: if(dict.count()==1) {
! 671: // optimized simple case
! 672:
! 673: Dictionary::Subst subst=dict.get(0);
! 674: while(char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 675: // prematch
1.238 ! misha 676: if(size_t prematch_length=p-prematch_begin) {
! 677: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 678: }
679:
1.174 paf 680: // match
1.238 ! misha 681: prematch_begin=p+subst.from_length;
1.174 paf 682:
1.184 paf 683: if(const String* b=subst.to) // are there any b?
1.174 paf 684: result<<*b;
1.238 ! misha 685: }
! 686:
! 687: } else {
! 688:
! 689: const char* current=old_cstr;
! 690: while(*current) {
! 691: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
! 692: // prematch
! 693: if(size_t prematch_length=current-prematch_begin) {
! 694: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
! 695: }
! 696:
! 697: // match
! 698: // skip 'a' in 'current'; move prematch_begin
! 699: current+=subst.from_length; prematch_begin=current;
! 700:
! 701: if(const String* b=subst.to) // are there any b?
! 702: result<<*b;
! 703: } else // simply advance
! 704: current++;
! 705: }
! 706:
1.174 paf 707: }
1.156 paf 708:
1.238 ! misha 709: if(prematch_begin==old_cstr) // not modified
! 710: return *this;
! 711:
1.174 paf 712: // postmatch
1.238 ! misha 713: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
! 714: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 715: }
1.156 paf 716:
1.174 paf 717: ASSERT_STRING_INVARIANT(result);
1.82 parser 718: return result;
1.61 paf 719: }
1.113 parser 720:
1.180 paf 721: static int serialize_body_char(char c, char** cur) {
722: *((*cur)++)=c;
723: return 0; // 0=continue
724: };
1.174 paf 725: static int serialize_body_piece(const char* s, char** cur) {
726: size_t length=strlen(s);
727: memcpy(*cur, s, length); *cur+=length;
1.178 paf 728: return 0; // 0=continue
1.174 paf 729: };
1.178 paf 730: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
731: // lang
1.191 paf 732: **cur=alang; (*cur)++;
733: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 734: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
735:
736: return 0; // 0=continue
737: }
1.174 paf 738: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 739: size_t fragments_count=langs.count();
1.202 paf 740: size_t body_length=body.length();
1.174 paf 741: size_t buf_length=
1.178 paf 742: prolog_length //1
743: +sizeof(size_t) //2
1.202 paf 744: +body_length //3
745: +1 // 4 for zero terminator used in deserialize
746: +sizeof(size_t) //5
747: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
748:
1.174 paf 749: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
750:
751: // 1: prolog
752: char *cur=result.str+prolog_length;
1.202 paf 753: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
754: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
755: // 3: letters
756: body.for_each(serialize_body_char, serialize_body_piece, &cur);
757: // 4: zero terminator
758: *cur++=0;
759: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 760: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 761: // 6: lang info
1.178 paf 762: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 763:
1.174 paf 764: return result;
1.113 parser 765: }
1.202 paf 766: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
767: size_t in_buf=buf_size;
768: if(in_buf<=prolog_size)
1.148 paf 769: return false;
1.202 paf 770: in_buf-=prolog_size;
1.135 paf 771:
1.174 paf 772: // 1: prolog
1.202 paf 773: const char* cur=(const char* )buf+prolog_size;
1.113 parser 774:
1.207 paf 775: // 2: chars.count
1.202 paf 776: size_t body_length;
777: if(in_buf<sizeof(body_length)) // body.length don't fit?
778: return false;
779: // [WARNING: not cast, addresses must be %4=0 on sparc]
780: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
781: in_buf-=sizeof(body_length);
782:
783: if(in_buf<body_length+1) // letters+terminator don't fit?
784: return false;
785: // 4: zero terminator
786: if(cur[body_length] != 0) // in place?
787: return false;
788: // 3: letters
1.227 misha 789: body=String::Body(*cur?cur:0);
1.229 misha 790: #ifdef STRING_LENGTH_CACHING
791: body.set_length(body_length);
792: #endif
1.202 paf 793: cur+=body_length+1;
794: in_buf-=body_length+1;
795:
796: // 5: langs.count
1.191 paf 797: size_t fragments_count;
1.202 paf 798: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 799: return false;
1.191 paf 800: // [WARNING: not cast, addresses must be %4=0 on sparc]
801: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 802: in_buf-=sizeof(fragments_count);
1.174 paf 803:
804: if(fragments_count) {
1.202 paf 805: // 6: lang info
1.174 paf 806: size_t total_length=0;
807: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 808: char lang;
809: size_t fragment_length;
810: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 811: if(in_buf<piece_length) // lang+length
1.174 paf 812: return false;
813:
1.191 paf 814: // lang
815: lang=*cur++;
816: // length [WARNING: not cast, addresses must be %4=0 on sparc]
817: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
818:
1.206 paf 819: size_t combined_length=total_length+fragment_length;
820: if(combined_length>body_length)
821: return false; // file curruption
1.191 paf 822: // uchar needed to prevent propagating 0x80 bit to upper bytes
823: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 824: total_length=combined_length;
1.202 paf 825: in_buf-=piece_length;
1.174 paf 826: }
1.128 paf 827:
1.202 paf 828: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 829: return false;
1.174 paf 830: }
1.202 paf 831: if(in_buf!=0) // some strange extra bytes
832: return false;
1.113 parser 833:
1.174 paf 834: ASSERT_STRING_INVARIANT(*this);
1.148 paf 835: return true;
1.176 paf 836: }
837:
838: const char* String::Body::v() const {
1.233 misha 839: return CORD_to_const_char_star(body, length());
1.176 paf 840: }
1.201 paf 841: void String::Body::dump() const {
842: CORD_dump(body);
843: }
844:
1.176 paf 845: const char* String::Languages::v() const {
1.177 paf 846: if(opt.is_not_just_lang)
1.233 misha 847: return CORD_to_const_char_star(langs, 0);
1.176 paf 848: else
849: return (const char*)&langs;
850: }
1.201 paf 851: void String::Languages::dump() const {
852: if(opt.is_not_just_lang)
853: CORD_dump(langs);
854: else
855: puts((const char*)&langs);
856: }
1.176 paf 857: const char* String::v() const {
1.198 paf 858: const uint LIMIT_VIEW=20;
1.176 paf 859: char* buf=(char*)malloc(MAX_STRING);
860: const char*body_view=body.v();
861: const char*langs_view=langs.v();
862: snprintf(buf, MAX_STRING,
1.178 paf 863: "%d:%.*s%s} "
1.176 paf 864: "{%d:%s",
1.178 paf 865: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 866: strlen(body_view), body_view
867: );
868:
869: return buf;
1.113 parser 870: }
1.229 misha 871:
1.201 paf 872: void String::dump() const {
873: body.dump();
874: langs.dump();
875: }
1.229 misha 876:
877: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 878: if(is_empty())
1.195 paf 879: return *this;
880:
881: size_t substr_begin, substr_length;
1.229 misha 882: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 883: if(new_body==body) // we received unchanged pointer, do likewise
884: return *this;
885: // new_body differs from body, adjust langs along
886:
887: String& result=*new String;
888: if(!new_body) // body.trim produced empty result
889: return result;
890: // body.trim produced nonempty result
891:
892: // first: their langs
893: result.langs.append(result.body, langs, substr_begin, substr_length);
894: // next: letters themselves
895: result.body=new_body;
896:
897: ASSERT_STRING_INVARIANT(result);
898: return result;
1.198 paf 899: }
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