Annotation of parser3/src/main/pa_string.C, revision 1.230
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:
1.230 ! misha 8: static const char * const IDENT_STRING_C="$Date: 2009-06-23 10:05:16 $";
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 {
221: const XMLByte* src_begin=(const XMLByte*)cstrm();
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;
229: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0)/* searching first UTF-8 byte */)
230: char_length++;
231:
232: bool found=false;
233: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
234: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
235: found=true;
236: break;
237: }
238:
239: if(found){
240: start+=char_length;
241: if(start==our_length)
242: return 0; // all chars are empty, just return empty string
243: } else
244: break;
245: }
246: }
1.196 paf 247:
1.229 misha 248: // from right..
249: if(kind!=TRIM_START) {
250: while(src_begin<src_end){
251: uint char_length=1;
252: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0)/* searching first UTF-8 byte */)
253: char_length++;
254:
255: bool found=false;
256: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
257: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
258: found=true;
259: break;
260: }
261:
262: if(found){
263: end-=char_length;
264: if(end==0)
265: return 0; // all chars are empty, just return empty string
266: } else
267: break;
268: }
1.196 paf 269: }
270: }
1.195 paf 271:
272: if(start==0 && end==our_length) // nobody moved a thing
273: return *this;
274:
275: if(out_start)
276: *out_start=start;
277: size_t new_length=end-start;
278: if(out_length)
279: *out_length=new_length;
280:
281: return mid(start, new_length);
282: }
283:
1.174 paf 284: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
285: uint& result=*static_cast<uint*>(client_data);
286: generic_hash_code(result, c);
287: return 0;
288: }
289: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
290: uint& result=*static_cast<uint*>(client_data);
291: generic_hash_code(result, s);
292: return 0;
293: };
1.225 misha 294: uint String::Body::get_hash_code() const {
295: #ifdef HASH_CODE_CACHING
296: if(hash_code)
297: return hash_code;
298: #else
299: uint hash_code=0;
300: #endif
1.220 misha 301: if (body && CORD_IS_STRING(body)){
1.225 misha 302: generic_hash_code(hash_code, body);
1.220 misha 303: } else {
304: CORD_iter5(body, 0,
305: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 306: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 307: }
1.225 misha 308: return hash_code;
1.94 parser 309: }
310:
1.174 paf 311: // String methods
312:
313: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
314: if(!known_length)
1.9 paf 315: return *this;
1.122 paf 316:
1.176 paf 317: // first: langs
318: langs.append(body, lang, known_length);
319: // next: letters themselves
1.174 paf 320: body.append_know_length(str, known_length);
1.1 paf 321:
1.174 paf 322: ASSERT_STRING_INVARIANT(*this);
1.1 paf 323: return *this;
324: }
1.174 paf 325: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
326: if(!str)
327: return *this;
328: size_t known_length=helper_length?helper_length:strlen(str);
329: if(!known_length)
330: return *this;
1.1 paf 331:
1.174 paf 332: return append_know_length(str, known_length, lang);
1.5 paf 333: }
1.226 misha 334: String::String(int value, char *format) : langs(L_CLEAN){
335: char buf[MAX_NUMBER];
336: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
337: }
1.174 paf 338: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
339: size_t known_length=helper_length?helper_length:strlen(str);
340: if(!known_length)
341: return *this;
1.5 paf 342:
1.176 paf 343: // first: langs
344: langs.append(body, lang, known_length);
345: // next: letters themselves
1.174 paf 346: body.append_strdup_know_length(str, known_length);
1.33 paf 347:
1.174 paf 348: ASSERT_STRING_INVARIANT(*this);
349: return *this;
1.5 paf 350: }
1.46 paf 351:
1.221 misha 352: int CORD_batched_len(const char* s, size_t* len){
353: (*len) += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
354: return 0;
355: }
356:
357: // can be called only for IS_FUNCTION(CORD) which are used only in Lang
358: int CORD_batched_len(const char, size_t *len){
359: (*len)++;
1.220 misha 360: return 0;
361: }
362:
1.210 misha 363: size_t String::length(Charset& charset) const {
364: if(charset.isUTF8()){
1.220 misha 365: size_t len=0;
1.224 misha 366: body.for_each<size_t *>(CORD_batched_len, CORD_batched_len, &len);
1.220 misha 367: return len;
1.210 misha 368: } else
369: return body.length();
370: }
371:
1.174 paf 372: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
373: String& String::mid(size_t substr_begin, size_t substr_end) const {
374: String& result=*new String;
375:
376: size_t self_length=length();
377: substr_begin=min(substr_begin, self_length);
378: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 379: size_t substr_length=substr_end-substr_begin;
380: if(!substr_length)
1.107 parser 381: return result;
1.53 paf 382:
1.176 paf 383: // first: their langs
384: result.langs.append(result.body, langs, substr_begin, substr_length);
385: // next: letters themselves
386: result.body=body.mid(substr_begin, substr_length);
1.174 paf 387:
388: ASSERT_STRING_INVARIANT(result);
1.53 paf 389: return result;
1.54 paf 390: }
391:
1.211 misha 392: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
393: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 394: String& result=*new String;
395:
1.211 misha 396: size_t self_length=(helper_length)?helper_length:length(charset);
397:
398: if(!self_length)
399: return result;
400:
401: from=min(min(to, from), self_length);
1.210 misha 402: to=min(max(to, from), self_length);
1.211 misha 403:
1.210 misha 404: size_t substr_length=to-from;
1.211 misha 405:
1.210 misha 406: if(!substr_length)
407: return result;
408:
409: if(charset.isUTF8()){
1.230 ! misha 410: const XMLByte* src_begin=(const XMLByte*)cstrm();
! 411: const XMLByte* src_end=src_begin+body.length();
1.210 misha 412:
1.212 misha 413: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 ! misha 414: from=getUTF8BytePos(src_begin, src_end, from);
! 415: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 416: if(!substr_length)
417: return result;
418: }
419:
420: // first: their langs
421: result.langs.append(result.body, langs, from, substr_length);
422: // next: letters themselves
423: result.body=body.mid(from, substr_length);
424:
425: ASSERT_STRING_INVARIANT(result);
426: return result;
427: }
428:
1.176 paf 429: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 430: size_t substr_length=substr.length();
431: while(true) {
432: size_t substr_begin=body.pos(substr, this_offset);
433:
434: if(substr_begin==CORD_NOT_FOUND)
435: return STRING_NOT_FOUND;
1.174 paf 436:
1.183 paf 437: if(langs.check_lang(lang, substr_begin, substr_length))
438: return substr_begin;
439:
440: this_offset=substr_begin+substr_length;
441: }
1.58 paf 442: }
443:
1.174 paf 444: size_t String::pos(const String& substr,
445: size_t this_offset, Language lang) const {
446: return pos(substr.body, this_offset, lang);
1.60 paf 447: }
448:
1.210 misha 449: size_t String::pos(Charset& charset, const String& substr,
450: size_t this_offset, Language lang) const {
451:
452: if(charset.isUTF8()){
453: const XMLByte* srcPtr=(const XMLByte*)cstrm();
1.212 misha 454: const XMLByte* srcEnd=srcPtr+body.length();
455:
456: // convert 'this_offset' from 'characters' to 'bytes'
457: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
458:
1.229 misha 459: size_t result=pos(substr.body, this_offset, lang);
460: return (result==CORD_NOT_FOUND)
461: ? STRING_NOT_FOUND
462: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 463: } else {
1.229 misha 464: size_t result=pos(substr.body, this_offset, lang);
465: return (result==CORD_NOT_FOUND)
466: ? STRING_NOT_FOUND
467: : result;
1.210 misha 468: }
469: }
470:
1.174 paf 471: void String::split(ArrayString& result,
472: size_t& pos_after,
473: const char* delim,
474: Language lang, int limit) const {
475: size_t self_length=length();
476: if(size_t delim_length=strlen(delim)) {
1.186 paf 477: size_t pos_before;
1.60 paf 478: // while we have 'delim'...
1.174 paf 479: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 480: result+=&mid(pos_after, pos_before);
1.174 paf 481: pos_after=pos_before+delim_length;
1.60 paf 482: }
483: // last piece
1.174 paf 484: if(pos_after<self_length && limit) {
485: result+=&mid(pos_after, self_length);
486: pos_after=self_length;
1.60 paf 487: }
488: } else { // empty delim
489: result+=this;
1.174 paf 490: pos_after+=self_length;
1.60 paf 491: }
492: }
493:
1.174 paf 494: void String::split(ArrayString& result,
495: size_t& pos_after,
496: const String& delim, Language lang,
497: int limit) const {
1.140 paf 498: if(!delim.is_empty()) {
1.186 paf 499: size_t pos_before;
1.60 paf 500: // while we have 'delim'...
1.174 paf 501: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 502: result+=&mid(pos_after, pos_before);
1.174 paf 503: pos_after=pos_before+delim.length();
1.60 paf 504: }
505: // last piece
1.174 paf 506: if(pos_after<length() && limit) {
507: result+=&mid(pos_after, length());
508: pos_after=length();
1.60 paf 509: }
510: } else { // empty delim
511: result+=this;
1.174 paf 512: pos_after+=length();
1.60 paf 513: }
1.61 paf 514: }
515:
1.222 misha 516: Table* String::match(VRegex* vregex,
517: Row_action row_action, void *info,
518: int& matches_count) const {
1.209 misha 519:
1.222 misha 520: // vregex->info(); // I have no idea what does it for?
1.63 paf 521:
1.222 misha 522: bool need_pre_post_match=vregex->is_pre_post_match_needed();
523: bool global=vregex->is_global_search();
1.63 paf 524:
1.174 paf 525: const char* subject=cstr();
526: size_t subject_length=strlen(subject);
1.222 misha 527: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
528: int ovector[ovector_size];
1.155 paf 529:
1.173 paf 530: Table::Action_options table_options;
1.174 paf 531: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 532:
1.154 paf 533: int prestart=0;
534: int poststart=0;
1.174 paf 535: int postfinish=length();
1.63 paf 536: while(true) {
1.222 misha 537: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 538:
1.222 misha 539: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
540: break;
1.63 paf 541:
1.154 paf 542: int prefinish=ovector[0];
543: poststart=ovector[1];
1.174 paf 544: ArrayString* row=new ArrayString;
545: if(need_pre_post_match) {
546: *row+=&mid(0, prefinish); // .prematch column value
547: *row+=&mid(prefinish, poststart); // .match
548: *row+=&mid(poststart, postfinish); // .postmatch
549: } else {
1.185 paf 550: *row+=&Empty; // .prematch column value
551: *row+=&Empty; // .match
552: *row+=&Empty; // .postmatch
1.174 paf 553: }
1.63 paf 554:
1.222 misha 555: for(int i=1; i<exec_result; i++) {
1.69 paf 556: // -1:-1 case handled peacefully by mid() itself
1.174 paf 557: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 558: }
559:
1.209 misha 560: matches_count++;
1.174 paf 561: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 562:
1.222 misha 563: if(!global || prestart==poststart) // last step
564: break;
565:
1.154 paf 566: prestart=poststart;
1.222 misha 567: }
1.63 paf 568:
1.222 misha 569: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
570: return vregex->is_just_count() ? 0 : &table;
1.82 parser 571: }
572:
1.174 paf 573: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
574: String& result=*new String();
575: if(is_empty())
576: return result;
577:
578: char* new_cstr=cstrm();
1.216 misha 579:
1.181 paf 580: if(source_charset.isUTF8()) {
1.216 misha 581: size_t new_cstr_len=length();
1.181 paf 582: switch(kind) {
583: case CC_UPPER:
1.192 paf 584: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 585: break;
586: case CC_LOWER:
1.192 paf 587: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 588: break;
589: default:
590: assert(!"unknown change case kind");
591: break; // never
592: }
593:
594: } else {
595: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 596:
1.181 paf 597: const unsigned char *a;
598: const unsigned char *b;
599: switch(kind) {
600: case CC_UPPER:
601: a=tables+lcc_offset;
602: b=tables+fcc_offset;
603: break;
604: case CC_LOWER:
605: a=tables+lcc_offset;
606: b=0;
607: break;
608: default:
609: assert(!"unknown change case kind");
610: a=b=0; // calm, compiler
611: break; // never
612: }
613:
1.192 paf 614: char *dest=new_cstr;
1.181 paf 615: unsigned char index;
1.190 paf 616: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 617: unsigned char c=a[index];
618: if(b)
619: c=b[c];
620:
621: *dest++=(char)c;
622: }
1.174 paf 623: }
1.176 paf 624: result.langs=langs;
1.174 paf 625: result.body=new_cstr;
1.89 parser 626:
1.101 parser 627: return result;
628: }
629:
1.213 misha 630: const String& String::escape(Charset& source_charset) const {
631: if(is_empty())
632: return *this;
633:
634: return Charset::escape(*this, source_charset);
635: }
636:
1.174 paf 637: const String& String::replace(const Dictionary& dict) const {
638: String& result=*new String();
639: const char* old_cstr=cstr();
640: const char* prematch_begin=old_cstr;
641:
642: const char* current=old_cstr;
643: while(*current) {
1.184 paf 644: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
1.174 paf 645: // prematch
646: if(size_t prematch_length=current-prematch_begin) {
1.179 paf 647: result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length);
1.174 paf 648: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 649: }
650:
1.174 paf 651: // match
652: // skip 'a' in 'current'; move prematch_begin
1.184 paf 653: current+=subst.from_length; prematch_begin=current;
1.174 paf 654:
1.184 paf 655: if(const String* b=subst.to) // are there any b?
1.174 paf 656: result<<*b;
657: } else // simply advance
658: current++;
659: }
1.156 paf 660:
1.174 paf 661: // postmatch
662: if(size_t postmatch_length=current-prematch_begin) {
1.179 paf 663: result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length);
1.174 paf 664: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
665: }
1.156 paf 666:
1.174 paf 667: ASSERT_STRING_INVARIANT(result);
1.82 parser 668: return result;
1.61 paf 669: }
1.113 parser 670:
1.180 paf 671: static int serialize_body_char(char c, char** cur) {
672: *((*cur)++)=c;
673: return 0; // 0=continue
674: };
1.174 paf 675: static int serialize_body_piece(const char* s, char** cur) {
676: size_t length=strlen(s);
677: memcpy(*cur, s, length); *cur+=length;
1.178 paf 678: return 0; // 0=continue
1.174 paf 679: };
1.178 paf 680: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
681: // lang
1.191 paf 682: **cur=alang; (*cur)++;
683: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 684: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
685:
686: return 0; // 0=continue
687: }
1.174 paf 688: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 689: size_t fragments_count=langs.count();
1.202 paf 690: size_t body_length=body.length();
1.174 paf 691: size_t buf_length=
1.178 paf 692: prolog_length //1
693: +sizeof(size_t) //2
1.202 paf 694: +body_length //3
695: +1 // 4 for zero terminator used in deserialize
696: +sizeof(size_t) //5
697: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
698:
1.174 paf 699: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
700:
701: // 1: prolog
702: char *cur=result.str+prolog_length;
1.202 paf 703: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
704: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
705: // 3: letters
706: body.for_each(serialize_body_char, serialize_body_piece, &cur);
707: // 4: zero terminator
708: *cur++=0;
709: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 710: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 711: // 6: lang info
1.178 paf 712: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 713:
1.174 paf 714: return result;
1.113 parser 715: }
1.202 paf 716: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
717: size_t in_buf=buf_size;
718: if(in_buf<=prolog_size)
1.148 paf 719: return false;
1.202 paf 720: in_buf-=prolog_size;
1.135 paf 721:
1.174 paf 722: // 1: prolog
1.202 paf 723: const char* cur=(const char* )buf+prolog_size;
1.113 parser 724:
1.207 paf 725: // 2: chars.count
1.202 paf 726: size_t body_length;
727: if(in_buf<sizeof(body_length)) // body.length don't fit?
728: return false;
729: // [WARNING: not cast, addresses must be %4=0 on sparc]
730: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
731: in_buf-=sizeof(body_length);
732:
733: if(in_buf<body_length+1) // letters+terminator don't fit?
734: return false;
735: // 4: zero terminator
736: if(cur[body_length] != 0) // in place?
737: return false;
738: // 3: letters
1.227 misha 739: body=String::Body(*cur?cur:0);
1.229 misha 740: #ifdef STRING_LENGTH_CACHING
741: body.set_length(body_length);
742: #endif
1.202 paf 743: cur+=body_length+1;
744: in_buf-=body_length+1;
745:
746: // 5: langs.count
1.191 paf 747: size_t fragments_count;
1.202 paf 748: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 749: return false;
1.191 paf 750: // [WARNING: not cast, addresses must be %4=0 on sparc]
751: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 752: in_buf-=sizeof(fragments_count);
1.174 paf 753:
754: if(fragments_count) {
1.202 paf 755: // 6: lang info
1.174 paf 756: size_t total_length=0;
757: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 758: char lang;
759: size_t fragment_length;
760: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 761: if(in_buf<piece_length) // lang+length
1.174 paf 762: return false;
763:
1.191 paf 764: // lang
765: lang=*cur++;
766: // length [WARNING: not cast, addresses must be %4=0 on sparc]
767: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
768:
1.206 paf 769: size_t combined_length=total_length+fragment_length;
770: if(combined_length>body_length)
771: return false; // file curruption
1.191 paf 772: // uchar needed to prevent propagating 0x80 bit to upper bytes
773: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 774: total_length=combined_length;
1.202 paf 775: in_buf-=piece_length;
1.174 paf 776: }
1.128 paf 777:
1.202 paf 778: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 779: return false;
1.174 paf 780: }
1.202 paf 781: if(in_buf!=0) // some strange extra bytes
782: return false;
1.113 parser 783:
1.174 paf 784: ASSERT_STRING_INVARIANT(*this);
1.148 paf 785: return true;
1.176 paf 786: }
787:
788: const char* String::Body::v() const {
789: return CORD_to_const_char_star(body);
790: }
1.201 paf 791: void String::Body::dump() const {
792: CORD_dump(body);
793: }
794:
1.176 paf 795: const char* String::Languages::v() const {
1.177 paf 796: if(opt.is_not_just_lang)
1.176 paf 797: return CORD_to_const_char_star(langs);
798: else
799: return (const char*)&langs;
800: }
1.201 paf 801: void String::Languages::dump() const {
802: if(opt.is_not_just_lang)
803: CORD_dump(langs);
804: else
805: puts((const char*)&langs);
806: }
1.176 paf 807: const char* String::v() const {
1.198 paf 808: const uint LIMIT_VIEW=20;
1.176 paf 809: char* buf=(char*)malloc(MAX_STRING);
810: const char*body_view=body.v();
811: const char*langs_view=langs.v();
812: snprintf(buf, MAX_STRING,
1.178 paf 813: "%d:%.*s%s} "
1.176 paf 814: "{%d:%s",
1.178 paf 815: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 816: strlen(body_view), body_view
817: );
818:
819: return buf;
1.113 parser 820: }
1.229 misha 821:
1.201 paf 822: void String::dump() const {
823: body.dump();
824: langs.dump();
825: }
1.229 misha 826:
827: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 828: if(is_empty())
1.195 paf 829: return *this;
830:
831: size_t substr_begin, substr_length;
1.229 misha 832: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 833: if(new_body==body) // we received unchanged pointer, do likewise
834: return *this;
835: // new_body differs from body, adjust langs along
836:
837: String& result=*new String;
838: if(!new_body) // body.trim produced empty result
839: return result;
840: // body.trim produced nonempty result
841:
842: // first: their langs
843: result.langs.append(result.body, langs, substr_begin, substr_length);
844: // next: letters themselves
845: result.body=new_body;
846:
847: ASSERT_STRING_INVARIANT(result);
848: return result;
1.198 paf 849: }
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