Annotation of parser3/src/main/pa_string.C, revision 1.175.2.3
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:
1.175.2.3! paf 8: static const char* IDENT_STRING_C="$Date: 2003/09/23 13:53:04 $";
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: }
1.175.2.3! paf 35: int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int not_c)
1.175.2.1 paf 36: {
37: chr_data d;
38:
39: d.countdown = i+n;
40: d.target = not_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);
141: if(substr_begin==substr_end)
1.107 parser 142: return result;
1.53 paf 143:
1.175.2.1 paf 144: // first: their langs
145: result.langs.append(result.body, langs, substr_begin, substr_end);
146: // next: letters themselves
1.174 paf 147: result.body=body.mid(substr_begin, substr_end-substr_begin);
148:
149: // SAPI::log("piece of '%s' from %d to %d is '%s'",
150: //cstr(), substr_begin, substr_end, result.cstr());
1.175.2.2 paf 151: ASSERT_STRING_INVARIANT(result);
1.53 paf 152: return result;
1.54 paf 153: }
154:
1.175.2.1 paf 155: size_t String::pos(const StringBody substr, size_t this_offset, Language lang) const {
1.174 paf 156: size_t substr_begin=body.pos(substr, this_offset);
1.175.2.3! paf 157: if(substr_begin==CORD_NOT_FOUND || !langs.check_lang(lang, substr_begin, substr.length()))
1.174 paf 158: return STRING_NOT_FOUND;
159:
1.175.2.1 paf 160: return substr_begin;
1.58 paf 161: }
162:
1.174 paf 163: size_t String::pos(const String& substr,
164: size_t this_offset, Language lang) const {
165: return pos(substr.body, this_offset, lang);
1.60 paf 166: }
167:
1.174 paf 168: void String::split(ArrayString& result,
169: size_t& pos_after,
170: const char* delim,
171: Language lang, int limit) const {
172: size_t self_length=length();
173: if(size_t delim_length=strlen(delim)) {
1.60 paf 174: int pos_before;
175: // while we have 'delim'...
1.174 paf 176: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 177: result+=&mid(pos_after, pos_before);
1.174 paf 178: pos_after=pos_before+delim_length;
1.60 paf 179: }
180: // last piece
1.174 paf 181: if(pos_after<self_length && limit) {
182: result+=&mid(pos_after, self_length);
183: pos_after=self_length;
1.60 paf 184: }
185: } else { // empty delim
186: result+=this;
1.174 paf 187: pos_after+=self_length;
1.60 paf 188: }
189: }
190:
1.174 paf 191: void String::split(ArrayString& result,
192: size_t& pos_after,
193: const String& delim, Language lang,
194: int limit) const {
1.140 paf 195: if(!delim.is_empty()) {
1.60 paf 196: int pos_before;
197: // while we have 'delim'...
1.174 paf 198: for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
1.69 paf 199: result+=&mid(pos_after, pos_before);
1.174 paf 200: pos_after=pos_before+delim.length();
1.60 paf 201: }
202: // last piece
1.174 paf 203: if(pos_after<length() && limit) {
204: result+=&mid(pos_after, length());
205: pos_after=length();
1.60 paf 206: }
207: } else { // empty delim
208: result+=this;
1.174 paf 209: pos_after+=length();
1.60 paf 210: }
1.61 paf 211: }
212:
1.174 paf 213: static void regex_options(const String* options, int *result, bool& need_pre_post_match){
1.63 paf 214: struct Regex_option {
1.174 paf 215: const char* keyL;
216: const char* keyU;
1.63 paf 217: int clear, set;
218: int *result;
1.154 paf 219: bool *flag;
1.63 paf 220: } regex_option[]={
1.153 paf 221: {"i", "I", 0, PCRE_CASELESS, result}, // a=A
222: {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default]
223: {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
224: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
225: {"g", "G", 0, true, result+1}, // many rows
1.154 paf 226: {"'", 0, 0, 0, 0, &need_pre_post_match},
227: {0}
1.63 paf 228: };
1.171 paf 229: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 230: result[1]=0;
231:
1.174 paf 232: if(options && !options->is_empty())
1.153 paf 233: for(Regex_option *o=regex_option; o->keyL; o++)
1.174 paf 234: if(options->pos(o->keyL)!=STRING_NOT_FOUND
235: || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) {
1.154 paf 236: if(o->flag)
237: *o->flag=true;
238: else { // result
239: *o->result &= ~o->clear;
240: *o->result |= o->set;
241: }
1.63 paf 242: }
243: }
244:
1.174 paf 245: Table* String::match(Charset& source_charset,
246: const String& regexp,
247: const String* options,
248: Row_action row_action, void *info,
249: bool& just_matched) const {
1.140 paf 250: if(regexp.is_empty())
1.149 paf 251: throw Exception(0,
1.174 paf 252: 0,
1.73 paf 253: "regexp is empty");
1.154 paf 254:
1.174 paf 255: const char* pattern=regexp.cstr();
256: const char* errptr;
1.62 paf 257: int erroffset;
1.173 paf 258: bool need_pre_post_match=false;
1.174 paf 259: int option_bits[2]={0}; regex_options(options, option_bits, need_pre_post_match);
260: bool global=option_bits[1]!=0;
1.63 paf 261: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 262: &errptr, &erroffset,
1.174 paf 263: source_charset.pcre_tables);
1.62 paf 264:
1.67 paf 265: if(!code)
1.149 paf 266: throw Exception(0,
1.174 paf 267: ®exp.mid(erroffset, regexp.length()),
1.74 paf 268: "regular expression syntax error - %s", errptr);
1.62 paf 269:
1.174 paf 270: int subpatterns=pcre_info(code, 0, 0);
271: if(subpatterns<0) {
1.100 parser 272: pcre_free(code);
1.149 paf 273: throw Exception(0,
1.174 paf 274: ®exp,
1.76 paf 275: "pcre_info error (%d)",
1.174 paf 276: subpatterns);
1.63 paf 277: }
278:
1.174 paf 279: const char* subject=cstr();
280: size_t subject_length=strlen(subject);
281: const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3;
282: int ovector[oveclength];
1.155 paf 283:
284: // create table
1.173 paf 285: Table::Action_options table_options;
1.174 paf 286: Table& table=*new Table(string_match_table_template, table_options);
1.63 paf 287:
1.64 paf 288: int exec_option_bits=0;
1.154 paf 289: int prestart=0;
290: int poststart=0;
1.174 paf 291: int postfinish=length();
1.63 paf 292: while(true) {
293: int exec_substrings=pcre_exec(code, 0,
1.174 paf 294: subject, subject_length, prestart,
295: exec_option_bits, ovector, oveclength);
1.63 paf 296:
297: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 298: pcre_free(code);
1.174 paf 299: row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
300: if(global || subpatterns)
301: return &table; // global or with subpatterns=true+result
302: else {
303: just_matched=false; return 0; // not global=no result
304: }
1.63 paf 305: }
306:
307: if(exec_substrings<0) {
1.100 parser 308: pcre_free(code);
1.149 paf 309: throw Exception(0,
1.174 paf 310: ®exp,
1.76 paf 311: "regular expression execute error (%d)",
1.63 paf 312: exec_substrings);
313: }
314:
1.154 paf 315: int prefinish=ovector[0];
316: poststart=ovector[1];
1.174 paf 317: ArrayString* row=new ArrayString;
318: if(need_pre_post_match) {
319: *row+=&mid(0, prefinish); // .prematch column value
320: *row+=&mid(prefinish, poststart); // .match
321: *row+=&mid(poststart, postfinish); // .postmatch
322: } else {
323: *row+=0; // .prematch column value
324: *row+=0; // .match
325: *row+=0; // .postmatch
326: }
1.63 paf 327:
328: for(int i=1; i<exec_substrings; i++) {
1.69 paf 329: // -1:-1 case handled peacefully by mid() itself
1.174 paf 330: *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 331: }
332:
1.174 paf 333: row_action(table, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 334:
1.174 paf 335: if(!global || prestart==poststart) { // not global | going to hang
1.100 parser 336: pcre_free(code);
1.174 paf 337: row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
338: return &table;
1.63 paf 339: }
1.154 paf 340: prestart=poststart;
1.63 paf 341:
342: /*
343: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 344: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 345: */
346: }
1.82 parser 347: }
348:
1.174 paf 349: String& String::change_case(Charset& source_charset, Change_case_kind kind) const {
350: String& result=*new String();
351: if(is_empty())
352: return result;
353:
354: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 355:
356: const unsigned char *a;
357: const unsigned char *b;
358: switch(kind) {
359: case CC_UPPER:
360: a=tables+lcc_offset;
361: b=tables+fcc_offset;
362: break;
363: case CC_LOWER:
364: a=tables+lcc_offset;
365: b=0;
366: break;
367: default:
1.149 paf 368: throw Exception(0,
1.82 parser 369: this,
370: "unknown change case kind #%d",
371: static_cast<int>(kind)); // never
372: a=b=0; // calm, compiler
373: break; // never
374: }
375:
1.174 paf 376: char* new_cstr=cstrm();
377: char *dest=new_cstr;
378: unsigned char index;
379: for(const char* current=new_cstr; index=(unsigned char)*current; current++) {
380: unsigned char c=a[index];
381: if(b)
382: c=b[c];
1.82 parser 383:
1.174 paf 384: *dest++=(char)c;
385: }
1.175.2.1 paf 386: result.langs=langs;
1.174 paf 387: result.body=new_cstr;
1.89 parser 388:
1.101 parser 389: return result;
390: }
391:
1.174 paf 392: const String& String::replace(const Dictionary& dict) const {
393: String& result=*new String();
394: const char* old_cstr=cstr();
395: const char* prematch_begin=old_cstr;
396:
397: const char* current=old_cstr;
398: while(*current) {
399: if(Table::element_type row=dict.first_that_begins(current)) {
400: // prematch
401: if(size_t prematch_length=current-prematch_begin) {
1.175.2.1 paf 402: result.langs.append(result.body, langs, prematch_begin-old_cstr, current-old_cstr);
1.174 paf 403: result.body.append_strdup_know_length(prematch_begin, prematch_length);
1.101 parser 404: }
405:
1.174 paf 406: // match
1.170 paf 407:
1.174 paf 408: const String* a=row->get(0);
409: // skip 'a' in 'current'; move prematch_begin
410: current+=a->length(); prematch_begin=current;
411:
412: if(row->count()>1) { // are there any b?
413: const String* b=row->get(1);
414: result<<*b;
415: }
416: } else // simply advance
417: current++;
418: }
1.156 paf 419:
1.174 paf 420: // postmatch
421: if(size_t postmatch_length=current-prematch_begin) {
1.175.2.1 paf 422: result.langs.append(result.body, langs, prematch_begin-old_cstr, current-old_cstr);
1.174 paf 423: result.body.append_strdup_know_length(prematch_begin, postmatch_length);
424: }
1.156 paf 425:
1.175.2.2 paf 426: ASSERT_STRING_INVARIANT(result);
1.89 parser 427: return result;
428: }
429:
1.90 parser 430: double String::as_double() const {
1.89 parser 431: double result;
1.174 paf 432: const char *str=cstr();
433:
434: while(*str && isspace(*str))
435: str++;
436: if(!*str)
1.162 paf 437: return 0;
1.161 paf 438:
1.102 parser 439: char *error_pos;
1.89 parser 440: // 0xABC
1.174 paf 441: if(str[0]=='0')
442: if(str[1]=='x' || str[1]=='X')
443: result=(double)(unsigned long)strtol(str, &error_pos, 0);
1.99 parser 444: else
1.174 paf 445: result=(double)strtod(str+1/*skip leading 0*/, &error_pos);
1.89 parser 446: else
1.174 paf 447: result=(double)strtod(str, &error_pos);
1.89 parser 448:
1.159 paf 449: while(char c=*error_pos++)
450: if(!isspace(c))
451: throw Exception("number.format",
452: this,
453: "invalid number (double)");
1.89 parser 454:
455: return result;
456: }
1.90 parser 457: int String::as_int() const {
1.89 parser 458: int result;
1.174 paf 459: const char *str=cstr();
460:
461: while(*str && isspace(*str))
462: str++;
463: if(!*str)
1.162 paf 464: return 0;
1.161 paf 465:
1.102 parser 466: char *error_pos;
1.89 parser 467: // 0xABC
1.174 paf 468: if(str[0]=='0')
469: if(str[1]=='x' || str[1]=='X')
470: result=(int)(unsigned long)strtol(str, &error_pos, 0);
1.99 parser 471: else
1.174 paf 472: result=(int)strtol(str+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 473: else
1.174 paf 474: result=(int)strtol(str, &error_pos, 0);
1.89 parser 475:
1.159 paf 476: while(char c=*error_pos++)
477: if(!isspace(c))
478: throw Exception("number.format",
479: this,
480: "invalid number (int)");
1.82 parser 481:
482: return result;
1.61 paf 483: }
1.113 parser 484:
1.174 paf 485: inline void uint2uchars(uint word, uchar *bytes) {
486: bytes[0]=word&0xFF;
487: bytes[1]=(word>>8)&0xFF;
488: bytes[2]=(word>>16)&0xFF;
489: bytes[3]=(word>>24)&0xFF;
490: }
491: inline uint uchars2uint(uchar *bytes) {
492: return bytes[3]<<24
493: | bytes[2]<<16
494: | bytes[1]<<8
495: | bytes[0];
496: }
497:
498: static int serialize_body_piece(const char* s, char** cur) {
499: size_t length=strlen(s);
500: memcpy(*cur, s, length); *cur+=length;
501: return 0;
502: };
503: String::Cm String::serialize(size_t prolog_length) const {
1.175.2.1 paf 504: #if TODO
1.175 paf 505: //_asm int 3;
1.174 paf 506: size_t buf_length=
507: prolog_length
1.175 paf 508: +sizeof(size_t)
1.175.2.1 paf 509: +langs.count()*(sizeof(Language)+sizeof(size_t))
1.174 paf 510: +length();
511: String::Cm result(new(PointerFreeGC) char[buf_length], buf_length);
512:
513: // 1: prolog
514: char *cur=result.str+prolog_length;
515:
516:
1.175.2.1 paf 517: // 2: langs.count
518: size_t fragments_count=langs.count();
1.174 paf 519: memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count);
520:
521: // 3: lang info
1.175.2.1 paf 522: for(Array_iterator<ArrayFragment::element_type> i(langs); i.has_next(); ) {
1.174 paf 523: const Fragment fragment=i.next();
524: // lang
525: memcpy(cur, &fragment.lang, sizeof(fragment.lang)); cur+=sizeof(fragment.lang);
526: // length
527: memcpy(cur, &fragment.length, sizeof(fragment.length)); cur+=sizeof(fragment.length);
528: }
529:
530: // 4: letters
531: body.for_each(serialize_body_piece, &cur);
1.113 parser 532:
1.174 paf 533: return result;
1.175.2.1 paf 534: #endif
535: return String::Cm(0, 0);
1.113 parser 536: }
1.174 paf 537: bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length) {
1.175.2.1 paf 538: #if TODO
1.174 paf 539: if(buf_length<=prolog_length)
1.148 paf 540: return false;
1.174 paf 541: buf_length-=prolog_length;
1.135 paf 542:
1.174 paf 543: // 1: prolog
544: const char* cur=(const char* )buf+prolog_length;
1.113 parser 545:
1.175.2.1 paf 546: // 2: langs.count
547: if(buf_length<sizeof(size_t)) // langs.count don't fit?
1.174 paf 548: return false;
549: size_t fragments_count=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t);
550: buf_length-=sizeof(size_t);
551:
552: if(fragments_count) {
553: // 3: lang info
554: size_t total_length=0;
555: for(size_t f=0; f<fragments_count; f++) {
556: size_t piece_length=sizeof(Language)+sizeof(size_t);
557: if(buf_length<piece_length) // lang+length
558: return false;
559:
560: Language lang=*reinterpret_cast<const Language *>(cur); cur+=sizeof(Language);
561: size_t fragment_length=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t);
1.175.2.1 paf 562: langs+=Fragment(lang, fragment_length);
1.174 paf 563: total_length+=fragment_length;
1.148 paf 564:
1.174 paf 565: buf_length-=piece_length;
566: }
1.128 paf 567:
1.174 paf 568: // 4: letters
569: if(buf_length!=total_length)
1.148 paf 570: return false;
571:
1.174 paf 572: body=StringBody(cur, buf_length);
573: }
1.113 parser 574:
1.174 paf 575: ASSERT_STRING_INVARIANT(*this);
1.148 paf 576: return true;
1.175.2.1 paf 577: #endif
578: return false;
1.113 parser 579: }
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