feat: add update_field_metadata to edit per-field metadata (#3482)

### Summary
Adds update_field_metadata to the client SDK (Rust core, Python, and
TypeScript) so clients can edit per-field (column) Arrow metadata
(schema.fields[].metadata)

### Testing
- added unit tests
- ran E2E against a local server on both local and remote tables (set →
merge → delete), across Python sync/async and TypeScript

### Next steps
- deprecate replace_field_metadata in the python lancedb favor of this
(typescript didn't have replace_field_metadata method). This matches
Lance's API direction (Lance already deprecated replace_field_metadata
for update_field_metadata)
This commit is contained in:
Brendan Clement
2026-06-02 07:00:00 -07:00
committed by GitHub
parent 7b874905fd
commit d065be0474
17 changed files with 537 additions and 7 deletions

View File

@@ -1571,6 +1571,33 @@ describe("schema evolution", function () {
expect(await table.schema()).toEqual(expectedSchema3);
});
it("can update field metadata", async function () {
const con = await connect(tmpDir.name);
const table = await con.createTable("fm", [
{ id: 1, category: "a" },
{ id: 2, category: "b" },
]);
const res = await table.updateFieldMetadata([
{ path: "category", metadata: { unit: "label", pii: "false" } },
]);
expect(res).toHaveProperty("version");
expect(res.version).toBe(2);
let cat = (await table.schema()).fields.find((f) => f.name === "category");
expect(cat?.metadata.get("unit")).toBe("label");
expect(cat?.metadata.get("pii")).toBe("false");
// merge: add a key, delete one via null, keep the rest
await table.updateFieldMetadata([
{ path: "category", metadata: { source: "import", pii: null } },
]);
cat = (await table.schema()).fields.find((f) => f.name === "category");
expect(cat?.metadata.get("unit")).toBe("label"); // preserved
expect(cat?.metadata.get("source")).toBe("import"); // added
expect(cat?.metadata.has("pii")).toBe(false); // deleted
});
it("can cast to various types", async function () {
const con = await connect(tmpDir.name);

View File

@@ -42,6 +42,7 @@ export {
AddResult,
AddColumnsResult,
AlterColumnsResult,
UpdateFieldMetadataResult,
DeleteResult,
DropColumnsResult,
UpdateResult,
@@ -117,6 +118,7 @@ export {
WriteProgress,
LsmWriteSpec,
ColumnAlteration,
FieldMetadataUpdate,
} from "./table";
export {

View File

@@ -32,6 +32,7 @@ import {
OptimizeStats,
TableStatistics,
Tags,
UpdateFieldMetadataResult,
UpdateResult,
Table as _NativeTable,
} from "./native";
@@ -508,6 +509,18 @@ export abstract class Table {
abstract alterColumns(
columnAlterations: ColumnAlteration[],
): Promise<AlterColumnsResult>;
/**
* Update per-field (column) metadata.
* @param {FieldMetadataUpdate[]} updates One or more per-field updates. Each
* update's metadata is merged into the field's existing metadata by default;
* a value of `null` deletes that key, and `replace: true` swaps the whole map.
* @returns {Promise<UpdateFieldMetadataResult>} resolves to the new table version.
*/
abstract updateFieldMetadata(
updates: FieldMetadataUpdate[],
): Promise<UpdateFieldMetadataResult>;
/**
* Drop one or more columns from the dataset
*
@@ -1037,6 +1050,12 @@ export class LocalTable extends Table {
return await this.inner.alterColumns(processedAlterations);
}
async updateFieldMetadata(
updates: FieldMetadataUpdate[],
): Promise<UpdateFieldMetadataResult> {
return await this.inner.updateFieldMetadata(updates);
}
async dropColumns(columnNames: string[]): Promise<DropColumnsResult> {
return await this.inner.dropColumns(columnNames);
}
@@ -1203,3 +1222,19 @@ export interface ColumnAlteration {
/** Set the new nullability. Note that a nullable column cannot be made non-nullable. */
nullable?: boolean;
}
/** A per-field metadata update, addressed by dot-path. */
export interface FieldMetadataUpdate {
/**
* Dot-separated path to the field. For a top-level column this is just its
* name; for a nested field it's the path, e.g. "a.b.c".
*/
path: string;
/**
* Metadata key/value pairs. Merged into the field's existing metadata by
* default; a value of `null` deletes that key.
*/
metadata: Record<string, string | null>;
/** If true, replace the field's entire metadata map instead of merging. */
replace?: boolean;
}

View File

@@ -5,8 +5,9 @@ use std::collections::HashMap;
use lancedb::ipc::{ipc_file_to_batches, ipc_file_to_schema};
use lancedb::table::{
AddDataMode, ColumnAlteration as LanceColumnAlteration, Duration, NewColumnTransform,
OptimizeAction, OptimizeOptions, Table as LanceDbTable,
AddDataMode, ColumnAlteration as LanceColumnAlteration, Duration,
FieldMetadataUpdate as LanceFieldMetadataUpdate, NewColumnTransform, OptimizeAction,
OptimizeOptions, Table as LanceDbTable,
};
use napi::bindgen_prelude::*;
use napi::threadsafe_function::{ThreadsafeFunction, ThreadsafeFunctionCallMode};
@@ -355,6 +356,23 @@ impl Table {
Ok(res.into())
}
#[napi(catch_unwind)]
pub async fn update_field_metadata(
&self,
updates: Vec<FieldMetadataUpdate>,
) -> napi::Result<UpdateFieldMetadataResult> {
let updates = updates
.into_iter()
.map(LanceFieldMetadataUpdate::from)
.collect::<Vec<_>>();
let res = self
.inner_ref()?
.update_field_metadata(&updates)
.await
.default_error()?;
Ok(res.into())
}
#[napi(catch_unwind)]
pub async fn drop_columns(&self, columns: Vec<String>) -> napi::Result<DropColumnsResult> {
let col_refs = columns.iter().map(String::as_str).collect::<Vec<_>>();
@@ -747,6 +765,29 @@ pub struct ColumnAlteration {
pub nullable: Option<bool>,
}
/// A per-field metadata update, addressed by dot-path. Merges into the field's
/// existing metadata by default; a `null` value deletes a key, and `replace`
/// swaps the field's entire metadata map.
#[napi(object)]
pub struct FieldMetadataUpdate {
/// Dot-separated path to the field (e.g. "embedding" or "a.b.c").
pub path: String,
/// Metadata keys to set; a `null` value deletes that key.
pub metadata: HashMap<String, Option<String>>,
/// If true, replace the field's entire metadata map instead of merging.
pub replace: Option<bool>,
}
impl From<FieldMetadataUpdate> for LanceFieldMetadataUpdate {
fn from(js: FieldMetadataUpdate) -> Self {
Self {
path: js.path,
metadata: js.metadata,
replace: js.replace.unwrap_or(false),
}
}
}
impl TryFrom<ColumnAlteration> for LanceColumnAlteration {
type Error = String;
fn try_from(js: ColumnAlteration) -> std::result::Result<Self, Self::Error> {
@@ -987,6 +1028,19 @@ impl From<lancedb::table::AlterColumnsResult> for AlterColumnsResult {
}
}
#[napi(object)]
pub struct UpdateFieldMetadataResult {
pub version: i64,
}
impl From<lancedb::table::UpdateFieldMetadataResult> for UpdateFieldMetadataResult {
fn from(value: lancedb::table::UpdateFieldMetadataResult) -> Self {
Self {
version: value.version as i64,
}
}
}
#[napi(object)]
pub struct DropColumnsResult {
pub version: i64,