Annotation of parser3/src/main/pa_string.C, revision 1.74
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:
1.74 ! paf 8: $Id: pa_string.C,v 1.73 2001/04/05 19:35:16 paf Exp $
1.4 paf 9: */
10:
1.48 paf 11: #include "pa_config_includes.h"
1.1 paf 12:
1.70 paf 13: #include "pcre.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"
1.60 paf 24:
25: //#include "pa_sapi.h"
1.1 paf 26:
1.18 paf 27: // String
28:
1.55 paf 29: String::String(Pool& apool, const char *src, bool tasize_ted) :
1.17 paf 30: Pooled(apool) {
1.28 paf 31: last_chunk=&head;
32: head.count=CR_PREALLOCATED_COUNT;
1.5 paf 33: append_here=head.rows;
1.2 paf 34: head.preallocated_link=0;
1.28 paf 35: link_row=&head.rows[head.count];
1.8 paf 36: fused_rows=fsize=0;
1.41 paf 37:
38: if(src)
1.55 paf 39: if(tasize_ted)
1.41 paf 40: APPEND_TAINTED(src, 0, 0, 0);
41: else
1.53 paf 42: APPEND_CONST(src);
1.1 paf 43: }
44:
45: void String::expand() {
1.55 paf 46: size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100;
1.28 paf 47: last_chunk=static_cast<Chunk *>(
1.55 paf 48: malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *)));
1.28 paf 49: last_chunk->count=new_chunk_count;
50: link_row->link=last_chunk;
51: append_here=last_chunk->rows;
52: link_row=&last_chunk->rows[last_chunk->count];
1.8 paf 53: link_row->link=0;
1.1 paf 54: }
55:
1.40 paf 56: String::String(const String& src) : Pooled(src.pool()) {
1.8 paf 57: head.count=CR_PREALLOCATED_COUNT;
58:
1.55 paf 59: size_t src_used_rows=src.fused_rows;
1.8 paf 60: if(src_used_rows<=head.count) {
1.55 paf 61: // all new rows fit size_to preallocated area
62: size_t curr_chunk_rows=head.count;
1.8 paf 63: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows);
64: append_here=&head.rows[src_used_rows];
65: link_row=&head.rows[curr_chunk_rows];
66: } else {
67: // warning:
1.10 paf 68: // heavily relies on the fact
69: // "preallocated area is the same for all strings"
1.8 paf 70: //
71: // info:
72: // allocating only enough mem to fit src string rows
73: // next append would allocate a new chunk
74: //
1.55 paf 75: // new rows don't fit size_to preallocated area: splitting size_to two chunks
1.8 paf 76: // preallocated chunk src to constructing head
77: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count);
1.55 paf 78: // remaining rows size_to new_chunk
79: size_t curr_chunk_rows=src_used_rows-head.count;
1.8 paf 80: Chunk *new_chunk=static_cast<Chunk *>(
1.55 paf 81: malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *)));
1.8 paf 82: new_chunk->count=curr_chunk_rows;
83: head.preallocated_link=new_chunk;
1.28 paf 84: append_here=link_row=&new_chunk->rows[new_chunk->count];
1.8 paf 85:
86: Chunk *old_chunk=src.head.preallocated_link;
87: Chunk::Row *new_rows=new_chunk->rows;
1.55 paf 88: size_t rows_left_to_copy=new_chunk->count;
1.8 paf 89: while(true) {
1.55 paf 90: size_t old_count=old_chunk->count;
1.8 paf 91: Chunk *next_chunk=old_chunk->rows[old_count].link;
92: if(next_chunk) {
93: // not last source chunk
94: // taking it all
95: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count);
96: new_rows+=old_count;
97: rows_left_to_copy-=old_count;
98:
99: old_chunk=next_chunk;
100: } else {
101: // the last source chunk
102: // taking only those rows of chunk that _left_to_copy
103: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy);
104: break;
105: }
106: }
1.5 paf 107: }
1.8 paf 108: link_row->link=0;
109: fused_rows=src_used_rows;
110: fsize=src.fsize;
1.5 paf 111: }
1.28 paf 112:
1.42 paf 113: String& String::append(const String& src, Untaint_lang lang, bool forced) {
1.60 paf 114: const Chunk *chunk=&src.head;
1.40 paf 115: do {
1.60 paf 116: const Chunk::Row *row=chunk->rows;
117: for(size_t i=0; i<chunk->count; i++, row++) {
118: if(row==src.append_here)
1.40 paf 119: goto break2;
1.60 paf 120:
121: APPEND(row->item.ptr, row->item.size,
122: (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang,
123: row->item.origin.file, row->item.origin.line);
1.40 paf 124: }
125: chunk=row->link;
126: } while(chunk);
127: break2:
1.60 paf 128: return *this;
1.34 paf 129: }
1.60 paf 130:
1.13 paf 131: String& String::real_append(STRING_APPEND_PARAMS) {
1.9 paf 132: if(!src)
133: return *this;
1.26 paf 134: if(!size)
135: size=strlen(src);
136: if(!size)
1.9 paf 137: return *this;
138:
1.1 paf 139: if(chunk_is_full())
140: expand();
141:
142: append_here->item.ptr=src;
1.26 paf 143: fsize+=append_here->item.size=size;
1.52 paf 144: append_here->item.lang=lang;
1.13 paf 145: #ifndef NO_STRING_ORIGIN
1.14 paf 146: append_here->item.origin.file=file;
147: append_here->item.origin.line=line;
1.13 paf 148: #endif
1.8 paf 149: append_here++; fused_rows++;
1.1 paf 150:
151: return *this;
152: }
153:
1.16 paf 154: uint String::hash_code() const {
1.7 paf 155: uint result=0;
1.5 paf 156:
1.16 paf 157: const Chunk *chunk=&head;
1.5 paf 158: do {
1.16 paf 159: const Chunk::Row *row=chunk->rows;
1.55 paf 160: for(size_t i=0; i<chunk->count; i++) {
1.5 paf 161: if(row==append_here)
162: goto break2;
163:
1.6 paf 164: result=Hash::generic_code(result, row->item.ptr, row->item.size);
1.5 paf 165: row++;
166: }
167: chunk=row->link;
168: } while(chunk);
169: break2:
170: return result;
171: }
172:
1.60 paf 173: /// @todo move 'lang' skipping to pos
174: int String::cmp(int& partial, const String& src,
175: size_t this_offset, Untaint_lang lang) const {
1.59 paf 176: partial=-1;
1.55 paf 177: this_offset=min(this_offset, size()-1);
178:
1.16 paf 179: const Chunk *a_chunk=&head;
180: const Chunk *b_chunk=&src.head;
181: const Chunk::Row *a_row=a_chunk->rows;
182: const Chunk::Row *b_row=b_chunk->rows;
1.55 paf 183: size_t a_offset=this_offset;
184: size_t b_offset=0;
1.9 paf 185: Chunk::Row *a_end=append_here;
186: Chunk::Row *b_end=src.append_here;
1.55 paf 187: size_t a_countdown=a_chunk->count;
188: size_t b_countdown=b_chunk->count;
1.9 paf 189: bool a_break=false;
190: bool b_break=false;
1.55 paf 191: size_t result;
1.60 paf 192: size_t pos=0;
193: while(true) {
1.33 paf 194: a_break=a_row==a_end;
195: b_break=b_row==b_end;
196: if(a_break || b_break)
197: break;
198:
1.55 paf 199: if(pos+a_row->item.size > this_offset) {
1.71 paf 200: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 201: return -1; // wrong lang -- bail out
202:
1.55 paf 203: int size_diff=
204: (a_row->item.size-a_offset)-
205: (b_row->item.size-b_offset);
206:
207: if(size_diff==0) { // a has same size as b
1.60 paf 208: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
209: a_row->item.size-a_offset);
1.55 paf 210: if(result)
211: return result;
1.60 paf 212: pos+=a_row->item.size;
1.55 paf 213: a_row++; a_countdown--; a_offset=0;
214: b_row++; b_countdown--; b_offset=0;
215: } else if (size_diff>0) { // a longer
1.60 paf 216: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
217: b_row->item.size-b_offset);
1.55 paf 218: if(result)
219: return result;
220: a_offset+=b_row->item.size-b_offset;
221: b_row++; b_countdown--; b_offset=0;
222: } else { // b longer
1.60 paf 223: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
224: a_row->item.size-a_offset);
1.55 paf 225: if(result)
226: return result;
227: b_offset+=a_row->item.size-a_offset;
1.60 paf 228: pos+=a_row->item.size;
1.55 paf 229: a_row++; a_countdown--; a_offset=0;
230: }
1.60 paf 231:
1.55 paf 232: if(!b_countdown) {
233: b_chunk=b_row->link;
234: b_row=b_chunk->rows;
235: b_countdown=b_chunk->count;
236: }
237: } else {
1.60 paf 238: a_offset-=a_row->item.size;
239: pos+=a_row->item.size;
240: a_row++; a_countdown--;
1.9 paf 241: }
242:
1.11 paf 243: if(!a_countdown) {
1.9 paf 244: a_chunk=a_row->link;
245: a_row=a_chunk->rows;
1.11 paf 246: a_countdown=a_chunk->count;
1.9 paf 247: }
1.27 paf 248: }
1.55 paf 249: if(a_break==b_break) { // ended simultaneously
250: partial=0; return 0;
251: } else if(a_break) { // first bytes equal, but a ended before b
252: partial=1; return -1;
253: } else {
254: partial=2; return +1;
255: }
1.27 paf 256: }
257:
1.60 paf 258: /// @todo move 'lang' skipping to pos
1.59 paf 259: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 260: size_t this_offset, Untaint_lang lang) const {
1.59 paf 261: partial=-1;
1.50 paf 262: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.59 paf 263: this_offset=min(this_offset, size()-1);
1.27 paf 264:
265: const Chunk *a_chunk=&head;
266: const Chunk::Row *a_row=a_chunk->rows;
1.59 paf 267: size_t a_offset=this_offset;
1.55 paf 268: size_t b_offset=0;
1.27 paf 269: Chunk::Row *a_end=append_here;
1.55 paf 270: size_t a_countdown=a_chunk->count;
1.27 paf 271: bool a_break=false;
272: bool b_break=false;
1.60 paf 273: size_t pos=0;
274: while(true) {
1.52 paf 275: a_break=a_row==a_end;
276: if(a_break || b_break)
277: break;
278:
1.59 paf 279: if(pos+a_row->item.size > this_offset) {
1.71 paf 280: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 281: return -1; // wrong lang -- bail out
282:
1.59 paf 283: int size_diff=
284: (a_row->item.size-a_offset)-
285: (b_size-b_offset);
286:
287: if(size_diff==0) { // a has same size as b
288: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
289: a_row->item.size-a_offset)!=0)
290: return result;
1.60 paf 291: pos+=a_row->item.size;
1.59 paf 292: a_row++; a_countdown--; a_offset=0;
293: b_break=true;
294: } else if (size_diff>0) { // a longer
295: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
296: b_size-b_offset)!=0)
297: return result;
298: a_offset+=b_size-b_offset;
299: b_break=true;
300: } else { // b longer
301: if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
302: a_row->item.size-a_offset)!=0)
303: return result;
304: b_offset+=a_row->item.size-a_offset;
1.60 paf 305: pos+=a_row->item.size;
1.59 paf 306: a_row++; a_countdown--; a_offset=0;
307: }
308: } else {
1.60 paf 309: a_offset-=a_row->item.size;
310: pos+=a_row->item.size;
311: a_row++; a_countdown--;
1.27 paf 312: }
313:
314: if(!a_countdown) {
315: a_chunk=a_row->link;
316: a_row=a_chunk->rows;
317: a_countdown=a_chunk->count;
1.9 paf 318: }
319: }
1.55 paf 320: if(a_break==b_break) { // ended simultaneously
321: partial=0; return 0;
322: } else if(a_break) { // first bytes equal, but a ended before b
323: partial=1; return -1;
324: } else {
325: partial=2; return +1;
326: }
1.5 paf 327: }
1.46 paf 328:
329: #ifndef NO_STRING_ORIGIN
330: const Origin& String::origin() const {
331: if(!fused_rows)
332: THROW(0, 0,
1.50 paf 333: 0,
334: "String::origin() of empty string called");
1.46 paf 335:
1.49 paf 336: // determining origin by last appended piece
1.50 paf 337: // because first one frequently constant.
338: // ex: ^load[/file] "document_root" + "/file"
1.49 paf 339: return append_here[-1].item.origin;
1.46 paf 340: }
341: #endif
1.53 paf 342:
1.69 paf 343: String& String::mid(size_t start, size_t finish) const {
1.53 paf 344: start=max(0, start);
345: finish=min(size(), finish);
1.60 paf 346: if(start==finish)
347: return *empty_string;
1.53 paf 348:
349: String& result=*NEW String(pool());
350:
351: size_t pos=0;
352: const Chunk *chunk=&head;
353: do {
354: const Chunk::Row *row=chunk->rows;
1.55 paf 355: for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) {
1.53 paf 356: if(row==append_here)
357: goto break2;
358:
1.60 paf 359: size_t item_finish=pos+row->item.size;
360: if(item_finish > start) { // started now or already?
361: bool started=result.size()==0; // started now?
362: bool finished=finish <= item_finish; // finished now?
1.53 paf 363: size_t offset=started?start-pos:0;
364: size_t size=finished?finish-pos:row->item.size;
365: result.APPEND(
366: row->item.ptr+offset, size-offset,
367: row->item.lang,
368: row->item.origin.file, row->item.origin.line);
369: if(finished)
370: goto break2;
371: }
372: }
373: chunk=row->link;
374: } while(chunk);
375: break2:
1.60 paf 376: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
377: //cstr(), start, finish, result.cstr());
1.53 paf 378: return result;
1.54 paf 379: }
380:
1.60 paf 381: int String::pos(const String& substr,
382: size_t result, Untaint_lang lang) const {
1.58 paf 383: for(; result<size(); result++) {
1.60 paf 384: int partial; cmp(partial, substr, result, lang);
1.58 paf 385: if(
386: partial==0 || // full match
387: partial==2) // 'substr' starts 'this'+'result'
388: return result;
389: }
390:
391: return -1;
392: }
393:
1.60 paf 394: int String::pos(const char *substr, size_t substr_size,
395: size_t result, Untaint_lang lang) const {
1.57 paf 396: for(; result<size(); result++) {
1.60 paf 397: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 398: if(
399: partial==0 || // full match
400: partial==2) // 'substr' starts 'this'+'result'
401: return result;
402: }
403:
404: return -1;
1.60 paf 405: }
406:
407: void String::split(Array& result,
408: size_t* pos_after_ref,
409: const char *delim, size_t delim_size,
410: Untaint_lang lang, int limit) const {
411: if(delim_size) {
412: size_t pos_after=pos_after_ref?*pos_after_ref:0;
413: int pos_before;
414: // while we have 'delim'...
415: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 416: result+=&mid(pos_after, pos_before);
1.60 paf 417: pos_after=pos_before+delim_size;
418: }
419: // last piece
420: if(pos_after<size() && limit) {
1.69 paf 421: result+=&mid(pos_after, size());
1.60 paf 422: pos_after=size();
423: }
424: if(pos_after_ref)
425: *pos_after_ref=pos_after;
426: } else { // empty delim
427: result+=this;
428: if(pos_after_ref)
429: *pos_after_ref+=size();
430: }
431: }
432:
433: void String::split(Array& result,
434: size_t* pos_after_ref,
435: const String& delim, Untaint_lang lang,
436: int limit) const {
437: if(delim.size()) {
438: size_t pos_after=pos_after_ref?*pos_after_ref:0;
439: int pos_before;
440: // while we have 'delim'...
441: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 442: result+=&mid(pos_after, pos_before);
1.60 paf 443: pos_after=pos_before+delim.size();
444: }
445: // last piece
446: if(pos_after<size() && limit) {
1.69 paf 447: result+=&mid(pos_after, size());
1.60 paf 448: pos_after=size();
449: }
450: if(pos_after_ref)
451: *pos_after_ref=pos_after;
452: } else { // empty delim
453: result+=this;
454: if(pos_after_ref)
455: *pos_after_ref+=size();
456: }
1.61 paf 457: }
458:
1.63 paf 459: /// @test really @b test: s x m [tested: i & g ]
460: static void regex_options(char *options, int *result){
461: struct Regex_option {
462: char key;
463: int clear, set;
464: int *result;
465: } regex_option[]={
466: {'i', 0, PCRE_CASELESS, result}, // a=A
467: {'s', 0, PCRE_DOTALL, result}, // \n\n$
468: {'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
469: {'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
470: {'g', 0, true, result+1}, // many rows
471: {0},
472: };
473: result[0]=PCRE_EXTRA | PCRE_DOTALL;
474: result[1]=0;
475:
476: if(options)
477: for(Regex_option *o=regex_option; o->key; o++)
478: if(
479: strchr(options, o->key) ||
480: strchr(options, toupper(o->key))) {
481: *(o->result)&=~o->clear;
482: *(o->result)|=o->set;
483: }
484: }
485:
1.66 paf 486: /**
487: returns true if fills table.
488: table format is defined and fixed[can be used by others]:
489: @verbatim
490: pre-match/match/post-match/1/2/3/...
491: @endverbatim
492: */
1.62 paf 493: bool String::match(const String *aorigin,
494: const String& regexp,
1.63 paf 495: const String *options,
1.64 paf 496: Table **table,
497: Row_action row_action, void *info) const {
498:
1.73 paf 499: if(!regexp.size())
500: THROW(0, 0,
501: aorigin,
502: "regexp is empty");
1.68 paf 503: const char *pattern=regexp.cstr(UL_AS_IS);
1.62 paf 504: const char *errptr;
505: int erroffset;
1.63 paf 506: int option_bits[2]; regex_options(options?options->cstr():0, option_bits);
507: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 508: &errptr, &erroffset,
1.74 ! paf 509: pcre_tables);
1.62 paf 510:
1.67 paf 511: if(!code)
1.62 paf 512: THROW(0, 0,
1.69 paf 513: ®exp.mid(erroffset, regexp.size()),
1.74 ! paf 514: "regular expression syntax error - %s", errptr);
1.62 paf 515:
1.63 paf 516: int info_substrings=pcre_info(code, 0, 0);
517: if(info_substrings<0) {
518: (*pcre_free)(code);
519: THROW(0, 0,
1.73 paf 520: aorigin,
521: "pcre_info error #%d",
522: info_substrings);
1.63 paf 523: }
524:
525: int startoffset=0;
1.68 paf 526: const char *subject=cstr(UL_AS_IS);
1.62 paf 527: int length=strlen(subject);
1.63 paf 528: int ovecsize;
529: int *ovector=(int *)malloc(sizeof(int)*
1.65 paf 530: (ovecsize=(1/*match*/+info_substrings)*3));
1.62 paf 531:
1.64 paf 532: { // create table
533: Array& columns=*NEW Array(pool());
534: columns+=string_pre_match_name;
535: columns+=string_match_name;
536: columns+=string_post_match_name;
537: for(int i=1; i<=info_substrings; i++) {
538: char *column=(char *)malloc(MAX_NUMBER);
539: snprintf(column, MAX_NUMBER, "%d", i);
540: columns+=NEW String(pool(), column); // .i column name
541: }
542: *table=NEW Table(pool(), aorigin, &columns);
1.62 paf 543: }
1.63 paf 544:
1.64 paf 545: int exec_option_bits=0;
1.63 paf 546: while(true) {
547: int exec_substrings=pcre_exec(code, 0,
548: subject, length, startoffset,
1.64 paf 549: exec_option_bits, ovector, ovecsize);
1.63 paf 550:
551: if(exec_substrings==PCRE_ERROR_NOMATCH) {
552: (*pcre_free)(code);
1.67 paf 553: (*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 554: return option_bits[1]!=0; // global=true+table, not global=false
555: }
556:
557: if(exec_substrings<0) {
558: (*pcre_free)(code);
559: THROW(0, 0,
560: aorigin,
1.74 ! paf 561: "regular expression execute error #%d",
1.63 paf 562: exec_substrings);
563: }
564:
565: Array& row=*NEW Array(pool());
1.69 paf 566: row+=&mid(0, ovector[0]); // .pre-match column value
567: row+=&mid(ovector[0], ovector[1]); // .match
568: row+=&mid(ovector[1], size()); // .post-match
1.63 paf 569:
570: for(int i=1; i<exec_substrings; i++) {
1.69 paf 571: // -1:-1 case handled peacefully by mid() itself
572: row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 573: }
574:
1.67 paf 575: (*row_action)(**table, &row, startoffset, ovector[0], info);
1.63 paf 576:
1.67 paf 577: if(!option_bits[1] || !(startoffset=ovector[1])) { // not global | going to hang
1.63 paf 578: (*pcre_free)(code);
1.67 paf 579: (*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 580: return true;
581: }
582:
583: /*
584: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 585: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 586: */
587: }
1.61 paf 588: }
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