Annotation of parser3/src/main/pa_string.C, revision 1.62
1.45 paf 1: /** @file
1.55 paf 2: Parser: string class. @see untasize_t.C.
1.46 paf 3:
1.36 paf 4: Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
1.46 paf 5:
1.37 paf 6: Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
1.36 paf 7:
1.62 ! paf 8: $Id: pa_string.C,v 1.61 2001/04/03 08:23:08 paf Exp $
1.4 paf 9: */
10:
1.48 paf 11: #include "pa_config_includes.h"
1.1 paf 12:
1.62 ! paf 13: #include <locale.h>
! 14:
1.13 paf 15: #include "pa_pool.h"
1.12 paf 16: #include "pa_string.h"
1.5 paf 17: #include "pa_hash.h"
1.22 paf 18: #include "pa_exception.h"
1.53 paf 19: #include "pa_common.h"
1.60 paf 20: #include "pa_array.h"
21: #include "pa_globals.h"
1.61 paf 22: #include "pa_table.h"
1.62 ! paf 23: #include "pa_threads.h"
! 24:
! 25: #include "pcre.h"
1.60 paf 26:
27: //#include "pa_sapi.h"
1.1 paf 28:
1.62 ! paf 29: // consts
! 30:
! 31: const int MAX_MATCH_COLUMNS=20;
! 32:
1.18 paf 33: // String
34:
1.55 paf 35: String::String(Pool& apool, const char *src, bool tasize_ted) :
1.17 paf 36: Pooled(apool) {
1.28 paf 37: last_chunk=&head;
38: head.count=CR_PREALLOCATED_COUNT;
1.5 paf 39: append_here=head.rows;
1.2 paf 40: head.preallocated_link=0;
1.28 paf 41: link_row=&head.rows[head.count];
1.8 paf 42: fused_rows=fsize=0;
1.41 paf 43:
44: if(src)
1.55 paf 45: if(tasize_ted)
1.41 paf 46: APPEND_TAINTED(src, 0, 0, 0);
47: else
1.53 paf 48: APPEND_CONST(src);
1.1 paf 49: }
50:
51: void String::expand() {
1.55 paf 52: size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100;
1.28 paf 53: last_chunk=static_cast<Chunk *>(
1.55 paf 54: malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *)));
1.28 paf 55: last_chunk->count=new_chunk_count;
56: link_row->link=last_chunk;
57: append_here=last_chunk->rows;
58: link_row=&last_chunk->rows[last_chunk->count];
1.8 paf 59: link_row->link=0;
1.1 paf 60: }
61:
1.40 paf 62: String::String(const String& src) : Pooled(src.pool()) {
1.8 paf 63: head.count=CR_PREALLOCATED_COUNT;
64:
1.55 paf 65: size_t src_used_rows=src.fused_rows;
1.8 paf 66: if(src_used_rows<=head.count) {
1.55 paf 67: // all new rows fit size_to preallocated area
68: size_t curr_chunk_rows=head.count;
1.8 paf 69: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows);
70: append_here=&head.rows[src_used_rows];
71: link_row=&head.rows[curr_chunk_rows];
72: } else {
73: // warning:
1.10 paf 74: // heavily relies on the fact
75: // "preallocated area is the same for all strings"
1.8 paf 76: //
77: // info:
78: // allocating only enough mem to fit src string rows
79: // next append would allocate a new chunk
80: //
1.55 paf 81: // new rows don't fit size_to preallocated area: splitting size_to two chunks
1.8 paf 82: // preallocated chunk src to constructing head
83: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count);
1.55 paf 84: // remaining rows size_to new_chunk
85: size_t curr_chunk_rows=src_used_rows-head.count;
1.8 paf 86: Chunk *new_chunk=static_cast<Chunk *>(
1.55 paf 87: malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *)));
1.8 paf 88: new_chunk->count=curr_chunk_rows;
89: head.preallocated_link=new_chunk;
1.28 paf 90: append_here=link_row=&new_chunk->rows[new_chunk->count];
1.8 paf 91:
92: Chunk *old_chunk=src.head.preallocated_link;
93: Chunk::Row *new_rows=new_chunk->rows;
1.55 paf 94: size_t rows_left_to_copy=new_chunk->count;
1.8 paf 95: while(true) {
1.55 paf 96: size_t old_count=old_chunk->count;
1.8 paf 97: Chunk *next_chunk=old_chunk->rows[old_count].link;
98: if(next_chunk) {
99: // not last source chunk
100: // taking it all
101: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count);
102: new_rows+=old_count;
103: rows_left_to_copy-=old_count;
104:
105: old_chunk=next_chunk;
106: } else {
107: // the last source chunk
108: // taking only those rows of chunk that _left_to_copy
109: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy);
110: break;
111: }
112: }
1.5 paf 113: }
1.8 paf 114: link_row->link=0;
115: fused_rows=src_used_rows;
116: fsize=src.fsize;
1.5 paf 117: }
1.28 paf 118:
1.42 paf 119: String& String::append(const String& src, Untaint_lang lang, bool forced) {
1.60 paf 120: const Chunk *chunk=&src.head;
1.40 paf 121: do {
1.60 paf 122: const Chunk::Row *row=chunk->rows;
123: for(size_t i=0; i<chunk->count; i++, row++) {
124: if(row==src.append_here)
1.40 paf 125: goto break2;
1.60 paf 126:
127: APPEND(row->item.ptr, row->item.size,
128: (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang,
129: row->item.origin.file, row->item.origin.line);
1.40 paf 130: }
131: chunk=row->link;
132: } while(chunk);
133: break2:
1.60 paf 134: return *this;
1.34 paf 135: }
1.60 paf 136:
1.13 paf 137: String& String::real_append(STRING_APPEND_PARAMS) {
1.9 paf 138: if(!src)
139: return *this;
1.26 paf 140: if(!size)
141: size=strlen(src);
142: if(!size)
1.9 paf 143: return *this;
144:
1.1 paf 145: if(chunk_is_full())
146: expand();
147:
148: append_here->item.ptr=src;
1.26 paf 149: fsize+=append_here->item.size=size;
1.52 paf 150: append_here->item.lang=lang;
1.13 paf 151: #ifndef NO_STRING_ORIGIN
1.14 paf 152: append_here->item.origin.file=file;
153: append_here->item.origin.line=line;
1.13 paf 154: #endif
1.8 paf 155: append_here++; fused_rows++;
1.1 paf 156:
157: return *this;
158: }
159:
1.16 paf 160: uint String::hash_code() const {
1.7 paf 161: uint result=0;
1.5 paf 162:
1.16 paf 163: const Chunk *chunk=&head;
1.5 paf 164: do {
1.16 paf 165: const Chunk::Row *row=chunk->rows;
1.55 paf 166: for(size_t i=0; i<chunk->count; i++) {
1.5 paf 167: if(row==append_here)
168: goto break2;
169:
1.6 paf 170: result=Hash::generic_code(result, row->item.ptr, row->item.size);
1.5 paf 171: row++;
172: }
173: chunk=row->link;
174: } while(chunk);
175: break2:
176: return result;
177: }
178:
1.60 paf 179: /// @todo move 'lang' skipping to pos
180: int String::cmp(int& partial, const String& src,
181: size_t this_offset, Untaint_lang lang) const {
1.59 paf 182: partial=-1;
1.55 paf 183: this_offset=min(this_offset, size()-1);
184:
1.16 paf 185: const Chunk *a_chunk=&head;
186: const Chunk *b_chunk=&src.head;
187: const Chunk::Row *a_row=a_chunk->rows;
188: const Chunk::Row *b_row=b_chunk->rows;
1.55 paf 189: size_t a_offset=this_offset;
190: size_t b_offset=0;
1.9 paf 191: Chunk::Row *a_end=append_here;
192: Chunk::Row *b_end=src.append_here;
1.55 paf 193: size_t a_countdown=a_chunk->count;
194: size_t b_countdown=b_chunk->count;
1.9 paf 195: bool a_break=false;
196: bool b_break=false;
1.55 paf 197: size_t result;
1.60 paf 198: size_t pos=0;
199: while(true) {
1.33 paf 200: a_break=a_row==a_end;
201: b_break=b_row==b_end;
202: if(a_break || b_break)
203: break;
204:
1.55 paf 205: if(pos+a_row->item.size > this_offset) {
1.60 paf 206: if(lang!=UL_UNKNOWN && a_row->item.lang!=lang)
207: return -1; // wrong lang -- bail out
208:
1.55 paf 209: int size_diff=
210: (a_row->item.size-a_offset)-
211: (b_row->item.size-b_offset);
212:
213: if(size_diff==0) { // a has same size as b
1.60 paf 214: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
215: a_row->item.size-a_offset);
1.55 paf 216: if(result)
217: return result;
1.60 paf 218: pos+=a_row->item.size;
1.55 paf 219: a_row++; a_countdown--; a_offset=0;
220: b_row++; b_countdown--; b_offset=0;
221: } else if (size_diff>0) { // a longer
1.60 paf 222: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
223: b_row->item.size-b_offset);
1.55 paf 224: if(result)
225: return result;
226: a_offset+=b_row->item.size-b_offset;
227: b_row++; b_countdown--; b_offset=0;
228: } else { // b longer
1.60 paf 229: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
230: a_row->item.size-a_offset);
1.55 paf 231: if(result)
232: return result;
233: b_offset+=a_row->item.size-a_offset;
1.60 paf 234: pos+=a_row->item.size;
1.55 paf 235: a_row++; a_countdown--; a_offset=0;
236: }
1.60 paf 237:
1.55 paf 238: if(!b_countdown) {
239: b_chunk=b_row->link;
240: b_row=b_chunk->rows;
241: b_countdown=b_chunk->count;
242: }
243: } else {
1.60 paf 244: a_offset-=a_row->item.size;
245: pos+=a_row->item.size;
246: a_row++; a_countdown--;
1.9 paf 247: }
248:
1.11 paf 249: if(!a_countdown) {
1.9 paf 250: a_chunk=a_row->link;
251: a_row=a_chunk->rows;
1.11 paf 252: a_countdown=a_chunk->count;
1.9 paf 253: }
1.27 paf 254: }
1.55 paf 255: if(a_break==b_break) { // ended simultaneously
256: partial=0; return 0;
257: } else if(a_break) { // first bytes equal, but a ended before b
258: partial=1; return -1;
259: } else {
260: partial=2; return +1;
261: }
1.27 paf 262: }
263:
1.60 paf 264: /// @todo move 'lang' skipping to pos
1.59 paf 265: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 266: size_t this_offset, Untaint_lang lang) const {
1.59 paf 267: partial=-1;
1.50 paf 268: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.59 paf 269: this_offset=min(this_offset, size()-1);
1.27 paf 270:
271: const Chunk *a_chunk=&head;
272: const Chunk::Row *a_row=a_chunk->rows;
1.59 paf 273: size_t a_offset=this_offset;
1.55 paf 274: size_t b_offset=0;
1.27 paf 275: Chunk::Row *a_end=append_here;
1.55 paf 276: size_t a_countdown=a_chunk->count;
1.27 paf 277: bool a_break=false;
278: bool b_break=false;
1.60 paf 279: size_t pos=0;
280: while(true) {
1.52 paf 281: a_break=a_row==a_end;
282: if(a_break || b_break)
283: break;
284:
1.59 paf 285: if(pos+a_row->item.size > this_offset) {
1.60 paf 286: if(lang!=UL_UNKNOWN && a_row->item.lang!=lang)
287: return -1; // wrong lang -- bail out
288:
1.59 paf 289: int size_diff=
290: (a_row->item.size-a_offset)-
291: (b_size-b_offset);
292:
293: if(size_diff==0) { // a has same size as b
294: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
295: a_row->item.size-a_offset)!=0)
296: return result;
1.60 paf 297: pos+=a_row->item.size;
1.59 paf 298: a_row++; a_countdown--; a_offset=0;
299: b_break=true;
300: } else if (size_diff>0) { // a longer
301: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
302: b_size-b_offset)!=0)
303: return result;
304: a_offset+=b_size-b_offset;
305: b_break=true;
306: } else { // b longer
307: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
308: a_row->item.size-a_offset)!=0)
309: return result;
310: b_offset+=a_row->item.size-a_offset;
1.60 paf 311: pos+=a_row->item.size;
1.59 paf 312: a_row++; a_countdown--; a_offset=0;
313: }
314: } else {
1.60 paf 315: a_offset-=a_row->item.size;
316: pos+=a_row->item.size;
317: a_row++; a_countdown--;
1.27 paf 318: }
319:
320: if(!a_countdown) {
321: a_chunk=a_row->link;
322: a_row=a_chunk->rows;
323: a_countdown=a_chunk->count;
1.9 paf 324: }
325: }
1.55 paf 326: if(a_break==b_break) { // ended simultaneously
327: partial=0; return 0;
328: } else if(a_break) { // first bytes equal, but a ended before b
329: partial=1; return -1;
330: } else {
331: partial=2; return +1;
332: }
1.5 paf 333: }
1.46 paf 334:
335: #ifndef NO_STRING_ORIGIN
336: const Origin& String::origin() const {
337: if(!fused_rows)
338: THROW(0, 0,
1.50 paf 339: 0,
340: "String::origin() of empty string called");
1.46 paf 341:
1.49 paf 342: // determining origin by last appended piece
1.50 paf 343: // because first one frequently constant.
344: // ex: ^load[/file] "document_root" + "/file"
1.49 paf 345: return append_here[-1].item.origin;
1.46 paf 346: }
347: #endif
1.53 paf 348:
349: String& String::piece(size_t start, size_t finish) const {
350: start=max(0, start);
351: finish=min(size(), finish);
1.60 paf 352: if(start==finish)
353: return *empty_string;
1.53 paf 354:
355: String& result=*NEW String(pool());
356:
357: size_t pos=0;
358: const Chunk *chunk=&head;
359: do {
360: const Chunk::Row *row=chunk->rows;
1.55 paf 361: for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) {
1.53 paf 362: if(row==append_here)
363: goto break2;
364:
1.60 paf 365: size_t item_finish=pos+row->item.size;
366: if(item_finish > start) { // started now or already?
367: bool started=result.size()==0; // started now?
368: bool finished=finish <= item_finish; // finished now?
1.53 paf 369: size_t offset=started?start-pos:0;
370: size_t size=finished?finish-pos:row->item.size;
371: result.APPEND(
372: row->item.ptr+offset, size-offset,
373: row->item.lang,
374: row->item.origin.file, row->item.origin.line);
375: if(finished)
376: goto break2;
377: }
378: }
379: chunk=row->link;
380: } while(chunk);
381: break2:
1.60 paf 382: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
383: //cstr(), start, finish, result.cstr());
1.53 paf 384: return result;
1.54 paf 385: }
386:
1.60 paf 387: int String::pos(const String& substr,
388: size_t result, Untaint_lang lang) const {
1.58 paf 389: for(; result<size(); result++) {
1.60 paf 390: int partial; cmp(partial, substr, result, lang);
1.58 paf 391: if(
392: partial==0 || // full match
393: partial==2) // 'substr' starts 'this'+'result'
394: return result;
395: }
396:
397: return -1;
398: }
399:
1.60 paf 400: int String::pos(const char *substr, size_t substr_size,
401: size_t result, Untaint_lang lang) const {
1.57 paf 402: for(; result<size(); result++) {
1.60 paf 403: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 404: if(
405: partial==0 || // full match
406: partial==2) // 'substr' starts 'this'+'result'
407: return result;
408: }
409:
410: return -1;
1.60 paf 411: }
412:
413: void String::split(Array& result,
414: size_t* pos_after_ref,
415: const char *delim, size_t delim_size,
416: Untaint_lang lang, int limit) const {
417: if(delim_size) {
418: size_t pos_after=pos_after_ref?*pos_after_ref:0;
419: int pos_before;
420: // while we have 'delim'...
421: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
422: result+=&piece(pos_after, pos_before);
423: pos_after=pos_before+delim_size;
424: }
425: // last piece
426: if(pos_after<size() && limit) {
427: result+=&piece(pos_after, size());
428: pos_after=size();
429: }
430: if(pos_after_ref)
431: *pos_after_ref=pos_after;
432: } else { // empty delim
433: result+=this;
434: if(pos_after_ref)
435: *pos_after_ref+=size();
436: }
437: }
438:
439: void String::split(Array& result,
440: size_t* pos_after_ref,
441: const String& delim, Untaint_lang lang,
442: int limit) const {
443: if(delim.size()) {
444: size_t pos_after=pos_after_ref?*pos_after_ref:0;
445: int pos_before;
446: // while we have 'delim'...
447: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
448: result+=&piece(pos_after, pos_before);
449: pos_after=pos_before+delim.size();
450: }
451: // last piece
452: if(pos_after<size() && limit) {
453: result+=&piece(pos_after, size());
454: pos_after=size();
455: }
456: if(pos_after_ref)
457: *pos_after_ref=pos_after;
458: } else { // empty delim
459: result+=this;
460: if(pos_after_ref)
461: *pos_after_ref+=size();
462: }
1.61 paf 463: }
464:
1.62 ! paf 465: /// @test setlocale param to auto.p @test pcre_malloc & pcre_free substs
! 466: bool String::match(const String *aorigin,
! 467: const String& regexp,
! 468: const String& options,
! 469: Table **table) const {
! 470: SYNCHRONIZED(true);
! 471: static const unsigned char *tables=0;
! 472: if(!tables) {
! 473: setlocale(LC_CTYPE, "ru");
! 474: tables = pcre_maketables();
! 475: }
! 476: const char *pattern=regexp.cstr();
! 477: const char *errptr;
! 478: int erroffset;
! 479: pcre *code=pcre_compile(pattern, 0,
! 480: &errptr, &erroffset,
! 481: tables);
! 482:
! 483: if(!code)
! 484: THROW(0, 0,
! 485: ®exp.piece(erroffset, regexp.size()),
! 486: errptr);
! 487:
! 488: int ovecsize;
! 489: int *ovector=(int *)malloc(sizeof(int)*(ovecsize=(1/*.match*/+MAX_MATCH_COLUMNS)*3));
! 490: const char *subject=cstr();
! 491: int length=strlen(subject);
! 492: int exec_result=pcre_exec(code, 0,
! 493: subject, length, 0/*startoffset*/,
! 494: 0/*options*/, ovector, ovecsize);
! 495:
! 496: if(exec_result==PCRE_ERROR_NOMATCH) {
! 497: *table=0;
! 498: return false;
! 499: }
! 500:
! 501: if(exec_result<0)
! 502: THROW(0, 0,
! 503: 0,
! 504: "pcre_exec failed");
! 505:
! 506: if(exec_result==0)
! 507: THROW(0, 0,
! 508: aorigin,
! 509: "produced more substrings than maximum handled by Parser, which is %d",
! 510: MAX_MATCH_COLUMNS);
! 511:
! 512: Array& columns=*NEW Array(pool());
! 513: columns+=string_match_name; // .match column name
! 514: Array& row=*NEW Array(pool());
! 515: row+=&piece(ovector[0], ovector[1]); // match column value
! 516:
! 517: for(int i=1; i<exec_result; i++) {
! 518: char *column=(char *)malloc(MAX_NUMBER);
! 519: snprintf(column, MAX_NUMBER, "%d", i);
! 520: columns+=NEW String(pool(), column); // .i column name
! 521: row+=&piece(ovector[i*2+0], ovector[i*2+1]); // .i column value
! 522: }
! 523:
! 524: *table=NEW Table(pool(), aorigin, &columns);
! 525: (**table)+=&row;
! 526: return true;
1.61 paf 527: }
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