apunkt/sqlite-vec
1
0
Fork 0
mirror of https://github.com/asg017/sqlite-vec.git synced 2026-04-25 00:36:56 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

add broken shadow table test

Browse source
This commit is contained in:
Alex Garcia 2024-11-19 22:03:31 -08:00
parent a657b3a216
commit e99e31feb7
5 changed files with 226 additions and 36 deletions
Download patch file Download diff file Expand all files Collapse all files

44
ARCHITECTURE.md
View file

@ -1,3 +1,13 @@
# `sqlite-vec` Architecture
Internal documentation for how `sqlite-vec` works under-the-hood. Not meant for
users of the `sqlite-vec` project, consult
[the official `sqlite-vec` documentation](https://alexgarcia.xyz/sqlite-vec) for
how-to-guides. Rather, this is for people interested in how `sqlite-vec` works
and some guidelines to any future contributors.
Very much a WIP.
## `vec0`
### Shadow Tables
@ -9,7 +19,6 @@
- `validity BLOB`
- `rowids BLOB`
#### `xyz_rowids`
- `rowid INTEGER`
@ -32,7 +41,6 @@
- `rowid INTEGER`
- `data BLOB`
#### `xyz_metadata_text_data_00`
- `rowid INTEGER`
@ -52,8 +60,11 @@ The "header" charcter denotes the type of query plan, as determined by the
| `VEC0_QUERY_PLAN_POINT` | `'2'` | Perform a single-lookup point query for the provided rowid |
| `VEC0_QUERY_PLAN_KNN` | `'3'` | Perform a KNN-style query on the provided query vector and parameters. |
Each 4-character "block" is associated with a corresponding value in `argv[]`. For example, the 1st block at byte offset `1-4` (inclusive) is the 1st block and is associated with `argv[1]`. The 2nd block at byte offset `5-8` (inclusive) is associated with `argv[2]` and so on. Each block describes what kind of value or filter the given `argv[i]` value is.
Each 4-character "block" is associated with a corresponding value in `argv[]`.
For example, the 1st block at byte offset `1-4` (inclusive) is the 1st block and
is associated with `argv[1]`. The 2nd block at byte offset `5-8` (inclusive) is
associated with `argv[2]` and so on. Each block describes what kind of value or
filter the given `argv[i]` value is.
#### `VEC0_IDXSTR_KIND_KNN_MATCH` (`'{'`)
@ -69,7 +80,8 @@ The remaining 3 characters of the block are `_` fillers.
#### `VEC0_IDXSTR_KIND_KNN_ROWID_IN` (`'['`)
`argv[i]` is the optional `rowid in (...)` value, and must be handled with [`sqlite3_vtab_in_first()` /
`argv[i]` is the optional `rowid in (...)` value, and must be handled with
[`sqlite3_vtab_in_first()` /
`sqlite3_vtab_in_next()`](https://www.sqlite.org/c3ref/vtab_in_first.html).
The remaining 3 characters of the block are `_` fillers.
@ -78,13 +90,16 @@ The remaining 3 characters of the block are `_` fillers.
`argv[i]` is a "constraint" on a specific partition key.
The second character of the block denotes which partition key to filter on, using `A` to denote the first partition key column, `B` for the second, etc. It is encoded with `'A' + partition_idx` and can be decoded with `c - 'A'`.
The second character of the block denotes which partition key to filter on,
using `A` to denote the first partition key column, `B` for the second, etc. It
is encoded with `'A' + partition_idx` and can be decoded with `c - 'A'`.
The third character of the block denotes which operator is used in the constraint. It will be one of the values of `enum vec0_partition_operator`, as only a subset of operations are supported on partition keys.
The third character of the block denotes which operator is used in the
constraint. It will be one of the values of `enum vec0_partition_operator`, as
only a subset of operations are supported on partition keys.
The fourth character of the block is a `_` filler.
#### `VEC0_IDXSTR_KIND_POINT_ID` (`'!'`)
`argv[i]` is the value of the rowid or id to match against for the point query.
@ -93,11 +108,16 @@ The remaining 3 characters of the block are `_` fillers.
#### `VEC0_IDXSTR_KIND_METADATA_CONSTRAINT` (`'&'`)
`argv[i]` is the value of the `WHERE` constraint for a metdata column in a KNN query.
`argv[i]` is the value of the `WHERE` constraint for a metdata column in a KNN
query.
The second character of the block denotes which metadata column the constraint belongs to, using `A` to denote the first metadata column column, `B` for the second, etc. It is encoded with `'A' + metadata_idx` and can be decoded with `c - 'A'`.
The second character of the block denotes which metadata column the constraint
belongs to, using `A` to denote the first metadata column column, `B` for the
second, etc. It is encoded with `'A' + metadata_idx` and can be decoded with
`c - 'A'`.
The third character of the block is the constraint operator. It will be one of `enum vec0_metadata_operator`, as only a subset of operators are supported on metadata column KNN filters.
The third character of the block is the constraint operator. It will be one of
`enum vec0_metadata_operator`, as only a subset of operators are supported on
metadata column KNN filters.
The foruth character of the block is a `_` filler.

38
TODO
View file

@ -1,25 +1,17 @@
# partition
- [ ] UPDATE on partition key values
- remove previous row from chunk, insert into new one?
- [ ] properly sqlite3_vtab_nochange / sqlite3_value_nochange handling
# auxiliary columns
- later:
- NOT NULL?
- perf: INSERT stmt should be cached on vec0_vtab
- perf: LEFT JOIN aux table to rowids query in vec0_cursor for rowid/point stmts, to avoid N lookup queries
# metadata filtering
- `v in (...)` handling
- [ ] test accessing aux values when rowid is different than 1,2,3 etc.
- [ ] add `xyz_info` shadow table with version etc.
- later
- null!
- date/datetime
- remaining TODO items
- skip invalid validity entries in knn filter?
- dictionary encoding?
- partition `x in (...)` handling
- [ ] partition: UPDATE support
- [ ] skip invalid validity entries in knn filter?
- [ ] nulls in metadata
- [ ] partition `x in (...)` handling
- [ ] blobs/date/datetime
- [ ] uuid/ulid perf
- [ ] Aux columns: `NOT NULL` constraint
- [ ] Metadata columns: `NOT NULL` constraint
- [ ] Partiion key: `NOT NULL` constraint
- [ ] dictionary encoding?
- [ ] properly sqlite3_vtab_nochange / sqlite3_value_nochange handling
- [ ] perf
- [ ] aux: cache INSERT
- [ ] aux: LEFT JOIN on `_rowids` queries to avoid N lookup queries

2
sqlite-vec.c
View file

@ -8759,7 +8759,7 @@ static int vec0Update(sqlite3_vtab *pVTab, int argc, sqlite3_value **argv,
}
static int vec0ShadowName(const char *zName) {
static const char *azName[] = {"rowids", "chunks", "auxiliary", "vector_chunks", "metadata_chunks"};
static const char *azName[] = {"rowids", "chunks", "auxiliary"};
for (size_t i = 0; i < sizeof(azName) / sizeof(azName[0]); i++) {
if (sqlite3_stricmp(zName, azName[i]) == 0)

123
tests/__snapshots__/test-general.ambr Normal file
View file

@ -0,0 +1,123 @@
# serializer version: 1
# name: test_shadow
OrderedDict({
'sql': 'select * from sqlite_master order by name',
'rows': list([
OrderedDict({
'type': 'index',
'name': 'sqlite_autoindex_v_metadata_chunks00_1',
'tbl_name': 'v_metadata_chunks00',
'rootpage': 8,
'sql': None,
}),
OrderedDict({
'type': 'index',
'name': 'sqlite_autoindex_v_metadata_text_data_00_1',
'tbl_name': 'v_metadata_text_data_00',
'rootpage': 10,
'sql': None,
}),
OrderedDict({
'type': 'index',
'name': 'sqlite_autoindex_v_vector_chunks00_1',
'tbl_name': 'v_vector_chunks00',
'rootpage': 6,
'sql': None,
}),
OrderedDict({
'type': 'table',
'name': 'sqlite_sequence',
'tbl_name': 'sqlite_sequence',
'rootpage': 3,
'sql': 'CREATE TABLE sqlite_sequence(name,seq)',
}),
OrderedDict({
'type': 'table',
'name': 'v',
'tbl_name': 'v',
'rootpage': 0,
'sql': 'CREATE VIRTUAL TABLE v using vec0(a float[1], partition text partition key, metadata text, +name text, chunk_size=8)',
}),
OrderedDict({
'type': 'table',
'name': 'v_auxiliary',
'tbl_name': 'v_auxiliary',
'rootpage': 11,
'sql': 'CREATE TABLE "v_auxiliary"( rowid integer PRIMARY KEY , value00)',
}),
OrderedDict({
'type': 'table',
'name': 'v_chunks',
'tbl_name': 'v_chunks',
'rootpage': 2,
'sql': 'CREATE TABLE "v_chunks"(chunk_id INTEGER PRIMARY KEY AUTOINCREMENT,size INTEGER NOT NULL,sequence_id integer,partition00,validity BLOB NOT NULL, rowids BLOB NOT NULL)',
}),
OrderedDict({
'type': 'table',
'name': 'v_metadata_chunks00',
'tbl_name': 'v_metadata_chunks00',
'rootpage': 7,
'sql': 'CREATE TABLE "v_metadata_chunks00"(rowid PRIMARY KEY, data BLOB NOT NULL)',
}),
OrderedDict({
'type': 'table',
'name': 'v_metadata_text_data_00',
'tbl_name': 'v_metadata_text_data_00',
'rootpage': 9,
'sql': 'CREATE TABLE "v_metadata_text_data_00"(rowid PRIMARY KEY, data TEXT)',
}),
OrderedDict({
'type': 'table',
'name': 'v_rowids',
'tbl_name': 'v_rowids',
'rootpage': 4,
'sql': 'CREATE TABLE "v_rowids"(rowid INTEGER PRIMARY KEY AUTOINCREMENT,id,chunk_id INTEGER,chunk_offset INTEGER)',
}),
OrderedDict({
'type': 'table',
'name': 'v_vector_chunks00',
'tbl_name': 'v_vector_chunks00',
'rootpage': 5,
'sql': 'CREATE TABLE "v_vector_chunks00"(rowid PRIMARY KEY,vectors BLOB NOT NULL)',
}),
]),
})
# ---
# name: test_shadow.1
OrderedDict({
'sql': "select * from pragma_table_list where type = 'shadow'",
'rows': list([
OrderedDict({
'schema': 'main',
'name': 'v_auxiliary',
'type': 'shadow',
'ncol': 2,
'wr': 0,
'strict': 0,
}),
OrderedDict({
'schema': 'main',
'name': 'v_rowids',
'type': 'shadow',
'ncol': 4,
'wr': 0,
'strict': 0,
}),
OrderedDict({
'schema': 'main',
'name': 'v_chunks',
'type': 'shadow',
'ncol': 6,
'wr': 0,
'strict': 0,
}),
]),
})
# ---
# name: test_shadow.2
OrderedDict({
'sql': "select * from pragma_table_list where type = 'shadow'",
'rows': list([
]),
})
# ---

55
tests/test-general.py Normal file
View file

@ -0,0 +1,55 @@
import sqlite3
from collections import OrderedDict
import pytest
@pytest.mark.skipif(
sqlite3.sqlite_version_info[1] < 37,
reason="pragma_table_list was added in SQLite 3.37",
)
def test_shadow(db, snapshot):
db.execute(
"create virtual table v using vec0(a float[1], partition text partition key, metadata text, +name text, chunk_size=8)"
)
assert exec(db, "select * from sqlite_master order by name") == snapshot()
assert (
exec(db, "select * from pragma_table_list where type = 'shadow'") == snapshot()
)
db.execute("drop table v;")
assert (
exec(db, "select * from pragma_table_list where type = 'shadow'") == snapshot()
)
def exec(db, sql, parameters=[]):
try:
rows = db.execute(sql, parameters).fetchall()
except (sqlite3.OperationalError, sqlite3.DatabaseError) as e:
return {
"error": e.__class__.__name__,
"message": str(e),
}
a = []
for row in rows:
o = OrderedDict()
for k in row.keys():
o[k] = row[k]
a.append(o)
result = OrderedDict()
result["sql"] = sql
result["rows"] = a
return result
def vec0_shadow_table_contents(db, v):
shadow_tables = [
row[0]
for row in db.execute(
"select name from sqlite_master where name like ? order by 1", [f"{v}_%"]
).fetchall()
]
o = {}
for shadow_table in shadow_tables:
o[shadow_table] = exec(db, f"select * from {shadow_table}")
return o