Annotation of parser3/src/main/pa_string.C, revision 1.128
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)
1.119 paf 5: Author: Alexander Petrosyan <paf@design.ru> (http://paf.design.ru)
1.46 paf 6:
1.128 ! paf 7: $Id: pa_string.C,v 1.125.2.1.2.2 2001/12/07 13:10:18 paf Exp $
1.4 paf 8: */
9:
1.48 paf 10: #include "pa_config_includes.h"
1.1 paf 11:
1.70 paf 12: #include "pcre.h"
1.82 parser 13: #include "internal.h"
1.70 paf 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"
1.101 parser 23: #include "pa_dictionary.h"
1.60 paf 24:
1.75 paf 25: String::String(Pool& apool, const char *src, size_t src_size, bool tainted) :
1.120 paf 26: Pooled(apool) {
1.28 paf 27: last_chunk=&head;
28: head.count=CR_PREALLOCATED_COUNT;
1.5 paf 29: append_here=head.rows;
1.123 paf 30: head_link=0;
1.28 paf 31: link_row=&head.rows[head.count];
1.41 paf 32:
33: if(src)
1.75 paf 34: if(tainted)
35: APPEND_TAINTED(src, src_size, 0, 0);
1.41 paf 36: else
1.75 paf 37: APPEND_CLEAN(src, src_size, 0, 0);
1.1 paf 38: }
39:
1.94 parser 40: String::String(const String& src) :
1.120 paf 41: Pooled(src.pool()) {
1.8 paf 42: head.count=CR_PREALLOCATED_COUNT;
43:
1.116 paf 44: uint src_used_rows=src.used_rows();
1.8 paf 45: if(src_used_rows<=head.count) {
1.55 paf 46: // all new rows fit size_to preallocated area
1.98 parser 47: last_chunk=&head;
1.116 paf 48: uint curr_chunk_rows=head.count;
1.8 paf 49: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows);
50: append_here=&head.rows[src_used_rows];
51: link_row=&head.rows[curr_chunk_rows];
52: } else {
53: // warning:
1.10 paf 54: // heavily relies on the fact
55: // "preallocated area is the same for all strings"
1.8 paf 56: //
57: // info:
58: // allocating only enough mem to fit src string rows
59: // next append would allocate a new chunk
60: //
1.55 paf 61: // new rows don't fit size_to preallocated area: splitting size_to two chunks
1.8 paf 62: // preallocated chunk src to constructing head
63: memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count);
1.55 paf 64: // remaining rows size_to new_chunk
1.116 paf 65: uint curr_chunk_rows=src_used_rows-head.count;
1.98 parser 66: last_chunk=static_cast<Chunk *>(
1.116 paf 67: malloc(sizeof(uint)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *), 9));
1.98 parser 68: last_chunk->count=curr_chunk_rows;
1.123 paf 69: head_link=last_chunk;
1.98 parser 70: append_here=link_row=&last_chunk->rows[last_chunk->count];
1.8 paf 71:
1.123 paf 72: Chunk *old_chunk=src.head_link;
1.98 parser 73: Chunk::Row *new_rows=last_chunk->rows;
1.116 paf 74: uint rows_left_to_copy=last_chunk->count;
1.8 paf 75: while(true) {
1.116 paf 76: uint old_count=old_chunk->count;
1.8 paf 77: Chunk *next_chunk=old_chunk->rows[old_count].link;
78: if(next_chunk) {
79: // not last source chunk
80: // taking it all
81: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count);
82: new_rows+=old_count;
83: rows_left_to_copy-=old_count;
84:
85: old_chunk=next_chunk;
86: } else {
87: // the last source chunk
88: // taking only those rows of chunk that _left_to_copy
89: memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy);
90: break;
91: }
92: }
1.5 paf 93: }
1.8 paf 94: link_row->link=0;
1.115 paf 95: src_used_rows;
1.120 paf 96: }
97:
98: size_t String::size() const {
99: size_t result=0;
1.123 paf 100: STRING_FOREACH_ROW(
1.120 paf 101: result+=row->item.size;
1.123 paf 102: );
1.120 paf 103: break2:
104: return result;
1.94 parser 105: }
106:
1.115 paf 107: /// @todo not very optimal
108: uint String::used_rows() const {
109: uint result=0;
1.123 paf 110: STRING_FOREACH_ROW(
111: result++;
112: );
1.115 paf 113: break2:
114: return result;
115: }
1.94 parser 116: void String::expand() {
1.116 paf 117: uint new_chunk_count=last_chunk->count+CR_GROW_COUNT;
1.122 paf 118: if(new_chunk_count>MAX_USHORT)
119: new_chunk_count=MAX_USHORT;
120:
1.94 parser 121: last_chunk=static_cast<Chunk *>(
1.116 paf 122: malloc(sizeof(uint)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10));
1.94 parser 123: last_chunk->count=new_chunk_count;
124: link_row->link=last_chunk;
125: append_here=last_chunk->rows;
126: link_row=&last_chunk->rows[last_chunk->count];
127: link_row->link=0;
1.5 paf 128: }
1.28 paf 129:
1.13 paf 130: String& String::real_append(STRING_APPEND_PARAMS) {
1.9 paf 131: if(!src)
132: return *this;
1.26 paf 133: if(!size)
134: size=strlen(src);
135: if(!size)
1.9 paf 136: return *this;
1.122 paf 137:
138: while(size>MAX_USHORT) {
139: if(chunk_is_full())
140: expand();
141:
142: append_here->item.ptr=src;
143: append_here->item.size=MAX_USHORT;
144: append_here->item.lang=lang;
145: #ifndef NO_STRING_ORIGIN
146: append_here->item.origin.file=file;
147: append_here->item.origin.line=line;
148: #endif
149: append_here++;
150:
151: src+=MAX_USHORT;
152: size-=MAX_USHORT;
153: }
1.9 paf 154:
1.1 paf 155: if(chunk_is_full())
156: expand();
157:
158: append_here->item.ptr=src;
1.121 paf 159: append_here->item.size=size;
1.52 paf 160: append_here->item.lang=lang;
1.13 paf 161: #ifndef NO_STRING_ORIGIN
1.14 paf 162: append_here->item.origin.file=file;
163: append_here->item.origin.line=line;
1.13 paf 164: #endif
1.115 paf 165: append_here++;
1.1 paf 166:
167: return *this;
1.97 parser 168: }
169:
170: char String::first_char() const {
1.120 paf 171: if(!used_rows())
1.112 parser 172: throw Exception(0, 0,
1.97 parser 173: this,
174: "getting first char of empty string");
175:
176: return *head.rows[0].item.ptr;
1.1 paf 177: }
178:
1.16 paf 179: uint String::hash_code() const {
1.7 paf 180: uint result=0;
1.123 paf 181: STRING_FOREACH_ROW(
1.6 paf 182: result=Hash::generic_code(result, row->item.ptr, row->item.size);
1.123 paf 183: );
1.5 paf 184: break2:
185: return result;
186: }
187:
1.60 paf 188: /// @todo move 'lang' skipping to pos
189: int String::cmp(int& partial, const String& src,
190: size_t this_offset, Untaint_lang lang) const {
1.59 paf 191: partial=-1;
1.125 paf 192: size_t a_size=size();
193: this_offset=min(this_offset, a_size-1);
1.55 paf 194:
1.16 paf 195: const Chunk *a_chunk=&head;
196: const Chunk *b_chunk=&src.head;
197: const Chunk::Row *a_row=a_chunk->rows;
198: const Chunk::Row *b_row=b_chunk->rows;
1.55 paf 199: size_t a_offset=this_offset;
200: size_t b_offset=0;
1.9 paf 201: Chunk::Row *a_end=append_here;
202: Chunk::Row *b_end=src.append_here;
1.116 paf 203: uint a_countdown=a_chunk->count;
204: uint b_countdown=b_chunk->count;
205: int result;
1.60 paf 206: size_t pos=0;
1.33 paf 207:
1.125 paf 208: bool a_break=a_size==0;
1.91 parser 209: bool b_break=src.size()==0;
1.83 parser 210: if(!(a_break || b_break)) while(true) {
1.55 paf 211: if(pos+a_row->item.size > this_offset) {
1.71 paf 212: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 213: return -1; // wrong lang -- bail out
214:
1.55 paf 215: int size_diff=
216: (a_row->item.size-a_offset)-
217: (b_row->item.size-b_offset);
218:
219: if(size_diff==0) { // a has same size as b
1.60 paf 220: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
221: a_row->item.size-a_offset);
1.55 paf 222: if(result)
223: return result;
1.60 paf 224: pos+=a_row->item.size;
1.55 paf 225: a_row++; a_countdown--; a_offset=0;
226: b_row++; b_countdown--; b_offset=0;
227: } else if (size_diff>0) { // a longer
1.60 paf 228: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
229: b_row->item.size-b_offset);
1.55 paf 230: if(result)
231: return result;
232: a_offset+=b_row->item.size-b_offset;
233: b_row++; b_countdown--; b_offset=0;
234: } else { // b longer
1.60 paf 235: result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset,
236: a_row->item.size-a_offset);
1.55 paf 237: if(result)
238: return result;
239: b_offset+=a_row->item.size-a_offset;
1.60 paf 240: pos+=a_row->item.size;
1.55 paf 241: a_row++; a_countdown--; a_offset=0;
242: }
1.83 parser 243: if(b_break=b_row==b_end) {
244: a_break=a_row==a_end;
245: break;
246: }
1.55 paf 247: if(!b_countdown) {
248: b_chunk=b_row->link;
249: b_row=b_chunk->rows;
250: b_countdown=b_chunk->count;
251: }
252: } else {
1.60 paf 253: a_offset-=a_row->item.size;
254: pos+=a_row->item.size;
255: a_row++; a_countdown--;
1.9 paf 256: }
257:
1.83 parser 258: if(a_break=a_row==a_end) {
259: b_break=b_row==b_end;
260: break;
261: }
1.11 paf 262: if(!a_countdown) {
1.9 paf 263: a_chunk=a_row->link;
264: a_row=a_chunk->rows;
1.11 paf 265: a_countdown=a_chunk->count;
1.9 paf 266: }
1.27 paf 267: }
1.55 paf 268: if(a_break==b_break) { // ended simultaneously
269: partial=0; return 0;
270: } else if(a_break) { // first bytes equal, but a ended before b
271: partial=1; return -1;
272: } else {
273: partial=2; return +1;
274: }
1.27 paf 275: }
276:
1.60 paf 277: /// @todo move 'lang' skipping to pos
1.59 paf 278: int String::cmp(int& partial, const char* b_ptr, size_t src_size,
1.60 paf 279: size_t this_offset, Untaint_lang lang) const {
1.59 paf 280: partial=-1;
1.125 paf 281: size_t a_size=size();
1.50 paf 282: size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0;
1.125 paf 283: this_offset=min(this_offset, a_size-1);
1.27 paf 284:
285: const Chunk *a_chunk=&head;
286: const Chunk::Row *a_row=a_chunk->rows;
1.59 paf 287: size_t a_offset=this_offset;
1.55 paf 288: size_t b_offset=0;
1.27 paf 289: Chunk::Row *a_end=append_here;
1.116 paf 290: uint a_countdown=a_chunk->count;
1.60 paf 291: size_t pos=0;
1.52 paf 292:
1.125 paf 293: bool a_break=a_size==0;
1.83 parser 294: bool b_break=b_size==0;
295: if(!(a_break || b_break)) while(true) {
1.59 paf 296: if(pos+a_row->item.size > this_offset) {
1.71 paf 297: if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang)
1.60 paf 298: return -1; // wrong lang -- bail out
299:
1.59 paf 300: int size_diff=
301: (a_row->item.size-a_offset)-
302: (b_size-b_offset);
303:
304: if(size_diff==0) { // a has same size as b
1.116 paf 305: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 306: a_row->item.size-a_offset)!=0)
307: return result;
1.60 paf 308: pos+=a_row->item.size;
1.59 paf 309: a_row++; a_countdown--; a_offset=0;
310: b_break=true;
311: } else if (size_diff>0) { // a longer
1.116 paf 312: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 313: b_size-b_offset)!=0)
314: return result;
315: a_offset+=b_size-b_offset;
316: b_break=true;
317: } else { // b longer
1.116 paf 318: if(int result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset,
1.59 paf 319: a_row->item.size-a_offset)!=0)
320: return result;
321: b_offset+=a_row->item.size-a_offset;
1.60 paf 322: pos+=a_row->item.size;
1.59 paf 323: a_row++; a_countdown--; a_offset=0;
324: }
325: } else {
1.60 paf 326: a_offset-=a_row->item.size;
327: pos+=a_row->item.size;
328: a_row++; a_countdown--;
1.27 paf 329: }
330:
1.86 parser 331: a_break=a_row==a_end;
332: if(a_break || b_break)
1.83 parser 333: break;
1.27 paf 334: if(!a_countdown) {
335: a_chunk=a_row->link;
336: a_row=a_chunk->rows;
337: a_countdown=a_chunk->count;
1.9 paf 338: }
339: }
1.55 paf 340: if(a_break==b_break) { // ended simultaneously
341: partial=0; return 0;
342: } else if(a_break) { // first bytes equal, but a ended before b
343: partial=1; return -1;
344: } else {
345: partial=2; return +1;
346: }
1.5 paf 347: }
1.46 paf 348:
349: #ifndef NO_STRING_ORIGIN
350: const Origin& String::origin() const {
1.120 paf 351: if(!used_rows()) {
1.96 parser 352: static const Origin empty_origin={"empty string"};
353: return empty_origin;
354: }
1.46 paf 355:
1.49 paf 356: // determining origin by last appended piece
1.50 paf 357: // because first one frequently constant.
358: // ex: ^load[/file] "document_root" + "/file"
1.80 paf 359: // when last peice is constant,
360: // ex: parser_root_auto_path{dynamic} / auto.p{const}
361: // using first piece
362: Origin& last_origin=append_here[-1].item.origin;
363: return last_origin.file ? last_origin : head.rows[0].item.origin;
1.46 paf 364: }
365: #endif
1.53 paf 366:
1.69 paf 367: String& String::mid(size_t start, size_t finish) const {
1.107 parser 368: String& result=*NEW String(pool());
369:
1.53 paf 370: start=max(0, start);
1.111 parser 371: finish=min(size(), finish);
1.60 paf 372: if(start==finish)
1.107 parser 373: return result;
1.53 paf 374:
375: size_t pos=0;
1.123 paf 376: STRING_FOREACH_ROW(
377: size_t item_finish=pos+row->item.size;
378: if(item_finish > start) { // started now or already?
379: bool started=result.size()==0; // started now?
380: bool finished=finish <= item_finish; // finished now?
381: size_t offset=started?start-pos:0;
382: size_t size=finished?finish-pos:row->item.size;
383: result.APPEND(
384: row->item.ptr+offset, size-offset,
385: row->item.lang,
386: row->item.origin.file, row->item.origin.line);
387: if(finished)
1.53 paf 388: goto break2;
389: }
1.123 paf 390: pos+=row->item.size;
391: );
1.53 paf 392: break2:
1.60 paf 393: // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'",
394: //cstr(), start, finish, result.cstr());
1.53 paf 395: return result;
1.54 paf 396: }
397:
1.60 paf 398: int String::pos(const String& substr,
1.116 paf 399: int result, Untaint_lang lang) const {
1.125 paf 400: size_t self_size=size();
401: for(; result<self_size; result++) {
1.60 paf 402: int partial; cmp(partial, substr, result, lang);
1.58 paf 403: if(
404: partial==0 || // full match
405: partial==2) // 'substr' starts 'this'+'result'
406: return result;
407: }
408:
409: return -1;
410: }
411:
1.60 paf 412: int String::pos(const char *substr, size_t substr_size,
1.116 paf 413: int result, Untaint_lang lang) const {
1.125 paf 414: size_t self_size=size();
415: for(; result<self_size; result++) {
1.60 paf 416: int partial; cmp(partial, substr, substr_size, result, lang);
1.55 paf 417: if(
418: partial==0 || // full match
419: partial==2) // 'substr' starts 'this'+'result'
420: return result;
421: }
422:
423: return -1;
1.60 paf 424: }
425:
426: void String::split(Array& result,
427: size_t* pos_after_ref,
428: const char *delim, size_t delim_size,
429: Untaint_lang lang, int limit) const {
1.125 paf 430: size_t self_size=size();
1.60 paf 431: if(delim_size) {
432: size_t pos_after=pos_after_ref?*pos_after_ref:0;
433: int pos_before;
434: // while we have 'delim'...
435: for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 436: result+=&mid(pos_after, pos_before);
1.60 paf 437: pos_after=pos_before+delim_size;
438: }
439: // last piece
1.124 paf 440: if(pos_after<self_size && limit) {
441: result+=&mid(pos_after, self_size);
442: pos_after=self_size;
1.60 paf 443: }
444: if(pos_after_ref)
445: *pos_after_ref=pos_after;
446: } else { // empty delim
447: result+=this;
448: if(pos_after_ref)
1.124 paf 449: *pos_after_ref+=self_size;
1.60 paf 450: }
451: }
452:
453: void String::split(Array& result,
454: size_t* pos_after_ref,
455: const String& delim, Untaint_lang lang,
456: int limit) const {
457: if(delim.size()) {
458: size_t pos_after=pos_after_ref?*pos_after_ref:0;
459: int pos_before;
460: // while we have 'delim'...
461: for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
1.69 paf 462: result+=&mid(pos_after, pos_before);
1.60 paf 463: pos_after=pos_before+delim.size();
464: }
465: // last piece
466: if(pos_after<size() && limit) {
1.69 paf 467: result+=&mid(pos_after, size());
1.60 paf 468: pos_after=size();
469: }
470: if(pos_after_ref)
471: *pos_after_ref=pos_after;
472: } else { // empty delim
473: result+=this;
474: if(pos_after_ref)
475: *pos_after_ref+=size();
476: }
1.61 paf 477: }
478:
1.63 paf 479: static void regex_options(char *options, int *result){
480: struct Regex_option {
481: char key;
482: int clear, set;
483: int *result;
484: } regex_option[]={
485: {'i', 0, PCRE_CASELESS, result}, // a=A
1.79 paf 486: {'s', 0, PCRE_DOTALL, result}, // \n\n$ [default]
1.63 paf 487: {'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
488: {'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
489: {'g', 0, true, result+1}, // many rows
490: {0},
491: };
492: result[0]=PCRE_EXTRA | PCRE_DOTALL;
493: result[1]=0;
494:
495: if(options)
496: for(Regex_option *o=regex_option; o->key; o++)
497: if(
498: strchr(options, o->key) ||
499: strchr(options, toupper(o->key))) {
500: *(o->result)&=~o->clear;
501: *(o->result)|=o->set;
502: }
503: }
504:
1.88 parser 505: /// @todo maybe need speedup: some option to remove pre/match/post string generation
1.77 paf 506: bool String::match(const unsigned char *pcre_tables,
507: const String *aorigin,
1.62 paf 508: const String& regexp,
1.63 paf 509: const String *options,
1.64 paf 510: Table **table,
1.95 parser 511: Row_action row_action, void *info,
512: bool *was_global) const {
1.64 paf 513:
1.73 paf 514: if(!regexp.size())
1.112 parser 515: throw Exception(0, 0,
1.73 paf 516: aorigin,
517: "regexp is empty");
1.118 paf 518: const char *pattern=regexp.cstr();
1.62 paf 519: const char *errptr;
520: int erroffset;
1.63 paf 521: int option_bits[2]; regex_options(options?options->cstr():0, option_bits);
1.95 parser 522: if(was_global)
523: *was_global=option_bits[1]!=0;
1.63 paf 524: pcre *code=pcre_compile(pattern, option_bits[0],
1.62 paf 525: &errptr, &erroffset,
1.74 paf 526: pcre_tables);
1.62 paf 527:
1.67 paf 528: if(!code)
1.112 parser 529: throw Exception(0, 0,
1.69 paf 530: ®exp.mid(erroffset, regexp.size()),
1.74 paf 531: "regular expression syntax error - %s", errptr);
1.62 paf 532:
1.63 paf 533: int info_substrings=pcre_info(code, 0, 0);
534: if(info_substrings<0) {
1.100 parser 535: pcre_free(code);
1.112 parser 536: throw Exception(0, 0,
1.73 paf 537: aorigin,
1.76 paf 538: "pcre_info error (%d)",
1.73 paf 539: info_substrings);
1.63 paf 540: }
541:
542: int startoffset=0;
1.118 paf 543: const char *subject=cstr();
1.62 paf 544: int length=strlen(subject);
1.63 paf 545: int ovecsize;
546: int *ovector=(int *)malloc(sizeof(int)*
1.114 paf 547: (ovecsize=(1/*match*/+info_substrings)*3), 11);
1.62 paf 548:
1.64 paf 549: { // create table
550: Array& columns=*NEW Array(pool());
551: columns+=string_pre_match_name;
552: columns+=string_match_name;
553: columns+=string_post_match_name;
554: for(int i=1; i<=info_substrings; i++) {
555: char *column=(char *)malloc(MAX_NUMBER);
556: snprintf(column, MAX_NUMBER, "%d", i);
557: columns+=NEW String(pool(), column); // .i column name
558: }
559: *table=NEW Table(pool(), aorigin, &columns);
1.62 paf 560: }
1.63 paf 561:
1.64 paf 562: int exec_option_bits=0;
1.63 paf 563: while(true) {
564: int exec_substrings=pcre_exec(code, 0,
565: subject, length, startoffset,
1.64 paf 566: exec_option_bits, ovector, ovecsize);
1.63 paf 567:
568: if(exec_substrings==PCRE_ERROR_NOMATCH) {
1.100 parser 569: pcre_free(code);
570: row_action(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 571: return option_bits[1]!=0; // global=true+table, not global=false
572: }
573:
574: if(exec_substrings<0) {
1.100 parser 575: pcre_free(code);
1.112 parser 576: throw Exception(0, 0,
1.63 paf 577: aorigin,
1.76 paf 578: "regular expression execute error (%d)",
1.63 paf 579: exec_substrings);
580: }
581:
582: Array& row=*NEW Array(pool());
1.81 paf 583: row+=&mid(0, ovector[0]); // .prematch column value
1.69 paf 584: row+=&mid(ovector[0], ovector[1]); // .match
1.81 paf 585: row+=&mid(ovector[1], size()); // .postmatch
1.63 paf 586:
587: for(int i=1; i<exec_substrings; i++) {
1.69 paf 588: // -1:-1 case handled peacefully by mid() itself
589: row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
1.63 paf 590: }
591:
1.100 parser 592: row_action(**table, &row, startoffset, ovector[0], info);
1.63 paf 593:
1.100 parser 594: if(!option_bits[1] || startoffset==ovector[1]) { // not global | going to hang
595: pcre_free(code);
596: row_action(**table, 0/*last time, no row*/, 0, 0, info);
1.63 paf 597: return true;
598: }
1.100 parser 599: startoffset=ovector[1];
1.63 paf 600:
601: /*
602: if(option_bits[0] & PCRE_MULTILINE)
1.64 paf 603: exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
1.63 paf 604: */
605: }
1.82 parser 606: }
607:
608: String& String::change_case(Pool& pool, const unsigned char *tables,
609: Change_case_kind kind) const {
610: String& result=*new(pool) String(pool);
611:
612: const unsigned char *a;
613: const unsigned char *b;
614: switch(kind) {
615: case CC_UPPER:
616: a=tables+lcc_offset;
617: b=tables+fcc_offset;
618: break;
619: case CC_LOWER:
620: a=tables+lcc_offset;
621: b=0;
622: break;
623: default:
1.112 parser 624: throw Exception(0, 0,
1.82 parser 625: this,
626: "unknown change case kind #%d",
627: static_cast<int>(kind)); // never
628: a=b=0; // calm, compiler
629: break; // never
630: }
631:
632: const Chunk *chunk=&head;
633: do {
634: const Chunk::Row *row=chunk->rows;
1.116 paf 635: for(uint i=0; i<chunk->count; i++, row++) {
1.82 parser 636: if(row==append_here)
637: goto break2;
638:
1.114 paf 639: char *new_cstr=(char *)pool.malloc(row->item.size, 12);
1.82 parser 640: char *dest=new_cstr;
641: const char *src=row->item.ptr;
642: for(int size=row->item.size; size--; src++) {
643: unsigned char c=a[(unsigned char)*src];
644: if(b)
645: c=b[c];
646:
647: *dest++=(char)c;
648: }
649:
650: result.APPEND(new_cstr, row->item.size,
1.123 paf 651: row->item.lang,
1.82 parser 652: row->item.origin.file, row->item.origin.line);
653: }
654: chunk=row->link;
655: } while(chunk);
656: break2:
1.89 parser 657:
1.101 parser 658: return result;
659: }
660:
1.108 parser 661: void String::join_chain(Pool& pool,
1.116 paf 662: uint& ai, const Chunk*& achunk, const Chunk::Row*& arow,
1.123 paf 663: uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const {
664: joined_lang=arow->item.lang;
1.108 parser 665:
666: // calc size
667: joined_size=0;
668: {
1.116 paf 669: uint start_i=ai;
1.108 parser 670: const Chunk::Row *start_row=arow;
671: const Chunk *chunk=achunk;
672: do {
673: const Chunk::Row *row=start_row;
1.116 paf 674: for(uint i=start_i; i<chunk->count; i++, row++) {
1.108 parser 675: if(row==append_here)
676: goto break21;
677:
678: if(row->item.lang==joined_lang)
679: joined_size+=row->item.size;
680: else
681: break;
682: }
683: if(chunk=row->link) {
684: start_i=0;
685: start_row=chunk->rows;
686: } else
687: break;
688: } while(true);
689: break21:;
690: }
691:
692: // if one row, return simply itself
693: if(joined_size==arow->item.size) {
694: joined_ptr=arow->item.ptr;
695: ai++; arow++;
696: if(ai==achunk->count)
697: achunk=arow->link;
698: } else {
699: // join adjacent rows
1.114 paf 700: char *ptr=(char *)pool.malloc(joined_size,13);
1.108 parser 701: joined_ptr=ptr;
1.116 paf 702: uint start_i=ai;
1.108 parser 703: const Chunk::Row *start_row=arow;
704: const Chunk *chunk=achunk;
1.116 paf 705: uint i;
1.108 parser 706: const Chunk::Row *row;
707: do {
708: row=start_row;
709: for(i=start_i; i<chunk->count; i++, row++) {
710: if(row==append_here)
711: goto break22;
712:
713: if(row->item.lang==joined_lang) {
714: memcpy(ptr, row->item.ptr, row->item.size);
715: ptr+=row->item.size;
716: } else
717: break;
718: }
719: if(chunk=row->link) {
720: start_i=0;
721: start_row=chunk->rows;
722: } else
723: break;
724: } while(true);
725: break22:;
726:
727: // return joined rows
728: ai=i;
729: arow=row;
730: achunk=chunk;
731: }
732: }
733:
734: String& String::reconstruct(Pool& pool) const {
735: //_asm int 3;
736: String& result=*new(pool) String(pool);
737: const Chunk *chunk=&head;
738: do {
739: const Chunk::Row *row=chunk->rows;
1.116 paf 740: for(uint i=0; i<chunk->count; ) {
1.108 parser 741: if(row==append_here)
742: goto break2;
743:
1.123 paf 744: uchar joined_lang;
1.109 parser 745: const char *joined_ptr;
1.108 parser 746: size_t joined_size;
747: join_chain(pool, i, chunk, row,
748: joined_lang, joined_ptr, joined_size);
749:
750: result.APPEND(joined_ptr, joined_size,
751: joined_lang,
752: row->item.origin.file, row->item.origin.line);
753: if(!chunk)
754: goto break2;
755: }
756: } while(true);
757: break2:
758:
759: return result;
760: };
761:
762: String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const {
1.106 parser 763: //_asm int 3;
1.101 parser 764: String& result=*new(pool) String(pool);
1.123 paf 765: STRING_FOREACH_ROW(
766: const char *src=row->item.ptr;
767: size_t src_size=row->item.size;
768: char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14);
769: char *dest=new_cstr;
770: while(src_size) {
771: // there is a row where first column starts 'src'
772: if(Table::Item *item=dict.first_that_starts(src, src_size)) {
773: // get a=>b values
774: const String& a=*static_cast<Array *>(item)->get_string(0);
775: const String& b=*static_cast<Array *>(item)->get_string(1);
776: // skip 'a' in 'src' && reduce work size
777: src+=a.size(); src_size-=a.size();
778: // write 'b' to 'dest' && skip 'b' in 'dest'
779: b.store_to(dest); dest+=b.size();
780: } else {
781: // write a char to b && reduce work size
782: *dest++=*src++; src_size--;
1.101 parser 783: }
784: }
785:
1.123 paf 786: result.APPEND(new_cstr, dest-new_cstr,
787: row->item.lang,
788: row->item.origin.file, row->item.origin.line);
789: );
1.101 parser 790: break2:
1.89 parser 791: return result;
1.108 parser 792: }
793:
794: String& String::replace(Pool& pool, Dictionary& dict) const {
795: return reconstruct(pool).replace_in_reconstructed(pool, dict);
1.89 parser 796: }
797:
1.90 parser 798: double String::as_double() const {
1.89 parser 799: double result;
1.114 paf 800: const char *cstr;
801: char buf[MAX_NUMBER];
1.115 paf 802: if(head.rows+1==append_here) {
1.114 paf 803: int size=min(head.rows[0].item.size, MAX_NUMBER-1);
804: memcpy(buf, head.rows[0].item.ptr, size);
805: buf[size]=0;
806: cstr=buf;
807: } else
808: cstr=this->cstr();
1.102 parser 809: char *error_pos;
1.89 parser 810: // 0xABC
1.99 parser 811: if(cstr[0]=='0')
812: if(cstr[1]=='x' || cstr[1]=='X')
813: result=(double)(unsigned long)strtol(cstr, &error_pos, 0);
814: else
1.102 parser 815: result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos);
1.89 parser 816: else
1.99 parser 817: result=(double)strtod(cstr, &error_pos);
1.89 parser 818:
1.103 parser 819: if(*error_pos/*not EOS*/)
1.112 parser 820: throw Exception(0, 0,
1.89 parser 821: this,
822: "invalid number (double)");
823:
824: return result;
825: }
1.90 parser 826: int String::as_int() const {
1.89 parser 827: int result;
1.114 paf 828: const char *cstr;
829: char buf[MAX_NUMBER];
1.115 paf 830: if(head.rows+1==append_here) {
1.114 paf 831: int size=min(head.rows[0].item.size, MAX_NUMBER-1);
832: memcpy(buf, head.rows[0].item.ptr, size);
833: buf[size]=0;
834: cstr=buf;
835: } else
836: cstr=this->cstr();
1.102 parser 837: char *error_pos;
1.89 parser 838: // 0xABC
1.99 parser 839: if(cstr[0]=='0')
840: if(cstr[1]=='x' || cstr[1]=='X')
841: result=(int)(unsigned long)strtol(cstr, &error_pos, 0);
842: else
1.102 parser 843: result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0);
1.89 parser 844: else
845: result=(int)strtol(cstr, &error_pos, 0);
846:
1.103 parser 847: if(*error_pos/*not EOS*/)
1.112 parser 848: throw Exception(0, 0,
1.89 parser 849: this,
850: "invalid number (int)");
1.82 parser 851:
852: return result;
1.61 paf 853: }
1.113 parser 854:
1.128 ! paf 855: inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) {
! 856: byte1=word&0xFF;
! 857: byte2=word>>8;
! 858: }
! 859: inline ushort uchars2ushort(uchar byte1, uchar byte2) {
! 860: return (byte2<<8) | byte1;
! 861: }
1.113 parser 862: /* @todo maybe network order worth spending some effort?
863: don't bothering myself with network byte order,
864: am not planning to be able to move resulting file across platforms
865: for now
866: */
867: void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const {
868: buf_size=
869: prolog_size
1.126 paf 870: +used_rows()*(sizeof(uchar)+sizeof(ushort))
1.113 parser 871: +size();
1.114 paf 872: buf=malloc(buf_size,15);
1.113 parser 873: char *cur=(char *)buf+prolog_size;
874:
1.123 paf 875: STRING_FOREACH_ROW(
876: // lang
1.126 paf 877: memcpy(cur, &row->item.lang, sizeof(uchar));
878: cur+=sizeof(uchar);
1.123 paf 879: // size
1.128 ! paf 880: uchar byte1; uchar byte2;
! 881: ushort2uchars(row->item.size, byte1, byte2);
! 882: memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar);
! 883: memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar);
1.123 paf 884: // bytes
885: memcpy(cur, row->item.ptr, row->item.size);
886: cur+=row->item.size;
887: );
1.113 parser 888: break2:
889: ;
890: }
891: void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) {
1.126 paf 892: char *cur=(char *)buf+prolog_size;
1.113 parser 893: buf_size-=prolog_size;
894:
895: while(buf_size) {
1.128 ! paf 896: uchar lang=*(uchar *)(cur);
! 897:
! 898: ushort size=uchars2ushort(
! 899: *(uchar*)(cur+sizeof(uchar)*1),
! 900: *(uchar*)(cur+sizeof(uchar)*2)
! 901: );
! 902:
! 903: const char *ptr=(const char*)(cur+sizeof(uchar)*3);
1.126 paf 904: APPEND(ptr, size, lang, file, 0);
1.113 parser 905:
1.126 paf 906: size_t piece_size=sizeof(uchar)+sizeof(ushort)+size;
1.113 parser 907: cur+=piece_size;
908: buf_size-=piece_size;
909: }
910: }
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