mirror of
https://github.com/pgvector/pgvector.git
synced 2026-07-22 03:57:34 +08:00
Compare commits
18 Commits
v0.6.2
...
vector-ran
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
4f4286f74e | ||
|
|
11ea3d8483 | ||
|
|
2c48e3edc2 | ||
|
|
7d63bb4b98 | ||
|
|
de410a2915 | ||
|
|
64aa99aa31 | ||
|
|
997fa167da | ||
|
|
67eec4edbf | ||
|
|
396090d8e0 | ||
|
|
ba18942fcf | ||
|
|
8e59455c3c | ||
|
|
bd50e3067d | ||
|
|
af9d4ad659 | ||
|
|
08abb63cbe | ||
|
|
06b8556a49 | ||
|
|
3f674c9994 | ||
|
|
31e41b3ba9 | ||
|
|
903a925662 |
34
README.md
34
README.md
@@ -415,7 +415,19 @@ You can use [Reciprocal Rank Fusion](https://github.com/pgvector/pgvector-python
|
||||
|
||||
### Tuning
|
||||
|
||||
Use a tool like [PgTune](https://pgtune.leopard.in.ua/) to set initial values for Postgres server parameters.
|
||||
Use a tool like [PgTune](https://pgtune.leopard.in.ua/) to set initial values for Postgres server parameters. For instance, `shared_buffers` should typically be 25% of the server’s memory. You can find the config file with:
|
||||
|
||||
```sql
|
||||
SHOW config_file;
|
||||
```
|
||||
|
||||
And check individual settings with:
|
||||
|
||||
```sql
|
||||
SHOW shared_buffers;
|
||||
```
|
||||
|
||||
Be sure to restart Postgres for changes to take effect.
|
||||
|
||||
### Loading
|
||||
|
||||
@@ -604,6 +616,18 @@ and query with:
|
||||
SELECT * FROM items ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;
|
||||
```
|
||||
|
||||
#### Are binary vectors supported?
|
||||
|
||||
You can store binary vectors and perform exact nearest neighbor search by Hamming distance in Postgres without an extension ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/hash_image_search.py)).
|
||||
|
||||
```tsql
|
||||
CREATE TABLE items (id bigserial PRIMARY KEY, embedding bit(3));
|
||||
INSERT INTO items (embedding) VALUES (B'000'), (B'111');
|
||||
SELECT * FROM items ORDER BY bit_count(embedding # B'101') LIMIT 5;
|
||||
```
|
||||
|
||||
Indexing is not currently supported.
|
||||
|
||||
#### Do indexes need to fit into memory?
|
||||
|
||||
No, but like other index types, you’ll likely see better performance if they do. You can get the size of an index with:
|
||||
@@ -659,6 +683,8 @@ ALTER TABLE items ALTER COLUMN embedding SET STORAGE PLAIN;
|
||||
|
||||
Results are limited by the size of the dynamic candidate list (`hnsw.ef_search`). There may be even less results due to dead tuples or filtering conditions in the query. We recommend setting `hnsw.ef_search` to at least twice the `LIMIT` of the query. If you need more than 500 results, use an IVFFlat index instead.
|
||||
|
||||
Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).
|
||||
|
||||
#### Why are there less results for a query after adding an IVFFlat index?
|
||||
|
||||
The index was likely created with too little data for the number of lists. Drop the index until the table has more data.
|
||||
@@ -669,11 +695,13 @@ DROP INDEX index_name;
|
||||
|
||||
Results can also be limited by the number of probes (`ivfflat.probes`).
|
||||
|
||||
Also, note that `NULL` vectors are not indexed (as well as zero vectors for cosine distance).
|
||||
|
||||
## Reference
|
||||
|
||||
### Vector Type
|
||||
|
||||
Each vector takes `4 * dimensions + 8` bytes of storage. Each element is a single precision floating-point number (like the `real` type in Postgres), and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Vectors can have up to 16,000 dimensions.
|
||||
Each vector takes `4 * dimensions + 8` bytes of storage. Each element is a single-precision floating-point number (like the `real` type in Postgres), and all elements must be finite (no `NaN`, `Infinity` or `-Infinity`). Vectors can have up to 16,000 dimensions.
|
||||
|
||||
### Vector Operators
|
||||
|
||||
@@ -697,7 +725,7 @@ l1_distance(vector, vector) → double precision | taxicab distance | 0.5.0
|
||||
vector_dims(vector) → integer | number of dimensions |
|
||||
vector_norm(vector) → double precision | Euclidean norm |
|
||||
|
||||
### Aggregate Functions
|
||||
### Vector Aggregate Functions
|
||||
|
||||
Function | Description | Added
|
||||
--- | --- | ---
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
|
||||
\echo Use "CREATE EXTENSION vector" to load this file. \quit
|
||||
|
||||
-- type
|
||||
-- vector type
|
||||
|
||||
CREATE TYPE vector;
|
||||
|
||||
@@ -29,7 +29,7 @@ CREATE TYPE vector (
|
||||
STORAGE = external
|
||||
);
|
||||
|
||||
-- functions
|
||||
-- vector functions
|
||||
|
||||
CREATE FUNCTION l2_distance(vector, vector) RETURNS float8
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
@@ -58,7 +58,7 @@ CREATE FUNCTION vector_sub(vector, vector) RETURNS vector
|
||||
CREATE FUNCTION vector_mul(vector, vector) RETURNS vector
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
-- private functions
|
||||
-- vector private functions
|
||||
|
||||
CREATE FUNCTION vector_lt(vector, vector) RETURNS bool
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
@@ -99,7 +99,7 @@ CREATE FUNCTION vector_avg(double precision[]) RETURNS vector
|
||||
CREATE FUNCTION vector_combine(double precision[], double precision[]) RETURNS double precision[]
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
-- aggregates
|
||||
-- vector aggregates
|
||||
|
||||
CREATE AGGREGATE avg(vector) (
|
||||
SFUNC = vector_accum,
|
||||
@@ -117,7 +117,7 @@ CREATE AGGREGATE sum(vector) (
|
||||
PARALLEL = SAFE
|
||||
);
|
||||
|
||||
-- cast functions
|
||||
-- vector cast functions
|
||||
|
||||
CREATE FUNCTION vector(vector, integer, boolean) RETURNS vector
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
@@ -137,7 +137,7 @@ CREATE FUNCTION array_to_vector(numeric[], integer, boolean) RETURNS vector
|
||||
CREATE FUNCTION vector_to_float4(vector, integer, boolean) RETURNS real[]
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
-- casts
|
||||
-- vector casts
|
||||
|
||||
CREATE CAST (vector AS vector)
|
||||
WITH FUNCTION vector(vector, integer, boolean) AS IMPLICIT;
|
||||
@@ -157,7 +157,7 @@ CREATE CAST (double precision[] AS vector)
|
||||
CREATE CAST (numeric[] AS vector)
|
||||
WITH FUNCTION array_to_vector(numeric[], integer, boolean) AS ASSIGNMENT;
|
||||
|
||||
-- operators
|
||||
-- vector operators
|
||||
|
||||
CREATE OPERATOR <-> (
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = l2_distance,
|
||||
@@ -240,7 +240,7 @@ CREATE ACCESS METHOD hnsw TYPE INDEX HANDLER hnswhandler;
|
||||
|
||||
COMMENT ON ACCESS METHOD hnsw IS 'hnsw index access method';
|
||||
|
||||
-- opclasses
|
||||
-- vector opclasses
|
||||
|
||||
CREATE OPERATOR CLASS vector_ops
|
||||
DEFAULT FOR TYPE vector USING btree AS
|
||||
|
||||
10
src/hnsw.h
10
src/hnsw.h
@@ -55,6 +55,11 @@
|
||||
#define HNSW_UPDATE_ENTRY_GREATER 1
|
||||
#define HNSW_UPDATE_ENTRY_ALWAYS 2
|
||||
|
||||
typedef enum HnswType
|
||||
{
|
||||
HNSW_TYPE_VECTOR
|
||||
} HnswType;
|
||||
|
||||
/* Build phases */
|
||||
/* PROGRESS_CREATEIDX_SUBPHASE_INITIALIZE is 1 */
|
||||
#define PROGRESS_HNSW_PHASE_LOAD 2
|
||||
@@ -242,6 +247,7 @@ typedef struct HnswBuildState
|
||||
Relation index;
|
||||
IndexInfo *indexInfo;
|
||||
ForkNumber forkNum;
|
||||
HnswType type;
|
||||
|
||||
/* Settings */
|
||||
int dimensions;
|
||||
@@ -262,7 +268,6 @@ typedef struct HnswBuildState
|
||||
HnswGraph *graph;
|
||||
double ml;
|
||||
int maxLevel;
|
||||
Vector *normvec;
|
||||
|
||||
/* Memory */
|
||||
MemoryContext graphCtx;
|
||||
@@ -367,7 +372,8 @@ typedef struct HnswVacuumState
|
||||
int HnswGetM(Relation index);
|
||||
int HnswGetEfConstruction(Relation index);
|
||||
FmgrInfo *HnswOptionalProcInfo(Relation index, uint16 procnum);
|
||||
bool HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
|
||||
HnswType HnswGetType(Relation index);
|
||||
bool HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, HnswType type);
|
||||
Buffer HnswNewBuffer(Relation index, ForkNumber forkNum);
|
||||
void HnswInitPage(Buffer buf, Page page);
|
||||
void HnswInit(void);
|
||||
|
||||
@@ -489,7 +489,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
|
||||
/* Normalize if needed */
|
||||
if (buildstate->normprocinfo != NULL)
|
||||
{
|
||||
if (!HnswNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->normvec))
|
||||
if (!HnswNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->type))
|
||||
return false;
|
||||
}
|
||||
|
||||
@@ -671,10 +671,13 @@ HnswSharedMemoryAlloc(Size size, void *state)
|
||||
static void
|
||||
InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, IndexInfo *indexInfo, ForkNumber forkNum)
|
||||
{
|
||||
int maxDimensions = HNSW_MAX_DIM;
|
||||
|
||||
buildstate->heap = heap;
|
||||
buildstate->index = index;
|
||||
buildstate->indexInfo = indexInfo;
|
||||
buildstate->forkNum = forkNum;
|
||||
buildstate->type = HnswGetType(index);
|
||||
|
||||
buildstate->m = HnswGetM(index);
|
||||
buildstate->efConstruction = HnswGetEfConstruction(index);
|
||||
@@ -684,8 +687,8 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
|
||||
if (buildstate->dimensions < 0)
|
||||
elog(ERROR, "column does not have dimensions");
|
||||
|
||||
if (buildstate->dimensions > HNSW_MAX_DIM)
|
||||
elog(ERROR, "column cannot have more than %d dimensions for hnsw index", HNSW_MAX_DIM);
|
||||
if (buildstate->dimensions > maxDimensions)
|
||||
elog(ERROR, "column cannot have more than %d dimensions for hnsw index", maxDimensions);
|
||||
|
||||
if (buildstate->efConstruction < 2 * buildstate->m)
|
||||
elog(ERROR, "ef_construction must be greater than or equal to 2 * m");
|
||||
@@ -703,9 +706,6 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
|
||||
buildstate->ml = HnswGetMl(buildstate->m);
|
||||
buildstate->maxLevel = HnswGetMaxLevel(buildstate->m);
|
||||
|
||||
/* Reuse for each tuple */
|
||||
buildstate->normvec = InitVector(buildstate->dimensions);
|
||||
|
||||
buildstate->graphCtx = GenerationContextCreate(CurrentMemoryContext,
|
||||
"Hnsw build graph context",
|
||||
#if PG_VERSION_NUM >= 150000
|
||||
@@ -729,7 +729,6 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
|
||||
static void
|
||||
FreeBuildState(HnswBuildState * buildstate)
|
||||
{
|
||||
pfree(buildstate->normvec);
|
||||
MemoryContextDelete(buildstate->graphCtx);
|
||||
MemoryContextDelete(buildstate->tmpCtx);
|
||||
}
|
||||
|
||||
@@ -622,7 +622,7 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
|
||||
normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
|
||||
if (normprocinfo != NULL)
|
||||
{
|
||||
if (!HnswNormValue(normprocinfo, collation, &value, NULL))
|
||||
if (!HnswNormValue(normprocinfo, collation, &value, HnswGetType(index)))
|
||||
return;
|
||||
}
|
||||
|
||||
|
||||
@@ -40,29 +40,6 @@ GetScanItems(IndexScanDesc scan, Datum q)
|
||||
return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
|
||||
}
|
||||
|
||||
/*
|
||||
* Get dimensions from metapage
|
||||
*/
|
||||
static int
|
||||
GetDimensions(Relation index)
|
||||
{
|
||||
Buffer buf;
|
||||
Page page;
|
||||
HnswMetaPage metap;
|
||||
int dimensions;
|
||||
|
||||
buf = ReadBuffer(index, HNSW_METAPAGE_BLKNO);
|
||||
LockBuffer(buf, BUFFER_LOCK_SHARE);
|
||||
page = BufferGetPage(buf);
|
||||
metap = HnswPageGetMeta(page);
|
||||
|
||||
dimensions = metap->dimensions;
|
||||
|
||||
UnlockReleaseBuffer(buf);
|
||||
|
||||
return dimensions;
|
||||
}
|
||||
|
||||
/*
|
||||
* Get scan value
|
||||
*/
|
||||
@@ -73,7 +50,7 @@ GetScanValue(IndexScanDesc scan)
|
||||
Datum value;
|
||||
|
||||
if (scan->orderByData->sk_flags & SK_ISNULL)
|
||||
value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
|
||||
value = PointerGetDatum(NULL);
|
||||
else
|
||||
{
|
||||
value = scan->orderByData->sk_argument;
|
||||
@@ -84,7 +61,7 @@ GetScanValue(IndexScanDesc scan)
|
||||
|
||||
/* Fine if normalization fails */
|
||||
if (so->normprocinfo != NULL)
|
||||
HnswNormValue(so->normprocinfo, so->collation, &value, NULL);
|
||||
HnswNormValue(so->normprocinfo, so->collation, &value, HnswGetType(scan->indexRelation));
|
||||
}
|
||||
|
||||
return value;
|
||||
|
||||
@@ -149,6 +149,15 @@ HnswOptionalProcInfo(Relation index, uint16 procnum)
|
||||
return index_getprocinfo(index, 1, procnum);
|
||||
}
|
||||
|
||||
/*
|
||||
* Get vector type
|
||||
*/
|
||||
HnswType
|
||||
HnswGetType(Relation index)
|
||||
{
|
||||
return HNSW_TYPE_VECTOR;
|
||||
}
|
||||
|
||||
/*
|
||||
* Divide by the norm
|
||||
*
|
||||
@@ -158,21 +167,25 @@ HnswOptionalProcInfo(Relation index, uint16 procnum)
|
||||
* if it's different than the original value
|
||||
*/
|
||||
bool
|
||||
HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result)
|
||||
HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, HnswType type)
|
||||
{
|
||||
double norm = DatumGetFloat8(FunctionCall1Coll(procinfo, collation, *value));
|
||||
|
||||
if (norm > 0)
|
||||
{
|
||||
Vector *v = DatumGetVector(*value);
|
||||
/* TODO Remove vector-specific code */
|
||||
if (type == HNSW_TYPE_VECTOR)
|
||||
{
|
||||
Vector *v = DatumGetVector(*value);
|
||||
Vector *result = InitVector(v->dim);
|
||||
|
||||
if (result == NULL)
|
||||
result = InitVector(v->dim);
|
||||
for (int i = 0; i < v->dim; i++)
|
||||
result->x[i] = v->x[i] / norm;
|
||||
|
||||
for (int i = 0; i < v->dim; i++)
|
||||
result->x[i] = v->x[i] / norm;
|
||||
|
||||
*value = PointerGetDatum(result);
|
||||
*value = PointerGetDatum(result);
|
||||
}
|
||||
else
|
||||
elog(ERROR, "Unsupported type");
|
||||
|
||||
return true;
|
||||
}
|
||||
@@ -575,7 +588,12 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
|
||||
|
||||
/* Calculate distance */
|
||||
if (distance != NULL)
|
||||
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
|
||||
{
|
||||
if (DatumGetPointer(*q) == NULL)
|
||||
*distance = 0;
|
||||
else
|
||||
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
|
||||
}
|
||||
|
||||
UnlockReleaseBuffer(buf);
|
||||
}
|
||||
|
||||
@@ -57,7 +57,7 @@ AddSample(Datum *values, IvfflatBuildState * buildstate)
|
||||
*/
|
||||
if (buildstate->kmeansnormprocinfo != NULL)
|
||||
{
|
||||
if (!IvfflatNormValue(buildstate->kmeansnormprocinfo, buildstate->collation, &value, buildstate->normvec))
|
||||
if (!IvfflatNormValue(buildstate->kmeansnormprocinfo, buildstate->collation, &value))
|
||||
return;
|
||||
}
|
||||
|
||||
@@ -105,7 +105,7 @@ SampleCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
|
||||
oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
|
||||
|
||||
/* Add sample */
|
||||
AddSample(values, state);
|
||||
AddSample(values, buildstate);
|
||||
|
||||
/* Reset memory context */
|
||||
MemoryContextSwitchTo(oldCtx);
|
||||
@@ -153,7 +153,7 @@ AddTupleToSort(Relation index, ItemPointer tid, Datum *values, IvfflatBuildState
|
||||
/* Normalize if needed */
|
||||
if (buildstate->normprocinfo != NULL)
|
||||
{
|
||||
if (!IvfflatNormValue(buildstate->normprocinfo, buildstate->collation, &value, buildstate->normvec))
|
||||
if (!IvfflatNormValue(buildstate->normprocinfo, buildstate->collation, &value))
|
||||
return;
|
||||
}
|
||||
|
||||
@@ -356,9 +356,6 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
|
||||
buildstate->centers = VectorArrayInit(buildstate->lists, buildstate->dimensions);
|
||||
buildstate->listInfo = palloc(sizeof(ListInfo) * buildstate->lists);
|
||||
|
||||
/* Reuse for each tuple */
|
||||
buildstate->normvec = InitVector(buildstate->dimensions);
|
||||
|
||||
buildstate->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
|
||||
"Ivfflat build temporary context",
|
||||
ALLOCSET_DEFAULT_SIZES);
|
||||
@@ -380,7 +377,6 @@ FreeBuildState(IvfflatBuildState * buildstate)
|
||||
{
|
||||
VectorArrayFree(buildstate->centers);
|
||||
pfree(buildstate->listInfo);
|
||||
pfree(buildstate->normvec);
|
||||
|
||||
#ifdef IVFFLAT_KMEANS_DEBUG
|
||||
pfree(buildstate->listSums);
|
||||
|
||||
@@ -172,7 +172,6 @@ typedef struct IvfflatBuildState
|
||||
VectorArray samples;
|
||||
VectorArray centers;
|
||||
ListInfo *listInfo;
|
||||
Vector *normvec;
|
||||
|
||||
#ifdef IVFFLAT_KMEANS_DEBUG
|
||||
double inertia;
|
||||
@@ -267,7 +266,7 @@ void VectorArrayFree(VectorArray arr);
|
||||
void PrintVectorArray(char *msg, VectorArray arr);
|
||||
void IvfflatKmeans(Relation index, VectorArray samples, VectorArray centers);
|
||||
FmgrInfo *IvfflatOptionalProcInfo(Relation index, uint16 procnum);
|
||||
bool IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
|
||||
bool IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value);
|
||||
int IvfflatGetLists(Relation index);
|
||||
void IvfflatGetMetaPageInfo(Relation index, int *lists, int *dimensions);
|
||||
void IvfflatUpdateList(Relation index, ListInfo listInfo, BlockNumber insertPage, BlockNumber originalInsertPage, BlockNumber startPage, ForkNumber forkNum);
|
||||
|
||||
@@ -85,7 +85,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
|
||||
normprocinfo = IvfflatOptionalProcInfo(index, IVFFLAT_NORM_PROC);
|
||||
if (normprocinfo != NULL)
|
||||
{
|
||||
if (!IvfflatNormValue(normprocinfo, index->rd_indcollation[0], &value, NULL))
|
||||
if (!IvfflatNormValue(normprocinfo, index->rd_indcollation[0], &value))
|
||||
return;
|
||||
}
|
||||
|
||||
|
||||
@@ -293,7 +293,7 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
|
||||
|
||||
/* Fine if normalization fails */
|
||||
if (so->normprocinfo != NULL)
|
||||
IvfflatNormValue(so->normprocinfo, so->collation, &value, NULL);
|
||||
IvfflatNormValue(so->normprocinfo, so->collation, &value);
|
||||
}
|
||||
|
||||
IvfflatBench("GetScanLists", GetScanLists(scan, value));
|
||||
|
||||
@@ -75,16 +75,14 @@ IvfflatOptionalProcInfo(Relation index, uint16 procnum)
|
||||
* if it's different than the original value
|
||||
*/
|
||||
bool
|
||||
IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result)
|
||||
IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value)
|
||||
{
|
||||
double norm = DatumGetFloat8(FunctionCall1Coll(procinfo, collation, *value));
|
||||
|
||||
if (norm > 0)
|
||||
{
|
||||
Vector *v = DatumGetVector(*value);
|
||||
|
||||
if (result == NULL)
|
||||
result = InitVector(v->dim);
|
||||
Vector *result = InitVector(v->dim);
|
||||
|
||||
for (int i = 0; i < v->dim; i++)
|
||||
result->x[i] = v->x[i] / norm;
|
||||
|
||||
16
src/vector.c
16
src/vector.c
@@ -197,6 +197,8 @@ vector_in(PG_FUNCTION_ARGS)
|
||||
|
||||
while (pt != NULL && *stringEnd != ']')
|
||||
{
|
||||
float val;
|
||||
|
||||
if (dim == VECTOR_MAX_DIM)
|
||||
ereport(ERROR,
|
||||
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
|
||||
@@ -212,15 +214,23 @@ vector_in(PG_FUNCTION_ARGS)
|
||||
errmsg("invalid input syntax for type vector: \"%s\"", lit)));
|
||||
|
||||
/* Use strtof like float4in to avoid a double-rounding problem */
|
||||
x[dim] = strtof(pt, &stringEnd);
|
||||
CheckElement(x[dim]);
|
||||
dim++;
|
||||
errno = 0;
|
||||
val = strtof(pt, &stringEnd);
|
||||
|
||||
if (stringEnd == pt)
|
||||
ereport(ERROR,
|
||||
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
|
||||
errmsg("invalid input syntax for type vector: \"%s\"", lit)));
|
||||
|
||||
/* Check for range error like float4in */
|
||||
if (errno == ERANGE && (val == 0 || isinf(val)))
|
||||
ereport(ERROR,
|
||||
(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
|
||||
errmsg("\"%s\" is out of range for type vector", pt)));
|
||||
|
||||
CheckElement(val);
|
||||
x[dim++] = val;
|
||||
|
||||
while (vector_isspace(*stringEnd))
|
||||
stringEnd++;
|
||||
|
||||
|
||||
@@ -12,14 +12,11 @@ SELECT * FROM t ORDER BY val <-> '[3,3,3]';
|
||||
[0,0,0]
|
||||
(4 rows)
|
||||
|
||||
SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector);
|
||||
val
|
||||
---------
|
||||
[0,0,0]
|
||||
[1,1,1]
|
||||
[1,2,3]
|
||||
[1,2,4]
|
||||
(4 rows)
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector)) t2;
|
||||
count
|
||||
-------
|
||||
4
|
||||
(1 row)
|
||||
|
||||
SELECT COUNT(*) FROM t;
|
||||
count
|
||||
|
||||
@@ -63,9 +63,21 @@ SELECT '[1.5e-38,-1.5e-38]'::vector;
|
||||
(1 row)
|
||||
|
||||
SELECT '[4e38,1]'::vector;
|
||||
ERROR: infinite value not allowed in vector
|
||||
ERROR: "4e38" is out of range for type vector
|
||||
LINE 1: SELECT '[4e38,1]'::vector;
|
||||
^
|
||||
SELECT '[-4e38,1]'::vector;
|
||||
ERROR: "-4e38" is out of range for type vector
|
||||
LINE 1: SELECT '[-4e38,1]'::vector;
|
||||
^
|
||||
SELECT '[1e-46,1]'::vector;
|
||||
ERROR: "1e-46" is out of range for type vector
|
||||
LINE 1: SELECT '[1e-46,1]'::vector;
|
||||
^
|
||||
SELECT '[-1e-46,1]'::vector;
|
||||
ERROR: "-1e-46" is out of range for type vector
|
||||
LINE 1: SELECT '[-1e-46,1]'::vector;
|
||||
^
|
||||
SELECT '[1,2,3'::vector;
|
||||
ERROR: malformed vector literal: "[1,2,3"
|
||||
LINE 1: SELECT '[1,2,3'::vector;
|
||||
@@ -116,8 +128,30 @@ SELECT '[1, ,3]'::vector;
|
||||
ERROR: invalid input syntax for type vector: "[1, ,3]"
|
||||
LINE 1: SELECT '[1, ,3]'::vector;
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(3);
|
||||
vector
|
||||
---------
|
||||
[1,2,3]
|
||||
(1 row)
|
||||
|
||||
SELECT '[1,2,3]'::vector(2);
|
||||
ERROR: expected 2 dimensions, not 3
|
||||
SELECT '[1,2,3]'::vector(3, 2);
|
||||
ERROR: invalid type modifier
|
||||
LINE 1: SELECT '[1,2,3]'::vector(3, 2);
|
||||
^
|
||||
SELECT '[1,2,3]'::vector('a');
|
||||
ERROR: invalid input syntax for type integer: "a"
|
||||
LINE 1: SELECT '[1,2,3]'::vector('a');
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(0);
|
||||
ERROR: dimensions for type vector must be at least 1
|
||||
LINE 1: SELECT '[1,2,3]'::vector(0);
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(16001);
|
||||
ERROR: dimensions for type vector cannot exceed 16000
|
||||
LINE 1: SELECT '[1,2,3]'::vector(16001);
|
||||
^
|
||||
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
|
||||
unnest
|
||||
---------
|
||||
|
||||
@@ -7,7 +7,7 @@ CREATE INDEX ON t USING hnsw (val vector_l2_ops);
|
||||
INSERT INTO t (val) VALUES ('[1,2,4]');
|
||||
|
||||
SELECT * FROM t ORDER BY val <-> '[3,3,3]';
|
||||
SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector);
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::vector)) t2;
|
||||
SELECT COUNT(*) FROM t;
|
||||
|
||||
TRUNCATE t;
|
||||
|
||||
@@ -11,6 +11,9 @@ SELECT '[1.5e38,-1.5e38]'::vector;
|
||||
SELECT '[1.5e+38,-1.5e+38]'::vector;
|
||||
SELECT '[1.5e-38,-1.5e-38]'::vector;
|
||||
SELECT '[4e38,1]'::vector;
|
||||
SELECT '[-4e38,1]'::vector;
|
||||
SELECT '[1e-46,1]'::vector;
|
||||
SELECT '[-1e-46,1]'::vector;
|
||||
SELECT '[1,2,3'::vector;
|
||||
SELECT '[1,2,3]9'::vector;
|
||||
SELECT '1,2,3'::vector;
|
||||
@@ -22,7 +25,13 @@ SELECT '[1,]'::vector;
|
||||
SELECT '[1a]'::vector;
|
||||
SELECT '[1,,3]'::vector;
|
||||
SELECT '[1, ,3]'::vector;
|
||||
|
||||
SELECT '[1,2,3]'::vector(3);
|
||||
SELECT '[1,2,3]'::vector(2);
|
||||
SELECT '[1,2,3]'::vector(3, 2);
|
||||
SELECT '[1,2,3]'::vector('a');
|
||||
SELECT '[1,2,3]'::vector(0);
|
||||
SELECT '[1,2,3]'::vector(16001);
|
||||
|
||||
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
|
||||
SELECT '{"[1,2,3]"}'::vector(2)[];
|
||||
|
||||
@@ -86,7 +86,7 @@ foreach (@queries)
|
||||
push(@expected, $res);
|
||||
}
|
||||
|
||||
test_recall(0.20, $limit, "before vacuum");
|
||||
test_recall(0.19, $limit, "before vacuum");
|
||||
test_recall(0.95, 100, "before vacuum");
|
||||
|
||||
# TODO Test concurrent inserts with vacuum
|
||||
|
||||
Reference in New Issue
Block a user