From 53014f55edf4dd969cfff6f100fcd979f5f0acb2 Mon Sep 17 00:00:00 2001 From: Gatefixer <313497061+lancedb-gatefixer[bot]@users.noreply.github.com> Date: Wed, 26 Aug 2026 05:38:46 +0000 Subject: [PATCH] fix: canonicalize materialized view filters --- rust/lancedb/src/materialized_view.rs | 13 +++++-- rust/lancedb/src/materialized_view/refresh.rs | 34 +++++++++++++++++++ 2 files changed, 45 insertions(+), 2 deletions(-) diff --git a/rust/lancedb/src/materialized_view.rs b/rust/lancedb/src/materialized_view.rs index b28d52931..08d6c921e 100644 --- a/rust/lancedb/src/materialized_view.rs +++ b/rust/lancedb/src/materialized_view.rs @@ -170,6 +170,15 @@ pub(crate) fn plan( filter: Option<&str>, limit: Option, ) -> Result<(MaterializedViewDefinition, Vec, Lineage)> { + let filter = filter + .map(crate::expr::canonicalize_sql_predicate) + .transpose() + .map_err(|err| match err { + Error::InvalidInput { message } => Error::InvalidInput { + message: format!("invalid view filter: {message}"), + }, + err => err, + })?; let projections: Vec<(String, String)> = if projections.is_empty() { source_schema .fields() @@ -274,7 +283,7 @@ pub(crate) fn plan( declared.push(output); } - if let Some(filter) = filter { + if let Some(filter) = filter.as_deref() { let expr = planner .parse_filter(filter) .map_err(|e| Error::InvalidInput { @@ -314,7 +323,7 @@ pub(crate) fn plan( .into_iter() .map(|(output, expression)| ViewProjection { output, expression }) .collect(), - filter: filter.map(String::from), + filter, limit, inputs, }; diff --git a/rust/lancedb/src/materialized_view/refresh.rs b/rust/lancedb/src/materialized_view/refresh.rs index 735751c27..39172724c 100644 --- a/rust/lancedb/src/materialized_view/refresh.rs +++ b/rust/lancedb/src/materialized_view/refresh.rs @@ -1585,6 +1585,40 @@ mod tests { assert_eq!(read(view.table(), "x").await, vec![20, 40]); } + #[tokio::test] + async fn test_mixed_case_filter_is_canonicalized_for_lineage_and_refresh() { + let conn = connect("memory://").execute().await.unwrap(); + let batch = record_batch!( + ("id", Int32, [1, 2, 3]), + ("PartyAbbrev", Utf8, ["D", "R", "D"]) + ) + .unwrap(); + conn.create_table("src", batch) + .write_options(crate::materialized_view::tests::stable_row_ids()) + .execute() + .await + .unwrap(); + conn.create_materialized_view("democrats", "src") + .select([("id", "id")]) + .only_if(r#""PartyAbbrev" = 'D'"#) + .execute() + .await + .unwrap(); + + // Reopen from schema metadata so these assertions cover the stored + // predicate and lineage, not only the declaration-time handle. + let view = conn.open_materialized_view("democrats").await.unwrap(); + assert_eq!( + view.definition().filter.as_deref(), + Some("`PartyAbbrev` = 'D'") + ); + assert_eq!(view.definition().inputs, ["PartyAbbrev", "id"]); + + let result = view.refresh().execute().await.unwrap(); + assert_eq!(result.rows_written, 2); + assert_eq!(read(view.table(), "id").await, vec![1, 3]); + } + #[tokio::test] async fn test_append_refreshes_incrementally() { let (_conn, source, view) = refreshed_doubled(vec![1, 2]).await;