Recovered from cvs revision 2007-07-26 14:33:00

ase/awk/Awk.cpp

@@ -1,5 +1,5 @@
 /*
- * $Id: Awk.cpp,v 1.46 2007/07/15 16:31:59 bacon Exp $
+ * $Id: Awk.cpp,v 1.47 2007/07/25 07:00:09 bacon Exp $
  */
 
 #include <ase/awk/Awk.hpp>
@@ -503,7 +503,7 @@ namespace ASE
 		}
 
 		functionMap = ase_awk_map_open (
-			this, 512, freeFunctionMapValue, awk);
+			this, 512, 70, freeFunctionMapValue, awk);
 		if (functionMap == ASE_NULL)
 		{
 			ase_awk_close (awk);

ase/awk/awk.c

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.c,v 1.7 2007/06/17 14:54:10 bacon Exp $ 
+ * $Id: awk.c,v 1.8 2007/07/25 07:00:09 bacon Exp $ 
  *
  * {License}
  */
@@ -58,7 +58,7 @@ ase_awk_t* ase_awk_open (const ase_awk_prmfns_t* prmfns, void* custom_data)
 		return ASE_NULL;	
 	}
 
-	awk->kwtab = ase_awk_map_open (awk, 512, free_kw, awk);
+	awk->kwtab = ase_awk_map_open (awk, 512, 70, free_kw, awk);
 	if (awk->kwtab == ASE_NULL)
 	{
 		ase_str_close (&awk->token.name);
@@ -67,7 +67,7 @@ ase_awk_t* ase_awk_open (const ase_awk_prmfns_t* prmfns, void* custom_data)
 	}
 
 	/* TODO: initial map size?? */
-	awk->tree.afns = ase_awk_map_open (awk, 512, free_afn, awk);
+	awk->tree.afns = ase_awk_map_open (awk, 512, 70, free_afn, awk);
 	if (awk->tree.afns == ASE_NULL) 
 	{
 		ase_awk_map_close (awk->kwtab);

ase/awk/map.c

@@ -1,15 +1,13 @@
 /*
- * $Id: map.c,v 1.6 2007/05/06 08:52:18 bacon Exp $
+ * $Id: map.c,v 1.9 2007/07/25 09:53:28 bacon Exp $
  *
  * {License}
  */
 
 #include <ase/awk/awk_i.h>
 
-/* TODO: improve the entire map routines.
-         support automatic bucket resizing and remaping, etc. */
-
 static ase_size_t hashkey (const ase_char_t* keyptr, ase_size_t keylen);
+static int rehash (ase_awk_map_t* map);
 
 #define FREE_PAIR(map,pair) \
 	do { \
@@ -20,7 +18,7 @@ static ase_size_t hashkey (const ase_char_t* keyptr, ase_size_t keylen);
 	} while (0)
 
 ase_awk_map_t* ase_awk_map_open (
-	void* owner, ase_size_t capa, 
+	void* owner, ase_size_t capa, unsigned int factor,
 	void(*freeval)(void*,void*), ase_awk_t* awk)
 {
 	ase_awk_map_t* map;
@@ -46,6 +44,9 @@ ase_awk_map_t* ase_awk_map_open (
 	map->freeval = freeval;
 	while (capa > 0) map->buck[--capa] = ASE_NULL;
 
+	map->factor = factor;
+	map->threshold = ((ase_size_t)map->factor) * map->capa / 100;
+
 	return map;
 }
 
@@ -93,7 +94,6 @@ ase_awk_pair_t* ase_awk_map_get (
 
 	while (pair != ASE_NULL) 
 	{
-
 		if (ase_strxncmp (
 			ASE_AWK_PAIR_KEYPTR(pair), ASE_AWK_PAIR_KEYLEN(pair), 
 			keyptr, keylen) == 0) return pair;
@@ -142,6 +142,15 @@ int ase_awk_map_putx (
 		pair = ASE_AWK_PAIR_LNK(pair);
 	}
 
+	if (map->threshold > 0 && 
+	    map->size >= map->threshold)
+	{
+		if (rehash(map) == 0) /* ignore the rehash error */
+		{
+			hc = hashkey(keyptr,keylen) % map->capa;
+		}
+	}
+
 	pair = (ase_awk_pair_t*) ASE_AWK_MALLOC (
 		map->awk, ASE_SIZEOF(ase_awk_pair_t));
 	if (pair == ASE_NULL) return -1; /* error */
@@ -286,3 +295,48 @@ static ase_size_t hashkey (const ase_char_t* keyptr, ase_size_t keylen)
 
 	return n;
 }
+
+static int rehash (ase_awk_map_t* map)
+{
+	ase_size_t i, hc, new_capa;
+	ase_awk_pair_t** new_buck;
+
+	new_capa = (map->capa >= 65536)? (map->capa + 65536): (map->capa << 1);
+
+	new_buck = (ase_awk_pair_t**) ASE_AWK_MALLOC (
+		map->awk, ASE_SIZEOF(ase_awk_pair_t*) * new_capa);
+	if (new_buck == ASE_NULL) 
+	{
+		/* once rehash fails, the rehashing is disabled */
+		map->threshold = 0;
+		return -1;
+	}
+
+	for (i = 0; i < new_capa; i++) new_buck[i] = ASE_NULL;
+
+	for (i = 0; i < map->capa; i++)
+	{
+		ase_awk_pair_t* pair = map->buck[i];
+
+		while (pair != ASE_NULL) 
+		{
+			ase_awk_pair_t* next = ASE_AWK_PAIR_LNK(pair);
+
+			hc = hashkey(
+				ASE_AWK_PAIR_KEYPTR(pair),
+				ASE_AWK_PAIR_KEYLEN(pair)) % new_capa;
+
+			ASE_AWK_PAIR_LNK(pair) = new_buck[hc];
+			new_buck[hc] = pair;
+
+			pair = next;
+		}
+	}
+
+	ASE_AWK_FREE (map->awk, map->buck);
+	map->buck = new_buck;
+	map->capa = new_capa;
+	map->threshold = ((ase_size_t)map->factor) * map->capa / 100;
+
+	return 0;
+}

ase/awk/map.h

@@ -1,5 +1,5 @@
 /*
- * $Id: map.h,v 1.6 2007/05/06 08:52:18 bacon Exp $
+ * $Id: map.h,v 1.7 2007/07/25 07:00:09 bacon Exp $
  *
  * {License}
  */
@@ -31,6 +31,8 @@ struct ase_awk_map_t
 	void* owner;
 	ase_size_t size;
 	ase_size_t capa;
+	unsigned int factor;
+	ase_size_t threshold;
 	ase_awk_pair_t** buck;
 	void (*freeval) (void*,void*);
 	ase_awk_t* awk;
@@ -46,7 +48,7 @@ extern "C" {
 #endif
 
 ase_awk_map_t* ase_awk_map_open (
-	void* owner, ase_size_t capa, 
+	void* owner, ase_size_t capa, unsigned int factor,
 	void(*freeval)(void*,void*), ase_awk_t* awk);
 
 void ase_awk_map_close (ase_awk_map_t* map);

ase/awk/run.c

@@ -1,5 +1,5 @@
 /*
- * $Id: run.c,v 1.8 2007/05/28 13:53:31 bacon Exp $
+ * $Id: run.c,v 1.9 2007/07/25 07:00:09 bacon Exp $
  *
  * {License}
  */
@@ -749,7 +749,7 @@ static int __init_run (
 	}
 
 	run->named = ase_awk_map_open (
-		run, DEF_BUF_CAPA, __free_namedval, run->awk);
+		run, 1024, 70, __free_namedval, run->awk);
 	if (run->named == ASE_NULL)
 	{
 		ase_str_close (&run->format.fmt);

ase/awk/val.c

@@ -1,5 +1,5 @@
 /*
- * $Id: val.c,v 1.6 2007/05/10 16:08:37 bacon Exp $
+ * $Id: val.c,v 1.7 2007/07/25 07:00:09 bacon Exp $
  *
  * {License}
  */
@@ -242,7 +242,7 @@ ase_awk_val_t* ase_awk_makemapval (ase_awk_run_t* run)
 
 	val->type = ASE_AWK_VAL_MAP;
 	val->ref = 0;
-	val->map = ase_awk_map_open (run, 256, free_map_val, run->awk);
+	val->map = ase_awk_map_open (run, 256, 70, free_map_val, run->awk);
 	if (val->map == ASE_NULL)
 	{
 		ASE_AWK_FREE (run->awk, val);

ase/change.log

@@ -15,6 +15,8 @@
 * changed test/awk/AseAwk.java to include an option(-w) to utilize
   the setWord method.
 
+* enhanced awk/map.c to support automatic rehashing
+
 [0.2.0]
 
 * fixed bug (nextofile shown as nextfile in source output)