Annotation of parser3/src/include/pa_hash.h, revision 1.60
1.28 paf 1: /** @file
1.29 paf 2: Parser: hash class decl.
3:
1.59 paf 4: Copyright (c) 2001-2003 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:
1.60 ! paf 20: static const char * const IDENT_HASH_H="$Date: 2003/07/24 11:31:21 $";
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:
1.29 paf 27: /**
1.59 paf 28: Simple hash.
1.29 paf 29:
1.59 paf 30: Automatically rehashed when almost is_full.
1.51 paf 31: Contains no 0 values.
32: get returning 0 means there were no such.
33: "put value 0" means "remove"
1.29 paf 34: */
1.59 paf 35: template<typename K, typename V> class Hash: public PA_Object {
1.1 paf 36: public:
37:
1.59 paf 38: typedef K key_type;
39: typedef V value_type;
1.3 paf 40:
1.59 paf 41: Hash() {
42: allocated=allocates[allocates_index=0];
43: threshold=allocated*THRESHOLD_PERCENT/100;
44: fpairs_count=fused_refs=0;
45: refs=new(UseGC) Pair*[allocated];
46: }
1.25 paf 47:
1.59 paf 48: Hash(const Hash& source) {
49: allocates_index=source.allocates_index;
50: allocated=source.allocated;
51: threshold=source.threshold;
52: fused_refs=source.fused_refs;
53: fpairs_count=source.fpairs_count;
54: refs=new(UseGC) Pair*[allocated];
55:
56: // clone & rehash
57: Pair **old_ref=source.refs;
58: for(int index=0; index<allocated; index++)
59: for(Pair *pair=*old_ref++; pair; ) {
60: Pair *next=pair->link;
1.45 paf 61:
1.59 paf 62: Pair **new_ref=&refs[index];
63: *new_ref=new Pair(pair->code, pair->key, pair->value, *new_ref);
1.38 paf 64:
1.59 paf 65: pair=next;
66: }
1.43 parser 67: }
68:
1.59 paf 69: /// put a [value] under the [key] @returns existed or not
70: bool put(K key, V value) {
71: if(!value) {
72: remove(key);
73: return false;
74: }
75: if(is_full())
76: expand();
77:
78: uint code=hash_code(key);
79: uint index=code%allocated;
80: Pair **ref=&refs[index];
81: for(Pair *pair=*ref; pair; pair=pair->link)
82: if(pair->code==code && pair->key==key) {
83: // found a pair with the same key
84: pair->value=value;
85: return true;
86: }
87:
88: // proper pair not found -- create&link_in new pair
89: if(!*ref) // root cell were fused_refs?
90: fused_refs++; // not, we'll use it and record the fact
91: *ref=new Pair(code, key, value, *ref);
92: fpairs_count++;
93: return false;
1.24 paf 94: }
1.10 paf 95:
1.59 paf 96: /// remove the [key] @returns existed or not
97: bool remove(K key) {
98: uint code=hash_code(key);
99: uint index=code%allocated;
100: for(Pair **ref=&refs[index]; *ref; ref=&(*ref)->link)
101: if((*ref)->code==code && (*ref)->key==key) {
102: // found a pair with the same key
103: Pair *next=(*ref)->link;
104: delete *ref;
105: *ref=next;
106: --fpairs_count;
107: return true;
108: }
1.8 paf 109:
1.59 paf 110: return false;
111: }
1.48 paf 112:
1.59 paf 113: /// get associated [value] by the [key]
114: V get(K key) const {
115: uint code=hash_code(key);
116: uint index=code%allocated;
117: for(Pair *pair=refs[index]; pair; pair=pair->link)
118: if(pair->code==code && pair->key==key)
119: return pair->value;
120:
121: return V(0);
1.33 paf 122: }
1.17 paf 123:
1.51 paf 124: /// put a [value] under the [key] if that [key] existed @returns existed or not
1.59 paf 125: bool put_replace(K key, V value) {
126: if(!value) {
127: remove(key);
128: return false;
129: }
130: uint code=hash_code(key);
131: uint index=code%allocated;
132: for(Pair *pair=refs[index]; pair; pair=pair->link)
133: if(pair->code==code && pair->key==key) {
134: // found a pair with the same key, replacing
135: pair->value=value;
136: return true;
137: }
138:
139: // proper pair not found
140: return false;
141: }
1.18 paf 142:
1.51 paf 143: /// put a [value] under the [key] if that [key] NOT existed @returns existed or not
1.59 paf 144: bool put_dont_replace(K key, V value) {
145: if(!value) {
146: remove(key);
147: return false;
148: }
149: if(is_full())
150: expand();
151:
152: uint code=hash_code(key);
153: uint index=code%allocated;
154: Pair **ref=&refs[index];
155: for(Pair *pair=*ref; pair; pair=pair->link)
156: if(pair->code==code && pair->key==key) {
157: // found a pair with the same key, NOT replacing
158: return true;
159: }
160:
161: // proper pair not found -- create&link_in new pair
162: if(!*ref) // root cell were fused_refs?
163: fused_refs++; // not, we'll use it and record the fact
164: *ref=new Pair(code, key, value, *ref);
165: fpairs_count++;
166: return false;
167: }
1.18 paf 168:
1.59 paf 169: /** put all 'src' values if NO with same key existed
170: @todo optimize this.allocated==src.allocated case
171: */
172: void merge_dont_replace(const Hash& src) {
173: for(int i=0; i<src.allocated; i++)
174: for(Pair *pair=src.refs[i]; pair; pair=pair->link)
175: put_dont_replace(pair->key, pair->value);
1.36 paf 176: }
1.11 paf 177:
1.29 paf 178: /// number of elements in hash
1.59 paf 179: int count() const { return fpairs_count; }
1.25 paf 180:
1.59 paf 181: /// iterate over all pairs
182: template<typename I> void for_each(void callback(K, V, I), I info) const {
183: Pair **ref=refs;
184: for(int index=0; index<allocated; index++)
185: for(Pair *pair=*ref++; pair; pair=pair->link)
186: callback(pair->key, pair->value, info);
187: }
1.45 paf 188:
1.59 paf 189: /// iterate over all pairs
190: template<typename I> void for_each_ref(void callback(K, V&, I), I info) const {
191: Pair **ref=refs;
192: for(int index=0; index<allocated; index++)
193: for(Pair *pair=*ref++; pair; pair=pair->link)
194: callback(pair->key, pair->value, info);
195: }
1.38 paf 196:
1.59 paf 197: /// iterate over all pairs until condition becomes true, return that element
198: template<typename I> V first_that(bool callback(K, V, I), I info) const {
199: Pair **ref=refs;
200: for(int index=0; index<allocated; index++)
201: for(Pair *pair=*ref++; pair; pair=pair->link)
202: if(callback(pair->key, pair->value, info))
203: return pair->value;
204:
205: return V(0);
206: }
1.27 paf 207:
1.29 paf 208: /// remove all elements
1.59 paf 209: void clear() {
210: memset(refs, 0, sizeof(*refs)*allocated);
211: fpairs_count=fused_refs=0;
212: }
1.15 paf 213:
1.1 paf 214: private:
215:
1.39 paf 216: /// expand when these %% of allocated exausted
1.1 paf 217: enum {
218: THRESHOLD_PERCENT=75
219: };
1.9 paf 220:
1.39 paf 221: /// the index of [allocated] in [allocates]
1.19 paf 222: int allocates_index;
1.1 paf 223:
1.39 paf 224: /// possible [allocates]. prime numbers
1.19 paf 225: static uint allocates[];
1.59 paf 226:
1.39 paf 227: /// number of allocated pairs
1.19 paf 228: int allocated;
1.1 paf 229:
1.59 paf 230: /// helper: expanding when fused_refs == threshold
1.1 paf 231: int threshold;
232:
1.39 paf 233: /// used pairs
1.59 paf 234: int fused_refs;
1.44 parser 235:
236: /// stored pairs total (including those by links)
1.59 paf 237: int fpairs_count;
1.1 paf 238:
1.39 paf 239: /// pair storage
1.59 paf 240: class Pair: public PA_Allocated {
241: public:
1.1 paf 242: uint code;
1.59 paf 243: K key;
244: V value;
1.1 paf 245: Pair *link;
1.2 paf 246:
1.59 paf 247: Pair(uint acode, K akey, V avalue, Pair *alink) :
1.1 paf 248: code(acode),
249: key(akey),
250: value(avalue),
1.2 paf 251: link(alink) {}
252: } **refs;
1.1 paf 253:
1.39 paf 254: /// filled to threshold: needs expanding
1.59 paf 255: bool is_full() { return fused_refs==threshold; }
1.5 paf 256:
1.39 paf 257: /// allocate larger buffer & rehash
1.59 paf 258: void expand() {
259: int old_allocated=allocated;
260: Pair **old_refs=refs;
261:
262: allocates_index=allocates_index+1<HASH_ALLOCATES_COUNT?allocates_index+1:HASH_ALLOCATES_COUNT-1;
263: // allocated bigger refs array
264: allocated=allocates[allocates_index];
265: threshold=allocated*THRESHOLD_PERCENT/100;
266: refs=new(UseGC) Pair*[allocated];
267:
268: // rehash
269: Pair **old_ref=old_refs;
270: for(int old_index=0; old_index<old_allocated; old_index++)
271: for(Pair *pair=*old_ref++; pair; ) {
272: Pair *next=pair->link;
273:
274: uint new_index=pair->code%allocated;
275: Pair **new_ref=&refs[new_index];
276: pair->link=*new_ref;
277: *new_ref=pair;
278:
279: pair=next;
280: }
281:
282: delete[] old_refs;
283: }
1.4 paf 284:
285: private: //disabled
286:
1.12 paf 287: Hash& operator = (const Hash&) { return *this; }
1.1 paf 288: };
1.49 paf 289:
1.59 paf 290: /* Zend comment: Generated on an Octa-ALPHA 300MHz CPU & 2.5GB RAM monster */
291: template<typename K, typename V>
292: uint Hash<K, V>::allocates[HASH_ALLOCATES_COUNT]={
293: 5, 11, 19, 53, 107, 223, 463, 983, 1979, 3907, 7963,
294: 16229, 32531, 65407, 130987, 262237, 524521, 1048793,
295: 2097397, 4194103, 8388857, 16777447, 33554201, 67108961,
296: 134217487, 268435697, 536870683, 1073741621, 2147483399};
297:
298: /// useful generic hash function
299: inline void generic_hash_code(uint& result, char c) {
300: result=(result<<4)+c;
301: if(uint g=(result&0xF0000000)) {
302: result=result^(g>>24);
303: result=result^g;
304: }
305: }
306: /// useful generic hash function
307: inline void generic_hash_code(uint& result, const char* s) {
308: while(char c=*s++) {
309: result=(result<<4)+c;
310: if(uint g=(result&0xF0000000)) {
311: result=result^(g>>24);
312: result=result^g;
313: }
314: }
315: }
316:
317: /// useful generic hash function
318: inline void generic_hash_code(uint& result, const char* buf, size_t size) {
319: const char* end=buf+size;
320: while(buf<end) {
321: result=(result<<4)+*buf++;
322: if(uint g=(result&0xF0000000)) {
323: result=result^(g>>24);
324: result=result^g;
325: }
326: }
327: }
328:
329: /// simple hash code of int. used by EXIF mapping
330: inline uint hash_code(int self) {
331: uint result=0;
332: generic_hash_code(result, (const char*)&self, sizeof(self));
333: return result;
334: }
335:
336: /// Auto-object used to temporarily substituting/removing hash values
337: template <typename K, typename V>
1.55 paf 338: class Temp_hash_value {
1.59 paf 339: Hash<K, V>& fhash;
340: K fname;
341: V saved_value;
1.55 paf 342: public:
1.59 paf 343: Temp_hash_value(Hash<K, V>& ahash, K aname, V avalue) :
1.55 paf 344: fhash(ahash),
345: fname(aname),
346: saved_value(ahash.get(aname)) {
347: fhash.put(aname, avalue);
348: }
349: ~Temp_hash_value() {
350: fhash.put(fname, saved_value);
351: }
352: };
1.1 paf 353:
354: #endif
E-mail:
![[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