Annotation of parser3/src/main/pa_string.C, revision 1.175.2.5
1.45 paf 1: /** @file
1.174 paf 2: Parser: string class. @see untalength_t.C.
1.46 paf 3:
1.174 paf 4: Copyright (c) 2001-2003 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:
1.175.2.5! paf 8: static const char* IDENT_STRING_C="$Date: 2003/09/24 08:02:09 $";
1.4 paf 9:
1.70 paf 10: #include "pcre.h"
11:
1.12 paf 12: #include "pa_string.h"
1.22 paf 13: #include "pa_exception.h"
1.61 paf 14: #include "pa_table.h"
1.101 parser 15: #include "pa_dictionary.h"
1.132 paf 16: #include "pa_charset.h"
1.60 paf 17:
1.175.2.1 paf 18: // cord lib extension
19:
20: #ifndef DOXYGEN
21: typedef struct {
22: ssize_t countdown;
23: char target; /* Character we're looking for */
24: } chr_data;
25: #endif
1.175.2.3 paf 26: static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
1.175.2.1 paf 27: {
28: register chr_data * d = (chr_data *)client_data;
29:
30: if (d -> countdown<=0) return(2);
31: d -> countdown -= size;
1.175.2.3 paf 32: if (c > d -> target) return(1);
1.175.2.1 paf 33: return(0);
34: }
1.175.2.4 paf 35: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c)
1.175.2.1 paf 36: {
37: chr_data d;
38:
1.175.2.4 paf 39: d.countdown = n;
40: d.target = c;
1.175.2.3 paf 41: return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/);
1.175.2.1 paf 42: }
43:
1.174 paf 44: // helpers
1.139 paf 45:
1.174 paf 46: /// String::match uses this as replace & global search table columns
1.139 paf 47:
1.174 paf 48: const int MAX_MATCH_GROUPS=100;
1.139 paf 49:
1.174 paf 50: class String_match_table_template_columns: public ArrayString {
51: public:
52: String_match_table_template_columns() {
53: *this+=new String("prematch");
54: *this+=new String("match");
55: *this+=new String("postmatch");
56: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
57: *this+=new String(StringBody::Format(1+i), String::L_CLEAN);
58: }
59: }
60: };
61:
62: Table string_match_table_template(new String_match_table_template_columns);
63:
64: // StringBody methods
1.140 paf 65:
1.174 paf 66: StringBody StringBody::Format(int value) {
67: char local[MAX_NUMBER];
68: size_t length=snprintf(local, MAX_NUMBER, "%d", value);
69: return StringBody(pa_strdup(local, length), length);
1.120 paf 70: }
71:
1.174 paf 72: static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) {
73: uint& result=*static_cast<uint*>(client_data);
74: generic_hash_code(result, c);
75: return 0;
76: }
77: static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) {
78: uint& result=*static_cast<uint*>(client_data);
79: generic_hash_code(result, s);
80: return 0;
81: };
82: uint StringBody::hash_code() const {
83: uint result=0;
84: CORD_iter5(body, 0,
85: CORD_batched_iter_fn_generic_hash_code,
86: CORD_batched_iter_fn_generic_hash_code, &result);
1.120 paf 87: return result;
1.94 parser 88: }
89:
1.174 paf 90: // String methods
91:
92: String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) {
93: append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN);
1.115 paf 94: }
1.174 paf 95: String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) {
96: append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN);
1.5 paf 97: }
1.28 paf 98:
1.174 paf 99: String& String::append_know_length(const char* str, size_t known_length, Language lang) {
100: if(!known_length)
1.9 paf 101: return *this;
1.122 paf 102:
1.175.2.1 paf 103: // first: langs
104: langs.append(body, lang, known_length);
105: // next: letters themselves
1.174 paf 106: body.append_know_length(str, known_length);
1.1 paf 107:
1.175.2.2 paf 108: ASSERT_STRING_INVARIANT(*this);
1.1 paf 109: return *this;
110: }
1.174 paf 111: String& String::append_help_length(const char* str, size_t helper_length, Language lang) {
112: if(!str)
113: return *this;
114: size_t known_length=helper_length?helper_length:strlen(str);
115: if(!known_length)
116: return *this;
1.1 paf 117:
1.174 paf 118: return append_know_length(str, known_length, lang);
1.5 paf 119: }
1.174 paf 120: String& String::append_strdup(const char* str, size_t helper_length, Language lang) {
121: size_t known_length=helper_length?helper_length:strlen(str);
122: if(!known_length)
123: return *this;
1.5 paf 124:
1.175.2.1 paf 125: // first: langs
126: langs.append(body, lang, known_length);
127: // next: letters themselves
1.174 paf 128: body.append_strdup_know_length(str, known_length);
1.33 paf 129:
1.175.2.2 paf 130: ASSERT_STRING_INVARIANT(*this);
1.174 paf 131: return *this;
1.5 paf 132: }
1.46 paf 133:
1.174 paf 134: /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab"
135: String& String::mid(size_t substr_begin, size_t substr_end) const {
136: String& result=*new String;
137:
138: size_t self_length=length();
139: substr_begin=min(substr_begin, self_length);
140: substr_end=min(max(substr_end, substr_begin), self_length);
1.175.2.5! paf 141: size_t substr_length=substr_end-substr_begin;
! 142: if(!substr_length)
1.107 parser 143: return result;
1.53 paf 144:
1.175.2.1 paf 145: // first: their langs
1.175.2.5! paf 146: result.langs.append(result.body, langs, substr_begin, substr_length);
1.175.2.1 paf 147: // next: letters themselves
1.175.2.5! paf 148: result.body=body.mid(substr_begin, substr_length);
1.174 paf 149:
150: // SAPI::log("piece of '%s' from %d to %d is '%s'",
151: //cstr(), substr_begin, substr_end, result.cstr());
1.175.2.2 paf 152: ASSERT_STRING_INVARIANT(result);
1.53 paf 153: return result;
1.54 paf 154: }
155:
1.175.2.1 paf 156: size_t String::pos(const StringBody substr, size_t this_offset, Language lang) const {
1.174 paf 157: size_t substr_begin=body.pos(substr, this_offset);
1.175.2.3 paf 158: if(substr_begin==CORD_NOT_FOUND || !langs.check_lang(lang, substr_begin, substr.length()))
1.174 paf 159: return STRING_NOT_FOUND;
160:
1.175.2.1 paf 161: return substr_begin;
1.58 paf 162: }
163:
1.174 paf 164: size_t String::pos(const String& substr,
165: size_t this_offset, Language lang) const {
166: return pos(substr.body, this_offset, lang);
1.60 paf 167: }
168:
1.174 paf 169: void String::split(ArrayString& result,
170: size_t& pos_after,
171: const char* delim,
172: Language lang, int limit) const {
173: size_t self_length=length();
174: if(size_t delim_length=strlen(delim)) {
1.60 paf 175: int pos_before;
176: // while we have 'delim'...
1.174 paf 177: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 178: result+=&mid(pos_after, pos_before);
1.174 paf 179: pos_after=pos_before+delim_length;
1.60 paf 180: }
181: // last piece
1.174 paf 182: if(pos_after<self_length && limit) {
183: result+=&mid(pos_after, self_length);
184: pos_after=self_length;
1.60 paf 185: }
186: } else { // empty delim
187: result+=this;
1.174 paf 188: pos_after+=self_length;
1.60 paf 189: }
190: }
191:
1.174 paf 192: void String::split(ArrayString& result,
193: size_t& pos_after,
194: const String& delim, Language lang,
195: int limit) const {
1.140 paf 196: if(!delim.is_empty()) {
1.60 paf 197: int pos_before;
198: // while we have 'delim'...
1.174 paf 199: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 200: result+=&mid(pos_after, pos_before);
1.174 paf 201: pos_after=pos_before+delim.length();
1.60 paf 202: }
203: // last piece
1.174 paf 204: if(pos_after<length() && limit) {
205: result+=&mid(pos_after, length());
206: pos_after=length();
1.60 paf 207: }
208: } else { // empty delim
209: result+=this;
1.174 paf 210: pos_after+=length();
1.60 paf 211: }
1.61 paf 212: }
213:
1.174 paf 214: static void regex_options(const String* options, int *result, bool& need_pre_post_match){
1.63 paf 215: struct Regex_option {
1.174 paf 216: const char* keyL;
217: const char* keyU;
1.63 paf 218: int clear, set;
219: int *result;
1.154 paf 220: bool *flag;
1.63 paf 221: } regex_option[]={
1.153 paf 222: {"i", "I", 0, PCRE_CASELESS, result}, // a=A
223: {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default]
224: {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
225: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
226: {"g", "G", 0, true, result+1}, // many rows
1.154 paf 227: {"'", 0, 0, 0, 0, &need_pre_post_match},
228: {0}
1.63 paf 229: };
1.171 paf 230: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 231: result[1]=0;
232:
1.174 paf 233: if(options && !options->is_empty())
1.153 paf 234: for(Regex_option *o=regex_option; o->keyL; o++)
1.174 paf 235: if(options->pos(o->keyL)!=STRING_NOT_FOUND
236: || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) {
1.154 paf 237: if(o->flag)
238: *o->flag=true;
239: else { // result
240: *o->result &= ~o->clear;
241: *o->result |= o->set;
242: }
1.63 paf 243: }
244: }
245:
1.174 paf 246: Table* String::match(Charset& source_charset,
247: const String& regexp,
248: const String* options,
249: Row_action row_action, void *info,
250: bool& just_matched) const {
1.140 paf 251: if(regexp.is_empty())
1.149 paf 252: throw Exception(0,
1.174 paf 253: 0,
1.73 paf 254: "regexp is empty");
1.154 paf 255:
1.174 paf 256: const char* pattern=regexp.cstr();
257: const char* errptr;
1.62 paf 258: int erroffset;
1.173 paf 259: bool need_pre_post_match=false;
1.174 paf 260: int option_bits[2]={0}; regex_options(options, option_bits, need_pre_post_match);
261: bool global=option_bits[1]!=0;
1.63 paf 262: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 263: &errptr, &erroffset,
1.174 paf 264: source_charset.pcre_tables);
1.62 paf 265:
1.67 paf 266: if(!code)
1.149 paf 267: throw Exception(0,
1.174 paf 268: ®exp.mid(erroffset, regexp.length()),
1.74 paf 269: "regular expression syntax error - %s", errptr);
1.62 paf 270:
1.174 paf 271: int subpatterns=pcre_info(code, 0, 0);
272: if(subpatterns<0) {
1.100 parser 273: pcre_free(code);
1.149 paf 274: throw Exception(0,
1.174 paf 275: ®exp,
1.76 paf 276: "pcre_info error (%d)",
1.174 paf 277: subpatterns);
1.63 paf 278: }
279:
1.174 paf 280: const char* subject=cstr();
281: size_t subject_length=strlen(subject);
282: const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3;
283: int ovector[oveclength];
1.155 paf 284:
285: // create table
1.173 paf 286: Table::Action_options table_options;
1.174 paf 287: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 288:
1.64 paf 289: int exec_option_bits=0;
1.154 paf 290: int prestart=0;
291: int poststart=0;
1.174 paf 292: int postfinish=length();
1.63 paf 293: while(true) {
294: int exec_substrings=pcre_exec(code, 0,
1.174 paf 295: subject, subject_length, prestart,
296: exec_option_bits, ovector, oveclength);
1.63 paf 297:
298: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 299: pcre_free(code);
1.174 paf 300: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
301: if(global || subpatterns)
302: return &table; // global or with subpatterns=true+result
303: else {
304: just_matched=false; return 0; // not global=no result
305: }
1.63 paf 306: }
307:
308: if(exec_substrings<0) {
1.100 parser 309: pcre_free(code);
1.149 paf 310: throw Exception(0,
1.174 paf 311: ®exp,
1.76 paf 312: "regular expression execute error (%d)",
1.63 paf 313: exec_substrings);
314: }
315:
1.154 paf 316: int prefinish=ovector[0];
317: poststart=ovector[1];
1.174 paf 318: ArrayString* row=new ArrayString;
319: if(need_pre_post_match) {
320: *row+=&mid(0, prefinish); // .prematch column value
321: *row+=&mid(prefinish, poststart); // .match
322: *row+=&mid(poststart, postfinish); // .postmatch
323: } else {
324: *row+=0; // .prematch column value
325: *row+=0; // .match
326: *row+=0; // .postmatch
327: }
1.63 paf 328:
329: for(int i=1; i<exec_substrings; i++) {
1.69 paf 330: // -1:-1 case handled peacefully by mid() itself
1.174 paf 331: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 332: }
333:
1.174 paf 334: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 335:
1.174 paf 336: if(!global || prestart==poststart) { // not global | going to hang
1.100 parser 337: pcre_free(code);
1.174 paf 338: row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
339: return &table;
1.63 paf 340: }
1.154 paf 341: prestart=poststart;
1.63 paf 342:
343: /*
344: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 345: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 346: */
347: }
1.82 parser 348: }
349:
1.174 paf 350: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
351: String& result=*new String();
352: if(is_empty())
353: return result;
354:
355: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 356:
357: const unsigned char *a;
358: const unsigned char *b;
359: switch(kind) {
360: case CC_UPPER:
361: a=tables+lcc_offset;
362: b=tables+fcc_offset;
363: break;
364: case CC_LOWER:
365: a=tables+lcc_offset;
366: b=0;
367: break;
368: default:
1.149 paf 369: throw Exception(0,
1.82 parser 370: this,
371: "unknown change case kind #%d",
372: static_cast<int>(kind)); // never
373: a=b=0; // calm, compiler
374: break; // never
375: }
376:
1.174 paf 377: char* new_cstr=cstrm();
378: char *dest=new_cstr;
379: unsigned char index;
380: for(const char* current=new_cstr; index=(unsigned char)*current; current++) {
381: unsigned char c=a[index];
382: if(b)
383: c=b[c];
1.82 parser 384:
1.174 paf 385: *dest++=(char)c;
386: }
1.175.2.1 paf 387: result.langs=langs;
1.174 paf 388: result.body=new_cstr;
1.89 parser 389:
1.101 parser 390: return result;
391: }
392:
1.174 paf 393: const String& String::replace(const Dictionary& dict) const {
394: String& result=*new String();
395: const char* old_cstr=cstr();
396: const char* prematch_begin=old_cstr;
397:
398: const char* current=old_cstr;
399: while(*current) {
400: if(Table::element_type row=dict.first_that_begins(current)) {
401: // prematch
402: if(size_t prematch_length=current-prematch_begin) {
1.175.2.1 paf 403: result.langs.append(result.body, langs, prematch_begin-old_cstr, current-old_cstr);
1.174 paf 404: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 405: }
406:
1.174 paf 407: // match
1.170 paf 408:
1.174 paf 409: const String* a=row->get(0);
410: // skip 'a' in 'current'; move prematch_begin
411: current+=a->length(); prematch_begin=current;
412:
413: if(row->count()>1) { // are there any b?
414: const String* b=row->get(1);
415: result<<*b;
416: }
417: } else // simply advance
418: current++;
419: }
1.156 paf 420:
1.174 paf 421: // postmatch
422: if(size_t postmatch_length=current-prematch_begin) {
1.175.2.1 paf 423: result.langs.append(result.body, langs, prematch_begin-old_cstr, current-old_cstr);
1.174 paf 424: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
425: }
1.156 paf 426:
1.175.2.2 paf 427: ASSERT_STRING_INVARIANT(result);
1.89 parser 428: return result;
429: }
430:
1.90 parser 431: double String::as_double() const {
1.89 parser 432: double result;
1.174 paf 433: const char *str=cstr();
434:
435: while(*str && isspace(*str))
436: str++;
437: if(!*str)
1.162 paf 438: return 0;
1.161 paf 439:
1.102 parser 440: char *error_pos;
1.89 parser 441: // 0xABC
1.174 paf 442: if(str[0]=='0')
443: if(str[1]=='x' || str[1]=='X')
444: result=(double)(unsigned long)strtol(str, &error_pos, 0);
1.99 parser 445: else
1.174 paf 446: result=(double)strtod(str+1/*skip leading 0*/, &error_pos);
1.89 parser 447: else
1.174 paf 448: result=(double)strtod(str, &error_pos);
1.89 parser 449:
1.159 paf 450: while(char c=*error_pos++)
451: if(!isspace(c))
452: throw Exception("number.format",
453: this,
454: "invalid number (double)");
1.89 parser 455:
456: return result;
457: }
1.90 parser 458: int String::as_int() const {
1.89 parser 459: int result;
1.174 paf 460: const char *str=cstr();
461:
462: while(*str && isspace(*str))
463: str++;
464: if(!*str)
1.162 paf 465: return 0;
1.161 paf 466:
1.102 parser 467: char *error_pos;
1.89 parser 468: // 0xABC
1.174 paf 469: if(str[0]=='0')
470: if(str[1]=='x' || str[1]=='X')
471: result=(int)(unsigned long)strtol(str, &error_pos, 0);
1.99 parser 472: else
1.174 paf 473: result=(int)strtol(str+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 474: else
1.174 paf 475: result=(int)strtol(str, &error_pos, 0);
1.89 parser 476:
1.159 paf 477: while(char c=*error_pos++)
478: if(!isspace(c))
479: throw Exception("number.format",
480: this,
481: "invalid number (int)");
1.82 parser 482:
483: return result;
1.61 paf 484: }
1.113 parser 485:
1.174 paf 486: inline void uint2uchars(uint word, uchar *bytes) {
487: bytes[0]=word&0xFF;
488: bytes[1]=(word>>8)&0xFF;
489: bytes[2]=(word>>16)&0xFF;
490: bytes[3]=(word>>24)&0xFF;
491: }
492: inline uint uchars2uint(uchar *bytes) {
493: return bytes[3]<<24
494: | bytes[2]<<16
495: | bytes[1]<<8
496: | bytes[0];
497: }
498:
499: static int serialize_body_piece(const char* s, char** cur) {
500: size_t length=strlen(s);
501: memcpy(*cur, s, length); *cur+=length;
502: return 0;
503: };
504: String::Cm String::serialize(size_t prolog_length) const {
1.175.2.1 paf 505: #if TODO
1.175 paf 506: //_asm int 3;
1.174 paf 507: size_t buf_length=
508: prolog_length
1.175 paf 509: +sizeof(size_t)
1.175.2.1 paf 510: +langs.count()*(sizeof(Language)+sizeof(size_t))
1.174 paf 511: +length();
512: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
513:
514: // 1: prolog
515: char *cur=result.str+prolog_length;
516:
517:
1.175.2.1 paf 518: // 2: langs.count
519: size_t fragments_count=langs.count();
1.174 paf 520: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
521:
522: // 3: lang info
1.175.2.1 paf 523: for(Array_iterator<ArrayFragment::element_type> i(langs); i.has_next(); ) {
1.174 paf 524: const Fragment fragment=i.next();
525: // lang
526: memcpy(cur, &fragment.lang, sizeof(fragment.lang)); cur+=sizeof(fragment.lang);
527: // length
528: memcpy(cur, &fragment.length, sizeof(fragment.length)); cur+=sizeof(fragment.length);
529: }
530:
531: // 4: letters
532: body.for_each(serialize_body_piece, &cur);
1.113 parser 533:
1.174 paf 534: return result;
1.175.2.1 paf 535: #endif
536: return String::Cm(0, 0);
1.113 parser 537: }
1.174 paf 538: bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length) {
1.175.2.1 paf 539: #if TODO
1.174 paf 540: if(buf_length<=prolog_length)
1.148 paf 541: return false;
1.174 paf 542: buf_length-=prolog_length;
1.135 paf 543:
1.174 paf 544: // 1: prolog
545: const char* cur=(const char* )buf+prolog_length;
1.113 parser 546:
1.175.2.1 paf 547: // 2: langs.count
548: if(buf_length<sizeof(size_t)) // langs.count don't fit?
1.174 paf 549: return false;
550: size_t fragments_count=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t);
551: buf_length-=sizeof(size_t);
552:
553: if(fragments_count) {
554: // 3: lang info
555: size_t total_length=0;
556: for(size_t f=0; f<fragments_count; f++) {
557: size_t piece_length=sizeof(Language)+sizeof(size_t);
558: if(buf_length<piece_length) // lang+length
559: return false;
560:
561: Language lang=*reinterpret_cast<const Language *>(cur); cur+=sizeof(Language);
562: size_t fragment_length=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t);
1.175.2.1 paf 563: langs+=Fragment(lang, fragment_length);
1.174 paf 564: total_length+=fragment_length;
1.148 paf 565:
1.174 paf 566: buf_length-=piece_length;
567: }
1.128 paf 568:
1.174 paf 569: // 4: letters
570: if(buf_length!=total_length)
1.148 paf 571: return false;
572:
1.174 paf 573: body=StringBody(cur, buf_length);
574: }
1.113 parser 575:
1.174 paf 576: ASSERT_STRING_INVARIANT(*this);
1.148 paf 577: return true;
1.175.2.1 paf 578: #endif
579: return false;
1.113 parser 580: }
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