Added support for bit vectors to HNSW

src/bitvector.c

@@ -0,0 +1,90 @@
+#include "postgres.h"
+
+#include "bitvector.h"
+#include "port/pg_bitutils.h"
+#include "utils/varbit.h"
+
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
+/*
+ * Allocate and initialize a new bit vector
+ */
+VarBit *
+InitBitVector(int dim)
+{
+	VarBit	   *result;
+	int			size;
+
+	size = VARBITTOTALLEN(dim);
+	result = (VarBit *) palloc0(size);
+	SET_VARSIZE(result, size);
+	VARBITLEN(result) = dim;
+
+	return result;
+}
+
+/*
+ * Ensure same number of bits
+ */
+static inline void
+CheckDims(VarBit *a, VarBit *b)
+{
+	if (VARBITLEN(a) != VARBITLEN(b))
+		ereport(ERROR,
+				(errcode(ERRCODE_DATA_EXCEPTION),
+				 errmsg("different bit lengths %u and %u", VARBITLEN(a), VARBITLEN(b))));
+}
+
+/*
+ * Get the Hamming distance between two bit strings
+ */
+PGDLLEXPORT PG_FUNCTION_INFO_V1(hamming_distance);
+Datum
+hamming_distance(PG_FUNCTION_ARGS)
+{
+	VarBit	   *a = PG_GETARG_VARBIT_P(0);
+	VarBit	   *b = PG_GETARG_VARBIT_P(1);
+	unsigned char *ax = VARBITS(a);
+	unsigned char *bx = VARBITS(b);
+	uint64		distance = 0;
+
+	CheckDims(a, b);
+
+	/* TODO Improve performance */
+	for (uint32 i = 0; i < VARBITBYTES(a); i++)
+		distance += pg_number_of_ones[ax[i] ^ bx[i]];
+
+	PG_RETURN_FLOAT8((double) distance);
+}
+
+/*
+ * Get the Jaccard distance between two bit strings
+ */
+PGDLLEXPORT PG_FUNCTION_INFO_V1(jaccard_distance);
+Datum
+jaccard_distance(PG_FUNCTION_ARGS)
+{
+	VarBit	   *a = PG_GETARG_VARBIT_P(0);
+	VarBit	   *b = PG_GETARG_VARBIT_P(1);
+	unsigned char *ax = VARBITS(a);
+	unsigned char *bx = VARBITS(b);
+	uint64		ab = 0;
+	uint64		aa;
+	uint64		bb;
+
+	CheckDims(a, b);
+
+	/* TODO Improve performance */
+	for (uint32 i = 0; i < VARBITBYTES(a); i++)
+		ab += pg_number_of_ones[ax[i] & bx[i]];
+
+	if (ab == 0)
+		PG_RETURN_FLOAT8(1);
+
+	aa = pg_popcount((char *) ax, VARBITBYTES(a));
+	bb = pg_popcount((char *) bx, VARBITBYTES(b));
+
+	PG_RETURN_FLOAT8(1 - (ab / ((double) (aa + bb - ab))));
+}

src/bitvector.h

@@ -0,0 +1,8 @@
+#ifndef BITVECTOR_H
+#define BITVECTOR_H
+
+#include "utils/varbit.h"
+
+VarBit	   *InitBitVector(int dim);
+
+#endif

src/hnsw.h

@@ -57,7 +57,8 @@
 
 typedef enum HnswType
 {
-	HNSW_TYPE_VECTOR
+	HNSW_TYPE_VECTOR,
+	HNSW_TYPE_BIT
 }			HnswType;
 
 /* Build phases */

src/hnswbuild.c

@@ -44,6 +44,7 @@
 #include "access/xact.h"
 #include "access/xloginsert.h"
 #include "catalog/index.h"
+#include "catalog/pg_type_d.h"
 #include "commands/progress.h"
 #include "hnsw.h"
 #include "miscadmin.h"
@@ -665,13 +666,27 @@ HnswSharedMemoryAlloc(Size size, void *state)
 	return chunk;
 }
 
+/*
+ * Get max dimensions
+ */
+static int
+GetMaxDimensions(HnswType type)
+{
+	int			maxDimensions = HNSW_MAX_DIM;
+
+	if (type == HNSW_TYPE_BIT)
+		maxDimensions *= 32;
+
+	return maxDimensions;
+}
+
 /*
  * Initialize the build state
  */
 static void
 InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, IndexInfo *indexInfo, ForkNumber forkNum)
 {
-	int			maxDimensions = HNSW_MAX_DIM;
+	int			maxDimensions;
 
 	buildstate->heap = heap;
 	buildstate->index = index;
@@ -683,6 +698,8 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
 	buildstate->efConstruction = HnswGetEfConstruction(index);
 	buildstate->dimensions = TupleDescAttr(index->rd_att, 0)->atttypmod;
 
+	maxDimensions = GetMaxDimensions(buildstate->type);
+
 	/* Require column to have dimensions to be indexed */
 	if (buildstate->dimensions < 0)
 		elog(ERROR, "column does not have dimensions");

src/hnswscan.c

@@ -1,6 +1,8 @@
 #include "postgres.h"
 
 #include "access/relscan.h"
+#include "bitvector.h"
+#include "catalog/pg_type_d.h"
 #include "hnsw.h"
 #include "pgstat.h"
 #include "storage/bufmgr.h"

src/vector.c

@@ -2,6 +2,7 @@
 
 #include <math.h>
 
+#include "bitvector.h"
 #include "catalog/pg_type.h"
 #include "common/shortest_dec.h"
 #include "fmgr.h"
@@ -858,6 +859,25 @@ vector_mul(PG_FUNCTION_ARGS)
 	PG_RETURN_POINTER(result);
 }
 
+/*
+ * Quantize a vector
+ */
+PGDLLEXPORT PG_FUNCTION_INFO_V1(quantize_binary);
+Datum
+quantize_binary(PG_FUNCTION_ARGS)
+{
+	Vector	   *a = PG_GETARG_VECTOR_P(0);
+	float	   *ax = a->x;
+	VarBit	   *result = InitBitVector(a->dim);
+	unsigned char *rx = VARBITS(result);
+
+	for (int i = 0; i < a->dim; i++)
+		rx[i / 8] |= (ax[i] > 0) << (7 - (i % 8));
+
+	PG_RETURN_VARBIT_P(result);
+}
+
+
 /*
  * Internal helper to compare vectors
  */