Annotation of parser3/src/main/pa_string.C, revision 1.71
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:
1.71 ! paf 8: $Id: pa_string.C,v 1.70 2001/04/05 13:19:43 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:
1.70 paf 15: #include "pcre.h"
16:
1.13 paf 17: #include "pa_pool.h"
1.12 paf 18: #include "pa_string.h"
1.5 paf 19: #include "pa_hash.h"
1.22 paf 20: #include "pa_exception.h"
1.53 paf 21: #include "pa_common.h"
1.60 paf 22: #include "pa_array.h"
23: #include "pa_globals.h"
1.61 paf 24: #include "pa_table.h"
1.62 paf 25: #include "pa_threads.h"
1.60 paf 26:
27: //#include "pa_sapi.h"
1.1 paf 28:
1.18 paf 29: // String
30:
1.55 paf 31: String::String(Pool& apool, const char *src, bool tasize_ted) :
1.17 paf 32: Pooled(apool) {
1.28 paf 33: last_chunk=&head;
34: head.count=CR_PREALLOCATED_COUNT;
1.5 paf 35: append_here=head.rows;
1.2 paf 36: head.preallocated_link=0;
1.28 paf 37: link_row=&head.rows[head.count];
1.8 paf 38: fused_rows=fsize=0;
1.41 paf 39:
40: if(src)
1.55 paf 41: if(tasize_ted)
1.41 paf 42: APPEND_TAINTED(src, 0, 0, 0);
43: else
1.53 paf 44: APPEND_CONST(src);
1.1 paf 45: }
46:
47: void String::expand() {
1.55 paf 48: size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100;
1.28 paf 49: last_chunk=static_cast<Chunk *>(
1.55 paf 50: malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *)));
1.28 paf 51: last_chunk->count=new_chunk_count;
52: link_row->link=last_chunk;
53: append_here=last_chunk->rows;
54: link_row=&last_chunk->rows[last_chunk->count];
1.8 paf 55: link_row->link=0;
1.1 paf 56: }
57:
1.40 paf 58: String::String(const String& src) : Pooled(src.pool()) {
1.8 paf 59: head.count=CR_PREALLOCATED_COUNT;
60:
1.55 paf 61: size_t src_used_rows=src.fused_rows;
1.8 paf 62: if(src_used_rows<=head.count) {
1.55 paf 63: // all new rows fit size_to preallocated area
64: size_t curr_chunk_rows=head.count;
1.8 paf 65: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows);
66: append_here=&head.rows[src_used_rows];
67: link_row=&head.rows[curr_chunk_rows];
68: } else {
69: // warning:
1.10 paf 70: // heavily relies on the fact
71: // "preallocated area is the same for all strings"
1.8 paf 72: //
73: // info:
74: // allocating only enough mem to fit src string rows
75: // next append would allocate a new chunk
76: //
1.55 paf 77: // new rows don't fit size_to preallocated area: splitting size_to two chunks
1.8 paf 78: // preallocated chunk src to constructing head
79: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count);
1.55 paf 80: // remaining rows size_to new_chunk
81: size_t curr_chunk_rows=src_used_rows-head.count;
1.8 paf 82: Chunk *new_chunk=static_cast<Chunk *>(
1.55 paf 83: malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *)));
1.8 paf 84: new_chunk->count=curr_chunk_rows;
85: head.preallocated_link=new_chunk;
1.28 paf 86: append_here=link_row=&new_chunk->rows[new_chunk->count];
1.8 paf 87:
88: Chunk *old_chunk=src.head.preallocated_link;
89: Chunk::Row *new_rows=new_chunk->rows;
1.55 paf 90: size_t rows_left_to_copy=new_chunk->count;
1.8 paf 91: while(true) {
1.55 paf 92: size_t old_count=old_chunk->count;
1.8 paf 93: Chunk *next_chunk=old_chunk->rows[old_count].link;
94: if(next_chunk) {
95: // not last source chunk
96: // taking it all
97: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count);
98: new_rows+=old_count;
99: rows_left_to_copy-=old_count;
100:
101: old_chunk=next_chunk;
102: } else {
103: // the last source chunk
104: // taking only those rows of chunk that _left_to_copy
105: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy);
106: break;
107: }
108: }
1.5 paf 109: }
1.8 paf 110: link_row->link=0;
111: fused_rows=src_used_rows;
112: fsize=src.fsize;
1.5 paf 113: }
1.28 paf 114:
1.42 paf 115: String& String::append(const String& src, Untaint_lang lang, bool forced) {
1.60 paf 116: const Chunk *chunk=&src.head;
1.40 paf 117: do {
1.60 paf 118: const Chunk::Row *row=chunk->rows;
119: for(size_t i=0; i<chunk->count; i++, row++) {
120: if(row==src.append_here)
1.40 paf 121: goto break2;
1.60 paf 122:
123: APPEND(row->item.ptr, row->item.size,
124: (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang,
125: row->item.origin.file, row->item.origin.line);
1.40 paf 126: }
127: chunk=row->link;
128: } while(chunk);
129: break2:
1.60 paf 130: return *this;
1.34 paf 131: }
1.60 paf 132:
1.13 paf 133: String& String::real_append(STRING_APPEND_PARAMS) {
1.9 paf 134: if(!src)
135: return *this;
1.26 paf 136: if(!size)
137: size=strlen(src);
138: if(!size)
1.9 paf 139: return *this;
140:
1.1 paf 141: if(chunk_is_full())
142: expand();
143:
144: append_here->item.ptr=src;
1.26 paf 145: fsize+=append_here->item.size=size;
1.52 paf 146: append_here->item.lang=lang;
1.13 paf 147: #ifndef NO_STRING_ORIGIN
1.14 paf 148: append_here->item.origin.file=file;
149: append_here->item.origin.line=line;
1.13 paf 150: #endif
1.8 paf 151: append_here++; fused_rows++;
1.1 paf 152:
153: return *this;
154: }
155:
1.16 paf 156: uint String::hash_code() const {
1.7 paf 157: uint result=0;
1.5 paf 158:
1.16 paf 159: const Chunk *chunk=&head;
1.5 paf 160: do {
1.16 paf 161: const Chunk::Row *row=chunk->rows;
1.55 paf 162: for(size_t i=0; i<chunk->count; i++) {
1.5 paf 163: if(row==append_here)
164: goto break2;
165:
1.6 paf 166: result=Hash::generic_code(result, row->item.ptr, row->item.size);
1.5 paf 167: row++;
168: }
169: chunk=row->link;
170: } while(chunk);
171: break2:
172: return result;
173: }
174:
1.60 paf 175: /// @todo move 'lang' skipping to pos
176: int String::cmp(int& partial, const String& src,
177: size_t this_offset, Untaint_lang lang) const {
1.59 paf 178: partial=-1;
1.55 paf 179: this_offset=min(this_offset, size()-1);
180:
1.16 paf 181: const Chunk *a_chunk=&head;
182: const Chunk *b_chunk=&src.head;
183: const Chunk::Row *a_row=a_chunk->rows;
184: const Chunk::Row *b_row=b_chunk->rows;
1.55 paf 185: size_t a_offset=this_offset;
186: size_t b_offset=0;
1.9 paf 187: Chunk::Row *a_end=append_here;
188: Chunk::Row *b_end=src.append_here;
1.55 paf 189: size_t a_countdown=a_chunk->count;
190: size_t b_countdown=b_chunk->count;
1.9 paf 191: bool a_break=false;
192: bool b_break=false;
1.55 paf 193: size_t result;
1.60 paf 194: size_t pos=0;
195: while(true) {
1.33 paf 196: a_break=a_row==a_end;
197: b_break=b_row==b_end;
198: if(a_break || b_break)
199: break;
200:
1.55 paf 201: if(pos+a_row->item.size > this_offset) {
1.71 ! paf 202: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 203: return -1; // wrong lang -- bail out
204:
1.55 paf 205: int size_diff=
206: (a_row->item.size-a_offset)-
207: (b_row->item.size-b_offset);
208:
209: if(size_diff==0) { // a has same size as b
1.60 paf 210: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
211: a_row->item.size-a_offset);
1.55 paf 212: if(result)
213: return result;
1.60 paf 214: pos+=a_row->item.size;
1.55 paf 215: a_row++; a_countdown--; a_offset=0;
216: b_row++; b_countdown--; b_offset=0;
217: } else if (size_diff>0) { // a longer
1.60 paf 218: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
219: b_row->item.size-b_offset);
1.55 paf 220: if(result)
221: return result;
222: a_offset+=b_row->item.size-b_offset;
223: b_row++; b_countdown--; b_offset=0;
224: } else { // b longer
1.60 paf 225: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
226: a_row->item.size-a_offset);
1.55 paf 227: if(result)
228: return result;
229: b_offset+=a_row->item.size-a_offset;
1.60 paf 230: pos+=a_row->item.size;
1.55 paf 231: a_row++; a_countdown--; a_offset=0;
232: }
1.60 paf 233:
1.55 paf 234: if(!b_countdown) {
235: b_chunk=b_row->link;
236: b_row=b_chunk->rows;
237: b_countdown=b_chunk->count;
238: }
239: } else {
1.60 paf 240: a_offset-=a_row->item.size;
241: pos+=a_row->item.size;
242: a_row++; a_countdown--;
1.9 paf 243: }
244:
1.11 paf 245: if(!a_countdown) {
1.9 paf 246: a_chunk=a_row->link;
247: a_row=a_chunk->rows;
1.11 paf 248: a_countdown=a_chunk->count;
1.9 paf 249: }
1.27 paf 250: }
1.55 paf 251: if(a_break==b_break) { // ended simultaneously
252: partial=0; return 0;
253: } else if(a_break) { // first bytes equal, but a ended before b
254: partial=1; return -1;
255: } else {
256: partial=2; return +1;
257: }
1.27 paf 258: }
259:
1.60 paf 260: /// @todo move 'lang' skipping to pos
1.59 paf 261: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 262: size_t this_offset, Untaint_lang lang) const {
1.59 paf 263: partial=-1;
1.50 paf 264: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.59 paf 265: this_offset=min(this_offset, size()-1);
1.27 paf 266:
267: const Chunk *a_chunk=&head;
268: const Chunk::Row *a_row=a_chunk->rows;
1.59 paf 269: size_t a_offset=this_offset;
1.55 paf 270: size_t b_offset=0;
1.27 paf 271: Chunk::Row *a_end=append_here;
1.55 paf 272: size_t a_countdown=a_chunk->count;
1.27 paf 273: bool a_break=false;
274: bool b_break=false;
1.60 paf 275: size_t pos=0;
276: while(true) {
1.52 paf 277: a_break=a_row==a_end;
278: if(a_break || b_break)
279: break;
280:
1.59 paf 281: if(pos+a_row->item.size > this_offset) {
1.71 ! paf 282: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 283: return -1; // wrong lang -- bail out
284:
1.59 paf 285: int size_diff=
286: (a_row->item.size-a_offset)-
287: (b_size-b_offset);
288:
289: if(size_diff==0) { // a has same size as b
290: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
291: a_row->item.size-a_offset)!=0)
292: return result;
1.60 paf 293: pos+=a_row->item.size;
1.59 paf 294: a_row++; a_countdown--; a_offset=0;
295: b_break=true;
296: } else if (size_diff>0) { // a longer
297: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
298: b_size-b_offset)!=0)
299: return result;
300: a_offset+=b_size-b_offset;
301: b_break=true;
302: } else { // b longer
303: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
304: a_row->item.size-a_offset)!=0)
305: return result;
306: b_offset+=a_row->item.size-a_offset;
1.60 paf 307: pos+=a_row->item.size;
1.59 paf 308: a_row++; a_countdown--; a_offset=0;
309: }
310: } else {
1.60 paf 311: a_offset-=a_row->item.size;
312: pos+=a_row->item.size;
313: a_row++; a_countdown--;
1.27 paf 314: }
315:
316: if(!a_countdown) {
317: a_chunk=a_row->link;
318: a_row=a_chunk->rows;
319: a_countdown=a_chunk->count;
1.9 paf 320: }
321: }
1.55 paf 322: if(a_break==b_break) { // ended simultaneously
323: partial=0; return 0;
324: } else if(a_break) { // first bytes equal, but a ended before b
325: partial=1; return -1;
326: } else {
327: partial=2; return +1;
328: }
1.5 paf 329: }
1.46 paf 330:
331: #ifndef NO_STRING_ORIGIN
332: const Origin& String::origin() const {
333: if(!fused_rows)
334: THROW(0, 0,
1.50 paf 335: 0,
336: "String::origin() of empty string called");
1.46 paf 337:
1.49 paf 338: // determining origin by last appended piece
1.50 paf 339: // because first one frequently constant.
340: // ex: ^load[/file] "document_root" + "/file"
1.49 paf 341: return append_here[-1].item.origin;
1.46 paf 342: }
343: #endif
1.53 paf 344:
1.69 paf 345: String& String::mid(size_t start, size_t finish) const {
1.53 paf 346: start=max(0, start);
347: finish=min(size(), finish);
1.60 paf 348: if(start==finish)
349: return *empty_string;
1.53 paf 350:
351: String& result=*NEW String(pool());
352:
353: size_t pos=0;
354: const Chunk *chunk=&head;
355: do {
356: const Chunk::Row *row=chunk->rows;
1.55 paf 357: for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) {
1.53 paf 358: if(row==append_here)
359: goto break2;
360:
1.60 paf 361: size_t item_finish=pos+row->item.size;
362: if(item_finish > start) { // started now or already?
363: bool started=result.size()==0; // started now?
364: bool finished=finish <= item_finish; // finished now?
1.53 paf 365: size_t offset=started?start-pos:0;
366: size_t size=finished?finish-pos:row->item.size;
367: result.APPEND(
368: row->item.ptr+offset, size-offset,
369: row->item.lang,
370: row->item.origin.file, row->item.origin.line);
371: if(finished)
372: goto break2;
373: }
374: }
375: chunk=row->link;
376: } while(chunk);
377: break2:
1.60 paf 378: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
379: //cstr(), start, finish, result.cstr());
1.53 paf 380: return result;
1.54 paf 381: }
382:
1.60 paf 383: int String::pos(const String& substr,
384: size_t result, Untaint_lang lang) const {
1.58 paf 385: for(; result<size(); result++) {
1.60 paf 386: int partial; cmp(partial, substr, result, lang);
1.58 paf 387: if(
388: partial==0 || // full match
389: partial==2) // 'substr' starts 'this'+'result'
390: return result;
391: }
392:
393: return -1;
394: }
395:
1.60 paf 396: int String::pos(const char *substr, size_t substr_size,
397: size_t result, Untaint_lang lang) const {
1.57 paf 398: for(; result<size(); result++) {
1.60 paf 399: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 400: if(
401: partial==0 || // full match
402: partial==2) // 'substr' starts 'this'+'result'
403: return result;
404: }
405:
406: return -1;
1.60 paf 407: }
408:
409: void String::split(Array& result,
410: size_t* pos_after_ref,
411: const char *delim, size_t delim_size,
412: Untaint_lang lang, int limit) const {
413: if(delim_size) {
414: size_t pos_after=pos_after_ref?*pos_after_ref:0;
415: int pos_before;
416: // while we have 'delim'...
417: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 418: result+=&mid(pos_after, pos_before);
1.60 paf 419: pos_after=pos_before+delim_size;
420: }
421: // last piece
422: if(pos_after<size() && limit) {
1.69 paf 423: result+=&mid(pos_after, size());
1.60 paf 424: pos_after=size();
425: }
426: if(pos_after_ref)
427: *pos_after_ref=pos_after;
428: } else { // empty delim
429: result+=this;
430: if(pos_after_ref)
431: *pos_after_ref+=size();
432: }
433: }
434:
435: void String::split(Array& result,
436: size_t* pos_after_ref,
437: const String& delim, Untaint_lang lang,
438: int limit) const {
439: if(delim.size()) {
440: size_t pos_after=pos_after_ref?*pos_after_ref:0;
441: int pos_before;
442: // while we have 'delim'...
443: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 444: result+=&mid(pos_after, pos_before);
1.60 paf 445: pos_after=pos_before+delim.size();
446: }
447: // last piece
448: if(pos_after<size() && limit) {
1.69 paf 449: result+=&mid(pos_after, size());
1.60 paf 450: pos_after=size();
451: }
452: if(pos_after_ref)
453: *pos_after_ref=pos_after;
454: } else { // empty delim
455: result+=this;
456: if(pos_after_ref)
457: *pos_after_ref+=size();
458: }
1.61 paf 459: }
460:
1.63 paf 461: /// @test really @b test: s x m [tested: i & g ]
462: static void regex_options(char *options, int *result){
463: struct Regex_option {
464: char key;
465: int clear, set;
466: int *result;
467: } regex_option[]={
468: {'i', 0, PCRE_CASELESS, result}, // a=A
469: {'s', 0, PCRE_DOTALL, result}, // \n\n$
470: {'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
471: {'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
472: {'g', 0, true, result+1}, // many rows
473: {0},
474: };
475: result[0]=PCRE_EXTRA | PCRE_DOTALL;
476: result[1]=0;
477:
478: if(options)
479: for(Regex_option *o=regex_option; o->key; o++)
480: if(
481: strchr(options, o->key) ||
482: strchr(options, toupper(o->key))) {
483: *(o->result)&=~o->clear;
484: *(o->result)|=o->set;
485: }
486: }
487:
1.66 paf 488: /**
489: returns true if fills table.
490: table format is defined and fixed[can be used by others]:
491: @verbatim
492: pre-match/match/post-match/1/2/3/...
493: @endverbatim
494: @test setlocale param to auto.p
495: */
1.62 paf 496: bool String::match(const String *aorigin,
497: const String& regexp,
1.63 paf 498: const String *options,
1.64 paf 499: Table **table,
500: Row_action row_action, void *info) const {
501:
1.63 paf 502: static const unsigned char *tables=0; { SYNCHRONIZED(true);
503: if(!tables) {
504: setlocale(LC_CTYPE, "ru");
505: tables=pcre_maketables();
506: }
1.62 paf 507: }
1.68 paf 508: const char *pattern=regexp.cstr(UL_AS_IS);
1.62 paf 509: const char *errptr;
510: int erroffset;
1.63 paf 511: int option_bits[2]; regex_options(options?options->cstr():0, option_bits);
512: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 513: &errptr, &erroffset,
514: tables);
515:
1.67 paf 516: if(!code)
1.62 paf 517: THROW(0, 0,
1.69 paf 518: ®exp.mid(erroffset, regexp.size()),
1.67 paf 519: "match error - %s", errptr);
1.62 paf 520:
1.63 paf 521: int info_substrings=pcre_info(code, 0, 0);
522: if(info_substrings<0) {
523: (*pcre_free)(code);
524: THROW(0, 0,
525: aorigin,
526: "pcre_info error #%d",
527: info_substrings);
528: }
529:
530: int startoffset=0;
1.68 paf 531: const char *subject=cstr(UL_AS_IS);
1.62 paf 532: int length=strlen(subject);
1.63 paf 533: int ovecsize;
534: int *ovector=(int *)malloc(sizeof(int)*
1.65 paf 535: (ovecsize=(1/*match*/+info_substrings)*3));
1.62 paf 536:
1.64 paf 537: { // create table
538: Array& columns=*NEW Array(pool());
539: columns+=string_pre_match_name;
540: columns+=string_match_name;
541: columns+=string_post_match_name;
542: for(int i=1; i<=info_substrings; i++) {
543: char *column=(char *)malloc(MAX_NUMBER);
544: snprintf(column, MAX_NUMBER, "%d", i);
545: columns+=NEW String(pool(), column); // .i column name
546: }
547: *table=NEW Table(pool(), aorigin, &columns);
1.62 paf 548: }
1.63 paf 549:
1.64 paf 550: int exec_option_bits=0;
1.63 paf 551: while(true) {
552: int exec_substrings=pcre_exec(code, 0,
553: subject, length, startoffset,
1.64 paf 554: exec_option_bits, ovector, ovecsize);
1.63 paf 555:
556: if(exec_substrings==PCRE_ERROR_NOMATCH) {
557: (*pcre_free)(code);
1.67 paf 558: (*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 559: return option_bits[1]!=0; // global=true+table, not global=false
560: }
561:
562: if(exec_substrings<0) {
563: (*pcre_free)(code);
564: THROW(0, 0,
565: aorigin,
566: "pcre_exec error #%d",
567: exec_substrings);
568: }
569:
570: Array& row=*NEW Array(pool());
1.69 paf 571: row+=&mid(0, ovector[0]); // .pre-match column value
572: row+=&mid(ovector[0], ovector[1]); // .match
573: row+=&mid(ovector[1], size()); // .post-match
1.63 paf 574:
575: for(int i=1; i<exec_substrings; i++) {
1.69 paf 576: // -1:-1 case handled peacefully by mid() itself
577: row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 578: }
579:
1.67 paf 580: (*row_action)(**table, &row, startoffset, ovector[0], info);
1.63 paf 581:
1.67 paf 582: if(!option_bits[1] || !(startoffset=ovector[1])) { // not global | going to hang
1.63 paf 583: (*pcre_free)(code);
1.67 paf 584: (*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 585: return true;
586: }
587:
588: /*
589: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 590: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 591: */
592: }
1.61 paf 593: }
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