Annotation of parser3/src/main/pa_string.C, revision 1.140
1.45 paf 1: /** @file
1.55 paf 2: Parser: string class. @see untasize_t.C.
1.46 paf 3:
1.137 paf 4: Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
1.138 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.46 paf 6:
1.140 ! paf 7: $Id: pa_string.C,v 1.139 2002/02/20 09:13:08 paf Exp $
1.4 paf 8: */
9:
1.70 paf 10: #include "pcre.h"
11:
1.13 paf 12: #include "pa_pool.h"
1.12 paf 13: #include "pa_string.h"
1.5 paf 14: #include "pa_hash.h"
1.22 paf 15: #include "pa_exception.h"
1.53 paf 16: #include "pa_common.h"
1.60 paf 17: #include "pa_array.h"
18: #include "pa_globals.h"
1.61 paf 19: #include "pa_table.h"
1.101 parser 20: #include "pa_dictionary.h"
1.132 paf 21: #include "pa_charset.h"
1.60 paf 22:
1.139 paf 23: #define DEBUG_STRING_APPENDS_VS_EXPANDS
24:
25:
26: #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS
27: ulong string_piece_appends=0;
28: #endif
29:
1.75 paf 30: String::String(Pool& apool, const char *src, size_t src_size, bool tainted) :
1.120 paf 31: Pooled(apool) {
1.28 paf 32: last_chunk=&head;
33: head.count=CR_PREALLOCATED_COUNT;
1.5 paf 34: append_here=head.rows;
1.140 ! paf 35: initial_head_link=0;
1.28 paf 36: link_row=&head.rows[head.count];
1.41 paf 37:
38: if(src)
1.75 paf 39: if(tainted)
40: APPEND_TAINTED(src, src_size, 0, 0);
1.41 paf 41: else
1.75 paf 42: APPEND_CLEAN(src, src_size, 0, 0);
1.1 paf 43: }
1.140 ! paf 44: /*
1.94 parser 45: String::String(const String& src) :
1.120 paf 46: Pooled(src.pool()) {
1.8 paf 47: head.count=CR_PREALLOCATED_COUNT;
48:
1.116 paf 49: uint src_used_rows=src.used_rows();
1.8 paf 50: if(src_used_rows<=head.count) {
1.55 paf 51: // all new rows fit size_to preallocated area
1.98 parser 52: last_chunk=&head;
1.116 paf 53: uint curr_chunk_rows=head.count;
1.8 paf 54: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows);
55: append_here=&head.rows[src_used_rows];
56: link_row=&head.rows[curr_chunk_rows];
57: } else {
58: // warning:
1.10 paf 59: // heavily relies on the fact
60: // "preallocated area is the same for all strings"
1.8 paf 61: //
62: // info:
63: // allocating only enough mem to fit src string rows
64: // next append would allocate a new chunk
65: //
1.55 paf 66: // new rows don't fit size_to preallocated area: splitting size_to two chunks
1.8 paf 67: // preallocated chunk src to constructing head
68: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count);
1.55 paf 69: // remaining rows size_to new_chunk
1.116 paf 70: uint curr_chunk_rows=src_used_rows-head.count;
1.98 parser 71: last_chunk=static_cast<Chunk *>(
1.139 paf 72: malloc(sizeof(Chunk::count_type)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *), 9));
1.98 parser 73: last_chunk->count=curr_chunk_rows;
74: append_here=link_row=&last_chunk->rows[last_chunk->count];
1.8 paf 75:
1.140 ! paf 76: Chunk *old_chunk=src.head.rows[src.head.count].link;
1.98 parser 77: Chunk::Row *new_rows=last_chunk->rows;
1.116 paf 78: uint rows_left_to_copy=last_chunk->count;
1.8 paf 79: while(true) {
1.139 paf 80: Chunk::count_type old_count=old_chunk->count;
1.8 paf 81: Chunk *next_chunk=old_chunk->rows[old_count].link;
82: if(next_chunk) {
83: // not last source chunk
84: // taking it all
85: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count);
86: new_rows+=old_count;
87: rows_left_to_copy-=old_count;
88:
89: old_chunk=next_chunk;
90: } else {
91: // the last source chunk
92: // taking only those rows of chunk that _left_to_copy
93: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy);
94: break;
95: }
96: }
1.5 paf 97: }
1.8 paf 98: link_row->link=0;
1.140 ! paf 99: }
! 100: */
! 101:
! 102: String::String(const String& src) :
! 103: Pooled(src.pool()) {
! 104: last_chunk=&head;
! 105: head.count=CR_PREALLOCATED_COUNT;
! 106: append_here=head.rows;
! 107: initial_head_link=0;
! 108: link_row=&head.rows[head.count];
! 109:
! 110: append(src, UL_UNSPECIFIED);
1.120 paf 111: }
112:
113: size_t String::size() const {
114: size_t result=0;
1.123 paf 115: STRING_FOREACH_ROW(
1.120 paf 116: result+=row->item.size;
1.123 paf 117: );
1.120 paf 118: break2:
119: return result;
1.94 parser 120: }
121:
1.115 paf 122: /// @todo not very optimal
123: uint String::used_rows() const {
124: uint result=0;
1.123 paf 125: STRING_FOREACH_ROW(
126: result++;
127: );
1.115 paf 128: break2:
129: return result;
130: }
1.94 parser 131: void String::expand() {
1.139 paf 132: Chunk::count_type new_chunk_count=last_chunk->count+CR_GROW_COUNT;
133: if(new_chunk_count>max_integral(Chunk::count_type))
134: new_chunk_count=max_integral(Chunk::count_type);
1.122 paf 135:
1.94 parser 136: last_chunk=static_cast<Chunk *>(
1.139 paf 137: malloc(sizeof(Chunk::count_type)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10));
1.94 parser 138: last_chunk->count=new_chunk_count;
139: link_row->link=last_chunk;
140: append_here=last_chunk->rows;
141: link_row=&last_chunk->rows[last_chunk->count];
142: link_row->link=0;
1.5 paf 143: }
1.28 paf 144:
1.13 paf 145: String& String::real_append(STRING_APPEND_PARAMS) {
1.139 paf 146: if(!last_chunk) // growth stopped [we're appended as string to somebody]
147: throw Exception(0, 0,
148: this,
149: "string growth stopped");
150:
1.9 paf 151: if(!src)
152: return *this;
1.26 paf 153: if(!size)
154: size=strlen(src);
155: if(!size)
1.9 paf 156: return *this;
1.122 paf 157:
1.139 paf 158: #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS
159: string_piece_appends++;
160: #endif
161:
162: // manually unrolled to avoid extra check
163: while(size>max_integral(Chunk::Row::item_size_type)) {
1.122 paf 164: if(chunk_is_full())
165: expand();
166:
167: append_here->item.ptr=src;
1.139 paf 168: append_here->item.size=max_integral(Chunk::Row::item_size_type);
1.122 paf 169: append_here->item.lang=lang;
170: #ifndef NO_STRING_ORIGIN
171: append_here->item.origin.file=file;
172: append_here->item.origin.line=line;
173: #endif
174: append_here++;
175:
1.139 paf 176: src+=max_integral(Chunk::Row::item_size_type);
177: size-=max_integral(Chunk::Row::item_size_type);
1.122 paf 178: }
1.9 paf 179:
1.1 paf 180: if(chunk_is_full())
181: expand();
182:
183: append_here->item.ptr=src;
1.121 paf 184: append_here->item.size=size;
1.52 paf 185: append_here->item.lang=lang;
1.13 paf 186: #ifndef NO_STRING_ORIGIN
1.14 paf 187: append_here->item.origin.file=file;
188: append_here->item.origin.line=line;
1.13 paf 189: #endif
1.115 paf 190: append_here++;
1.1 paf 191:
192: return *this;
1.97 parser 193: }
194:
195: char String::first_char() const {
1.140 ! paf 196: if(is_empty())
1.112 parser 197: throw Exception(0, 0,
1.97 parser 198: this,
199: "getting first char of empty string");
200:
201: return *head.rows[0].item.ptr;
1.1 paf 202: }
203:
1.16 paf 204: uint String::hash_code() const {
1.7 paf 205: uint result=0;
1.123 paf 206: STRING_FOREACH_ROW(
1.6 paf 207: result=Hash::generic_code(result, row->item.ptr, row->item.size);
1.123 paf 208: );
1.5 paf 209: break2:
210: return result;
211: }
212:
1.60 paf 213: /// @todo move 'lang' skipping to pos
214: int String::cmp(int& partial, const String& src,
215: size_t this_offset, Untaint_lang lang) const {
1.59 paf 216: partial=-1;
1.125 paf 217: size_t a_size=size();
218: this_offset=min(this_offset, a_size-1);
1.55 paf 219:
1.16 paf 220: const Chunk *a_chunk=&head;
221: const Chunk *b_chunk=&src.head;
222: const Chunk::Row *a_row=a_chunk->rows;
223: const Chunk::Row *b_row=b_chunk->rows;
1.55 paf 224: size_t a_offset=this_offset;
225: size_t b_offset=0;
1.9 paf 226: Chunk::Row *a_end=append_here;
227: Chunk::Row *b_end=src.append_here;
1.116 paf 228: uint a_countdown=a_chunk->count;
229: uint b_countdown=b_chunk->count;
230: int result;
1.60 paf 231: size_t pos=0;
1.33 paf 232:
1.125 paf 233: bool a_break=a_size==0;
1.140 ! paf 234: bool b_break=src.is_empty();
1.83 parser 235: if(!(a_break || b_break)) while(true) {
1.55 paf 236: if(pos+a_row->item.size > this_offset) {
1.136 paf 237: if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang)
1.60 paf 238: return -1; // wrong lang -- bail out
239:
1.55 paf 240: int size_diff=
241: (a_row->item.size-a_offset)-
242: (b_row->item.size-b_offset);
243:
244: if(size_diff==0) { // a has same size as b
1.60 paf 245: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
246: a_row->item.size-a_offset);
1.55 paf 247: if(result)
248: return result;
1.60 paf 249: pos+=a_row->item.size;
1.55 paf 250: a_row++; a_countdown--; a_offset=0;
251: b_row++; b_countdown--; b_offset=0;
252: } else if (size_diff>0) { // a longer
1.60 paf 253: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
254: b_row->item.size-b_offset);
1.55 paf 255: if(result)
256: return result;
257: a_offset+=b_row->item.size-b_offset;
258: b_row++; b_countdown--; b_offset=0;
259: } else { // b longer
1.60 paf 260: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
261: a_row->item.size-a_offset);
1.55 paf 262: if(result)
263: return result;
264: b_offset+=a_row->item.size-a_offset;
1.60 paf 265: pos+=a_row->item.size;
1.55 paf 266: a_row++; a_countdown--; a_offset=0;
267: }
1.83 parser 268: if(b_break=b_row==b_end) {
269: a_break=a_row==a_end;
270: break;
271: }
1.55 paf 272: if(!b_countdown) {
273: b_chunk=b_row->link;
274: b_row=b_chunk->rows;
275: b_countdown=b_chunk->count;
276: }
277: } else {
1.60 paf 278: a_offset-=a_row->item.size;
279: pos+=a_row->item.size;
280: a_row++; a_countdown--;
1.9 paf 281: }
282:
1.83 parser 283: if(a_break=a_row==a_end) {
284: b_break=b_row==b_end;
285: break;
286: }
1.11 paf 287: if(!a_countdown) {
1.9 paf 288: a_chunk=a_row->link;
289: a_row=a_chunk->rows;
1.11 paf 290: a_countdown=a_chunk->count;
1.9 paf 291: }
1.27 paf 292: }
1.55 paf 293: if(a_break==b_break) { // ended simultaneously
294: partial=0; return 0;
295: } else if(a_break) { // first bytes equal, but a ended before b
296: partial=1; return -1;
297: } else {
298: partial=2; return +1;
299: }
1.27 paf 300: }
301:
1.60 paf 302: /// @todo move 'lang' skipping to pos
1.59 paf 303: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 304: size_t this_offset, Untaint_lang lang) const {
1.59 paf 305: partial=-1;
1.125 paf 306: size_t a_size=size();
1.50 paf 307: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.125 paf 308: this_offset=min(this_offset, a_size-1);
1.27 paf 309:
310: const Chunk *a_chunk=&head;
311: const Chunk::Row *a_row=a_chunk->rows;
1.59 paf 312: size_t a_offset=this_offset;
1.55 paf 313: size_t b_offset=0;
1.27 paf 314: Chunk::Row *a_end=append_here;
1.116 paf 315: uint a_countdown=a_chunk->count;
1.60 paf 316: size_t pos=0;
1.52 paf 317:
1.125 paf 318: bool a_break=a_size==0;
1.83 parser 319: bool b_break=b_size==0;
320: if(!(a_break || b_break)) while(true) {
1.59 paf 321: if(pos+a_row->item.size > this_offset) {
1.136 paf 322: if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang)
1.60 paf 323: return -1; // wrong lang -- bail out
324:
1.59 paf 325: int size_diff=
326: (a_row->item.size-a_offset)-
327: (b_size-b_offset);
328:
329: if(size_diff==0) { // a has same size as b
1.116 paf 330: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 331: a_row->item.size-a_offset)!=0)
332: return result;
1.60 paf 333: pos+=a_row->item.size;
1.59 paf 334: a_row++; a_countdown--; a_offset=0;
335: b_break=true;
336: } else if (size_diff>0) { // a longer
1.116 paf 337: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 338: b_size-b_offset)!=0)
339: return result;
340: a_offset+=b_size-b_offset;
341: b_break=true;
342: } else { // b longer
1.116 paf 343: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 344: a_row->item.size-a_offset)!=0)
345: return result;
346: b_offset+=a_row->item.size-a_offset;
1.60 paf 347: pos+=a_row->item.size;
1.59 paf 348: a_row++; a_countdown--; a_offset=0;
349: }
350: } else {
1.60 paf 351: a_offset-=a_row->item.size;
352: pos+=a_row->item.size;
353: a_row++; a_countdown--;
1.27 paf 354: }
355:
1.86 parser 356: a_break=a_row==a_end;
357: if(a_break || b_break)
1.83 parser 358: break;
1.27 paf 359: if(!a_countdown) {
360: a_chunk=a_row->link;
361: a_row=a_chunk->rows;
362: a_countdown=a_chunk->count;
1.9 paf 363: }
364: }
1.55 paf 365: if(a_break==b_break) { // ended simultaneously
366: partial=0; return 0;
367: } else if(a_break) { // first bytes equal, but a ended before b
368: partial=1; return -1;
369: } else {
370: partial=2; return +1;
371: }
1.5 paf 372: }
1.46 paf 373:
374: #ifndef NO_STRING_ORIGIN
375: const Origin& String::origin() const {
1.140 ! paf 376: if(is_empty()) {
1.96 parser 377: static const Origin empty_origin={"empty string"};
378: return empty_origin;
379: }
1.46 paf 380:
1.49 paf 381: // determining origin by last appended piece
1.50 paf 382: // because first one frequently constant.
383: // ex: ^load[/file] "document_root" + "/file"
1.80 paf 384: // when last peice is constant,
385: // ex: parser_root_auto_path{dynamic} / auto.p{const}
386: // using first piece
387: Origin& last_origin=append_here[-1].item.origin;
388: return last_origin.file ? last_origin : head.rows[0].item.origin;
1.46 paf 389: }
390: #endif
1.53 paf 391:
1.69 paf 392: String& String::mid(size_t start, size_t finish) const {
1.107 parser 393: String& result=*NEW String(pool());
394:
1.139 paf 395: start=max(size_t(0), start);
1.111 parser 396: finish=min(size(), finish);
1.60 paf 397: if(start==finish)
1.107 parser 398: return result;
1.53 paf 399:
400: size_t pos=0;
1.123 paf 401: STRING_FOREACH_ROW(
402: size_t item_finish=pos+row->item.size;
403: if(item_finish > start) { // started now or already?
1.140 ! paf 404: bool started=result.is_empty(); // started now?
1.123 paf 405: bool finished=finish <= item_finish; // finished now?
406: size_t offset=started?start-pos:0;
407: size_t size=finished?finish-pos:row->item.size;
408: result.APPEND(
409: row->item.ptr+offset, size-offset,
410: row->item.lang,
411: row->item.origin.file, row->item.origin.line);
412: if(finished)
1.53 paf 413: goto break2;
414: }
1.123 paf 415: pos+=row->item.size;
416: );
1.53 paf 417: break2:
1.60 paf 418: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
419: //cstr(), start, finish, result.cstr());
1.53 paf 420: return result;
1.54 paf 421: }
422:
1.60 paf 423: int String::pos(const String& substr,
1.116 paf 424: int result, Untaint_lang lang) const {
1.125 paf 425: size_t self_size=size();
1.131 paf 426: for(; size_t(result)<self_size; result++) {
1.60 paf 427: int partial; cmp(partial, substr, result, lang);
1.58 paf 428: if(
429: partial==0 || // full match
430: partial==2) // 'substr' starts 'this'+'result'
431: return result;
432: }
433:
434: return -1;
435: }
436:
1.60 paf 437: int String::pos(const char *substr, size_t substr_size,
1.116 paf 438: int result, Untaint_lang lang) const {
1.125 paf 439: size_t self_size=size();
1.131 paf 440: for(; size_t(result)<self_size; result++) {
1.60 paf 441: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 442: if(
443: partial==0 || // full match
444: partial==2) // 'substr' starts 'this'+'result'
445: return result;
446: }
447:
448: return -1;
1.60 paf 449: }
450:
451: void String::split(Array& result,
452: size_t* pos_after_ref,
453: const char *delim, size_t delim_size,
454: Untaint_lang lang, int limit) const {
1.125 paf 455: size_t self_size=size();
1.60 paf 456: if(delim_size) {
457: size_t pos_after=pos_after_ref?*pos_after_ref:0;
458: int pos_before;
459: // while we have 'delim'...
460: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 461: result+=&mid(pos_after, pos_before);
1.60 paf 462: pos_after=pos_before+delim_size;
463: }
464: // last piece
1.124 paf 465: if(pos_after<self_size && limit) {
466: result+=&mid(pos_after, self_size);
467: pos_after=self_size;
1.60 paf 468: }
469: if(pos_after_ref)
470: *pos_after_ref=pos_after;
471: } else { // empty delim
472: result+=this;
473: if(pos_after_ref)
1.124 paf 474: *pos_after_ref+=self_size;
1.60 paf 475: }
476: }
477:
478: void String::split(Array& result,
479: size_t* pos_after_ref,
480: const String& delim, Untaint_lang lang,
481: int limit) const {
1.140 ! paf 482: if(!delim.is_empty()) {
1.60 paf 483: size_t pos_after=pos_after_ref?*pos_after_ref:0;
484: int pos_before;
485: // while we have 'delim'...
486: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 487: result+=&mid(pos_after, pos_before);
1.60 paf 488: pos_after=pos_before+delim.size();
489: }
490: // last piece
491: if(pos_after<size() && limit) {
1.69 paf 492: result+=&mid(pos_after, size());
1.60 paf 493: pos_after=size();
494: }
495: if(pos_after_ref)
496: *pos_after_ref=pos_after;
497: } else { // empty delim
498: result+=this;
499: if(pos_after_ref)
500: *pos_after_ref+=size();
501: }
1.61 paf 502: }
503:
1.63 paf 504: static void regex_options(char *options, int *result){
505: struct Regex_option {
506: char key;
507: int clear, set;
508: int *result;
509: } regex_option[]={
510: {'i', 0, PCRE_CASELESS, result}, // a=A
1.79 paf 511: {'s', 0, PCRE_DOTALL, result}, // \n\n$ [default]
1.63 paf 512: {'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
513: {'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
514: {'g', 0, true, result+1}, // many rows
515: {0},
516: };
517: result[0]=PCRE_EXTRA | PCRE_DOTALL;
518: result[1]=0;
519:
520: if(options)
521: for(Regex_option *o=regex_option; o->key; o++)
522: if(
523: strchr(options, o->key) ||
524: strchr(options, toupper(o->key))) {
525: *(o->result)&=~o->clear;
526: *(o->result)|=o->set;
527: }
528: }
529:
1.88 parser 530: /// @todo maybe need speedup: some option to remove pre/match/post string generation
1.132 paf 531: bool String::match(
1.77 paf 532: const String *aorigin,
1.62 paf 533: const String& regexp,
1.63 paf 534: const String *options,
1.64 paf 535: Table **table,
1.95 parser 536: Row_action row_action, void *info,
537: bool *was_global) const {
1.64 paf 538:
1.140 ! paf 539: if(regexp.is_empty())
1.112 parser 540: throw Exception(0, 0,
1.73 paf 541: aorigin,
542: "regexp is empty");
1.118 paf 543: const char *pattern=regexp.cstr();
1.62 paf 544: const char *errptr;
545: int erroffset;
1.63 paf 546: int option_bits[2]; regex_options(options?options->cstr():0, option_bits);
1.95 parser 547: if(was_global)
548: *was_global=option_bits[1]!=0;
1.63 paf 549: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 550: &errptr, &erroffset,
1.132 paf 551: pool().get_source_charset().pcre_tables);
1.62 paf 552:
1.67 paf 553: if(!code)
1.112 parser 554: throw Exception(0, 0,
1.69 paf 555: ®exp.mid(erroffset, regexp.size()),
1.74 paf 556: "regular expression syntax error - %s", errptr);
1.62 paf 557:
1.63 paf 558: int info_substrings=pcre_info(code, 0, 0);
559: if(info_substrings<0) {
1.100 parser 560: pcre_free(code);
1.112 parser 561: throw Exception(0, 0,
1.73 paf 562: aorigin,
1.76 paf 563: "pcre_info error (%d)",
1.73 paf 564: info_substrings);
1.63 paf 565: }
566:
567: int startoffset=0;
1.118 paf 568: const char *subject=cstr();
1.62 paf 569: int length=strlen(subject);
1.63 paf 570: int ovecsize;
571: int *ovector=(int *)malloc(sizeof(int)*
1.114 paf 572: (ovecsize=(1/*match*/+info_substrings)*3), 11);
1.62 paf 573:
1.64 paf 574: { // create table
575: Array& columns=*NEW Array(pool());
576: columns+=string_pre_match_name;
577: columns+=string_match_name;
578: columns+=string_post_match_name;
579: for(int i=1; i<=info_substrings; i++) {
580: char *column=(char *)malloc(MAX_NUMBER);
581: snprintf(column, MAX_NUMBER, "%d", i);
582: columns+=NEW String(pool(), column); // .i column name
583: }
584: *table=NEW Table(pool(), aorigin, &columns);
1.62 paf 585: }
1.63 paf 586:
1.64 paf 587: int exec_option_bits=0;
1.63 paf 588: while(true) {
589: int exec_substrings=pcre_exec(code, 0,
590: subject, length, startoffset,
1.64 paf 591: exec_option_bits, ovector, ovecsize);
1.63 paf 592:
593: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 594: pcre_free(code);
595: row_action(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 596: return option_bits[1]!=0; // global=true+table, not global=false
597: }
598:
599: if(exec_substrings<0) {
1.100 parser 600: pcre_free(code);
1.112 parser 601: throw Exception(0, 0,
1.63 paf 602: aorigin,
1.76 paf 603: "regular expression execute error (%d)",
1.63 paf 604: exec_substrings);
605: }
606:
607: Array& row=*NEW Array(pool());
1.81 paf 608: row+=&mid(0, ovector[0]); // .prematch column value
1.69 paf 609: row+=&mid(ovector[0], ovector[1]); // .match
1.81 paf 610: row+=&mid(ovector[1], size()); // .postmatch
1.63 paf 611:
612: for(int i=1; i<exec_substrings; i++) {
1.69 paf 613: // -1:-1 case handled peacefully by mid() itself
614: row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 615: }
616:
1.100 parser 617: row_action(**table, &row, startoffset, ovector[0], info);
1.63 paf 618:
1.100 parser 619: if(!option_bits[1] || startoffset==ovector[1]) { // not global | going to hang
620: pcre_free(code);
621: row_action(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 622: return true;
623: }
1.100 parser 624: startoffset=ovector[1];
1.63 paf 625:
626: /*
627: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 628: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 629: */
630: }
1.82 parser 631: }
632:
1.132 paf 633: String& String::change_case(Pool& pool,
1.82 parser 634: Change_case_kind kind) const {
1.132 paf 635: const unsigned char *tables=pool.get_source_charset().pcre_tables;
1.82 parser 636: String& result=*new(pool) String(pool);
637:
638: const unsigned char *a;
639: const unsigned char *b;
640: switch(kind) {
641: case CC_UPPER:
642: a=tables+lcc_offset;
643: b=tables+fcc_offset;
644: break;
645: case CC_LOWER:
646: a=tables+lcc_offset;
647: b=0;
648: break;
649: default:
1.112 parser 650: throw Exception(0, 0,
1.82 parser 651: this,
652: "unknown change case kind #%d",
653: static_cast<int>(kind)); // never
654: a=b=0; // calm, compiler
655: break; // never
656: }
657:
658: const Chunk *chunk=&head;
659: do {
660: const Chunk::Row *row=chunk->rows;
1.139 paf 661: for(Chunk::count_type i=0; i<chunk->count; i++, row++) {
1.82 parser 662: if(row==append_here)
663: goto break2;
664:
1.114 paf 665: char *new_cstr=(char *)pool.malloc(row->item.size, 12);
1.82 parser 666: char *dest=new_cstr;
667: const char *src=row->item.ptr;
668: for(int size=row->item.size; size--; src++) {
669: unsigned char c=a[(unsigned char)*src];
670: if(b)
671: c=b[c];
672:
673: *dest++=(char)c;
674: }
675:
676: result.APPEND(new_cstr, row->item.size,
1.123 paf 677: row->item.lang,
1.82 parser 678: row->item.origin.file, row->item.origin.line);
679: }
680: chunk=row->link;
681: } while(chunk);
682: break2:
1.89 parser 683:
1.101 parser 684: return result;
685: }
686:
1.108 parser 687: void String::join_chain(Pool& pool,
1.116 paf 688: uint& ai, const Chunk*& achunk, const Chunk::Row*& arow,
1.123 paf 689: uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
690: joined_lang=arow->item.lang;
1.108 parser 691:
692: // calc size
693: joined_size=0;
694: {
1.116 paf 695: uint start_i=ai;
1.108 parser 696: const Chunk::Row *start_row=arow;
697: const Chunk *chunk=achunk;
698: do {
699: const Chunk::Row *row=start_row;
1.116 paf 700: for(uint i=start_i; i<chunk->count; i++, row++) {
1.108 parser 701: if(row==append_here)
702: goto break21;
703:
704: if(row->item.lang==joined_lang)
705: joined_size+=row->item.size;
706: else
1.129 paf 707: goto break21;
1.108 parser 708: }
709: if(chunk=row->link) {
710: start_i=0;
711: start_row=chunk->rows;
712: } else
713: break;
714: } while(true);
715: break21:;
716: }
717:
718: // if one row, return simply itself
719: if(joined_size==arow->item.size) {
720: joined_ptr=arow->item.ptr;
721: ai++; arow++;
1.133 paf 722: if(ai==achunk->count) {
723: if(achunk=arow->link) {
724: ai=0;
1.134 paf 725: arow=achunk->rows;
1.133 paf 726: }
727: }
1.108 parser 728: } else {
729: // join adjacent rows
1.114 paf 730: char *ptr=(char *)pool.malloc(joined_size,13);
1.108 parser 731: joined_ptr=ptr;
1.116 paf 732: uint start_i=ai;
1.108 parser 733: const Chunk::Row *start_row=arow;
734: const Chunk *chunk=achunk;
1.116 paf 735: uint i;
1.108 parser 736: const Chunk::Row *row;
737: do {
738: row=start_row;
739: for(i=start_i; i<chunk->count; i++, row++) {
740: if(row==append_here)
741: goto break22;
742:
743: if(row->item.lang==joined_lang) {
744: memcpy(ptr, row->item.ptr, row->item.size);
745: ptr+=row->item.size;
746: } else
1.129 paf 747: goto break22;
1.108 parser 748: }
749: if(chunk=row->link) {
750: start_i=0;
751: start_row=chunk->rows;
752: } else
753: break;
754: } while(true);
755: break22:;
756:
757: // return joined rows
758: ai=i;
759: arow=row;
760: achunk=chunk;
761: }
762: }
763:
764: String& String::reconstruct(Pool& pool) const {
765: //_asm int 3;
766: String& result=*new(pool) String(pool);
767: const Chunk *chunk=&head;
1.133 paf 768: const Chunk::Row *row=chunk->rows;
769: for(uint i=0; i<chunk->count; ) {
770: if(row==append_here)
771: break;
1.108 parser 772:
1.133 paf 773: uchar joined_lang;
774: const char *joined_ptr;
775: size_t joined_size;
1.130 paf 776: #ifndef NO_STRING_ORIGIN
1.133 paf 777: const char *joined_origin_file=row->item.origin.file;
778: const size_t joined_origin_line=row->item.origin.line;
1.130 paf 779: #endif
1.133 paf 780: join_chain(pool, i, chunk, row,
781: joined_lang, joined_ptr, joined_size);
782:
783: result.APPEND(joined_ptr, joined_size, joined_lang,
784: joined_origin_file, joined_origin_line);
1.108 parser 785:
1.133 paf 786: if(!chunk)
787: break;
1.130 paf 788: }
1.108 parser 789:
790: return result;
791: };
792:
793: String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const {
1.106 parser 794: //_asm int 3;
1.101 parser 795: String& result=*new(pool) String(pool);
1.123 paf 796: STRING_FOREACH_ROW(
797: const char *src=row->item.ptr;
798: size_t src_size=row->item.size;
799: char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14);
800: char *dest=new_cstr;
801: while(src_size) {
802: // there is a row where first column starts 'src'
803: if(Table::Item *item=dict.first_that_starts(src, src_size)) {
804: // get a=>b values
805: const String& a=*static_cast<Array *>(item)->get_string(0);
806: const String& b=*static_cast<Array *>(item)->get_string(1);
807: // skip 'a' in 'src' && reduce work size
808: src+=a.size(); src_size-=a.size();
809: // write 'b' to 'dest' && skip 'b' in 'dest'
810: b.store_to(dest); dest+=b.size();
811: } else {
812: // write a char to b && reduce work size
813: *dest++=*src++; src_size--;
1.101 parser 814: }
815: }
816:
1.123 paf 817: result.APPEND(new_cstr, dest-new_cstr,
818: row->item.lang,
819: row->item.origin.file, row->item.origin.line);
820: );
1.101 parser 821: break2:
1.89 parser 822: return result;
1.108 parser 823: }
824:
825: String& String::replace(Pool& pool, Dictionary& dict) const {
826: return reconstruct(pool).replace_in_reconstructed(pool, dict);
1.89 parser 827: }
828:
1.90 parser 829: double String::as_double() const {
1.89 parser 830: double result;
1.114 paf 831: const char *cstr;
832: char buf[MAX_NUMBER];
1.115 paf 833: if(head.rows+1==append_here) {
1.114 paf 834: int size=min(head.rows[0].item.size, MAX_NUMBER-1);
835: memcpy(buf, head.rows[0].item.ptr, size);
836: buf[size]=0;
837: cstr=buf;
838: } else
839: cstr=this->cstr();
1.102 parser 840: char *error_pos;
1.89 parser 841: // 0xABC
1.99 parser 842: if(cstr[0]=='0')
843: if(cstr[1]=='x' || cstr[1]=='X')
844: result=(double)(unsigned long)strtol(cstr, &error_pos, 0);
845: else
1.102 parser 846: result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos);
1.89 parser 847: else
1.99 parser 848: result=(double)strtod(cstr, &error_pos);
1.89 parser 849:
1.103 parser 850: if(*error_pos/*not EOS*/)
1.112 parser 851: throw Exception(0, 0,
1.89 parser 852: this,
853: "invalid number (double)");
854:
855: return result;
856: }
1.90 parser 857: int String::as_int() const {
1.89 parser 858: int result;
1.114 paf 859: const char *cstr;
860: char buf[MAX_NUMBER];
1.115 paf 861: if(head.rows+1==append_here) {
1.114 paf 862: int size=min(head.rows[0].item.size, MAX_NUMBER-1);
863: memcpy(buf, head.rows[0].item.ptr, size);
864: buf[size]=0;
865: cstr=buf;
866: } else
867: cstr=this->cstr();
1.102 parser 868: char *error_pos;
1.89 parser 869: // 0xABC
1.99 parser 870: if(cstr[0]=='0')
871: if(cstr[1]=='x' || cstr[1]=='X')
872: result=(int)(unsigned long)strtol(cstr, &error_pos, 0);
873: else
1.102 parser 874: result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 875: else
876: result=(int)strtol(cstr, &error_pos, 0);
877:
1.103 parser 878: if(*error_pos/*not EOS*/)
1.112 parser 879: throw Exception(0, 0,
1.89 parser 880: this,
881: "invalid number (int)");
1.82 parser 882:
883: return result;
1.61 paf 884: }
1.113 parser 885:
1.128 paf 886: inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) {
887: byte1=word&0xFF;
888: byte2=word>>8;
889: }
890: inline ushort uchars2ushort(uchar byte1, uchar byte2) {
891: return (byte2<<8) | byte1;
892: }
1.113 parser 893: /* @todo maybe network order worth spending some effort?
894: don't bothering myself with network byte order,
895: am not planning to be able to move resulting file across platforms
896: for now
897: */
898: void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const {
899: buf_size=
900: prolog_size
1.126 paf 901: +used_rows()*(sizeof(uchar)+sizeof(ushort))
1.113 parser 902: +size();
1.114 paf 903: buf=malloc(buf_size,15);
1.113 parser 904: char *cur=(char *)buf+prolog_size;
905:
1.123 paf 906: STRING_FOREACH_ROW(
907: // lang
1.126 paf 908: memcpy(cur, &row->item.lang, sizeof(uchar));
909: cur+=sizeof(uchar);
1.123 paf 910: // size
1.128 paf 911: uchar byte1; uchar byte2;
912: ushort2uchars(row->item.size, byte1, byte2);
913: memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar);
914: memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar);
1.123 paf 915: // bytes
916: memcpy(cur, row->item.ptr, row->item.size);
917: cur+=row->item.size;
918: );
1.113 parser 919: break2:
920: ;
921: }
922: void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) {
1.135 paf 923: if(buf_size<=prolog_size)
924: return;
925:
1.126 paf 926: char *cur=(char *)buf+prolog_size;
1.113 parser 927: buf_size-=prolog_size;
928:
929: while(buf_size) {
1.128 paf 930: uchar lang=*(uchar *)(cur);
931:
932: ushort size=uchars2ushort(
933: *(uchar*)(cur+sizeof(uchar)*1),
934: *(uchar*)(cur+sizeof(uchar)*2)
935: );
936:
937: const char *ptr=(const char*)(cur+sizeof(uchar)*3);
1.126 paf 938: APPEND(ptr, size, lang, file, 0);
1.113 parser 939:
1.126 paf 940: size_t piece_size=sizeof(uchar)+sizeof(ushort)+size;
1.113 parser 941: cur+=piece_size;
942: buf_size-=piece_size;
943: }
944: }
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