Annotation of parser3/src/main/pa_string.C, revision 1.172.2.13
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:
1.172.2.13! paf 8: static const char* IDENT_STRING_C="$Date: 2003/02/04 12:08:57 $";
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:
1.172.2.10 paf 24: class String_match_table_template_columns: public Array<StringPtr> {
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
1.1 paf 76:
77: return *this;
78: }
79:
1.16 paf 80: uint String::hash_code() const {
1.7 paf 81: uint result=0;
1.172.2.4 paf 82: STRING_FOREACH_FRAGMENT(
83: result=generic_hash_code(result, fragment->ptr, fragment->size);
1.123 paf 84: );
1.5 paf 85: return result;
86: }
87:
1.60 paf 88: /// @todo move 'lang' skipping to pos
89: int String::cmp(int& partial, const String& src,
90: size_t this_offset, Untaint_lang lang) const {
1.59 paf 91: partial=-1;
1.125 paf 92: size_t a_size=size();
93: this_offset=min(this_offset, a_size-1);
1.55 paf 94:
1.172.2.4 paf 95: const String_fragment *a_current=felements;
96: const String_fragment *b_current=src.felements;
1.55 paf 97: size_t a_offset=this_offset;
98: size_t b_offset=0;
1.172.2.4 paf 99: String_fragment *a_end=felements+fused;
100: String_fragment *b_end=src.felements+src.fused;
1.116 paf 101: int result;
1.60 paf 102: size_t pos=0;
1.33 paf 103:
1.172.2.4 paf 104: bool a_break;
105: bool b_break;
106: while(!(
107: (a_break=(a_current==a_end))
108: || (b_break=(b_current==b_end)))) {
109: if(pos+a_current->size > this_offset) {
1.172.2.6 paf 110: if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang)
1.60 paf 111: return -1; // wrong lang -- bail out
112:
1.55 paf 113: int size_diff=
1.172.2.4 paf 114: (a_current->size-a_offset)-
115: (b_current->size-b_offset);
1.55 paf 116:
117: if(size_diff==0) { // a has same size as b
1.172.2.4 paf 118: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
119: a_current->size-a_offset);
1.55 paf 120: if(result)
121: return result;
1.172.2.4 paf 122: pos+=a_current->size;
123: a_current++; a_offset=0;
124: b_current++; b_offset=0;
125: } else if(size_diff>0) { // a longer
126: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
127: b_current->size-b_offset);
1.55 paf 128: if(result)
129: return result;
1.172.2.4 paf 130: a_offset+=b_current->size-b_offset;
131: b_current++; b_offset=0;
1.55 paf 132: } else { // b longer
1.172.2.4 paf 133: result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset,
134: a_current->size-a_offset);
1.55 paf 135: if(result)
136: return result;
1.172.2.4 paf 137: b_offset+=a_current->size-a_offset;
138: pos+=a_current->size;
139: a_current++; a_offset=0;
1.55 paf 140: }
141: } else {
1.172.2.4 paf 142: a_offset-=a_current->size;
143: pos+=a_current->size;
144: a_current++;
1.9 paf 145: }
1.27 paf 146: }
1.55 paf 147: if(a_break==b_break) { // ended simultaneously
148: partial=0; return 0;
149: } else if(a_break) { // first bytes equal, but a ended before b
150: partial=1; return -1;
151: } else {
152: partial=2; return +1;
153: }
1.27 paf 154: }
155:
1.60 paf 156: /// @todo move 'lang' skipping to pos
1.59 paf 157: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 158: size_t this_offset, Untaint_lang lang) const {
1.59 paf 159: partial=-1;
1.125 paf 160: size_t a_size=size();
1.50 paf 161: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.125 paf 162: this_offset=min(this_offset, a_size-1);
1.27 paf 163:
1.172.2.4 paf 164: const String_fragment *a_current=felements;
1.59 paf 165: size_t a_offset=this_offset;
1.55 paf 166: size_t b_offset=0;
1.172.2.4 paf 167: String_fragment *a_end=felements+fused;
1.60 paf 168: size_t pos=0;
1.52 paf 169:
1.172.2.4 paf 170: bool a_break;
1.83 parser 171: bool b_break=b_size==0;
1.172.2.4 paf 172: while(!(
173: (a_break=(a_current==a_end))
174: || b_break)) {
175: if(pos+a_current->size > this_offset) {
1.172.2.6 paf 176: if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang)
1.60 paf 177: return -1; // wrong lang -- bail out
178:
1.59 paf 179: int size_diff=
1.172.2.4 paf 180: (a_current->size-a_offset)-
1.59 paf 181: (b_size-b_offset);
182:
183: if(size_diff==0) { // a has same size as b
1.172.2.4 paf 184: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
185: a_current->size-a_offset)!=0)
1.59 paf 186: return result;
1.172.2.4 paf 187: pos+=a_current->size;
188: a_current++; a_offset=0;
1.59 paf 189: b_break=true;
190: } else if (size_diff>0) { // a longer
1.172.2.4 paf 191: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
1.59 paf 192: b_size-b_offset)!=0)
193: return result;
194: a_offset+=b_size-b_offset;
195: b_break=true;
196: } else { // b longer
1.172.2.4 paf 197: if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset,
198: a_current->size-a_offset)!=0)
1.59 paf 199: return result;
1.172.2.4 paf 200: b_offset+=a_current->size-a_offset;
201: pos+=a_current->size;
202: a_current++; a_offset=0;
1.59 paf 203: }
204: } else {
1.172.2.4 paf 205: a_offset-=a_current->size;
206: pos+=a_current->size;
207: a_current++;
1.9 paf 208: }
209: }
1.55 paf 210: if(a_break==b_break) { // ended simultaneously
211: partial=0; return 0;
212: } else if(a_break) { // first bytes equal, but a ended before b
213: partial=1; return -1;
214: } else {
215: partial=2; return +1;
216: }
1.5 paf 217: }
1.46 paf 218:
219: #ifndef NO_STRING_ORIGIN
1.172.2.4 paf 220: const String_fragment::Origin& String::origin() const {
1.140 paf 221: if(is_empty()) {
1.172.2.4 paf 222: static const String_fragment::Origin empty_origin={"empty string"};
1.96 parser 223: return empty_origin;
224: }
1.46 paf 225:
1.147 paf 226: // determining origin by first piece or last appended piece
227: // because any of them can be constant=without origin:
1.50 paf 228: // ex: ^load[/file] "document_root" + "/file"
1.80 paf 229: // when last peice is constant,
230: // ex: parser_root_auto_path{dynamic} / auto.p{const}
231: // using first piece
1.172.2.4 paf 232: String_fragment::Origin& first_origin=felements[0].origin;
233: return first_origin.file ? first_origin : felements[fused-1].origin;
1.46 paf 234: }
235: #endif
1.53 paf 236:
1.172.2.3 paf 237: StringPtr String::mid(size_t start, size_t finish) const {
238: StringPtr result(new String());
1.107 parser 239:
1.166 paf 240: start=min(start, size());
1.167 paf 241: finish=max(start, finish);
1.60 paf 242: if(start==finish)
1.107 parser 243: return result;
1.53 paf 244:
245: size_t pos=0;
1.172.2.4 paf 246: STRING_FOREACH_FRAGMENT(
247: size_t item_finish=pos+fragment->size;
1.123 paf 248: if(item_finish > start) { // started now or already?
1.172.2.4 paf 249: bool started=result->is_empty(); // started now?
1.123 paf 250: bool finished=finish <= item_finish; // finished now?
251: size_t offset=started?start-pos:0;
1.172.2.4 paf 252: size_t size=finished?finish-pos:fragment->size;
253: result->APPEND(
254: fragment->ptr+offset, size-offset,
255: fragment->lang,
256: fragment->origin.file, fragment->origin.line);
1.123 paf 257: if(finished)
1.53 paf 258: goto break2;
259: }
1.172.2.4 paf 260: pos+=fragment->size;
1.123 paf 261: );
1.53 paf 262: break2:
1.60 paf 263: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
264: //cstr(), start, finish, result.cstr());
1.53 paf 265: return result;
1.54 paf 266: }
267:
1.60 paf 268: int String::pos(const String& substr,
1.116 paf 269: int result, Untaint_lang lang) const {
1.125 paf 270: size_t self_size=size();
1.131 paf 271: for(; size_t(result)<self_size; result++) {
1.60 paf 272: int partial; cmp(partial, substr, result, lang);
1.58 paf 273: if(
274: partial==0 || // full match
275: partial==2) // 'substr' starts 'this'+'result'
276: return result;
277: }
278:
279: return -1;
280: }
281:
1.172.2.11 paf 282: int String::pos(const char* substr, size_t substr_size,
1.116 paf 283: int result, Untaint_lang lang) const {
1.125 paf 284: size_t self_size=size();
1.131 paf 285: for(; size_t(result)<self_size; result++) {
1.60 paf 286: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 287: if(
288: partial==0 || // full match
289: partial==2) // 'substr' starts 'this'+'result'
290: return result;
291: }
292:
293: return -1;
1.60 paf 294: }
295:
1.172.2.10 paf 296: void String::split(Array<StringPtr>& result,
1.60 paf 297: size_t* pos_after_ref,
1.172.2.11 paf 298: const char* delim, size_t delim_size,
1.172.2.10 paf 299: Untaint_lang lang, int limit) {
1.125 paf 300: size_t self_size=size();
1.60 paf 301: if(delim_size) {
302: size_t pos_after=pos_after_ref?*pos_after_ref:0;
303: int pos_before;
304: // while we have 'delim'...
305: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.172.2.4 paf 306: result+=mid(pos_after, pos_before);
1.60 paf 307: pos_after=pos_before+delim_size;
308: }
309: // last piece
1.124 paf 310: if(pos_after<self_size && limit) {
1.172.2.4 paf 311: result+=mid(pos_after, self_size);
1.124 paf 312: pos_after=self_size;
1.60 paf 313: }
314: if(pos_after_ref)
315: *pos_after_ref=pos_after;
316: } else { // empty delim
1.172.2.10 paf 317: result+=StringPtr(this);
1.60 paf 318: if(pos_after_ref)
1.124 paf 319: *pos_after_ref+=self_size;
1.60 paf 320: }
321: }
322:
1.172.2.10 paf 323: void String::split(Array<StringPtr>& result,
1.60 paf 324: size_t* pos_after_ref,
325: const String& delim, Untaint_lang lang,
1.172.2.10 paf 326: int limit) {
1.140 paf 327: if(!delim.is_empty()) {
1.60 paf 328: size_t pos_after=pos_after_ref?*pos_after_ref:0;
329: int pos_before;
330: // while we have 'delim'...
331: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.172.2.4 paf 332: result+=mid(pos_after, pos_before);
1.60 paf 333: pos_after=pos_before+delim.size();
334: }
335: // last piece
336: if(pos_after<size() && limit) {
1.172.2.4 paf 337: result+=mid(pos_after, size());
1.60 paf 338: pos_after=size();
339: }
340: if(pos_after_ref)
341: *pos_after_ref=pos_after;
342: } else { // empty delim
1.172.2.10 paf 343: result+=StringPtr(this);
1.60 paf 344: if(pos_after_ref)
345: *pos_after_ref+=size();
346: }
1.61 paf 347: }
348:
1.172.2.10 paf 349: static void regex_options(StringPtr options, int *result, bool& need_pre_post_match){
1.63 paf 350: struct Regex_option {
1.172.2.11 paf 351: const char* keyL;
352: const char* keyU;
1.63 paf 353: int clear, set;
354: int *result;
1.154 paf 355: bool *flag;
1.63 paf 356: } regex_option[]={
1.153 paf 357: {"i", "I", 0, PCRE_CASELESS, result}, // a=A
358: {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default]
359: {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
360: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
361: {"g", "G", 0, true, result+1}, // many rows
1.154 paf 362: {"'", 0, 0, 0, 0, &need_pre_post_match},
363: {0}
1.63 paf 364: };
1.171 paf 365: result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY;
1.63 paf 366: result[1]=0;
367:
368: if(options)
1.153 paf 369: for(Regex_option *o=regex_option; o->keyL; o++)
1.154 paf 370: if(options->pos(o->keyL)>=0
371: || (o->keyU && options->pos(o->keyU)>=0)) {
372: if(o->flag)
373: *o->flag=true;
374: else { // result
375: *o->result &= ~o->clear;
376: *o->result |= o->set;
377: }
1.63 paf 378: }
379: }
380:
1.172.2.4 paf 381: TablePtr String::match(Charset& source_charset,
1.172.2.10 paf 382: StringPtr aorigin,
1.172.2.4 paf 383: const String& regexp,
1.172.2.10 paf 384: StringPtr options,
1.172.2.4 paf 385: Row_action row_action, void *info,
386: bool *was_global) const {
1.64 paf 387:
1.140 paf 388: if(regexp.is_empty())
1.149 paf 389: throw Exception(0,
1.73 paf 390: aorigin,
391: "regexp is empty");
1.154 paf 392:
1.172.2.4 paf 393: CharPtr pattern=regexp.cstr();
1.172.2.11 paf 394: const char* errptr;
1.62 paf 395: int erroffset;
1.154 paf 396: bool need_pre_post_match=false;
397: int option_bits[2]; regex_options(options, option_bits, need_pre_post_match);
1.95 parser 398: if(was_global)
399: *was_global=option_bits[1]!=0;
1.172.2.4 paf 400: pcre *code=pcre_compile(pattern.get(), option_bits[0],
1.62 paf 401: &errptr, &erroffset,
1.172.2.1 paf 402: source_charset.pcre_tables);
1.62 paf 403:
1.67 paf 404: if(!code)
1.149 paf 405: throw Exception(0,
1.172.2.4 paf 406: regexp.mid(erroffset, regexp.size()),
1.74 paf 407: "regular expression syntax error - %s", errptr);
1.62 paf 408:
1.63 paf 409: int info_substrings=pcre_info(code, 0, 0);
410: if(info_substrings<0) {
1.100 parser 411: pcre_free(code);
1.149 paf 412: throw Exception(0,
1.73 paf 413: aorigin,
1.76 paf 414: "pcre_info error (%d)",
1.73 paf 415: info_substrings);
1.63 paf 416: }
417:
1.172.2.4 paf 418: CharPtr subject=cstr();
419: int length=size();
420: const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3;
1.155 paf 421: int ovector[ovecsize];
422:
423: // create table
1.172.2.4 paf 424: TablePtr result(new Table(string_match_table_template));
1.63 paf 425:
1.64 paf 426: int exec_option_bits=0;
1.154 paf 427: int prestart=0;
428: int poststart=0;
429: int postfinish=size();
1.63 paf 430: while(true) {
431: int exec_substrings=pcre_exec(code, 0,
1.172.2.4 paf 432: subject.get(), length, prestart,
1.64 paf 433: exec_option_bits, ovector, ovecsize);
1.63 paf 434:
435: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 436: pcre_free(code);
1.172.2.4 paf 437: row_action(*result.get(), 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
438: if(option_bits[1])
439: return result; // global=true+result
440: else
441: return TablePtr(0);// not global=no result
1.63 paf 442: }
443:
444: if(exec_substrings<0) {
1.100 parser 445: pcre_free(code);
1.149 paf 446: throw Exception(0,
1.63 paf 447: aorigin,
1.76 paf 448: "regular expression execute error (%d)",
1.63 paf 449: exec_substrings);
450: }
451:
1.154 paf 452: int prefinish=ovector[0];
453: poststart=ovector[1];
1.172.2.13! paf 454: object_ptr<Array<StringPtr> > row(new Array<StringPtr>);
1.172.2.4 paf 455: if(need_pre_post_match) {
456: *row+=mid(0, prefinish); // .prematch column value
457: *row+=mid(prefinish, poststart); // .match
458: *row+=mid(poststart, postfinish); // .postmatch
459: } else {
1.172.2.10 paf 460: *row+=StringPtr(0); // .prematch column value
461: *row+=StringPtr(0); // .match
462: *row+=StringPtr(0); // .postmatch
1.172.2.4 paf 463: }
1.63 paf 464:
465: for(int i=1; i<exec_substrings; i++) {
1.69 paf 466: // -1:-1 case handled peacefully by mid() itself
1.172.2.4 paf 467: *row+=mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 468: }
469:
1.172.2.4 paf 470: row_action(*result, row.get(), prestart, prefinish, poststart, postfinish, info);
1.63 paf 471:
1.154 paf 472: if(!option_bits[1] || prestart==poststart) { // not global | going to hang
1.100 parser 473: pcre_free(code);
1.172.2.4 paf 474: row_action(*result, 0/*last time, no row*/, 0, 0, poststart, postfinish, info);
475: return result;
1.63 paf 476: }
1.154 paf 477: prestart=poststart;
1.63 paf 478:
479: /*
480: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 481: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 482: */
483: }
1.82 parser 484: }
485:
1.172.2.10 paf 486: StringPtr String::change_case(Pool& pool, Charset& source_charset, Change_case_kind kind) {
1.172.2.4 paf 487: StringPtr result(new String());
488:
1.172.2.1 paf 489: const unsigned char *tables=source_charset.pcre_tables;
1.82 parser 490:
491: const unsigned char *a;
492: const unsigned char *b;
493: switch(kind) {
494: case CC_UPPER:
495: a=tables+lcc_offset;
496: b=tables+fcc_offset;
497: break;
498: case CC_LOWER:
499: a=tables+lcc_offset;
500: b=0;
501: break;
502: default:
1.149 paf 503: throw Exception(0,
1.172.2.10 paf 504: StringPtr(this),
1.82 parser 505: "unknown change case kind #%d",
506: static_cast<int>(kind)); // never
507: a=b=0; // calm, compiler
508: break; // never
509: }
510:
1.172.2.4 paf 511: STRING_FOREACH_FRAGMENT(
512: char *new_cstr=new(pool) char[fragment->size];
1.143 paf 513: char *dest=new_cstr;
1.172.2.11 paf 514: const char* end=fragment->ptr+fragment->size;
515: for(const char* current=fragment->ptr; current<end; current++) {
1.172.2.4 paf 516: unsigned char c=a[(unsigned char)*current];
1.143 paf 517: if(b)
518: c=b[c];
1.82 parser 519:
1.143 paf 520: *dest++=(char)c;
1.82 parser 521: }
1.143 paf 522:
1.172.2.4 paf 523: result->APPEND(new_cstr, fragment->size,
524: fragment->lang,
525: fragment->origin.file, fragment->origin.line);
1.143 paf 526: );
1.89 parser 527:
1.101 parser 528: return result;
529: }
530:
1.150 paf 531: /// @test if in some piece were found no dict words, append it, not it's duplicate
1.172.2.4 paf 532: StringPtr String::replace(Pool& pool, Dictionary& dict) const {
533: StringPtr result(new String());
534: CharPtr lcstr(cstr());
1.172.2.11 paf 535: const char* current=lcstr.get();
1.170 paf 536:
1.172.2.4 paf 537: STRING_FOREACH_FRAGMENT(
1.170 paf 538: IFNDEF_NO_STRING_ORIGIN(
1.172.2.11 paf 539: const char* joined_origin_file=fragment->origin.file;
1.172.2.4 paf 540: const size_t joined_origin_line=fragment->origin.line;
1.170 paf 541: );
1.172.2.4 paf 542: uchar joined_lang=fragment->lang;
1.172.2.11 paf 543: const char* joined_ptr=current;
1.170 paf 544: // calc size
545: size_t joined_size=0;
1.172.2.4 paf 546: STRING_PREPARED_FOREACH_FRAGMENT(
547: if(fragment->lang==joined_lang)
548: joined_size+=fragment->size;
1.170 paf 549: else
550: break; // before non-ours
551: );
552: current+=joined_size;
553:
554: // pointers are after joined piece
1.172.2.4 paf 555: // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT
556: --fragment;
1.170 paf 557:
1.172.2.4 paf 558: char *new_cstr=new(pool) char[(size_t)ceil(joined_size*dict.max_ratio())];
1.123 paf 559: char *dest=new_cstr;
1.170 paf 560: while(joined_size) {
1.172.2.4 paf 561: // there is a fragment where first column starts 'joined_ptr'
562: if(Table::element_type row=dict.first_that_starts(joined_ptr, joined_size)) {
1.123 paf 563: // get a=>b values
1.172.2.10 paf 564: StringPtr a=row->get(0);
565: StringPtr b=row->get(1);
1.170 paf 566: // skip 'a' in 'joined_ptr' && reduce work size
1.172.2.4 paf 567: joined_ptr+=a->size(); joined_size-=a->size();
1.123 paf 568: // write 'b' to 'dest' && skip 'b' in 'dest'
1.172.2.5 paf 569: b->store_to(dest, String::UL_AS_IS); dest+=b->size();
1.123 paf 570: } else {
571: // write a char to b && reduce work size
1.170 paf 572: *dest++=*joined_ptr++; joined_size--;
1.101 parser 573: }
574: }
575:
1.172.2.4 paf 576: result->APPEND(new_cstr, dest-new_cstr, joined_lang,
1.170 paf 577: joined_origin_file, joined_origin_line);
1.156 paf 578: );
1.170 paf 579:
1.156 paf 580: return result;
581: }
582:
1.172.2.4 paf 583: StringPtr String::join_chains(Pool& pool) const {
584: StringPtr result(new String());
585:
586: char *pooled_buf=new(pool) char[cstr_bufsize()];
1.172.2.5 paf 587: store_to(pooled_buf, String::UL_AS_IS);
1.172.2.11 paf 588: const char* current=pooled_buf;
1.156 paf 589:
1.172.2.4 paf 590: STRING_FOREACH_FRAGMENT(
1.156 paf 591: IFNDEF_NO_STRING_ORIGIN(
1.172.2.11 paf 592: const char* joined_origin_file=fragment->origin.file;
1.172.2.4 paf 593: const size_t joined_origin_line=fragment->origin.line;
1.156 paf 594: );
1.172.2.4 paf 595: uchar joined_lang=fragment->lang;
1.172.2.11 paf 596: const char* joined_ptr=current;
1.156 paf 597: // calc size
598: size_t joined_size=0;
1.172.2.4 paf 599: STRING_PREPARED_FOREACH_FRAGMENT(
600: if(fragment->lang==joined_lang)
601: joined_size+=fragment->size;
1.156 paf 602: else
603: break; // before non-ours
604: );
605: current+=joined_size;
606:
607: // pointers are after joined piece
1.172.2.4 paf 608: // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT
609: --fragment;
1.156 paf 610:
1.172.2.4 paf 611: result->APPEND(joined_ptr, joined_size, joined_lang,
1.150 paf 612: joined_origin_file, joined_origin_line);
1.123 paf 613: );
1.156 paf 614:
1.89 parser 615: return result;
616: }
617:
1.172.2.10 paf 618: double String::as_double() {
1.89 parser 619: double result;
1.172.2.4 paf 620: char buf[MAX_STRING];
621: if(size()>MAX_STRING-1)
622: throw Exception("number.format",
1.172.2.10 paf 623: StringPtr(this),
1.172.2.4 paf 624: "invalid number too long a string (%u>%u)", size(), MAX_STRING-1);
1.172.2.5 paf 625: char *eol=store_to(buf, String::UL_AS_IS); *eol=0;
1.172.2.11 paf 626: const char* cstr=buf;
1.172.2.4 paf 627:
1.161 paf 628: while(*cstr && isspace(*cstr))
629: cstr++;
630: if(!*cstr)
1.162 paf 631: return 0;
1.161 paf 632:
1.102 parser 633: char *error_pos;
1.89 parser 634: // 0xABC
1.99 parser 635: if(cstr[0]=='0')
636: if(cstr[1]=='x' || cstr[1]=='X')
637: result=(double)(unsigned long)strtol(cstr, &error_pos, 0);
638: else
1.102 parser 639: result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos);
1.89 parser 640: else
1.99 parser 641: result=(double)strtod(cstr, &error_pos);
1.89 parser 642:
1.159 paf 643: while(char c=*error_pos++)
644: if(!isspace(c))
645: throw Exception("number.format",
1.172.2.10 paf 646: StringPtr(this),
1.159 paf 647: "invalid number (double)");
1.89 parser 648:
649: return result;
650: }
1.172.2.10 paf 651: int String::as_int() {
1.89 parser 652: int result;
1.172.2.4 paf 653: char buf[MAX_STRING];
654: if(size()>MAX_STRING-1)
655: throw Exception("number.format",
1.172.2.10 paf 656: StringPtr(this),
1.172.2.4 paf 657: "invalid number too long a string (%u>%u)", size(), MAX_STRING-1);
1.172.2.5 paf 658: char *eol=store_to(buf, String::UL_AS_IS); *eol=0;
1.172.2.11 paf 659: const char* cstr=buf;
1.172.2.4 paf 660:
1.161 paf 661: while(*cstr && isspace(*cstr))
662: cstr++;
663: if(!*cstr)
1.162 paf 664: return 0;
1.161 paf 665:
1.102 parser 666: char *error_pos;
1.89 parser 667: // 0xABC
1.99 parser 668: if(cstr[0]=='0')
669: if(cstr[1]=='x' || cstr[1]=='X')
670: result=(int)(unsigned long)strtol(cstr, &error_pos, 0);
671: else
1.102 parser 672: result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 673: else
674: result=(int)strtol(cstr, &error_pos, 0);
675:
1.159 paf 676: while(char c=*error_pos++)
677: if(!isspace(c))
678: throw Exception("number.format",
1.172.2.10 paf 679: StringPtr(this),
1.159 paf 680: "invalid number (int)");
1.82 parser 681:
682: return result;
1.61 paf 683: }
1.113 parser 684:
1.172.2.4 paf 685: inline void uint2uchars(uint word, uchar *bytes) {
686: bytes[0]=word&0xFF;
687: bytes[1]=(word>>8)&0xFF;
688: bytes[2]=(word>>16)&0xFF;
689: bytes[3]=(word>>24)&0xFF;
690: }
691: inline uint uchars2uint(uchar *bytes) {
692: return bytes[3]<<24
693: | bytes[2]<<16
694: | bytes[1]<<8
695: | bytes[0];
696: }
697:
698: void String::serialize(Pool& pool, size_t prolog_size, char *& buf, size_t& buf_size) const {
1.113 parser 699: buf_size=
700: prolog_size
1.172.2.4 paf 701: +fused*(sizeof(uchar)+sizeof(size_t))
1.113 parser 702: +size();
1.172.2.4 paf 703: buf=new(pool) char[buf_size];
704: char *cur=buf+prolog_size;
1.113 parser 705:
1.172.2.4 paf 706: STRING_FOREACH_FRAGMENT(
1.123 paf 707: // lang
1.172.2.4 paf 708: memcpy(cur, &fragment->lang, sizeof(fragment->lang));
709: cur+=sizeof(fragment->lang);
1.123 paf 710: // size
1.172.2.4 paf 711: // bug on some sparc platform [you can't work with integers on odd pointers]
712: // forces us to use byte array instead
713: uchar bytes[4];
714: uint2uchars(fragment->size, bytes);
715: memcpy(cur, &bytes, sizeof(bytes)); cur+=sizeof(bytes);
1.123 paf 716: // bytes
1.172.2.4 paf 717: memcpy(cur, fragment->ptr, fragment->size);
718: cur+=fragment->size;
1.123 paf 719: );
1.113 parser 720: }
1.172.2.11 paf 721: bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) {
1.135 paf 722: if(buf_size<=prolog_size)
1.148 paf 723: return false;
1.135 paf 724:
1.126 paf 725: char *cur=(char *)buf+prolog_size;
1.113 parser 726: buf_size-=prolog_size;
727:
728: while(buf_size) {
1.172.2.4 paf 729: if(sizeof(uchar)+sizeof(size_t)>buf_size) // lang+size
1.148 paf 730: return false;
731:
1.172.2.6 paf 732: String_UL lang=*(String_UL *)(cur);
1.172.2.4 paf 733: size_t size=uchars2uint((uchar *)cur);
1.128 paf 734:
1.172.2.4 paf 735: size_t piece_size=sizeof(uchar)+sizeof(size_t)+size;
1.148 paf 736: if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files
737: return false;
738:
1.172.2.11 paf 739: const char* ptr=(const char*)(cur+sizeof(uchar)+sizeof(size_t));
1.126 paf 740: APPEND(ptr, size, lang, file, 0);
1.113 parser 741:
742: cur+=piece_size;
743: buf_size-=piece_size;
744: }
1.148 paf 745: return true;
1.113 parser 746: }
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