Annotation of parser3/src/main/pa_string.C, revision 1.172.2.16
1.45 paf 1: /** @file
1.55 paf 2: Parser: string class. @see untasize_t.C.
1.46 paf 3:
1.172.2.11 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:
1.172.2.16! paf 8: static const char* IDENT_STRING_C="$Date: 2003/02/14 12:59:10 $";
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:
1.172.2.2 paf 18: // helpers
1.139 paf 19:
1.172.2.2 paf 20: /// String::match uses this as replace & global search table columns
1.139 paf 21:
1.172.2.4 paf 22: const int MAX_MATCH_GROUPS=100;
23:
1.172.2.14 paf 24: class String_match_table_template_columns: public ArrayString {
1.172.2.4 paf 25: Pool pool;
1.172.2.2 paf 26: public:
1.172.2.4 paf 27: String_match_table_template_columns() {
1.172.2.10 paf 28: *this+=StringPtr(new String("prematch"));
29: *this+=StringPtr(new String("match"));
30: *this+=StringPtr(new String("postmatch"));
1.172.2.4 paf 31: for(int i=0; i<MAX_MATCH_GROUPS; i++) {
32: char *cname=new(pool) char[3/*strlen("100")*/+1/*terminating 0*/];
1.172.2.10 paf 33: StringPtr sname(new String(cname, sprintf(cname, "%d", 1+i)));
1.172.2.4 paf 34: *this+=sname;
1.172.2.2 paf 35: }
36: }
1.172.2.4 paf 37: };
38:
39: Table string_match_table_template(
1.172.2.13 paf 40: Exception::undefined_source,
1.172.2.4 paf 41: Table::columns_type(new String_match_table_template_columns));
1.172.2.2 paf 42:
43: // methods
44:
1.172.2.11 paf 45: String::String(const char* src, size_t src_size, bool tainted): Array<String_fragment>(1), fsize(0) {
1.41 paf 46: if(src)
1.75 paf 47: if(tainted)
48: APPEND_TAINTED(src, src_size, 0, 0);
1.41 paf 49: else
1.75 paf 50: APPEND_CLEAN(src, src_size, 0, 0);
1.1 paf 51: }
1.140 paf 52:
1.172.2.4 paf 53: String::String(const String& src): Array<String_fragment>(src.count()) {
1.169 paf 54: append(src, UL_PASS_APPENDED);
1.120 paf 55: }
1.28 paf 56:
1.13 paf 57: String& String::real_append(STRING_APPEND_PARAMS) {
1.9 paf 58: if(!src)
59: return *this;
1.26 paf 60: if(!size)
61: size=strlen(src);
62: if(!size)
1.9 paf 63: return *this;
1.122 paf 64:
1.172.2.2 paf 65: if(is_full())
66: expand(fdelta);
1.1 paf 67:
1.172.2.4 paf 68: String_fragment *fragment=&felements[fused++];
69: fragment->ptr=src;
70: fragment->size=size;
71: fragment->lang=lang;
1.13 paf 72: #ifndef NO_STRING_ORIGIN
1.172.2.4 paf 73: fragment->origin.file=file;
74: fragment->origin.line=line;
1.13 paf 75: #endif
1.172.2.16! paf 76:
! 77: fsize+=size;
1.1 paf 78:
79: return *this;
80: }
81:
1.16 paf 82: uint String::hash_code() const {
1.7 paf 83: uint result=0;
1.172.2.4 paf 84: STRING_FOREACH_FRAGMENT(
85: result=generic_hash_code(result, fragment->ptr, fragment->size);
1.123 paf 86: );
1.5 paf 87: return result;
88: }
89:
1.60 paf 90: /// @todo move 'lang' skipping to pos
91: int String::cmp(int& partial, const String& src,
92: size_t this_offset, Untaint_lang lang) const {
1.59 paf 93: partial=-1;
1.125 paf 94: size_t a_size=size();
95: this_offset=min(this_offset, a_size-1);
1.55 paf 96:
1.172.2.4 paf 97: const String_fragment *a_current=felements;
98: const String_fragment *b_current=src.felements;
1.55 paf 99: size_t a_offset=this_offset;
100: size_t b_offset=0;
1.172.2.4 paf 101: String_fragment *a_end=felements+fused;
102: String_fragment *b_end=src.felements+src.fused;
1.116 paf 103: int result;
1.60 paf 104: size_t pos=0;
1.33 paf 105:
1.172.2.4 paf 106: bool a_break;
107: bool b_break;
108: while(!(
109: (a_break=(a_current==a_end))
110: || (b_break=(b_current==b_end)))) {
111: if(pos+a_current->size > this_offset) {
1.172.2.6 paf 112: if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang)
1.60 paf 113: return -1; // wrong lang -- bail out
114:
1.55 paf 115: int size_diff=
1.172.2.4 paf 116: (a_current->size-a_offset)-
117: (b_current->size-b_offset);
1.55 paf 118:
119: if(size_diff==0) { // a has same size as b
1.172.2.4 paf 120: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
121: a_current->size-a_offset);
1.55 paf 122: if(result)
123: return result;
1.172.2.4 paf 124: pos+=a_current->size;
125: a_current++; a_offset=0;
126: b_current++; b_offset=0;
127: } else if(size_diff>0) { // a longer
128: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
129: b_current->size-b_offset);
1.55 paf 130: if(result)
131: return result;
1.172.2.4 paf 132: a_offset+=b_current->size-b_offset;
133: b_current++; b_offset=0;
1.55 paf 134: } else { // b longer
1.172.2.4 paf 135: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
136: a_current->size-a_offset);
1.55 paf 137: if(result)
138: return result;
1.172.2.4 paf 139: b_offset+=a_current->size-a_offset;
140: pos+=a_current->size;
141: a_current++; a_offset=0;
1.55 paf 142: }
143: } else {
1.172.2.4 paf 144: a_offset-=a_current->size;
145: pos+=a_current->size;
146: a_current++;
1.9 paf 147: }
1.27 paf 148: }
1.55 paf 149: if(a_break==b_break) { // ended simultaneously
150: partial=0; return 0;
151: } else if(a_break) { // first bytes equal, but a ended before b
152: partial=1; return -1;
153: } else {
154: partial=2; return +1;
155: }
1.27 paf 156: }
157:
1.60 paf 158: /// @todo move 'lang' skipping to pos
1.59 paf 159: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 160: size_t this_offset, Untaint_lang lang) const {
1.59 paf 161: partial=-1;
1.125 paf 162: size_t a_size=size();
1.50 paf 163: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.125 paf 164: this_offset=min(this_offset, a_size-1);
1.27 paf 165:
1.172.2.4 paf 166: const String_fragment *a_current=felements;
1.59 paf 167: size_t a_offset=this_offset;
1.55 paf 168: size_t b_offset=0;
1.172.2.4 paf 169: String_fragment *a_end=felements+fused;
1.60 paf 170: size_t pos=0;
1.52 paf 171:
1.172.2.4 paf 172: bool a_break;
1.83 parser 173: bool b_break=b_size==0;
1.172.2.4 paf 174: while(!(
175: (a_break=(a_current==a_end))
176: || b_break)) {
177: if(pos+a_current->size > this_offset) {
1.172.2.6 paf 178: if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang)
1.60 paf 179: return -1; // wrong lang -- bail out
180:
1.59 paf 181: int size_diff=
1.172.2.4 paf 182: (a_current->size-a_offset)-
1.59 paf 183: (b_size-b_offset);
184:
185: if(size_diff==0) { // a has same size as b
1.172.2.4 paf 186: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
187: a_current->size-a_offset)!=0)
1.59 paf 188: return result;
1.172.2.4 paf 189: pos+=a_current->size;
190: a_current++; a_offset=0;
1.59 paf 191: b_break=true;
192: } else if (size_diff>0) { // a longer
1.172.2.4 paf 193: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
1.59 paf 194: b_size-b_offset)!=0)
195: return result;
196: a_offset+=b_size-b_offset;
197: b_break=true;
198: } else { // b longer
1.172.2.4 paf 199: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
200: a_current->size-a_offset)!=0)
1.59 paf 201: return result;
1.172.2.4 paf 202: b_offset+=a_current->size-a_offset;
203: pos+=a_current->size;
204: a_current++; a_offset=0;
1.59 paf 205: }
206: } else {
1.172.2.4 paf 207: a_offset-=a_current->size;
208: pos+=a_current->size;
209: a_current++;
1.9 paf 210: }
211: }
1.55 paf 212: if(a_break==b_break) { // ended simultaneously
213: partial=0; return 0;
214: } else if(a_break) { // first bytes equal, but a ended before b
215: partial=1; return -1;
216: } else {
217: partial=2; return +1;
218: }
1.5 paf 219: }
1.46 paf 220:
221: #ifndef NO_STRING_ORIGIN
1.172.2.4 paf 222: const String_fragment::Origin& String::origin() const {
1.140 paf 223: if(is_empty()) {
1.172.2.4 paf 224: static const String_fragment::Origin empty_origin={"empty string"};
1.96 parser 225: return empty_origin;
226: }
1.46 paf 227:
1.147 paf 228: // determining origin by first piece or last appended piece
229: // because any of them can be constant=without origin:
1.50 paf 230: // ex: ^load[/file] "document_root" + "/file"
1.80 paf 231: // when last peice is constant,
232: // ex: parser_root_auto_path{dynamic} / auto.p{const}
233: // using first piece
1.172.2.4 paf 234: String_fragment::Origin& first_origin=felements[0].origin;
235: return first_origin.file ? first_origin : felements[fused-1].origin;
1.46 paf 236: }
237: #endif
1.53 paf 238:
1.172.2.3 paf 239: StringPtr String::mid(size_t start, size_t finish) const {
240: StringPtr result(new String());
1.107 parser 241:
1.166 paf 242: start=min(start, size());
1.167 paf 243: finish=max(start, finish);
1.60 paf 244: if(start==finish)
1.107 parser 245: return result;
1.53 paf 246:
247: size_t pos=0;
1.172.2.4 paf 248: STRING_FOREACH_FRAGMENT(
249: size_t item_finish=pos+fragment->size;
1.123 paf 250: if(item_finish > start) { // started now or already?
1.172.2.4 paf 251: bool started=result->is_empty(); // started now?
1.123 paf 252: bool finished=finish <= item_finish; // finished now?
253: size_t offset=started?start-pos:0;
1.172.2.4 paf 254: size_t size=finished?finish-pos:fragment->size;
255: result->APPEND(
256: fragment->ptr+offset, size-offset,
257: fragment->lang,
258: fragment->origin.file, fragment->origin.line);
1.123 paf 259: if(finished)
1.53 paf 260: goto break2;
261: }
1.172.2.4 paf 262: pos+=fragment->size;
1.123 paf 263: );
1.53 paf 264: break2:
1.60 paf 265: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
266: //cstr(), start, finish, result.cstr());
1.53 paf 267: return result;
1.54 paf 268: }
269:
1.60 paf 270: int String::pos(const String& substr,
1.116 paf 271: int result, Untaint_lang lang) const {
1.125 paf 272: size_t self_size=size();
1.131 paf 273: for(; size_t(result)<self_size; result++) {
1.60 paf 274: int partial; cmp(partial, substr, result, lang);
1.58 paf 275: if(
276: partial==0 || // full match
277: partial==2) // 'substr' starts 'this'+'result'
278: return result;
279: }
280:
281: return -1;
282: }
283:
1.172.2.11 paf 284: int String::pos(const char* substr, size_t substr_size,
1.116 paf 285: int result, Untaint_lang lang) const {
1.125 paf 286: size_t self_size=size();
1.131 paf 287: for(; size_t(result)<self_size; result++) {
1.60 paf 288: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 289: if(
290: partial==0 || // full match
291: partial==2) // 'substr' starts 'this'+'result'
292: return result;
293: }
294:
295: return -1;
1.60 paf 296: }
297:
1.172.2.14 paf 298: void String::split(ArrayString& result,
1.60 paf 299: size_t* pos_after_ref,
1.172.2.11 paf 300: const char* delim, size_t delim_size,
1.172.2.10 paf 301: Untaint_lang lang, int limit) {
1.125 paf 302: size_t self_size=size();
1.60 paf 303: if(delim_size) {
304: size_t pos_after=pos_after_ref?*pos_after_ref:0;
305: int pos_before;
306: // while we have 'delim'...
307: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.172.2.4 paf 308: result+=mid(pos_after, pos_before);
1.60 paf 309: pos_after=pos_before+delim_size;
310: }
311: // last piece
1.124 paf 312: if(pos_after<self_size && limit) {
1.172.2.4 paf 313: result+=mid(pos_after, self_size);
1.124 paf 314: pos_after=self_size;
1.60 paf 315: }
316: if(pos_after_ref)
317: *pos_after_ref=pos_after;
318: } else { // empty delim
1.172.2.10 paf 319: result+=StringPtr(this);
1.60 paf 320: if(pos_after_ref)
1.124 paf 321: *pos_after_ref+=self_size;
1.60 paf 322: }
323: }
324:
1.172.2.14 paf 325: void String::split(ArrayString& result,
1.60 paf 326: size_t* pos_after_ref,
327: const String& delim, Untaint_lang lang,
1.172.2.10 paf 328: int limit) {
1.140 paf 329: if(!delim.is_empty()) {
1.60 paf 330: size_t pos_after=pos_after_ref?*pos_after_ref:0;
331: int pos_before;
332: // while we have 'delim'...
333: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.172.2.4 paf 334: result+=mid(pos_after, pos_before);
1.60 paf 335: pos_after=pos_before+delim.size();
336: }
337: // last piece
338: if(pos_after<size() && limit) {
1.172.2.4 paf 339: result+=mid(pos_after, size());
1.60 paf 340: pos_after=size();
341: }
342: if(pos_after_ref)
343: *pos_after_ref=pos_after;
344: } else { // empty delim
1.172.2.10 paf 345: result+=StringPtr(this);
1.60 paf 346: if(pos_after_ref)
347: *pos_after_ref+=size();
348: }
1.61 paf 349: }
350:
1.172.2.10 paf 351: static void regex_options(StringPtr options, int *result, bool& need_pre_post_match){
1.63 paf 352: struct Regex_option {
1.172.2.11 paf 353: const char* keyL;
354: const char* keyU;
1.63 paf 355: int clear, set;
356: int *result;
1.154 paf 357: bool *flag;
1.63 paf 358: } regex_option[]={
1.153 paf 359: {"i", "I", 0, PCRE_CASELESS, result}, // a=A
360: {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default]
361: {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
362: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
363: {"g", "G", 0, true, result+1}, // many rows
1.154 paf 364: {"'", 0, 0, 0, 0, &need_pre_post_match},
365: {0}
1.63 paf 366: };
1.171 paf 367: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 368: result[1]=0;
369:
370: if(options)
1.153 paf 371: for(Regex_option *o=regex_option; o->keyL; o++)
1.154 paf 372: if(options->pos(o->keyL)>=0
373: || (o->keyU && options->pos(o->keyU)>=0)) {
374: if(o->flag)
375: *o->flag=true;
376: else { // result
377: *o->result &= ~o->clear;
378: *o->result |= o->set;
379: }
1.63 paf 380: }
381: }
382:
1.172.2.4 paf 383: TablePtr String::match(Charset& source_charset,
1.172.2.10 paf 384: StringPtr aorigin,
1.172.2.4 paf 385: const String& regexp,
1.172.2.10 paf 386: StringPtr options,
1.172.2.4 paf 387: Row_action row_action, void *info,
388: bool *was_global) const {
1.64 paf 389:
1.140 paf 390: if(regexp.is_empty())
1.149 paf 391: throw Exception(0,
1.73 paf 392: aorigin,
393: "regexp is empty");
1.154 paf 394:
1.172.2.4 paf 395: CharPtr pattern=regexp.cstr();
1.172.2.11 paf 396: const char* errptr;
1.62 paf 397: int erroffset;
1.154 paf 398: bool need_pre_post_match=false;
399: int option_bits[2]; regex_options(options, option_bits, need_pre_post_match);
1.95 parser 400: if(was_global)
401: *was_global=option_bits[1]!=0;
1.172.2.4 paf 402: pcre *code=pcre_compile(pattern.get(), option_bits[0],
1.62 paf 403: &errptr, &erroffset,
1.172.2.1 paf 404: source_charset.pcre_tables);
1.62 paf 405:
1.67 paf 406: if(!code)
1.149 paf 407: throw Exception(0,
1.172.2.4 paf 408: regexp.mid(erroffset, regexp.size()),
1.74 paf 409: "regular expression syntax error - %s", errptr);
1.62 paf 410:
1.63 paf 411: int info_substrings=pcre_info(code, 0, 0);
412: if(info_substrings<0) {
1.100 parser 413: pcre_free(code);
1.149 paf 414: throw Exception(0,
1.73 paf 415: aorigin,
1.76 paf 416: "pcre_info error (%d)",
1.73 paf 417: info_substrings);
1.63 paf 418: }
419:
1.172.2.4 paf 420: CharPtr subject=cstr();
421: int length=size();
422: const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3;
1.155 paf 423: int ovector[ovecsize];
424:
425: // create table
1.172.2.4 paf 426: TablePtr result(new Table(string_match_table_template));
1.63 paf 427:
1.64 paf 428: int exec_option_bits=0;
1.154 paf 429: int prestart=0;
430: int poststart=0;
431: int postfinish=size();
1.63 paf 432: while(true) {
433: int exec_substrings=pcre_exec(code, 0,
1.172.2.4 paf 434: subject.get(), length, prestart,
1.64 paf 435: exec_option_bits, ovector, ovecsize);
1.63 paf 436:
437: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 438: pcre_free(code);
1.172.2.15 paf 439: row_action(result, ArrayStringPtr(0)/*last time, no raw*/, 0, 0, poststart, postfinish, info);
1.172.2.4 paf 440: if(option_bits[1])
441: return result; // global=true+result
442: else
443: return TablePtr(0);// not global=no result
1.63 paf 444: }
445:
446: if(exec_substrings<0) {
1.100 parser 447: pcre_free(code);
1.149 paf 448: throw Exception(0,
1.63 paf 449: aorigin,
1.76 paf 450: "regular expression execute error (%d)",
1.63 paf 451: exec_substrings);
452: }
453:
1.154 paf 454: int prefinish=ovector[0];
455: poststart=ovector[1];
1.172.2.14 paf 456: object_ptr<ArrayString> row(new ArrayString);
1.172.2.4 paf 457: if(need_pre_post_match) {
458: *row+=mid(0, prefinish); // .prematch column value
459: *row+=mid(prefinish, poststart); // .match
460: *row+=mid(poststart, postfinish); // .postmatch
461: } else {
1.172.2.10 paf 462: *row+=StringPtr(0); // .prematch column value
463: *row+=StringPtr(0); // .match
464: *row+=StringPtr(0); // .postmatch
1.172.2.4 paf 465: }
1.63 paf 466:
467: for(int i=1; i<exec_substrings; i++) {
1.69 paf 468: // -1:-1 case handled peacefully by mid() itself
1.172.2.4 paf 469: *row+=mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 470: }
471:
1.172.2.15 paf 472: row_action(result, row, prestart, prefinish, poststart, postfinish, info);
1.63 paf 473:
1.154 paf 474: if(!option_bits[1] || prestart==poststart) { // not global | going to hang
1.100 parser 475: pcre_free(code);
1.172.2.15 paf 476: row_action(result, ArrayStringPtr(0)/*last time, no row*/, 0, 0, poststart, postfinish, info);
1.172.2.4 paf 477: return result;
1.63 paf 478: }
1.154 paf 479: prestart=poststart;
1.63 paf 480:
481: /*
482: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 483: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 484: */
485: }
1.82 parser 486: }
487:
1.172.2.10 paf 488: StringPtr String::change_case(Pool& pool, Charset& source_charset, Change_case_kind kind) {
1.172.2.4 paf 489: StringPtr result(new String());
490:
1.172.2.1 paf 491: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 492:
493: const unsigned char *a;
494: const unsigned char *b;
495: switch(kind) {
496: case CC_UPPER:
497: a=tables+lcc_offset;
498: b=tables+fcc_offset;
499: break;
500: case CC_LOWER:
501: a=tables+lcc_offset;
502: b=0;
503: break;
504: default:
1.149 paf 505: throw Exception(0,
1.172.2.10 paf 506: StringPtr(this),
1.82 parser 507: "unknown change case kind #%d",
508: static_cast<int>(kind)); // never
509: a=b=0; // calm, compiler
510: break; // never
511: }
512:
1.172.2.4 paf 513: STRING_FOREACH_FRAGMENT(
514: char *new_cstr=new(pool) char[fragment->size];
1.143 paf 515: char *dest=new_cstr;
1.172.2.11 paf 516: const char* end=fragment->ptr+fragment->size;
517: for(const char* current=fragment->ptr; current<end; current++) {
1.172.2.4 paf 518: unsigned char c=a[(unsigned char)*current];
1.143 paf 519: if(b)
520: c=b[c];
1.82 parser 521:
1.143 paf 522: *dest++=(char)c;
1.82 parser 523: }
1.143 paf 524:
1.172.2.4 paf 525: result->APPEND(new_cstr, fragment->size,
526: fragment->lang,
527: fragment->origin.file, fragment->origin.line);
1.143 paf 528: );
1.89 parser 529:
1.101 parser 530: return result;
531: }
532:
1.150 paf 533: /// @test if in some piece were found no dict words, append it, not it's duplicate
1.172.2.4 paf 534: StringPtr String::replace(Pool& pool, Dictionary& dict) const {
535: StringPtr result(new String());
536: CharPtr lcstr(cstr());
1.172.2.11 paf 537: const char* current=lcstr.get();
1.170 paf 538:
1.172.2.4 paf 539: STRING_FOREACH_FRAGMENT(
1.170 paf 540: IFNDEF_NO_STRING_ORIGIN(
1.172.2.11 paf 541: const char* joined_origin_file=fragment->origin.file;
1.172.2.4 paf 542: const size_t joined_origin_line=fragment->origin.line;
1.170 paf 543: );
1.172.2.4 paf 544: uchar joined_lang=fragment->lang;
1.172.2.11 paf 545: const char* joined_ptr=current;
1.170 paf 546: // calc size
547: size_t joined_size=0;
1.172.2.4 paf 548: STRING_PREPARED_FOREACH_FRAGMENT(
549: if(fragment->lang==joined_lang)
550: joined_size+=fragment->size;
1.170 paf 551: else
552: break; // before non-ours
553: );
554: current+=joined_size;
555:
556: // pointers are after joined piece
1.172.2.4 paf 557: // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT
558: --fragment;
1.170 paf 559:
1.172.2.4 paf 560: char *new_cstr=new(pool) char[(size_t)ceil(joined_size*dict.max_ratio())];
1.123 paf 561: char *dest=new_cstr;
1.170 paf 562: while(joined_size) {
1.172.2.4 paf 563: // there is a fragment where first column starts 'joined_ptr'
564: if(Table::element_type row=dict.first_that_starts(joined_ptr, joined_size)) {
1.123 paf 565: // get a=>b values
1.172.2.10 paf 566: StringPtr a=row->get(0);
567: StringPtr b=row->get(1);
1.170 paf 568: // skip 'a' in 'joined_ptr' && reduce work size
1.172.2.4 paf 569: joined_ptr+=a->size(); joined_size-=a->size();
1.123 paf 570: // write 'b' to 'dest' && skip 'b' in 'dest'
1.172.2.5 paf 571: b->store_to(dest, String::UL_AS_IS); dest+=b->size();
1.123 paf 572: } else {
573: // write a char to b && reduce work size
1.170 paf 574: *dest++=*joined_ptr++; joined_size--;
1.101 parser 575: }
576: }
577:
1.172.2.4 paf 578: result->APPEND(new_cstr, dest-new_cstr, joined_lang,
1.170 paf 579: joined_origin_file, joined_origin_line);
1.156 paf 580: );
1.170 paf 581:
1.156 paf 582: return result;
583: }
584:
1.172.2.4 paf 585: StringPtr String::join_chains(Pool& pool) const {
586: StringPtr result(new String());
587:
588: char *pooled_buf=new(pool) char[cstr_bufsize()];
1.172.2.5 paf 589: store_to(pooled_buf, String::UL_AS_IS);
1.172.2.11 paf 590: const char* current=pooled_buf;
1.156 paf 591:
1.172.2.4 paf 592: STRING_FOREACH_FRAGMENT(
1.156 paf 593: IFNDEF_NO_STRING_ORIGIN(
1.172.2.11 paf 594: const char* joined_origin_file=fragment->origin.file;
1.172.2.4 paf 595: const size_t joined_origin_line=fragment->origin.line;
1.156 paf 596: );
1.172.2.4 paf 597: uchar joined_lang=fragment->lang;
1.172.2.11 paf 598: const char* joined_ptr=current;
1.156 paf 599: // calc size
600: size_t joined_size=0;
1.172.2.4 paf 601: STRING_PREPARED_FOREACH_FRAGMENT(
602: if(fragment->lang==joined_lang)
603: joined_size+=fragment->size;
1.156 paf 604: else
605: break; // before non-ours
606: );
607: current+=joined_size;
608:
609: // pointers are after joined piece
1.172.2.4 paf 610: // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT
611: --fragment;
1.156 paf 612:
1.172.2.4 paf 613: result->APPEND(joined_ptr, joined_size, joined_lang,
1.150 paf 614: joined_origin_file, joined_origin_line);
1.123 paf 615: );
1.156 paf 616:
1.89 parser 617: return result;
618: }
619:
1.172.2.10 paf 620: double String::as_double() {
1.89 parser 621: double result;
1.172.2.4 paf 622: char buf[MAX_STRING];
623: if(size()>MAX_STRING-1)
624: throw Exception("number.format",
1.172.2.10 paf 625: StringPtr(this),
1.172.2.4 paf 626: "invalid number too long a string (%u>%u)", size(), MAX_STRING-1);
1.172.2.5 paf 627: char *eol=store_to(buf, String::UL_AS_IS); *eol=0;
1.172.2.11 paf 628: const char* cstr=buf;
1.172.2.4 paf 629:
1.161 paf 630: while(*cstr && isspace(*cstr))
631: cstr++;
632: if(!*cstr)
1.162 paf 633: return 0;
1.161 paf 634:
1.102 parser 635: char *error_pos;
1.89 parser 636: // 0xABC
1.99 parser 637: if(cstr[0]=='0')
638: if(cstr[1]=='x' || cstr[1]=='X')
639: result=(double)(unsigned long)strtol(cstr, &error_pos, 0);
640: else
1.102 parser 641: result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos);
1.89 parser 642: else
1.99 parser 643: result=(double)strtod(cstr, &error_pos);
1.89 parser 644:
1.159 paf 645: while(char c=*error_pos++)
646: if(!isspace(c))
647: throw Exception("number.format",
1.172.2.10 paf 648: StringPtr(this),
1.159 paf 649: "invalid number (double)");
1.89 parser 650:
651: return result;
652: }
1.172.2.10 paf 653: int String::as_int() {
1.89 parser 654: int result;
1.172.2.4 paf 655: char buf[MAX_STRING];
656: if(size()>MAX_STRING-1)
657: throw Exception("number.format",
1.172.2.10 paf 658: StringPtr(this),
1.172.2.4 paf 659: "invalid number too long a string (%u>%u)", size(), MAX_STRING-1);
1.172.2.5 paf 660: char *eol=store_to(buf, String::UL_AS_IS); *eol=0;
1.172.2.11 paf 661: const char* cstr=buf;
1.172.2.4 paf 662:
1.161 paf 663: while(*cstr && isspace(*cstr))
664: cstr++;
665: if(!*cstr)
1.162 paf 666: return 0;
1.161 paf 667:
1.102 parser 668: char *error_pos;
1.89 parser 669: // 0xABC
1.99 parser 670: if(cstr[0]=='0')
671: if(cstr[1]=='x' || cstr[1]=='X')
672: result=(int)(unsigned long)strtol(cstr, &error_pos, 0);
673: else
1.102 parser 674: result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 675: else
676: result=(int)strtol(cstr, &error_pos, 0);
677:
1.159 paf 678: while(char c=*error_pos++)
679: if(!isspace(c))
680: throw Exception("number.format",
1.172.2.10 paf 681: StringPtr(this),
1.159 paf 682: "invalid number (int)");
1.82 parser 683:
684: return result;
1.61 paf 685: }
1.113 parser 686:
1.172.2.4 paf 687: inline void uint2uchars(uint word, uchar *bytes) {
688: bytes[0]=word&0xFF;
689: bytes[1]=(word>>8)&0xFF;
690: bytes[2]=(word>>16)&0xFF;
691: bytes[3]=(word>>24)&0xFF;
692: }
693: inline uint uchars2uint(uchar *bytes) {
694: return bytes[3]<<24
695: | bytes[2]<<16
696: | bytes[1]<<8
697: | bytes[0];
698: }
699:
700: void String::serialize(Pool& pool, size_t prolog_size, char *& buf, size_t& buf_size) const {
1.113 parser 701: buf_size=
702: prolog_size
1.172.2.4 paf 703: +fused*(sizeof(uchar)+sizeof(size_t))
1.113 parser 704: +size();
1.172.2.4 paf 705: buf=new(pool) char[buf_size];
706: char *cur=buf+prolog_size;
1.113 parser 707:
1.172.2.4 paf 708: STRING_FOREACH_FRAGMENT(
1.123 paf 709: // lang
1.172.2.4 paf 710: memcpy(cur, &fragment->lang, sizeof(fragment->lang));
711: cur+=sizeof(fragment->lang);
1.123 paf 712: // size
1.172.2.4 paf 713: // bug on some sparc platform [you can't work with integers on odd pointers]
714: // forces us to use byte array instead
715: uchar bytes[4];
716: uint2uchars(fragment->size, bytes);
717: memcpy(cur, &bytes, sizeof(bytes)); cur+=sizeof(bytes);
1.123 paf 718: // bytes
1.172.2.4 paf 719: memcpy(cur, fragment->ptr, fragment->size);
720: cur+=fragment->size;
1.123 paf 721: );
1.113 parser 722: }
1.172.2.11 paf 723: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) {
1.135 paf 724: if(buf_size<=prolog_size)
1.148 paf 725: return false;
1.135 paf 726:
1.126 paf 727: char *cur=(char *)buf+prolog_size;
1.113 parser 728: buf_size-=prolog_size;
729:
730: while(buf_size) {
1.172.2.4 paf 731: if(sizeof(uchar)+sizeof(size_t)>buf_size) // lang+size
1.148 paf 732: return false;
733:
1.172.2.6 paf 734: String_UL lang=*(String_UL *)(cur);
1.172.2.4 paf 735: size_t size=uchars2uint((uchar *)cur);
1.128 paf 736:
1.172.2.4 paf 737: size_t piece_size=sizeof(uchar)+sizeof(size_t)+size;
1.148 paf 738: if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files
739: return false;
740:
1.172.2.11 paf 741: const char* ptr=(const char*)(cur+sizeof(uchar)+sizeof(size_t));
1.126 paf 742: APPEND(ptr, size, lang, file, 0);
1.113 parser 743:
744: cur+=piece_size;
745: buf_size-=piece_size;
746: }
1.148 paf 747: return true;
1.113 parser 748: }
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