--- parser3/src/include/pa_hash.h	2005/07/26 12:43:05	1.63
+++ parser3/src/include/pa_hash.h	2017/02/07 22:00:34	1.96
@@ -1,7 +1,7 @@
 /** @file
 	Parser: hash class decl.
 
-	Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com)
+	Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com)
 
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 */
@@ -17,10 +17,11 @@
 #ifndef PA_HASH_H
 #define PA_HASH_H
 
-static const char * const IDENT_HASH_H="$Date: 2005/07/26 12:43:05 $";
+#define IDENT_PA_HASH_H "$Id: pa_hash.h,v 1.96 2017/02/07 22:00:34 moko Exp $"
 
 #include "pa_memory.h"
 #include "pa_types.h"
+#include "pa_string.h"
 
 const int HASH_ALLOCATES_COUNT=29;
 
@@ -34,6 +35,48 @@ static uint Hash_allocates[HASH_ALLOCATE
 	2097397, 4194103, 8388857, 16777447, 33554201, 67108961, 
 	134217487, 268435697, 536870683, 1073741621, 2147483399};
 
+/// useful generic hash function
+inline void generic_hash_code(uint& result, char c) {
+	result=(result<<4)+c;
+	if(uint g=(result&0xF0000000)) {
+		result=result^(g>>24);
+		result=result^g;
+	}
+}
+/// useful generic hash function
+inline void generic_hash_code(uint& result, const char* s) {
+	while(char c=*s++) {
+		result=(result<<4)+c;
+		if(uint g=(result&0xF0000000)) {
+			result=result^(g>>24);
+			result=result^g;
+		}
+	}
+}
+
+/// useful generic hash function
+inline void generic_hash_code(uint& result, const char* buf, size_t size) {
+	const char* end=buf+size;
+	while(buf<end) {
+		result=(result<<4)+*buf++;
+		if(uint g=(result&0xF0000000)) {
+			result=result^(g>>24);
+			result=result^g;
+		}
+	}
+}
+
+/// simple hash code of int. used by EXIF mapping
+inline uint hash_code(int self) {
+	uint result=0;
+	generic_hash_code(result, (const char*)&self, sizeof(self));
+	return result;
+}
+
+#endif // PA_HASH_H
+
+#ifndef PA_HASH_CLASS
+#define PA_HASH_CLASS
 /** 
 	Simple hash.
 
@@ -42,39 +85,87 @@ static uint Hash_allocates[HASH_ALLOCATE
 		get returning 0 means there were no such.
 		"put value 0" means "remove"
 */
-template<typename K, typename V> class Hash: public PA_Object {
-public:
+#ifdef HASH_ORDER
 
+#undef HASH
+#undef HASH_STRING
+#undef HASH_NEW_PAIR
+#undef HASH_ORDER_CLEAR
+#undef HASH_FOR_EACH
+
+#define HASH OrderedHash
+#define HASH_STRING OrderedHashString
+#define HASH_NEW_PAIR(code, key, value) *ref=new Pair(code, key, value, *ref, this->last); this->last=&((*ref)->next)
+#define HASH_ORDER_CLEAR() first=0; last=&first
+
+#define HASH_FOR_EACH \
+	for(Pair *pair=this->first; pair; pair=pair->next)
+
+#else
+
+#define HASH Hash
+#define HASH_STRING HashString
+#define HASH_NEW_PAIR(code, key, value) *ref=new Pair(code, key, value, *ref)
+#define HASH_ORDER_CLEAR()
+
+#define HASH_FOR_EACH \
+	Pair **ref=this->refs; \
+	for(int index=0; index<this->allocated; index++) \
+		for(Pair *pair=*ref++; pair; pair=pair->link)
+
+#endif
+
+template<typename K, typename V> class HASH: public PA_Object {
+protected:
+	class Pair;
+public:
 	typedef K key_type;
 	typedef V value_type;
 
-	Hash() { 
+	HASH() { 
 		allocated=Hash_allocates[allocates_index=0];
-		threshold=allocated*THRESHOLD_PERCENT/100;
 		fpairs_count=fused_refs=0;
-		refs=new(UseGC) Pair*[allocated];
+		refs=new Pair*[allocated];
+		HASH_ORDER_CLEAR();
 	}
 
-	Hash(const Hash& source) {
+	HASH(const HASH& source) {
 		allocates_index=source.allocates_index;
 		allocated=source.allocated;
-		threshold=source.threshold;
 		fused_refs=source.fused_refs;
 		fpairs_count=source.fpairs_count;
-		refs=new(UseGC) Pair*[allocated];
-
+		refs=new Pair*[allocated];
 		// clone & rehash
-		Pair **old_ref=source.refs;
+#ifdef HASH_ORDER
+		HASH_ORDER_CLEAR();
+		for(Pair *pair=source.first; pair; pair=pair->next)
+		{
+			uint index=pair->code%allocated;
+			Pair **ref=&refs[index];
+			HASH_NEW_PAIR(pair->code, pair->key, pair->value);
+		}
+#else
+		for(int i=0; i<source.allocated; i++)
+			for(Pair *pair=source.refs[i]; pair; pair=pair->link)
+			{
+				Pair **ref=&refs[i];
+				HASH_NEW_PAIR(pair->code, pair->key, pair->value);
+			}
+#endif
+	}
+
+#ifdef USE_DESTRUCTORS
+	~HASH() {
+		Pair **ref=refs;
 		for(int index=0; index<allocated; index++)
-			for(Pair *pair=*old_ref++; pair; ) {
+			for(Pair *pair=*ref++; pair;){
 				Pair *next=pair->link;
-
-				Pair **new_ref=&refs[index];
-				*new_ref=new Pair(pair->code, pair->key, pair->value, *new_ref);
-
+				delete pair;
 				pair=next;
 			}
+		delete[] refs;
 	}
+#endif
 
 	/// put a [value] under the [key] @returns existed or not
 	bool put(K key, V value) {
@@ -98,56 +189,44 @@ public:
 		// proper pair not found -- create&link_in new pair
 		if(!*ref) // root cell were fused_refs?
 			fused_refs++; // not, we'll use it and record the fact
-		*ref=new Pair(code, key, value, *ref);
+		HASH_NEW_PAIR(code, key, value);
 		fpairs_count++;
 		return false;
 	}
 
-	/// put a [value] under the [key] @returns existed or not
-	template<typename R, typename F> R maybe_put(K key, V value, F prevent) {
-		if(!value) {
-			remove(key);
-			return 0;
-		}
-		if(is_full()) 
-			expand();
-
+	/// remove the [key] @returns existed or not
+	bool remove(K key) {
 		uint code=hash_code(key);
 		uint index=code%allocated;
-		Pair **ref=&refs[index];
-		for(Pair *pair=*ref; pair; pair=pair->link)
+		for(Pair **ref=&refs[index]; *ref; ref=&(*ref)->link){
+			Pair *pair=*ref;
 			if(pair->code==code && pair->key==key) {
 				// found a pair with the same key
-
-				// prevent-function intercepted put?
-				if(R result=prevent(pair->value))
-					return result;
-				
-				pair->value=value;
-				return reinterpret_cast<R>(1);
+				Pair *next=pair->link;
+#ifdef HASH_ORDER
+				*(pair->prev)=pair->next;
+				if(pair->next)
+					pair->next->prev=pair->prev;
+				else
+					last=pair->prev;
+#endif
+				*ref=next;
+				--fpairs_count;
+				return true;
 			}
-		
-		// proper pair not found -- create&link_in new pair
-		if(!*ref) // root cell were fused_refs?
-			fused_refs++; // not, we'll use it and record the fact
-		*ref=new Pair(code, key, value, *ref);
-		fpairs_count++;
-		return 0;
+		}
+
+		return false;
 	}
 
-	/// remove the [key] @returns existed or not
-	bool remove(K key) {
+	/// return true if key exists
+	bool contains(K key){
 		uint code=hash_code(key);
 		uint index=code%allocated;
-		for(Pair **ref=&refs[index]; *ref; ref=&(*ref)->link)
-			if((*ref)->code==code && (*ref)->key==key) {
-				// found a pair with the same key
-				Pair *next=(*ref)->link;
-				delete *ref;
-				*ref=next;
-				--fpairs_count;
+		for(Pair *pair=refs[index]; pair; pair=pair->link){
+			if(pair->code==code && pair->key==key)
 				return true;
-			}
+		}
 
 		return false;
 	}
@@ -163,6 +242,49 @@ public:
 		return V(0);
 	}
 
+#ifdef HASH_ORDER
+	String::Body first_key() const {
+#ifdef HASH_CODE_CACHING
+		return (first) ? String::Body(first->key, first->code) : String::Body();
+#else
+		return (first) ? first->key : String::Body();
+#endif
+	}
+
+	V first_value() const {
+		return (first) ? first->value : V(0);
+	}
+
+	String::Body last_key() const {
+		if (fpairs_count) {
+			Pair* pair = (Pair*)((char *)last - offsetof(Pair, next));
+#ifdef HASH_CODE_CACHING
+			return String::Body(pair->key, pair->code);
+#else
+			return pair->key;
+#endif
+		} else {
+			return String::Body();
+		}
+	}
+
+	V last_value() const {
+		return (fpairs_count) ? ((Pair *)((char *)last - offsetof(Pair, next)))->value : V(0);
+	}
+
+	void order_clear() {
+		HASH_ORDER_CLEAR();
+	}
+
+	void order_next(Pair* pair) {
+		pair->prev=last;
+		pair->next=0;
+		*last=pair;
+		last=&(pair->next);
+	}
+
+#endif //HASH_ORDER
+
 	/// put a [value] under the [key] if that [key] existed @returns existed or not
 	bool put_replaced(K key, V value) {
 		if(!value) {
@@ -203,17 +325,19 @@ public:
 		// proper pair not found -- create&link_in new pair
 		if(!*ref) // root cell were fused_refs?
 			fused_refs++; // not, we'll use it and record the fact
-		*ref=new Pair(code, key, value, *ref);
+		HASH_NEW_PAIR(code, key, value);
 		fpairs_count++;
 		return false;
 	}
 
-	/** put all 'src' values if NO with same key existed
-		@todo optimize this.allocated==src.allocated case
-	*/
-	void merge_dont_replace(const Hash& src) {
+	/// put all 'src' values if NO with same key existed
+	void merge_dont_replace(const HASH& src) {
+#ifdef HASH_ORDER
+		for(Pair *pair=src.first; pair; pair=pair->next)
+#else
 		for(int i=0; i<src.allocated; i++)
 			for(Pair *pair=src.refs[i]; pair; pair=pair->link)
+#endif
 				put_dont_replace(pair->key, pair->value);
 	}
 
@@ -222,43 +346,32 @@ public:
 
 	/// iterate over all pairs
 	template<typename I> void for_each(void callback(K, V, I), I info) const {
-		Pair **ref=refs;
-		for(int index=0; index<allocated; index++)
-			for(Pair *pair=*ref++; pair; pair=pair->link)
-				callback(pair->key, pair->value, info);
+		HASH_FOR_EACH
+			callback(pair->key, pair->value, info);
 	}
 
 	/// iterate over all pairs
 	template<typename I> void for_each_ref(void callback(K, V&, I), I info) const {
-		Pair **ref=refs;
-		for(int index=0; index<allocated; index++)
-			for(Pair *pair=*ref++; pair; pair=pair->link)
-				callback(pair->key, pair->value, info);
+		HASH_FOR_EACH
+			callback(pair->key, pair->value, info);
 	}
 
 	/// iterate over all pairs until condition becomes true, return that element
 	template<typename I> V first_that(bool callback(K, V, I), I info) const {
-		Pair **ref=refs;
-		for(int index=0; index<allocated; index++)
-			for(Pair *pair=*ref++; pair; pair=pair->link)
-				if(callback(pair->key, pair->value, info))
-					return pair->value;
-		
+		HASH_FOR_EACH
+			if(callback(pair->key, pair->value, info))
+				return pair->value;
 		return V(0);
 	}
 
 	/// remove all elements
 	void clear() {
 		memset(refs, 0, sizeof(*refs)*allocated);
-		fpairs_count=fused_refs=0;	
+		fpairs_count=fused_refs=0;
+		HASH_ORDER_CLEAR();
 	}
 
-private:
-
-	/// expand when these %% of allocated exausted
-	enum {
-		THRESHOLD_PERCENT=75
-	};
+protected:
 
 	/// the index of [allocated] in [Hash_allocates]
 	int allocates_index;
@@ -266,9 +379,6 @@ private:
 	/// number of allocated pairs
 	int allocated;
 
-	/// helper: expanding when fused_refs == threshold
-	int threshold;
-
 	/// used pairs
 	int fused_refs;
 
@@ -282,27 +392,34 @@ private:
 		K key;
 		V value;
 		Pair *link;
-		
-		Pair(uint acode, K akey, V avalue, Pair *alink) :
-			code(acode),
-			key(akey),
-			value(avalue),
-			link(alink) {}
+#ifdef HASH_ORDER
+		Pair **prev;
+		Pair *next;
+
+		Pair(uint acode, K akey, V avalue, Pair *alink, Pair **aprev) : code(acode), key(akey), value(avalue), link(alink), 
+			prev(aprev), next(0) { *aprev=this; }
+#else
+		Pair(uint acode, K akey, V avalue, Pair *alink) : code(acode), key(akey), value(avalue), link(alink) {}
+#endif
 	} **refs;
 
-	/// filled to threshold: needs expanding
-	bool is_full() { return fused_refs==threshold; }
+#ifdef HASH_ORDER
+	Pair *first;
+	Pair **last;
+#endif
+
+	/// filled to threshold (THRESHOLD_PERCENT=75), needs expanding
+	bool is_full() { return fused_refs + allocated/4 >= allocated; }
 
 	/// allocate larger buffer & rehash
 	void expand() {
 		int old_allocated=allocated;
 		Pair **old_refs=refs;
 
-		allocates_index=allocates_index+1<HASH_ALLOCATES_COUNT?allocates_index+1:HASH_ALLOCATES_COUNT-1;
+		if (allocates_index<HASH_ALLOCATES_COUNT-1) allocates_index++;
 		// allocated bigger refs array
 		allocated=Hash_allocates[allocates_index];
-		threshold=allocated*THRESHOLD_PERCENT/100;
-		refs=new(UseGC) Pair*[allocated];
+		refs=new Pair*[allocated];
 
 		// rehash
 		Pair **old_ref=old_refs;
@@ -323,63 +440,282 @@ private:
 
 private: //disabled
 
-	Hash& operator = (const Hash&) { return *this; }
+	HASH& operator = (const HASH&) { return *this; }
 };
 
-/// useful generic hash function
-inline void generic_hash_code(uint& result, char c) {
-	result=(result<<4)+c;
-	if(uint g=(result&0xF0000000)) {
-		result=result^(g>>24);
-		result=result^g;
+/** 
+	Simple String::body hash.
+	Allows hash code caching
+*/
+
+#ifdef HASH_CODE_CACHING
+
+template<typename V> class HASH_STRING: public HASH<const CORD,V> {
+public:
+
+	typedef typename HASH<const CORD,V>::Pair Pair; 
+	typedef const String::Body &K;
+
+	typedef K key_type;
+	
+	/// put a [value] under the [key] @returns existed or not
+	bool put(K str, V value) {
+		if(!value) {
+			remove(str);
+			return false;
+		}
+		if(this->is_full()) 
+			this->expand();
+
+		CORD key=str.get_cord();
+
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		Pair **ref=&this->refs[index];
+		for(Pair *pair=*ref; pair; pair=pair->link)
+			if(pair->code==code && CORD_cmp(pair->key,key)==0) {
+				// found a pair with the same key
+				pair->value=value;
+				return true;
+			}
+
+		// proper pair not found -- create&link_in new pair
+		if(!*ref) // root cell were fused_refs?
+			this->fused_refs++; // not, we'll use it and record the fact
+		HASH_NEW_PAIR(code, key, value);
+		this->fpairs_count++;
+		return false;
 	}
-}
-/// useful generic hash function
-inline void generic_hash_code(uint& result, const char* s) {
-	while(char c=*s++) {
-		result=(result<<4)+c;
-		if(uint g=(result&0xF0000000)) {
-			result=result^(g>>24);
-			result=result^g;
+
+	/// remove the [key] @returns existed or not
+	bool remove(K str) {
+		CORD key=str.get_cord();
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		for(Pair **ref=&this->refs[index]; *ref; ref=&(*ref)->link){
+			Pair *pair=*ref;
+			if(pair->code==code && CORD_cmp(pair->key,key)==0) {
+				// found a pair with the same key
+				Pair *next=pair->link;
+#ifdef HASH_ORDER
+				*(pair->prev)=pair->next;
+				if(pair->next)
+					pair->next->prev=pair->prev;
+				else
+					this->last=pair->prev;
+#endif
+				*ref=next;
+				--this->fpairs_count;
+				return true;
+			}
 		}
+
+		return false;
 	}
-}
 
-/// useful generic hash function
-inline void generic_hash_code(uint& result, const char* buf, size_t size) {
-	const char* end=buf+size;
-	while(buf<end) {
-		result=(result<<4)+*buf++;
-		if(uint g=(result&0xF0000000)) {
-			result=result^(g>>24);
-			result=result^g;
+	/// return true if key exists
+	bool contains(K str){
+		CORD key=str.get_cord();
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		for(Pair *pair=this->refs[index]; pair; pair=pair->link){
+			if(pair->code==code && CORD_cmp(pair->key,key)==0)
+				return true;
 		}
+
+		return false;
 	}
-}
 
-/// simple hash code of int. used by EXIF mapping
-inline uint hash_code(int self) {
-	uint result=0;
-	generic_hash_code(result, (const char*)&self, sizeof(self));
-	return result;
-}
+	/// get associated [value] by the [key]
+	V get(K str) const {
+		CORD key=str.get_cord();
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		for(Pair *pair=this->refs[index]; pair; pair=pair->link)
+			if(pair->code==code && CORD_cmp(pair->key,key)==0)
+				return pair->value;
+
+		return V(0);
+	}
+
+	/// get associated [value] by the [key], optimized
+	V get(const char *key) const {
+		uint code=0;
+		if(key && *key){
+			generic_hash_code(code, key);
+		} else {
+			key=0;
+		}
+		uint index=code%this->allocated;
+		for(Pair *pair=this->refs[index]; pair; pair=pair->link)
+			if(pair->code==code && CORD_cmp(pair->key,(CORD)key)==0)
+				return pair->value;
+
+		return V(0);
+	}
+
+	/// put a [value] under the [key] if that [key] existed @returns existed or not
+	bool put_replaced(K str, V value) {
+		if(!value) {
+			remove(str);
+			return false;
+		}
+
+		CORD key=str.get_cord();
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		for(Pair *pair=this->refs[index]; pair; pair=pair->link)
+			if(pair->code==code && CORD_cmp(pair->key,key)==0) {
+				// found a pair with the same key, replacing
+				pair->value=value;
+				return true;
+			}
+
+		// proper pair not found 
+		return false;
+	}
+
+	/// put a [value] under the [key] if that [key] NOT existed @returns existed or not
+	bool put_dont_replace(K str, V value) {
+		if(!value) {
+			remove(str);
+			return false;
+		}
+		if(this->is_full()) 
+			this->expand();
 
-///	Auto-object used to temporarily substituting/removing hash values
-template <typename K, typename V>
+		CORD key=str.get_cord();
+		uint code=str.get_hash_code();
+		uint index=code%this->allocated;
+		Pair **ref=&this->refs[index];
+		for(Pair *pair=*ref; pair; pair=pair->link)
+			if(pair->code==code && CORD_cmp(pair->key,key)==0) {
+				// found a pair with the same key, NOT replacing
+				return true;
+			}
+
+		// proper pair not found -- create&link_in new pair
+		if(!*ref) // root cell were fused_refs?
+			this->fused_refs++; // not, we'll use it and record the fact
+		HASH_NEW_PAIR(code, key, value);
+		this->fpairs_count++;
+		return false;
+	}
+
+	/// put all 'src' values if NO with same key existed
+	void merge_dont_replace(const HASH_STRING& src) {
+#ifdef HASH_ORDER
+		for(Pair *pair=src.first; pair; pair=pair->next)
+#else
+		for(int i=0; i<src.allocated; i++)
+			for(Pair *pair=src.refs[i]; pair; pair=pair->link)
+#endif
+				put_dont_replace(String::Body(pair->key, pair->code), pair->value);
+	}
+
+	/// iterate over all pairs
+	template<typename I> void for_each(void callback(K, V, I), I info) const {
+		HASH_FOR_EACH
+			callback(String::Body(pair->key, pair->code), pair->value, info);
+	}
+
+	/// iterate over all pairs
+	template<typename I> void for_each_ref(void callback(K, V&, I), I info) const {
+		HASH_FOR_EACH
+			callback(String::Body(pair->key, pair->code), pair->value, info);
+	}
+
+	/// iterate over all pairs until condition becomes true, return that element
+	template<typename I> V first_that(bool callback(K, V, I), I info) const {
+		HASH_FOR_EACH
+			if(callback(String::Body(pair->key, pair->code), pair->value, info))
+				return pair->value;
+		return V(0);
+	}
+
+#else //HASH_CODE_CACHING
+
+template<typename V> class HASH_STRING: public HASH<const String::Body,V>{
+public:
+	typedef typename HASH<const String::Body,V>::Pair Pair;
+
+#endif //HASH_CODE_CACHING
+
+	/// simple hash iterator
+	class Iterator {
+		const HASH_STRING<V>& fhash;
+		Pair *fcurrent;
+#ifndef HASH_ORDER
+		int i;
+#endif
+	public:
+#ifdef HASH_ORDER
+		Iterator(const HASH_STRING<V>& ahash): fhash(ahash) {
+			fcurrent=fhash.first;
+		}
+
+		void next() {
+			fcurrent=fcurrent->next;
+		}
+#else
+		Iterator(const HASH_STRING<V>& ahash): fhash(ahash) {
+			fcurrent=0;
+			for(i=0; i<fhash.allocated; i++)
+				if (fcurrent=fhash.refs[i])
+					break;
+		}
+
+		void next() {
+			if(fcurrent=fcurrent->link)
+				return;
+			for(i++; i<fhash.allocated; i++)
+				if (fcurrent=fhash.refs[i])
+					break;
+		}
+#endif
+
+		operator bool () {
+			return fcurrent != 0;
+		}
+
+		String::Body key(){
+#ifdef HASH_CODE_CACHING
+			return String::Body(fcurrent->key, fcurrent->code);
+#else
+			return fcurrent->key;
+#endif
+		}
+
+		V value(){
+			return fcurrent->value;
+		}
+
+		Pair *pair(){
+			return fcurrent;
+		}
+	};
+
+};
+
+#ifndef HASH_ORDER
+///    Auto-object used to temporarily substituting/removing string hash values
+template <typename H, typename V>
 class Temp_hash_value {
-	Hash<K, V>& fhash;
-	K fname;
+	H *fhash;
+	String::Body fname;
 	V saved_value;
 public:
-	Temp_hash_value(Hash<K, V>& ahash, K aname, V avalue) : 
-		fhash(ahash),
-		fname(aname),
-		saved_value(ahash.get(aname)) {
-		fhash.put(aname, avalue);
+	Temp_hash_value(H *ahash, String::Body aname, V avalue) : fhash(ahash), fname(aname) {
+		if(fhash){
+			saved_value=fhash->get(aname);
+			fhash->put(aname, avalue);
+		}
 	}
-	~Temp_hash_value() { 
-		fhash.put(fname, saved_value);
+	~Temp_hash_value() {
+		if(fhash)
+			fhash->put(fname, saved_value);
 	}
 };
-
 #endif
+
+#endif //PA_HASH_CLASS