![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_hash.h CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / include|
Return to pa_hash.h CVS log | Up to [parser3project] / parser3 / src / include |
1.28 paf 1: /** @file
1.29 paf 2: Parser: hash class decl.
3:
1.66 paf 4: Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com)
1.29 paf 5:
1.54 paf 6: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1 paf 7: */
8:
1.59 paf 9: /*
10: The prime numbers used from zend_hash.c,
11: the part of Zend scripting engine library,
12: Copyrighted (C) 1999-2000 Zend Technologies Ltd.
13: http://www.zend.com/license/0_92.txt
14: For more information about Zend please visit http://www.zend.com/
15: */
16:
1.1 paf 17: #ifndef PA_HASH_H
18: #define PA_HASH_H
1.56 paf 19:
1.69 ! misha 20: static const char * const IDENT_HASH_H="$Date: 2007/06/19 16:40:14 $";
1.1 paf 21:
1.59 paf 22: #include "pa_memory.h"
1.1 paf 23: #include "pa_types.h"
1.59 paf 24:
25: const int HASH_ALLOCATES_COUNT=29;
1.1 paf 26:
1.61 paf 27: /** Zend comment: Generated on an Octa-ALPHA 300MHz CPU & 2.5GB RAM monster
28:
29: paf: HPUX ld could not handle static member: unsatisfied symbols
30: */
31: static uint Hash_allocates[HASH_ALLOCATES_COUNT]={
32: 5, 11, 19, 53, 107, 223, 463, 983, 1979, 3907, 7963,
33: 16229, 32531, 65407, 130987, 262237, 524521, 1048793,
34: 2097397, 4194103, 8388857, 16777447, 33554201, 67108961,
35: 134217487, 268435697, 536870683, 1073741621, 2147483399};
36:
1.68 misha 37: /// useful generic hash function
38: inline void generic_hash_code(uint& result, char c) {
39: result=(result<<4)+c;
40: if(uint g=(result&0xF0000000)) {
41: result=result^(g>>24);
42: result=result^g;
43: }
44: }
45: /// useful generic hash function
46: inline void generic_hash_code(uint& result, const char* s) {
47: while(char c=*s++) {
48: result=(result<<4)+c;
49: if(uint g=(result&0xF0000000)) {
50: result=result^(g>>24);
51: result=result^g;
52: }
53: }
54: }
55:
56: /// useful generic hash function
57: inline void generic_hash_code(uint& result, const char* buf, size_t size) {
58: const char* end=buf+size;
59: while(buf<end) {
60: result=(result<<4)+*buf++;
61: if(uint g=(result&0xF0000000)) {
62: result=result^(g>>24);
63: result=result^g;
64: }
65: }
66: }
67:
68: /// simple hash code of int. used by EXIF mapping
69: inline uint hash_code(int self) {
70: uint result=0;
71: generic_hash_code(result, (const char*)&self, sizeof(self));
72: return result;
73: }
74:
1.29 paf 75: /**
1.59 paf 76: Simple hash.
1.29 paf 77:
1.59 paf 78: Automatically rehashed when almost is_full.
1.51 paf 79: Contains no 0 values.
80: get returning 0 means there were no such.
81: "put value 0" means "remove"
1.29 paf 82: */
1.59 paf 83: template<typename K, typename V> class Hash: public PA_Object {
1.1 paf 84: public:
85:
1.59 paf 86: typedef K key_type;
87: typedef V value_type;
1.3 paf 88:
1.59 paf 89: Hash() {
1.61 paf 90: allocated=Hash_allocates[allocates_index=0];
1.59 paf 91: threshold=allocated*THRESHOLD_PERCENT/100;
92: fpairs_count=fused_refs=0;
93: refs=new(UseGC) Pair*[allocated];
94: }
1.25 paf 95:
1.59 paf 96: Hash(const Hash& source) {
97: allocates_index=source.allocates_index;
98: allocated=source.allocated;
99: threshold=source.threshold;
100: fused_refs=source.fused_refs;
101: fpairs_count=source.fpairs_count;
102: refs=new(UseGC) Pair*[allocated];
103:
104: // clone & rehash
105: Pair **old_ref=source.refs;
106: for(int index=0; index<allocated; index++)
107: for(Pair *pair=*old_ref++; pair; ) {
108: Pair *next=pair->link;
1.45 paf 109:
1.59 paf 110: Pair **new_ref=&refs[index];
111: *new_ref=new Pair(pair->code, pair->key, pair->value, *new_ref);
1.38 paf 112:
1.59 paf 113: pair=next;
114: }
1.43 parser 115: }
116:
1.59 paf 117: /// put a [value] under the [key] @returns existed or not
118: bool put(K key, V value) {
119: if(!value) {
120: remove(key);
121: return false;
122: }
123: if(is_full())
124: expand();
125:
126: uint code=hash_code(key);
127: uint index=code%allocated;
128: Pair **ref=&refs[index];
129: for(Pair *pair=*ref; pair; pair=pair->link)
130: if(pair->code==code && pair->key==key) {
131: // found a pair with the same key
132: pair->value=value;
133: return true;
134: }
135:
136: // proper pair not found -- create&link_in new pair
137: if(!*ref) // root cell were fused_refs?
138: fused_refs++; // not, we'll use it and record the fact
139: *ref=new Pair(code, key, value, *ref);
140: fpairs_count++;
141: return false;
1.24 paf 142: }
1.10 paf 143:
1.63 paf 144: /// put a [value] under the [key] @returns existed or not
1.65 paf 145: template<typename R, typename F, typename I> R replace_maybe_append(K key, V value, F prevent, I info) {
146: if(!value) {
147: // they can come here from somewhere (true with maybe_replace_maybe_append, keeping parallel)
148: remove(key);
149: // this has nothing to do with properties, doing no special property handling here
150: return 0;
151: }
1.64 paf 152:
1.63 paf 153: if(is_full())
154: expand();
155:
156: uint code=hash_code(key);
157: uint index=code%allocated;
158: Pair **ref=&refs[index];
159: for(Pair *pair=*ref; pair; pair=pair->link)
160: if(pair->code==code && pair->key==key) {
161: // found a pair with the same key
1.65 paf 162: pair->value=value;
163: return reinterpret_cast<R>(1);
164: }
165:
166: // proper pair not found
167: // prevent-function intercepted append?
168: if(R result=prevent(value, info))
169: return result;
170:
171: //create&link_in new pair
172: if(!*ref) // root cell were fused_refs?
173: fused_refs++; // not, we'll use it and record the fact
174: *ref=new Pair(code, key, value, *ref);
175: fpairs_count++;
176: return 0;
177: }
1.63 paf 178:
1.65 paf 179: /// put a [value] under the [key] @returns existed or not
180: template<typename R, typename F1, typename F2, typename I>
181: R maybe_replace_maybe_append(K key, V value, F1 prevent_replace, F2 prevent_append, I info)
182: {
183: if(!value) {
184: // they can come here from Temp_value_element::dctor to restore some empty value
185: remove(key);
186: // this has nothing to do with properties, doing no special property handling here
187: return 0;
188: }
189:
190: if(is_full())
191: expand();
192:
193: uint code=hash_code(key);
194: uint index=code%allocated;
195: Pair **ref=&refs[index];
196: for(Pair *pair=*ref; pair; pair=pair->link)
197: if(pair->code==code && pair->key==key) {
198: // found a pair with the same key
199:
200: // prevent-function intercepted replace?
201: if(R result=prevent_replace(pair->value, info))
1.63 paf 202: return result;
203:
204: pair->value=value;
205: return reinterpret_cast<R>(1);
206: }
207:
1.65 paf 208: // proper pair not found
209: // prevent-function intercepted append?
210: if(R result=prevent_append(value, info))
211: return result;
212:
213: //create&link_in new pair
1.63 paf 214: if(!*ref) // root cell were fused_refs?
215: fused_refs++; // not, we'll use it and record the fact
216: *ref=new Pair(code, key, value, *ref);
217: fpairs_count++;
218: return 0;
219: }
220:
1.65 paf 221: /// put a [value] under the [key] @returns existed or not
222: template<typename R, typename F1, typename I>
223: R maybe_replace_never_append(K key, V value, F1 prevent_replace, I info)
224: {
225: if(!value) {
226: // they can come here from somewhere (true with maybe_replace_maybe_append, keeping parallel)
227: remove(key);
228: // this has nothing to do with properties, doing no special property handling here
229: return 0;
230: }
231:
232: if(is_full())
233: expand();
234:
235: uint code=hash_code(key);
236: uint index=code%allocated;
237: Pair **ref=&refs[index];
238: for(Pair *pair=*ref; pair; pair=pair->link)
239: if(pair->code==code && pair->key==key) {
240: // found a pair with the same key
241:
242: // prevent-function intercepted replace?
243: if(R result=prevent_replace(pair->value, info))
244: return result;
245:
246: pair->value=value;
247: return reinterpret_cast<R>(1);
248: }
249:
250: return 0;
251: }
252:
1.59 paf 253: /// remove the [key] @returns existed or not
254: bool remove(K key) {
255: uint code=hash_code(key);
256: uint index=code%allocated;
257: for(Pair **ref=&refs[index]; *ref; ref=&(*ref)->link)
258: if((*ref)->code==code && (*ref)->key==key) {
259: // found a pair with the same key
260: Pair *next=(*ref)->link;
261: delete *ref;
262: *ref=next;
263: --fpairs_count;
264: return true;
265: }
1.8 paf 266:
1.59 paf 267: return false;
268: }
1.48 paf 269:
1.67 misha 270: /// returns exist key or not
1.69 ! misha 271: bool contains(K key){
1.67 misha 272: uint code=hash_code(key);
273: uint index=code%allocated;
274: for(Pair **ref=&refs[index]; *ref; ref=&(*ref)->link){
275: if((*ref)->code==code && (*ref)->key==key) {
276: return true;
277: }
278: }
279:
280: return false;
281: }
282:
283:
1.59 paf 284: /// get associated [value] by the [key]
285: V get(K key) const {
286: uint code=hash_code(key);
287: uint index=code%allocated;
288: for(Pair *pair=refs[index]; pair; pair=pair->link)
289: if(pair->code==code && pair->key==key)
290: return pair->value;
291:
292: return V(0);
1.33 paf 293: }
1.65 paf 294:
1.51 paf 295: /// put a [value] under the [key] if that [key] existed @returns existed or not
1.63 paf 296: bool put_replaced(K key, V value) {
1.59 paf 297: if(!value) {
298: remove(key);
299: return false;
300: }
301: uint code=hash_code(key);
302: uint index=code%allocated;
303: for(Pair *pair=refs[index]; pair; pair=pair->link)
304: if(pair->code==code && pair->key==key) {
305: // found a pair with the same key, replacing
306: pair->value=value;
307: return true;
308: }
309:
310: // proper pair not found
311: return false;
1.64 paf 312: }
313:
314: /// put a [value] under the [key] if that [key] existed @returns existed or not
315: template<typename R, typename F> R maybe_put_replaced(K key, V value, F prevent) {
1.65 paf 316: if(!value) {
317: // they can come here from Temp_value_element::dctor to restore some empty value
318: remove(key);
319: // this has nothing to do with properties, doing no special property handling here
320: return 0;
321: }
1.64 paf 322:
323: uint code=hash_code(key);
324: uint index=code%allocated;
325: for(Pair *pair=refs[index]; pair; pair=pair->link)
326: if(pair->code==code && pair->key==key) {
327: // found a pair with the same key, replacing
328: // prevent-function intercepted put?
329: if(R result=prevent(pair->value))
330: return result;
331:
332: pair->value=value;
333: return reinterpret_cast<R>(1);
334: }
335:
336: // proper pair not found
337: return 0;
1.59 paf 338: }
1.18 paf 339:
1.51 paf 340: /// put a [value] under the [key] if that [key] NOT existed @returns existed or not
1.59 paf 341: bool put_dont_replace(K key, V value) {
342: if(!value) {
343: remove(key);
344: return false;
345: }
346: if(is_full())
347: expand();
348:
349: uint code=hash_code(key);
350: uint index=code%allocated;
351: Pair **ref=&refs[index];
352: for(Pair *pair=*ref; pair; pair=pair->link)
353: if(pair->code==code && pair->key==key) {
354: // found a pair with the same key, NOT replacing
355: return true;
356: }
357:
358: // proper pair not found -- create&link_in new pair
359: if(!*ref) // root cell were fused_refs?
360: fused_refs++; // not, we'll use it and record the fact
361: *ref=new Pair(code, key, value, *ref);
362: fpairs_count++;
363: return false;
364: }
1.18 paf 365:
1.59 paf 366: /** put all 'src' values if NO with same key existed
367: @todo optimize this.allocated==src.allocated case
368: */
369: void merge_dont_replace(const Hash& src) {
370: for(int i=0; i<src.allocated; i++)
371: for(Pair *pair=src.refs[i]; pair; pair=pair->link)
372: put_dont_replace(pair->key, pair->value);
1.36 paf 373: }
1.11 paf 374:
1.29 paf 375: /// number of elements in hash
1.59 paf 376: int count() const { return fpairs_count; }
1.25 paf 377:
1.59 paf 378: /// iterate over all pairs
379: template<typename I> void for_each(void callback(K, V, I), I info) const {
380: Pair **ref=refs;
381: for(int index=0; index<allocated; index++)
382: for(Pair *pair=*ref++; pair; pair=pair->link)
383: callback(pair->key, pair->value, info);
384: }
1.45 paf 385:
1.59 paf 386: /// iterate over all pairs
387: template<typename I> void for_each_ref(void callback(K, V&, I), I info) const {
388: Pair **ref=refs;
389: for(int index=0; index<allocated; index++)
390: for(Pair *pair=*ref++; pair; pair=pair->link)
391: callback(pair->key, pair->value, info);
392: }
1.38 paf 393:
1.59 paf 394: /// iterate over all pairs until condition becomes true, return that element
395: template<typename I> V first_that(bool callback(K, V, I), I info) const {
396: Pair **ref=refs;
397: for(int index=0; index<allocated; index++)
398: for(Pair *pair=*ref++; pair; pair=pair->link)
399: if(callback(pair->key, pair->value, info))
400: return pair->value;
401:
402: return V(0);
403: }
1.27 paf 404:
1.29 paf 405: /// remove all elements
1.59 paf 406: void clear() {
407: memset(refs, 0, sizeof(*refs)*allocated);
408: fpairs_count=fused_refs=0;
409: }
1.15 paf 410:
1.1 paf 411: private:
412:
1.39 paf 413: /// expand when these %% of allocated exausted
1.1 paf 414: enum {
415: THRESHOLD_PERCENT=75
416: };
1.9 paf 417:
1.61 paf 418: /// the index of [allocated] in [Hash_allocates]
1.19 paf 419: int allocates_index;
1.1 paf 420:
1.39 paf 421: /// number of allocated pairs
1.19 paf 422: int allocated;
1.1 paf 423:
1.59 paf 424: /// helper: expanding when fused_refs == threshold
1.1 paf 425: int threshold;
426:
1.39 paf 427: /// used pairs
1.59 paf 428: int fused_refs;
1.44 parser 429:
430: /// stored pairs total (including those by links)
1.59 paf 431: int fpairs_count;
1.1 paf 432:
1.39 paf 433: /// pair storage
1.59 paf 434: class Pair: public PA_Allocated {
435: public:
1.1 paf 436: uint code;
1.59 paf 437: K key;
438: V value;
1.1 paf 439: Pair *link;
1.2 paf 440:
1.59 paf 441: Pair(uint acode, K akey, V avalue, Pair *alink) :
1.1 paf 442: code(acode),
443: key(akey),
444: value(avalue),
1.2 paf 445: link(alink) {}
446: } **refs;
1.1 paf 447:
1.39 paf 448: /// filled to threshold: needs expanding
1.59 paf 449: bool is_full() { return fused_refs==threshold; }
1.5 paf 450:
1.39 paf 451: /// allocate larger buffer & rehash
1.59 paf 452: void expand() {
453: int old_allocated=allocated;
454: Pair **old_refs=refs;
455:
456: allocates_index=allocates_index+1<HASH_ALLOCATES_COUNT?allocates_index+1:HASH_ALLOCATES_COUNT-1;
457: // allocated bigger refs array
1.61 paf 458: allocated=Hash_allocates[allocates_index];
1.59 paf 459: threshold=allocated*THRESHOLD_PERCENT/100;
460: refs=new(UseGC) Pair*[allocated];
461:
462: // rehash
463: Pair **old_ref=old_refs;
464: for(int old_index=0; old_index<old_allocated; old_index++)
465: for(Pair *pair=*old_ref++; pair; ) {
466: Pair *next=pair->link;
467:
468: uint new_index=pair->code%allocated;
469: Pair **new_ref=&refs[new_index];
470: pair->link=*new_ref;
471: *new_ref=pair;
472:
473: pair=next;
474: }
475:
476: delete[] old_refs;
477: }
1.4 paf 478:
479: private: //disabled
480:
1.12 paf 481: Hash& operator = (const Hash&) { return *this; }
1.1 paf 482: };
1.59 paf 483:
484: /// Auto-object used to temporarily substituting/removing hash values
485: template <typename K, typename V>
1.55 paf 486: class Temp_hash_value {
1.59 paf 487: Hash<K, V>& fhash;
488: K fname;
489: V saved_value;
1.55 paf 490: public:
1.59 paf 491: Temp_hash_value(Hash<K, V>& ahash, K aname, V avalue) :
1.55 paf 492: fhash(ahash),
493: fname(aname),
494: saved_value(ahash.get(aname)) {
495: fhash.put(aname, avalue);
496: }
497: ~Temp_hash_value() {
498: fhash.put(fname, saved_value);
499: }
500: };
1.1 paf 501:
502: #endif