Annotation of parser3/src/main/pa_string.C, revision 1.248
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.240 moko 4: Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com)
1.138 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.164 paf 6: */
1.46 paf 7:
1.12 paf 8: #include "pa_string.h"
1.22 paf 9: #include "pa_exception.h"
1.61 paf 10: #include "pa_table.h"
1.101 parser 11: #include "pa_dictionary.h"
1.132 paf 12: #include "pa_charset.h"
1.222 misha 13: #include "pa_vregex.h"
1.215 misha 14:
1.248 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.247 2015/04/08 18:08:53 moko Exp $" IDENT_PA_STRING_H;
1.240 moko 16:
1.185 paf 17: const String String::Empty;
18:
1.242 moko 19: // pa_atoui is based on Manuel Novoa III _strto_l for uClibc
20:
21: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
22: unsigned int result = 0;
23: const char *pos = str;
24:
25: while (isspace(*pos)) /* skip leading whitespace */
26: ++pos;
27:
28: if (base == 16 && *pos == '0') { /* handle option prefix */
29: ++pos;
30: if (*pos == 'x' || *pos == 'X') {
31: ++pos;
32: }
33: }
34:
35: if (base == 0) { /* dynamic base */
36: base = 10; /* default is 10 */
37: if (*pos == '0') {
38: ++pos;
1.245 moko 39: if (*pos == 'x' || *pos == 'X'){
1.242 moko 40: ++pos;
41: base=16;
1.245 moko 42: }
1.242 moko 43: }
44: }
45:
46: if (base < 2 || base > 16) { /* illegal base */
47: throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
48: }
49:
50: unsigned int cutoff = UINT_MAX / base;
51: int cutoff_digit = UINT_MAX - cutoff * base;
52:
53: while(true) {
54: int digit;
55:
56: if ((*pos >= '0') && (*pos <= '9')) {
57: digit = (*pos - '0');
58: } else if (*pos >= 'a') {
59: digit = (*pos - 'a' + 10);
60: } else if (*pos >= 'A') {
61: digit = (*pos - 'A' + 10);
62: } else break;
63:
64: if (digit >= base) {
65: break;
66: }
67:
68: ++pos;
69:
70: /* adjust number, with overflow check */
71: if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
72: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
73: } else {
74: result = result * base + digit;
75: }
76: }
77:
78: while(char c=*pos++)
79: if(!isspace(c))
80: throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
81:
82: return result;
83: }
84:
1.193 paf 85: int pa_atoi(const char* str, const String* problem_source) {
86: if(!str)
87: return 0;
88:
1.242 moko 89: while(isspace(*str))
1.193 paf 90: str++;
1.242 moko 91:
1.193 paf 92: if(!*str)
93: return 0;
94:
1.200 paf 95: bool negative=false;
96: if(str[0]=='-') {
97: negative=true;
98: str++;
99: } else if(str[0]=='+') {
100: str++;
101: }
1.193 paf 102:
1.242 moko 103: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 104:
1.242 moko 105: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 106: return -(int)result;
1.242 moko 107:
108: if(result<=INT_MAX)
1.243 moko 109: return (int)result;
1.242 moko 110:
111: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 112: }
113:
114: double pa_atod(const char* str, const String* problem_source) {
115: if(!str)
116: return 0;
117:
1.242 moko 118: while(isspace(*str))
1.193 paf 119: str++;
1.242 moko 120:
1.193 paf 121: if(!*str)
122: return 0;
123:
1.200 paf 124: bool negative=false;
125: if(str[0]=='-') {
126: negative=true;
127: str++;
128: } else if(str[0]=='+') {
129: str++;
130: }
1.242 moko 131:
132: double result;
1.247 moko 133: if(str[0]=='0') {
134: if(str[1]=='x' || str[1]=='X') {
1.242 moko 135: // 0xABC
136: result=(double)pa_atoui(str, 0, problem_source);
137: return negative ? -result : result;
138: } else {
1.200 paf 139: // skip leading 0000, to disable octal interpretation
1.242 moko 140: do str++; while(*str=='0');
1.200 paf 141: }
1.247 moko 142: }
1.242 moko 143:
144: char *error_pos;
145: result=strtod(str, &error_pos);
1.193 paf 146:
147: while(char c=*error_pos++)
1.199 paf 148: if(!isspace((unsigned char)c))
1.242 moko 149: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 150:
1.242 moko 151: return negative ? -result : result;
1.193 paf 152: }
153:
1.176 paf 154: // cord lib extension
155:
156: #ifndef DOXYGEN
157: typedef struct {
158: ssize_t countdown;
1.204 paf 159: int target; /* Character we're looking for */
1.176 paf 160: } chr_data;
161: #endif
1.248 ! moko 162:
1.176 paf 163: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
164: {
165: register chr_data * d = (chr_data *)client_data;
166:
167: if (d -> countdown<=0) return(2);
168: d -> countdown -= size;
169: if (c > d -> target) return(1);
170: return(0);
171: }
1.248 ! moko 172:
1.176 paf 173: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
174: {
175: chr_data d;
176:
177: d.countdown = n;
178: d.target = c;
179: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
180: }
181:
1.187 paf 182: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 183: {
184: int* result=(int*)client_data;
185: (*result)++;
186: return(0); // 0=continue
187: }
1.248 ! moko 188:
1.178 paf 189: size_t CORD_block_count(CORD x)
190: {
191: size_t result=0;
192: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
193: return result;
194: }
195:
1.174 paf 196: // helpers
1.139 paf 197:
1.174 paf 198: /// String::match uses this as replace & global search table columns
1.139 paf 199:
1.174 paf 200: const int MAX_MATCH_GROUPS=100;
1.139 paf 201:
1.174 paf 202: class String_match_table_template_columns: public ArrayString {
203: public:
204: String_match_table_template_columns() {
205: *this+=new String("prematch");
206: *this+=new String("match");
207: *this+=new String("postmatch");
208: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 209: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 210: }
211: }
212: };
213:
214: Table string_match_table_template(new String_match_table_template_columns);
215:
1.176 paf 216: // String::Body methods
1.140 paf 217:
1.176 paf 218: String::Body String::Body::Format(int value) {
1.174 paf 219: char local[MAX_NUMBER];
220: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 221: return String::Body(pa_strdup(local, length));
1.120 paf 222: }
223:
1.195 paf 224: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 225: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 226: size_t our_length=length();
227: if(!our_length)
228: return *this;
1.229 misha 229:
230: // check if any UTF-8 in chars
231: bool fast=true;
232: if(chars && source_charset && source_charset->isUTF8()){
233: const char* pos=chars;
234: while(unsigned char c=*pos++)
235: if(c>127){
236: fast=false;
237: break;
238: }
239: }
240:
241: size_t start=0;
242: size_t end=our_length;
1.195 paf 243: if(!chars)
244: chars=" \t\n"; // white space
245:
1.229 misha 246: if(fast){
247: // from left...
248: if(kind!=TRIM_END) {
249: CORD_pos pos; set_pos(pos, 0);
250: while(true) {
251: char c=CORD_pos_fetch(pos);
252: if(strchr(chars, c)) {
253: if(++start==our_length)
254: return 0; // all chars are empty, just return empty string
255: } else
256: break;
257:
258: CORD_next(pos);
259: }
260: }
261:
262: // from right..
263: if(kind!=TRIM_START) {
264: CORD_pos pos; set_pos(pos, end-1);
265: while(true) {
266: char c=CORD_pos_fetch(pos);
267: if(strchr(chars, c)) {
268: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
269: return 0; // all chars are empty, just return empty string
270: } else
271: break;
272:
273: CORD_prev(pos);
274: }
275: }
276: } else {
1.234 misha 277: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 278: const XMLByte* src_end=src_begin+our_length;
279:
280: // from left...
281: if(kind!=TRIM_END) {
282: while(src_begin<src_end){
283: uint char_length=1;
284: const XMLByte* ptr=src_begin;
1.231 misha 285: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
286: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 287: char_length++;
288:
289: bool found=false;
290: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
291: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
292: found=true;
293: break;
294: }
295:
296: if(found){
297: start+=char_length;
298: if(start==our_length)
299: return 0; // all chars are empty, just return empty string
300: } else
301: break;
302: }
303: }
1.196 paf 304:
1.229 misha 305: // from right..
306: if(kind!=TRIM_START) {
307: while(src_begin<src_end){
308: uint char_length=1;
1.231 misha 309: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
310: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 311: char_length++;
312:
313: bool found=false;
314: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
315: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
316: found=true;
317: break;
318: }
319:
320: if(found){
321: end-=char_length;
322: if(end==0)
323: return 0; // all chars are empty, just return empty string
324: } else
325: break;
326: }
1.196 paf 327: }
328: }
1.195 paf 329:
330: if(start==0 && end==our_length) // nobody moved a thing
331: return *this;
332:
333: if(out_start)
334: *out_start=start;
335: size_t new_length=end-start;
336: if(out_length)
337: *out_length=new_length;
338:
339: return mid(start, new_length);
340: }
341:
1.174 paf 342: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
343: uint& result=*static_cast<uint*>(client_data);
344: generic_hash_code(result, c);
345: return 0;
346: }
1.248 ! moko 347:
1.174 paf 348: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
349: uint& result=*static_cast<uint*>(client_data);
350: generic_hash_code(result, s);
351: return 0;
1.248 ! moko 352: }
! 353:
1.225 misha 354: uint String::Body::get_hash_code() const {
355: #ifdef HASH_CODE_CACHING
356: if(hash_code)
357: return hash_code;
358: #else
359: uint hash_code=0;
360: #endif
1.220 misha 361: if (body && CORD_IS_STRING(body)){
1.225 misha 362: generic_hash_code(hash_code, body);
1.220 misha 363: } else {
364: CORD_iter5(body, 0,
365: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 366: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 367: }
1.225 misha 368: return hash_code;
1.94 parser 369: }
370:
1.241 misha 371: struct CORD_pos_info {
372: const char* chars;
373: size_t left;
374: size_t pos;
375: };
376:
377: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
378: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
379: if(info->pos < info->left){
380: info->pos=STRING_NOT_FOUND;
381: return 1;
382: }
383: if(strchr(info->chars, c))
384: return 1;
385: --(info->pos);
386: return 0;
387: }
388:
389: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
390: if(is_empty() || !chars || !strlen(chars))
391: return STRING_NOT_FOUND;
392: CORD_pos_info info={chars, left, right};
393: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
394: return info.pos;
395: else
396: return STRING_NOT_FOUND;
397: }
398:
399:
400: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
401: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
402: if(info->pos < info->left) {
403: info->pos=STRING_NOT_FOUND;
404: return 1;
405: }
406: if(!strchr(info->chars, c))
407: return 1;
408: --(info->pos);
409: return 0;
410: }
411:
412: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
413: if(is_empty() || !chars || !strlen(chars))
414: return STRING_NOT_FOUND;
415: CORD_pos_info info={chars, left, right};
416: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
417: return info.pos;
418: else
419: return STRING_NOT_FOUND;
420: }
421:
1.174 paf 422: // String methods
423:
424: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
425: if(!known_length)
1.9 paf 426: return *this;
1.122 paf 427:
1.176 paf 428: // first: langs
429: langs.append(body, lang, known_length);
430: // next: letters themselves
1.174 paf 431: body.append_know_length(str, known_length);
1.1 paf 432:
1.174 paf 433: ASSERT_STRING_INVARIANT(*this);
1.1 paf 434: return *this;
435: }
1.248 ! moko 436:
1.174 paf 437: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
438: if(!str)
439: return *this;
440: size_t known_length=helper_length?helper_length:strlen(str);
441: if(!known_length)
442: return *this;
1.1 paf 443:
1.174 paf 444: return append_know_length(str, known_length, lang);
1.5 paf 445: }
1.248 ! moko 446:
1.244 moko 447: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 448: char buf[MAX_NUMBER];
449: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
450: }
1.248 ! moko 451:
1.174 paf 452: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
453: size_t known_length=helper_length?helper_length:strlen(str);
454: if(!known_length)
455: return *this;
1.5 paf 456:
1.176 paf 457: // first: langs
458: langs.append(body, lang, known_length);
459: // next: letters themselves
1.174 paf 460: body.append_strdup_know_length(str, known_length);
1.33 paf 461:
1.174 paf 462: ASSERT_STRING_INVARIANT(*this);
463: return *this;
1.5 paf 464: }
1.46 paf 465:
1.234 misha 466: struct CORD_length_info {
467: size_t len;
468: size_t skip;
469: };
470:
471: int CORD_batched_len(const char* s, CORD_length_info* info){
472: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 473: }
474:
1.234 misha 475: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
476: int CORD_batched_len(const char c, CORD_length_info* info){
477: if (info->skip==0){
478: info->len++;
479: info->skip = lengthUTF8Char(c)-1;
480: } else {
481: info->skip--;
482: }
1.220 misha 483: return 0;
484: }
485:
1.210 misha 486: size_t String::length(Charset& charset) const {
487: if(charset.isUTF8()){
1.234 misha 488: CORD_length_info info = {0, 0};
489: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
490: return info.len;
1.210 misha 491: } else
492: return body.length();
493: }
494:
1.174 paf 495: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
496: String& String::mid(size_t substr_begin, size_t substr_end) const {
497: String& result=*new String;
498:
499: size_t self_length=length();
500: substr_begin=min(substr_begin, self_length);
501: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 502: size_t substr_length=substr_end-substr_begin;
503: if(!substr_length)
1.107 parser 504: return result;
1.53 paf 505:
1.176 paf 506: // first: their langs
507: result.langs.append(result.body, langs, substr_begin, substr_length);
508: // next: letters themselves
509: result.body=body.mid(substr_begin, substr_length);
1.174 paf 510:
511: ASSERT_STRING_INVARIANT(result);
1.53 paf 512: return result;
1.54 paf 513: }
514:
1.211 misha 515: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
516: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 517: String& result=*new String;
518:
1.211 misha 519: size_t self_length=(helper_length)?helper_length:length(charset);
520:
521: if(!self_length)
522: return result;
523:
524: from=min(min(to, from), self_length);
1.210 misha 525: to=min(max(to, from), self_length);
1.211 misha 526:
1.210 misha 527: size_t substr_length=to-from;
1.211 misha 528:
1.210 misha 529: if(!substr_length)
530: return result;
531:
532: if(charset.isUTF8()){
1.234 misha 533: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 534: const XMLByte* src_end=src_begin+body.length();
1.210 misha 535:
1.212 misha 536: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 537: from=getUTF8BytePos(src_begin, src_end, from);
538: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 539: if(!substr_length)
540: return result;
541: }
542:
543: // first: their langs
544: result.langs.append(result.body, langs, from, substr_length);
545: // next: letters themselves
546: result.body=body.mid(from, substr_length);
547:
548: ASSERT_STRING_INVARIANT(result);
549: return result;
550: }
551:
1.176 paf 552: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 553: size_t substr_length=substr.length();
554: while(true) {
555: size_t substr_begin=body.pos(substr, this_offset);
556:
557: if(substr_begin==CORD_NOT_FOUND)
558: return STRING_NOT_FOUND;
1.174 paf 559:
1.183 paf 560: if(langs.check_lang(lang, substr_begin, substr_length))
561: return substr_begin;
562:
563: this_offset=substr_begin+substr_length;
564: }
1.58 paf 565: }
566:
1.174 paf 567: size_t String::pos(const String& substr,
568: size_t this_offset, Language lang) const {
569: return pos(substr.body, this_offset, lang);
1.60 paf 570: }
571:
1.210 misha 572: size_t String::pos(Charset& charset, const String& substr,
573: size_t this_offset, Language lang) const {
574:
575: if(charset.isUTF8()){
1.234 misha 576: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 577: const XMLByte* srcEnd=srcPtr+body.length();
578:
579: // convert 'this_offset' from 'characters' to 'bytes'
580: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
581:
1.229 misha 582: size_t result=pos(substr.body, this_offset, lang);
583: return (result==CORD_NOT_FOUND)
584: ? STRING_NOT_FOUND
585: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 586: } else {
1.229 misha 587: size_t result=pos(substr.body, this_offset, lang);
588: return (result==CORD_NOT_FOUND)
589: ? STRING_NOT_FOUND
590: : result;
1.210 misha 591: }
592: }
593:
1.174 paf 594: void String::split(ArrayString& result,
595: size_t& pos_after,
596: const char* delim,
597: Language lang, int limit) const {
1.237 misha 598: if(is_empty())
599: return;
1.174 paf 600: size_t self_length=length();
1.237 misha 601: if(size_t delim_length=strlen(delim)) {
1.186 paf 602: size_t pos_before;
1.60 paf 603: // while we have 'delim'...
1.174 paf 604: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 605: result+=&mid(pos_after, pos_before);
1.174 paf 606: pos_after=pos_before+delim_length;
1.60 paf 607: }
608: // last piece
1.174 paf 609: if(pos_after<self_length && limit) {
610: result+=&mid(pos_after, self_length);
611: pos_after=self_length;
1.60 paf 612: }
613: } else { // empty delim
614: result+=this;
1.174 paf 615: pos_after+=self_length;
1.60 paf 616: }
617: }
618:
1.174 paf 619: void String::split(ArrayString& result,
620: size_t& pos_after,
621: const String& delim, Language lang,
622: int limit) const {
1.237 misha 623: if(is_empty())
624: return;
625: if(!delim.is_empty()) {
1.186 paf 626: size_t pos_before;
1.60 paf 627: // while we have 'delim'...
1.174 paf 628: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 629: result+=&mid(pos_after, pos_before);
1.174 paf 630: pos_after=pos_before+delim.length();
1.60 paf 631: }
632: // last piece
1.174 paf 633: if(pos_after<length() && limit) {
634: result+=&mid(pos_after, length());
635: pos_after=length();
1.60 paf 636: }
637: } else { // empty delim
638: result+=this;
1.174 paf 639: pos_after+=length();
1.60 paf 640: }
1.61 paf 641: }
642:
1.222 misha 643: Table* String::match(VRegex* vregex,
644: Row_action row_action, void *info,
645: int& matches_count) const {
1.209 misha 646:
1.222 misha 647: // vregex->info(); // I have no idea what does it for?
1.63 paf 648:
1.222 misha 649: bool need_pre_post_match=vregex->is_pre_post_match_needed();
650: bool global=vregex->is_global_search();
1.63 paf 651:
1.174 paf 652: const char* subject=cstr();
1.234 misha 653: size_t subject_length=length();
1.222 misha 654: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
655: int ovector[ovector_size];
1.155 paf 656:
1.173 paf 657: Table::Action_options table_options;
1.174 paf 658: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 659:
1.154 paf 660: int prestart=0;
661: int poststart=0;
1.174 paf 662: int postfinish=length();
1.63 paf 663: while(true) {
1.222 misha 664: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 665:
1.222 misha 666: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
667: break;
1.63 paf 668:
1.154 paf 669: int prefinish=ovector[0];
670: poststart=ovector[1];
1.236 moko 671:
672: if (prestart==poststart && subject[poststart]=='\n'){
673: prestart++;
674: continue;
675: }
676:
1.232 misha 677: ArrayString* row=new ArrayString(3);
1.174 paf 678: if(need_pre_post_match) {
679: *row+=&mid(0, prefinish); // .prematch column value
680: *row+=&mid(prefinish, poststart); // .match
681: *row+=&mid(poststart, postfinish); // .postmatch
682: } else {
1.185 paf 683: *row+=&Empty; // .prematch column value
684: *row+=&Empty; // .match
685: *row+=&Empty; // .postmatch
1.174 paf 686: }
1.63 paf 687:
1.222 misha 688: for(int i=1; i<exec_result; i++) {
1.69 paf 689: // -1:-1 case handled peacefully by mid() itself
1.232 misha 690: *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 691: }
692:
1.209 misha 693: matches_count++;
1.174 paf 694: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 695:
1.222 misha 696: if(!global || prestart==poststart) // last step
697: break;
698:
1.154 paf 699: prestart=poststart;
1.222 misha 700: }
1.63 paf 701:
1.222 misha 702: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
703: return vregex->is_just_count() ? 0 : &table;
1.82 parser 704: }
705:
1.174 paf 706: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
707: String& result=*new String();
708: if(is_empty())
709: return result;
710:
711: char* new_cstr=cstrm();
1.216 misha 712:
1.181 paf 713: if(source_charset.isUTF8()) {
1.216 misha 714: size_t new_cstr_len=length();
1.181 paf 715: switch(kind) {
716: case CC_UPPER:
1.192 paf 717: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 718: break;
719: case CC_LOWER:
1.192 paf 720: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 721: break;
722: default:
723: assert(!"unknown change case kind");
724: break; // never
725: }
726:
727: } else {
728: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 729:
1.181 paf 730: const unsigned char *a;
731: const unsigned char *b;
732: switch(kind) {
733: case CC_UPPER:
734: a=tables+lcc_offset;
735: b=tables+fcc_offset;
736: break;
737: case CC_LOWER:
738: a=tables+lcc_offset;
739: b=0;
740: break;
741: default:
742: assert(!"unknown change case kind");
743: a=b=0; // calm, compiler
744: break; // never
745: }
746:
1.192 paf 747: char *dest=new_cstr;
1.181 paf 748: unsigned char index;
1.190 paf 749: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 750: unsigned char c=a[index];
751: if(b)
752: c=b[c];
753:
754: *dest++=(char)c;
755: }
1.174 paf 756: }
1.176 paf 757: result.langs=langs;
1.174 paf 758: result.body=new_cstr;
1.89 parser 759:
1.101 parser 760: return result;
761: }
762:
1.213 misha 763: const String& String::escape(Charset& source_charset) const {
764: if(is_empty())
765: return *this;
766:
767: return Charset::escape(*this, source_charset);
768: }
769:
1.238 misha 770: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
771: result.langs.append(result.body, langs, langs_offset, length); \
772: result.body.append_strdup_know_length(from_cstr, length);
773:
1.174 paf 774: const String& String::replace(const Dictionary& dict) const {
1.238 misha 775: if(!dict.count() || is_empty())
776: return *this;
777:
1.174 paf 778: String& result=*new String();
779: const char* old_cstr=cstr();
780: const char* prematch_begin=old_cstr;
781:
1.238 misha 782: if(dict.count()==1) {
783: // optimized simple case
784:
785: Dictionary::Subst subst=dict.get(0);
1.239 moko 786: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 787: // prematch
1.238 misha 788: if(size_t prematch_length=p-prematch_begin) {
789: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 790: }
791:
1.174 paf 792: // match
1.238 misha 793: prematch_begin=p+subst.from_length;
1.174 paf 794:
1.184 paf 795: if(const String* b=subst.to) // are there any b?
1.174 paf 796: result<<*b;
1.238 misha 797: }
798:
799: } else {
800:
801: const char* current=old_cstr;
802: while(*current) {
803: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
804: // prematch
805: if(size_t prematch_length=current-prematch_begin) {
806: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
807: }
808:
809: // match
810: // skip 'a' in 'current'; move prematch_begin
811: current+=subst.from_length; prematch_begin=current;
812:
813: if(const String* b=subst.to) // are there any b?
814: result<<*b;
815: } else // simply advance
816: current++;
817: }
818:
1.174 paf 819: }
1.156 paf 820:
1.238 misha 821: if(prematch_begin==old_cstr) // not modified
822: return *this;
823:
1.174 paf 824: // postmatch
1.238 misha 825: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
826: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 827: }
1.156 paf 828:
1.174 paf 829: ASSERT_STRING_INVARIANT(result);
1.82 parser 830: return result;
1.61 paf 831: }
1.113 parser 832:
1.180 paf 833: static int serialize_body_char(char c, char** cur) {
834: *((*cur)++)=c;
835: return 0; // 0=continue
1.248 ! moko 836: }
! 837:
1.174 paf 838: static int serialize_body_piece(const char* s, char** cur) {
839: size_t length=strlen(s);
840: memcpy(*cur, s, length); *cur+=length;
1.178 paf 841: return 0; // 0=continue
1.248 ! moko 842: }
! 843:
1.178 paf 844: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
845: // lang
1.191 paf 846: **cur=alang; (*cur)++;
847: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 848: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
849:
850: return 0; // 0=continue
851: }
1.248 ! moko 852:
1.174 paf 853: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 854: size_t fragments_count=langs.count();
1.202 paf 855: size_t body_length=body.length();
1.174 paf 856: size_t buf_length=
1.178 paf 857: prolog_length //1
858: +sizeof(size_t) //2
1.202 paf 859: +body_length //3
860: +1 // 4 for zero terminator used in deserialize
861: +sizeof(size_t) //5
862: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
863:
1.174 paf 864: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
865:
866: // 1: prolog
867: char *cur=result.str+prolog_length;
1.202 paf 868: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
869: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
870: // 3: letters
871: body.for_each(serialize_body_char, serialize_body_piece, &cur);
872: // 4: zero terminator
873: *cur++=0;
874: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 875: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 876: // 6: lang info
1.178 paf 877: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 878:
1.174 paf 879: return result;
1.113 parser 880: }
1.248 ! moko 881:
1.202 paf 882: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
883: size_t in_buf=buf_size;
884: if(in_buf<=prolog_size)
1.148 paf 885: return false;
1.202 paf 886: in_buf-=prolog_size;
1.135 paf 887:
1.174 paf 888: // 1: prolog
1.202 paf 889: const char* cur=(const char* )buf+prolog_size;
1.113 parser 890:
1.207 paf 891: // 2: chars.count
1.202 paf 892: size_t body_length;
893: if(in_buf<sizeof(body_length)) // body.length don't fit?
894: return false;
895: // [WARNING: not cast, addresses must be %4=0 on sparc]
896: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
897: in_buf-=sizeof(body_length);
898:
899: if(in_buf<body_length+1) // letters+terminator don't fit?
900: return false;
901: // 4: zero terminator
902: if(cur[body_length] != 0) // in place?
903: return false;
904: // 3: letters
1.227 misha 905: body=String::Body(*cur?cur:0);
1.229 misha 906: #ifdef STRING_LENGTH_CACHING
907: body.set_length(body_length);
908: #endif
1.202 paf 909: cur+=body_length+1;
910: in_buf-=body_length+1;
911:
912: // 5: langs.count
1.191 paf 913: size_t fragments_count;
1.202 paf 914: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 915: return false;
1.191 paf 916: // [WARNING: not cast, addresses must be %4=0 on sparc]
917: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 918: in_buf-=sizeof(fragments_count);
1.174 paf 919:
920: if(fragments_count) {
1.202 paf 921: // 6: lang info
1.174 paf 922: size_t total_length=0;
923: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 924: char lang;
925: size_t fragment_length;
926: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 927: if(in_buf<piece_length) // lang+length
1.174 paf 928: return false;
929:
1.191 paf 930: // lang
931: lang=*cur++;
932: // length [WARNING: not cast, addresses must be %4=0 on sparc]
933: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
934:
1.206 paf 935: size_t combined_length=total_length+fragment_length;
936: if(combined_length>body_length)
937: return false; // file curruption
1.191 paf 938: // uchar needed to prevent propagating 0x80 bit to upper bytes
939: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 940: total_length=combined_length;
1.202 paf 941: in_buf-=piece_length;
1.174 paf 942: }
1.128 paf 943:
1.202 paf 944: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 945: return false;
1.174 paf 946: }
1.202 paf 947: if(in_buf!=0) // some strange extra bytes
948: return false;
1.113 parser 949:
1.174 paf 950: ASSERT_STRING_INVARIANT(*this);
1.148 paf 951: return true;
1.176 paf 952: }
953:
954: const char* String::Body::v() const {
1.233 misha 955: return CORD_to_const_char_star(body, length());
1.176 paf 956: }
1.248 ! moko 957:
1.201 paf 958: void String::Body::dump() const {
959: CORD_dump(body);
960: }
961:
1.176 paf 962: const char* String::Languages::v() const {
1.177 paf 963: if(opt.is_not_just_lang)
1.233 misha 964: return CORD_to_const_char_star(langs, 0);
1.176 paf 965: else
966: return (const char*)&langs;
967: }
1.248 ! moko 968:
1.201 paf 969: void String::Languages::dump() const {
970: if(opt.is_not_just_lang)
971: CORD_dump(langs);
972: else
973: puts((const char*)&langs);
974: }
1.248 ! moko 975:
1.176 paf 976: const char* String::v() const {
1.198 paf 977: const uint LIMIT_VIEW=20;
1.246 moko 978: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 979: const char*body_view=body.v();
980: const char*langs_view=langs.v();
981: snprintf(buf, MAX_STRING,
1.178 paf 982: "%d:%.*s%s} "
1.176 paf 983: "{%d:%s",
1.178 paf 984: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 985: strlen(body_view), body_view
986: );
987:
988: return buf;
1.113 parser 989: }
1.229 misha 990:
1.201 paf 991: void String::dump() const {
992: body.dump();
993: langs.dump();
994: }
1.229 misha 995:
996: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 997: if(is_empty())
1.195 paf 998: return *this;
999:
1000: size_t substr_begin, substr_length;
1.229 misha 1001: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1002: if(new_body==body) // we received unchanged pointer, do likewise
1003: return *this;
1004: // new_body differs from body, adjust langs along
1005:
1006: String& result=*new String;
1007: if(!new_body) // body.trim produced empty result
1008: return result;
1009: // body.trim produced nonempty result
1010:
1011: // first: their langs
1012: result.langs.append(result.body, langs, substr_begin, substr_length);
1013: // next: letters themselves
1014: result.body=new_body;
1015:
1016: ASSERT_STRING_INVARIANT(result);
1017: return result;
1.198 paf 1018: }
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