Annotation of parser3/src/main/pa_string.C, revision 1.251
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:
1.251 ! moko 15: volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.250 2015/10/08 18:29:16 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:
1.249 moko 21: template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
22: T result = 0;
1.242 moko 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:
1.249 moko 50: T cutoff = max / base;
51: int cutoff_digit = (int)(max - cutoff * base);
1.242 moko 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:
1.249 moko 85: unsigned int pa_atoui(const char *str, int base, const String* problem_source){
86: return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
87: }
88:
89: unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
90: return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
91: }
92:
1.193 paf 93: int pa_atoi(const char* str, const String* problem_source) {
94: if(!str)
95: return 0;
96:
1.242 moko 97: while(isspace(*str))
1.193 paf 98: str++;
1.242 moko 99:
1.193 paf 100: if(!*str)
101: return 0;
102:
1.200 paf 103: bool negative=false;
104: if(str[0]=='-') {
105: negative=true;
106: str++;
107: } else if(str[0]=='+') {
108: str++;
109: }
1.193 paf 110:
1.242 moko 111: unsigned int result=pa_atoui(str, 0, problem_source);
1.193 paf 112:
1.242 moko 113: if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
1.243 moko 114: return -(int)result;
1.242 moko 115:
116: if(result<=INT_MAX)
1.243 moko 117: return (int)result;
1.242 moko 118:
119: throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
1.193 paf 120: }
121:
122: double pa_atod(const char* str, const String* problem_source) {
123: if(!str)
124: return 0;
125:
1.242 moko 126: while(isspace(*str))
1.193 paf 127: str++;
1.242 moko 128:
1.193 paf 129: if(!*str)
130: return 0;
131:
1.200 paf 132: bool negative=false;
133: if(str[0]=='-') {
134: negative=true;
135: str++;
136: } else if(str[0]=='+') {
137: str++;
138: }
1.242 moko 139:
140: double result;
1.247 moko 141: if(str[0]=='0') {
142: if(str[1]=='x' || str[1]=='X') {
1.242 moko 143: // 0xABC
1.249 moko 144: result=(double)pa_atoul(str, 0, problem_source);
1.242 moko 145: return negative ? -result : result;
146: } else {
1.200 paf 147: // skip leading 0000, to disable octal interpretation
1.242 moko 148: do str++; while(*str=='0');
1.200 paf 149: }
1.247 moko 150: }
1.242 moko 151:
152: char *error_pos;
153: result=strtod(str, &error_pos);
1.193 paf 154:
155: while(char c=*error_pos++)
1.199 paf 156: if(!isspace((unsigned char)c))
1.242 moko 157: throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
1.193 paf 158:
1.242 moko 159: return negative ? -result : result;
1.193 paf 160: }
161:
1.176 paf 162: // cord lib extension
163:
164: #ifndef DOXYGEN
165: typedef struct {
166: ssize_t countdown;
1.204 paf 167: int target; /* Character we're looking for */
1.176 paf 168: } chr_data;
169: #endif
1.248 moko 170:
1.176 paf 171: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
172: {
173: register chr_data * d = (chr_data *)client_data;
174:
175: if (d -> countdown<=0) return(2);
176: d -> countdown -= size;
177: if (c > d -> target) return(1);
178: return(0);
179: }
1.248 moko 180:
1.176 paf 181: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
182: {
183: chr_data d;
184:
185: d.countdown = n;
186: d.target = c;
187: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
188: }
189:
1.187 paf 190: static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data)
1.178 paf 191: {
192: int* result=(int*)client_data;
193: (*result)++;
194: return(0); // 0=continue
195: }
1.248 moko 196:
1.178 paf 197: size_t CORD_block_count(CORD x)
198: {
199: size_t result=0;
200: CORD_block_iter(x, 0, CORD_block_count_proc, &result);
201: return result;
202: }
203:
1.174 paf 204: // helpers
1.139 paf 205:
1.174 paf 206: /// String::match uses this as replace & global search table columns
1.139 paf 207:
1.174 paf 208: const int MAX_MATCH_GROUPS=100;
1.139 paf 209:
1.174 paf 210: class String_match_table_template_columns: public ArrayString {
211: public:
212: String_match_table_template_columns() {
213: *this+=new String("prematch");
214: *this+=new String("match");
215: *this+=new String("postmatch");
216: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
1.176 paf 217: *this+=new String(String::Body::Format(1+i), String::L_CLEAN);
1.174 paf 218: }
219: }
220: };
221:
222: Table string_match_table_template(new String_match_table_template_columns);
223:
1.176 paf 224: // String::Body methods
1.140 paf 225:
1.176 paf 226: String::Body String::Body::Format(int value) {
1.174 paf 227: char local[MAX_NUMBER];
228: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
1.224 misha 229: return String::Body(pa_strdup(local, length));
1.120 paf 230: }
231:
1.195 paf 232: String::Body String::Body::trim(String::Trim_kind kind, const char* chars,
1.229 misha 233: size_t* out_start, size_t* out_length, Charset* source_charset) const {
1.195 paf 234: size_t our_length=length();
235: if(!our_length)
236: return *this;
1.229 misha 237:
238: // check if any UTF-8 in chars
239: bool fast=true;
240: if(chars && source_charset && source_charset->isUTF8()){
241: const char* pos=chars;
242: while(unsigned char c=*pos++)
243: if(c>127){
244: fast=false;
245: break;
246: }
247: }
248:
249: size_t start=0;
250: size_t end=our_length;
1.195 paf 251: if(!chars)
252: chars=" \t\n"; // white space
253:
1.229 misha 254: if(fast){
255: // from left...
256: if(kind!=TRIM_END) {
257: CORD_pos pos; set_pos(pos, 0);
258: while(true) {
259: char c=CORD_pos_fetch(pos);
260: if(strchr(chars, c)) {
261: if(++start==our_length)
262: return 0; // all chars are empty, just return empty string
263: } else
264: break;
265:
266: CORD_next(pos);
267: }
268: }
269:
270: // from right..
271: if(kind!=TRIM_START) {
272: CORD_pos pos; set_pos(pos, end-1);
273: while(true) {
274: char c=CORD_pos_fetch(pos);
275: if(strchr(chars, c)) {
276: if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
277: return 0; // all chars are empty, just return empty string
278: } else
279: break;
280:
281: CORD_prev(pos);
282: }
283: }
284: } else {
1.234 misha 285: const XMLByte* src_begin=(const XMLByte*)cstr();
1.229 misha 286: const XMLByte* src_end=src_begin+our_length;
287:
288: // from left...
289: if(kind!=TRIM_END) {
290: while(src_begin<src_end){
291: uint char_length=1;
292: const XMLByte* ptr=src_begin;
1.231 misha 293: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
294: while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0))
1.229 misha 295: char_length++;
296:
297: bool found=false;
298: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++)
299: if(strncmp(chars_byte, (const char*)ptr, char_length)==0){
300: found=true;
301: break;
302: }
303:
304: if(found){
305: start+=char_length;
306: if(start==our_length)
307: return 0; // all chars are empty, just return empty string
308: } else
309: break;
310: }
311: }
1.196 paf 312:
1.229 misha 313: // from right..
314: if(kind!=TRIM_START) {
315: while(src_begin<src_end){
316: uint char_length=1;
1.231 misha 317: // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3
318: while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0))
1.229 misha 319: char_length++;
320:
321: bool found=false;
322: for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++)
323: if(strncmp(chars_byte, (const char*)src_end, char_length)==0){
324: found=true;
325: break;
326: }
327:
328: if(found){
329: end-=char_length;
330: if(end==0)
331: return 0; // all chars are empty, just return empty string
332: } else
333: break;
334: }
1.196 paf 335: }
336: }
1.195 paf 337:
338: if(start==0 && end==our_length) // nobody moved a thing
339: return *this;
340:
341: if(out_start)
342: *out_start=start;
343: size_t new_length=end-start;
344: if(out_length)
345: *out_length=new_length;
346:
347: return mid(start, new_length);
348: }
349:
1.174 paf 350: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
351: uint& result=*static_cast<uint*>(client_data);
352: generic_hash_code(result, c);
353: return 0;
354: }
1.248 moko 355:
1.174 paf 356: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
357: uint& result=*static_cast<uint*>(client_data);
358: generic_hash_code(result, s);
359: return 0;
1.248 moko 360: }
361:
1.225 misha 362: uint String::Body::get_hash_code() const {
363: #ifdef HASH_CODE_CACHING
364: if(hash_code)
365: return hash_code;
366: #else
367: uint hash_code=0;
368: #endif
1.220 misha 369: if (body && CORD_IS_STRING(body)){
1.250 moko 370: generic_hash_code(hash_code, (const char *)body);
1.220 misha 371: } else {
372: CORD_iter5(body, 0,
373: CORD_batched_iter_fn_generic_hash_code,
1.225 misha 374: CORD_batched_iter_fn_generic_hash_code, &hash_code);
1.220 misha 375: }
1.225 misha 376: return hash_code;
1.94 parser 377: }
378:
1.241 misha 379: struct CORD_pos_info {
380: const char* chars;
381: size_t left;
382: size_t pos;
383: };
384:
385: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk
386: static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) {
387: if(info->pos < info->left){
388: info->pos=STRING_NOT_FOUND;
389: return 1;
390: }
391: if(strchr(info->chars, c))
392: return 1;
393: --(info->pos);
394: return 0;
395: }
396:
397: size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const {
398: if(is_empty() || !chars || !strlen(chars))
399: return STRING_NOT_FOUND;
400: CORD_pos_info info={chars, left, right};
401: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info))
402: return info.pos;
403: else
404: return STRING_NOT_FOUND;
405: }
406:
407:
408: // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars
409: static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) {
410: if(info->pos < info->left) {
411: info->pos=STRING_NOT_FOUND;
412: return 1;
413: }
414: if(!strchr(info->chars, c))
415: return 1;
416: --(info->pos);
417: return 0;
418: }
419:
420: size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const {
421: if(is_empty() || !chars || !strlen(chars))
422: return STRING_NOT_FOUND;
423: CORD_pos_info info={chars, left, right};
424: if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info))
425: return info.pos;
426: else
427: return STRING_NOT_FOUND;
428: }
429:
1.174 paf 430: // String methods
431:
432: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
433: if(!known_length)
1.9 paf 434: return *this;
1.122 paf 435:
1.176 paf 436: // first: langs
437: langs.append(body, lang, known_length);
438: // next: letters themselves
1.174 paf 439: body.append_know_length(str, known_length);
1.1 paf 440:
1.174 paf 441: ASSERT_STRING_INVARIANT(*this);
1.1 paf 442: return *this;
443: }
1.248 moko 444:
1.174 paf 445: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
446: if(!str)
447: return *this;
448: size_t known_length=helper_length?helper_length:strlen(str);
449: if(!known_length)
450: return *this;
1.1 paf 451:
1.174 paf 452: return append_know_length(str, known_length, lang);
1.5 paf 453: }
1.248 moko 454:
1.244 moko 455: String::String(int value, const char *format) : langs(L_CLEAN){
1.226 misha 456: char buf[MAX_NUMBER];
457: body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value));
458: }
1.248 moko 459:
1.174 paf 460: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
461: size_t known_length=helper_length?helper_length:strlen(str);
462: if(!known_length)
463: return *this;
1.5 paf 464:
1.176 paf 465: // first: langs
466: langs.append(body, lang, known_length);
467: // next: letters themselves
1.174 paf 468: body.append_strdup_know_length(str, known_length);
1.33 paf 469:
1.174 paf 470: ASSERT_STRING_INVARIANT(*this);
471: return *this;
1.5 paf 472: }
1.46 paf 473:
1.234 misha 474: struct CORD_length_info {
475: size_t len;
476: size_t skip;
477: };
478:
479: int CORD_batched_len(const char* s, CORD_length_info* info){
480: info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0;
1.221 misha 481: }
482:
1.234 misha 483: // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
484: int CORD_batched_len(const char c, CORD_length_info* info){
485: if (info->skip==0){
486: info->len++;
487: info->skip = lengthUTF8Char(c)-1;
488: } else {
489: info->skip--;
490: }
1.220 misha 491: return 0;
492: }
493:
1.210 misha 494: size_t String::length(Charset& charset) const {
495: if(charset.isUTF8()){
1.234 misha 496: CORD_length_info info = {0, 0};
497: body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info);
498: return info.len;
1.210 misha 499: } else
500: return body.length();
501: }
502:
1.174 paf 503: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
504: String& String::mid(size_t substr_begin, size_t substr_end) const {
505: String& result=*new String;
506:
507: size_t self_length=length();
508: substr_begin=min(substr_begin, self_length);
509: substr_end=min(max(substr_end, substr_begin), self_length);
1.176 paf 510: size_t substr_length=substr_end-substr_begin;
511: if(!substr_length)
1.107 parser 512: return result;
1.53 paf 513:
1.176 paf 514: // first: their langs
515: result.langs.append(result.body, langs, substr_begin, substr_length);
516: // next: letters themselves
517: result.body=body.mid(substr_begin, substr_length);
1.174 paf 518:
519: ASSERT_STRING_INVARIANT(result);
1.53 paf 520: return result;
1.54 paf 521: }
522:
1.211 misha 523: // from, to and helper_length in characters, not in bytes (it's important for utf-8)
524: String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
1.210 misha 525: String& result=*new String;
526:
1.211 misha 527: size_t self_length=(helper_length)?helper_length:length(charset);
528:
529: if(!self_length)
530: return result;
531:
532: from=min(min(to, from), self_length);
1.210 misha 533: to=min(max(to, from), self_length);
1.211 misha 534:
1.210 misha 535: size_t substr_length=to-from;
1.211 misha 536:
1.210 misha 537: if(!substr_length)
538: return result;
539:
540: if(charset.isUTF8()){
1.234 misha 541: const XMLByte* src_begin=(const XMLByte*)cstr();
1.230 misha 542: const XMLByte* src_end=src_begin+body.length();
1.210 misha 543:
1.212 misha 544: // convert 'from' and 'substr_length' from 'characters' to 'bytes'
1.230 misha 545: from=getUTF8BytePos(src_begin, src_end, from);
546: substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length);
1.210 misha 547: if(!substr_length)
548: return result;
549: }
550:
551: // first: their langs
552: result.langs.append(result.body, langs, from, substr_length);
553: // next: letters themselves
554: result.body=body.mid(from, substr_length);
555:
556: ASSERT_STRING_INVARIANT(result);
557: return result;
558: }
559:
1.176 paf 560: size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const {
1.183 paf 561: size_t substr_length=substr.length();
562: while(true) {
563: size_t substr_begin=body.pos(substr, this_offset);
564:
565: if(substr_begin==CORD_NOT_FOUND)
566: return STRING_NOT_FOUND;
1.174 paf 567:
1.183 paf 568: if(langs.check_lang(lang, substr_begin, substr_length))
569: return substr_begin;
570:
571: this_offset=substr_begin+substr_length;
572: }
1.58 paf 573: }
574:
1.174 paf 575: size_t String::pos(const String& substr,
576: size_t this_offset, Language lang) const {
577: return pos(substr.body, this_offset, lang);
1.60 paf 578: }
579:
1.210 misha 580: size_t String::pos(Charset& charset, const String& substr,
581: size_t this_offset, Language lang) const {
582:
583: if(charset.isUTF8()){
1.234 misha 584: const XMLByte* srcPtr=(const XMLByte*)cstr();
1.212 misha 585: const XMLByte* srcEnd=srcPtr+body.length();
586:
587: // convert 'this_offset' from 'characters' to 'bytes'
588: this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset);
589:
1.229 misha 590: size_t result=pos(substr.body, this_offset, lang);
591: return (result==CORD_NOT_FOUND)
592: ? STRING_NOT_FOUND
593: : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters'
1.212 misha 594: } else {
1.229 misha 595: size_t result=pos(substr.body, this_offset, lang);
596: return (result==CORD_NOT_FOUND)
597: ? STRING_NOT_FOUND
598: : result;
1.210 misha 599: }
600: }
601:
1.174 paf 602: void String::split(ArrayString& result,
603: size_t& pos_after,
604: const char* delim,
605: Language lang, int limit) const {
1.237 misha 606: if(is_empty())
607: return;
1.174 paf 608: size_t self_length=length();
1.237 misha 609: if(size_t delim_length=strlen(delim)) {
1.186 paf 610: size_t pos_before;
1.60 paf 611: // while we have 'delim'...
1.174 paf 612: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 613: result+=&mid(pos_after, pos_before);
1.174 paf 614: pos_after=pos_before+delim_length;
1.60 paf 615: }
616: // last piece
1.174 paf 617: if(pos_after<self_length && limit) {
618: result+=&mid(pos_after, self_length);
619: pos_after=self_length;
1.60 paf 620: }
621: } else { // empty delim
622: result+=this;
1.174 paf 623: pos_after+=self_length;
1.60 paf 624: }
625: }
626:
1.174 paf 627: void String::split(ArrayString& result,
628: size_t& pos_after,
629: const String& delim, Language lang,
630: int limit) const {
1.237 misha 631: if(is_empty())
632: return;
633: if(!delim.is_empty()) {
1.186 paf 634: size_t pos_before;
1.60 paf 635: // while we have 'delim'...
1.174 paf 636: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 637: result+=&mid(pos_after, pos_before);
1.174 paf 638: pos_after=pos_before+delim.length();
1.60 paf 639: }
640: // last piece
1.174 paf 641: if(pos_after<length() && limit) {
642: result+=&mid(pos_after, length());
643: pos_after=length();
1.60 paf 644: }
645: } else { // empty delim
646: result+=this;
1.174 paf 647: pos_after+=length();
1.60 paf 648: }
1.61 paf 649: }
650:
1.222 misha 651: Table* String::match(VRegex* vregex,
652: Row_action row_action, void *info,
653: int& matches_count) const {
1.209 misha 654:
1.222 misha 655: // vregex->info(); // I have no idea what does it for?
1.63 paf 656:
1.222 misha 657: bool need_pre_post_match=vregex->is_pre_post_match_needed();
658: bool global=vregex->is_global_search();
1.63 paf 659:
1.174 paf 660: const char* subject=cstr();
1.234 misha 661: size_t subject_length=length();
1.222 misha 662: const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
663: int ovector[ovector_size];
1.155 paf 664:
1.173 paf 665: Table::Action_options table_options;
1.174 paf 666: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 667:
1.154 paf 668: int prestart=0;
669: int poststart=0;
1.174 paf 670: int postfinish=length();
1.63 paf 671: while(true) {
1.222 misha 672: int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
1.63 paf 673:
1.222 misha 674: if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception
675: break;
1.63 paf 676:
1.154 paf 677: int prefinish=ovector[0];
678: poststart=ovector[1];
1.236 moko 679:
680: if (prestart==poststart && subject[poststart]=='\n'){
681: prestart++;
682: continue;
683: }
684:
1.232 misha 685: ArrayString* row=new ArrayString(3);
1.174 paf 686: if(need_pre_post_match) {
687: *row+=&mid(0, prefinish); // .prematch column value
688: *row+=&mid(prefinish, poststart); // .match
689: *row+=&mid(poststart, postfinish); // .postmatch
690: } else {
1.185 paf 691: *row+=&Empty; // .prematch column value
692: *row+=&Empty; // .match
693: *row+=&Empty; // .postmatch
1.174 paf 694: }
1.63 paf 695:
1.222 misha 696: for(int i=1; i<exec_result; i++) {
1.69 paf 697: // -1:-1 case handled peacefully by mid() itself
1.232 misha 698: *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 699: }
700:
1.209 misha 701: matches_count++;
1.174 paf 702: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 703:
1.222 misha 704: if(!global || prestart==poststart) // last step
705: break;
706:
1.154 paf 707: prestart=poststart;
1.222 misha 708: }
1.63 paf 709:
1.222 misha 710: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
711: return vregex->is_just_count() ? 0 : &table;
1.82 parser 712: }
713:
1.174 paf 714: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
715: String& result=*new String();
716: if(is_empty())
717: return result;
718:
719: char* new_cstr=cstrm();
1.216 misha 720:
1.181 paf 721: if(source_charset.isUTF8()) {
1.216 misha 722: size_t new_cstr_len=length();
1.181 paf 723: switch(kind) {
724: case CC_UPPER:
1.192 paf 725: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper);
1.181 paf 726: break;
727: case CC_LOWER:
1.192 paf 728: change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower);
1.181 paf 729: break;
730: default:
731: assert(!"unknown change case kind");
732: break; // never
733: }
734:
735: } else {
736: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 737:
1.181 paf 738: const unsigned char *a;
739: const unsigned char *b;
740: switch(kind) {
741: case CC_UPPER:
742: a=tables+lcc_offset;
743: b=tables+fcc_offset;
744: break;
745: case CC_LOWER:
746: a=tables+lcc_offset;
747: b=0;
748: break;
749: default:
750: assert(!"unknown change case kind");
751: a=b=0; // calm, compiler
752: break; // never
753: }
754:
1.192 paf 755: char *dest=new_cstr;
1.181 paf 756: unsigned char index;
1.190 paf 757: for(const char* current=new_cstr; (index=(unsigned char)*current); current++) {
1.181 paf 758: unsigned char c=a[index];
759: if(b)
760: c=b[c];
761:
762: *dest++=(char)c;
763: }
1.174 paf 764: }
1.176 paf 765: result.langs=langs;
1.174 paf 766: result.body=new_cstr;
1.89 parser 767:
1.101 parser 768: return result;
769: }
770:
1.213 misha 771: const String& String::escape(Charset& source_charset) const {
772: if(is_empty())
773: return *this;
774:
775: return Charset::escape(*this, source_charset);
776: }
777:
1.238 misha 778: #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \
779: result.langs.append(result.body, langs, langs_offset, length); \
780: result.body.append_strdup_know_length(from_cstr, length);
781:
1.174 paf 782: const String& String::replace(const Dictionary& dict) const {
1.238 misha 783: if(!dict.count() || is_empty())
784: return *this;
785:
1.174 paf 786: String& result=*new String();
787: const char* old_cstr=cstr();
788: const char* prematch_begin=old_cstr;
789:
1.238 misha 790: if(dict.count()==1) {
791: // optimized simple case
792:
793: Dictionary::Subst subst=dict.get(0);
1.239 moko 794: while(const char* p=strstr(prematch_begin, subst.from)) {
1.174 paf 795: // prematch
1.238 misha 796: if(size_t prematch_length=p-prematch_begin) {
797: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
1.101 parser 798: }
799:
1.174 paf 800: // match
1.238 misha 801: prematch_begin=p+subst.from_length;
1.174 paf 802:
1.184 paf 803: if(const String* b=subst.to) // are there any b?
1.174 paf 804: result<<*b;
1.238 misha 805: }
806:
807: } else {
808:
809: const char* current=old_cstr;
810: while(*current) {
811: if(Dictionary::Subst subst=dict.first_that_begins(current)) {
812: // prematch
813: if(size_t prematch_length=current-prematch_begin) {
814: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length)
815: }
816:
817: // match
818: // skip 'a' in 'current'; move prematch_begin
819: current+=subst.from_length; prematch_begin=current;
820:
821: if(const String* b=subst.to) // are there any b?
822: result<<*b;
823: } else // simply advance
824: current++;
825: }
826:
1.174 paf 827: }
1.156 paf 828:
1.238 misha 829: if(prematch_begin==old_cstr) // not modified
830: return *this;
831:
1.174 paf 832: // postmatch
1.238 misha 833: if(size_t postmatch_length=old_cstr+length()-prematch_begin) {
834: STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, postmatch_length)
1.174 paf 835: }
1.156 paf 836:
1.174 paf 837: ASSERT_STRING_INVARIANT(result);
1.82 parser 838: return result;
1.61 paf 839: }
1.113 parser 840:
1.180 paf 841: static int serialize_body_char(char c, char** cur) {
842: *((*cur)++)=c;
843: return 0; // 0=continue
1.248 moko 844: }
845:
1.174 paf 846: static int serialize_body_piece(const char* s, char** cur) {
847: size_t length=strlen(s);
848: memcpy(*cur, s, length); *cur+=length;
1.178 paf 849: return 0; // 0=continue
1.248 moko 850: }
851:
1.178 paf 852: static int serialize_lang_piece(char alang, size_t asize, char** cur) {
853: // lang
1.191 paf 854: **cur=alang; (*cur)++;
855: // length [WARNING: not cast, addresses must be %4=0 on sparc]
1.178 paf 856: memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize);
857:
858: return 0; // 0=continue
859: }
1.248 moko 860:
1.174 paf 861: String::Cm String::serialize(size_t prolog_length) const {
1.178 paf 862: size_t fragments_count=langs.count();
1.202 paf 863: size_t body_length=body.length();
1.174 paf 864: size_t buf_length=
1.178 paf 865: prolog_length //1
866: +sizeof(size_t) //2
1.202 paf 867: +body_length //3
868: +1 // 4 for zero terminator used in deserialize
869: +sizeof(size_t) //5
870: +fragments_count*(sizeof(char)+sizeof(size_t)); //6
871:
1.174 paf 872: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
873:
874: // 1: prolog
875: char *cur=result.str+prolog_length;
1.202 paf 876: // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc]
877: memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length);
878: // 3: letters
879: body.for_each(serialize_body_char, serialize_body_piece, &cur);
880: // 4: zero terminator
881: *cur++=0;
882: // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc]
1.174 paf 883: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 884: // 6: lang info
1.178 paf 885: langs.for_each(body, serialize_lang_piece, &cur);
1.113 parser 886:
1.174 paf 887: return result;
1.113 parser 888: }
1.248 moko 889:
1.202 paf 890: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) {
891: size_t in_buf=buf_size;
892: if(in_buf<=prolog_size)
1.148 paf 893: return false;
1.202 paf 894: in_buf-=prolog_size;
1.135 paf 895:
1.174 paf 896: // 1: prolog
1.202 paf 897: const char* cur=(const char* )buf+prolog_size;
1.113 parser 898:
1.207 paf 899: // 2: chars.count
1.202 paf 900: size_t body_length;
901: if(in_buf<sizeof(body_length)) // body.length don't fit?
902: return false;
903: // [WARNING: not cast, addresses must be %4=0 on sparc]
904: memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length);
905: in_buf-=sizeof(body_length);
906:
907: if(in_buf<body_length+1) // letters+terminator don't fit?
908: return false;
909: // 4: zero terminator
910: if(cur[body_length] != 0) // in place?
911: return false;
912: // 3: letters
1.251 ! moko 913: body=String::Body(String::C(cur, body_length));
1.202 paf 914: cur+=body_length+1;
915: in_buf-=body_length+1;
916:
917: // 5: langs.count
1.191 paf 918: size_t fragments_count;
1.202 paf 919: if(in_buf<sizeof(fragments_count)) // langs.count don't fit?
1.174 paf 920: return false;
1.191 paf 921: // [WARNING: not cast, addresses must be %4=0 on sparc]
922: memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count);
1.202 paf 923: in_buf-=sizeof(fragments_count);
1.174 paf 924:
925: if(fragments_count) {
1.202 paf 926: // 6: lang info
1.174 paf 927: size_t total_length=0;
928: for(size_t f=0; f<fragments_count; f++) {
1.191 paf 929: char lang;
930: size_t fragment_length;
931: size_t piece_length=sizeof(lang)+sizeof(fragment_length);
1.202 paf 932: if(in_buf<piece_length) // lang+length
1.174 paf 933: return false;
934:
1.191 paf 935: // lang
936: lang=*cur++;
937: // length [WARNING: not cast, addresses must be %4=0 on sparc]
938: memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length);
939:
1.206 paf 940: size_t combined_length=total_length+fragment_length;
941: if(combined_length>body_length)
942: return false; // file curruption
1.191 paf 943: // uchar needed to prevent propagating 0x80 bit to upper bytes
944: langs.append(total_length, (String::Language)(uchar)lang, fragment_length);
1.206 paf 945: total_length=combined_length;
1.202 paf 946: in_buf-=piece_length;
1.174 paf 947: }
1.128 paf 948:
1.202 paf 949: if(total_length!=body_length) // length(all language fragments) vs length(letters)
1.148 paf 950: return false;
1.174 paf 951: }
1.202 paf 952: if(in_buf!=0) // some strange extra bytes
953: return false;
1.113 parser 954:
1.174 paf 955: ASSERT_STRING_INVARIANT(*this);
1.148 paf 956: return true;
1.176 paf 957: }
958:
959: const char* String::Body::v() const {
1.233 misha 960: return CORD_to_const_char_star(body, length());
1.176 paf 961: }
1.248 moko 962:
1.201 paf 963: void String::Body::dump() const {
964: CORD_dump(body);
965: }
966:
1.176 paf 967: const char* String::Languages::v() const {
1.177 paf 968: if(opt.is_not_just_lang)
1.233 misha 969: return CORD_to_const_char_star(langs, 0);
1.176 paf 970: else
971: return (const char*)&langs;
972: }
1.248 moko 973:
1.201 paf 974: void String::Languages::dump() const {
975: if(opt.is_not_just_lang)
976: CORD_dump(langs);
977: else
978: puts((const char*)&langs);
979: }
1.248 moko 980:
1.176 paf 981: const char* String::v() const {
1.198 paf 982: const uint LIMIT_VIEW=20;
1.246 moko 983: char* buf=(char*)pa_malloc(MAX_STRING);
1.176 paf 984: const char*body_view=body.v();
985: const char*langs_view=langs.v();
986: snprintf(buf, MAX_STRING,
1.178 paf 987: "%d:%.*s%s} "
1.176 paf 988: "{%d:%s",
1.178 paf 989: langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
1.176 paf 990: strlen(body_view), body_view
991: );
992:
993: return buf;
1.113 parser 994: }
1.229 misha 995:
1.201 paf 996: void String::dump() const {
997: body.dump();
998: langs.dump();
999: }
1.229 misha 1000:
1001: const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
1.223 misha 1002: if(is_empty())
1.195 paf 1003: return *this;
1004:
1005: size_t substr_begin, substr_length;
1.229 misha 1006: Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset);
1.195 paf 1007: if(new_body==body) // we received unchanged pointer, do likewise
1008: return *this;
1009: // new_body differs from body, adjust langs along
1010:
1011: String& result=*new String;
1012: if(!new_body) // body.trim produced empty result
1013: return result;
1014: // body.trim produced nonempty result
1015:
1016: // first: their langs
1017: result.langs.append(result.body, langs, substr_begin, substr_length);
1018: // next: letters themselves
1019: result.body=new_body;
1020:
1021: ASSERT_STRING_INVARIANT(result);
1022: return result;
1.198 paf 1023: }
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