feat(compiler): truncate Date/DateTime group_by keys to bucket units (4ab65f85) · Commits · GitLab.org / orbit / GitLab Knowledge Graph

config/QUERY_DSL_VERSION

+1 −1

Original line number	Diff line number	Diff line
		2.0.0
		2.1.0

config/schemas/graph_query.schema.json

+19 −0

Original line number	Diff line number	Diff line
		@@ -634,6 +634,10 @@
		"$ref": "#/$defs/Identifier",
		"description": "Property name to group by"
		},
		"transform": {
		"$ref": "#/$defs/PropertyTransform",
		"description": "Optional transform applied to the property before grouping (e.g. date truncation)."
		},
		"alias": {
		"$ref": "#/$defs/Identifier",
		"description": "Output column alias for this group key"
		@@ -643,6 +647,21 @@
		}
		]
		},
		"PropertyTransform": {
		"description": "Tagged enum: add new variants by `kind`.",
		"oneOf": [
		{
		"type": "object",
		"description": "Truncate a Date/DateTime property. minute/hour require node_ids or a filter on the property.",
		"required": ["kind", "unit"],
		"properties": {
		"kind": {"const": "truncate"},
		"unit": {"enum": ["minute", "hour", "day", "week", "month", "quarter", "year"]}
		},
		"additionalProperties": false
		}
		]
		},
		"AggFunction": {
		"type": "string",
		"description": "Available aggregation functions",

crates/gkg-server/src/tools/schema.rs

+2 −2

Original line number	Diff line number	Diff line
		@@ -106,8 +106,8 @@ mod tests {
		let condensed = condensed_query_schema().expect("Should condense");

		assert!(
		condensed.len() < 20000,
		"Condensed schema should be under 20KB, got {} bytes",
		condensed.len() < 22000,
		"Condensed schema should be under 22KB, got {} bytes",
		condensed.len()
		);
		}

crates/integration-tests/tests/compiler/dialects/clickhouse.rs

+163 −0

Original line number	Diff line number	Diff line
		@@ -86,6 +86,169 @@ fn aggregation_query() {
		assert!(rendered.contains("GROUP BY"));
		}

		#[test]
		fn group_by_property_truncate_month_wraps_column() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "filters": {"confidential": {"op": "eq", "value": false}}}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "month"}}
		],
		"limit": 50
		}"#;
		let result = compile(json, &test_ontology(), &test_ctx()).unwrap();
		let rendered = result.base.render();
		assert!(
		rendered.contains("toStartOfMonth(u.created_at)"),
		"expected toStartOfMonth wrapper; got:\n{rendered}"
		);
		assert!(
		rendered.contains("toStartOfMonth(u.created_at) AS created_at_month"),
		"expected default alias `created_at_month`; got:\n{rendered}"
		);
		}

		#[test]
		fn group_by_property_truncate_all_units_compile() {
		for unit in ["minute", "hour", "day", "week", "month", "quarter", "year"] {
		let json = format!(
		r#"{{
		"query_type": "aggregation",
		"nodes": [
		{{"id": "u", "entity": "Note", "node_ids": [1]}}
		],
		"aggregations": [{{"function": "count", "target": "u", "alias": "n"}}],
		"group_by": [
		{{"kind": "property", "node": "u", "property": "created_at", "transform": {{"kind": "truncate", "unit": "{unit}"}}}}
		],
		"limit": 10
		}}"#
		);
		let result = compile(&json, &test_ontology(), &test_ctx())
		.unwrap_or_else(\|e\| panic!("compile failed for unit {unit}: {e:?}"));
		let rendered = result.base.render();
		let expected = match unit {
		"minute" => "toStartOfMinute",
		"hour" => "toStartOfHour",
		"day" => "toStartOfDay",
		"week" => "toStartOfWeek",
		"month" => "toStartOfMonth",
		"quarter" => "toStartOfQuarter",
		"year" => "toStartOfYear",
		_ => unreachable!(),
		};
		assert!(
		rendered.contains(expected),
		"unit {unit}: expected {expected} in SQL; got:\n{rendered}"
		);
		}
		}

		#[test]
		fn group_by_truncate_minute_without_selectivity_rejected() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note"}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "minute"}}
		],
		"limit": 10
		}"#;
		let err = compile(json, &test_ontology(), &test_ctx()).unwrap_err();
		let msg = format!("{err:?}");
		assert!(
		msg.contains("requires either node_ids") && msg.contains("minute"),
		"expected cardinality-guard rejection; got: {msg}"
		);
		}

		#[test]
		fn group_by_truncate_minute_with_node_ids_accepted() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "node_ids": [1, 2]}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "minute"}}
		],
		"limit": 10
		}"#;
		let result = compile(json, &test_ontology(), &test_ctx()).unwrap();
		assert!(
		result
		.base
		.render()
		.contains("toStartOfMinute(u.created_at)")
		);
		}

		#[test]
		fn group_by_truncate_hour_with_property_filter_accepted() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "filters": {"created_at": {"op": "gte", "value": "2026-04-01T00:00:00Z"}}}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "hour"}}
		],
		"limit": 50
		}"#;
		let result = compile(json, &test_ontology(), &test_ctx()).unwrap();
		assert!(result.base.render().contains("toStartOfHour(u.created_at)"));
		}

		#[test]
		fn group_by_truncate_on_non_date_property_rejected() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "node_ids": [1]}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "confidential", "transform": {"kind": "truncate", "unit": "month"}}
		],
		"limit": 10
		}"#;
		let err = compile(json, &test_ontology(), &test_ctx()).unwrap_err();
		let msg = format!("{err:?}");
		assert!(
		msg.contains("requires a Date or DateTime property"),
		"expected data-type rejection; got: {msg}"
		);
		}

		#[test]
		fn group_by_truncate_custom_alias_preserved() {
		let json = r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "node_ids": [1]}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "month"}, "alias": "bucket"}
		],
		"limit": 10
		}"#;
		let result = compile(json, &test_ontology(), &test_ctx()).unwrap();
		let rendered = result.base.render();
		assert!(
		rendered.contains("toStartOfMonth(u.created_at) AS bucket"),
		"expected alias `bucket`; got:\n{rendered}"
		);
		}

		#[test]
		fn path_finding_query() {
		let json = r#"{

crates/integration-tests/tests/compiler/dialects/duckdb.rs

+55 −0

Original line number	Diff line number	Diff line
		@@ -140,6 +140,61 @@ fn neighbors() {
		assert_eq!(sql.limit_value(), Some(10));
		}

		#[test]
		fn group_by_truncate_emits_duckdb_date_trunc() {
		let result = compile_local(
		r#"{
		"query_type": "aggregation",
		"nodes": [
		{"id": "u", "entity": "Note", "node_ids": [1]}
		],
		"aggregations": [{"function": "count", "target": "u", "alias": "n"}],
		"group_by": [
		{"kind": "property", "node": "u", "property": "created_at", "transform": {"kind": "truncate", "unit": "month"}, "alias": "bucket"}
		],
		"limit": 10
		}"#,
		&test_ontology(),
		)
		.unwrap();
		let rendered = result.base.render();
		assert!(
		rendered.contains("date_trunc('month', u.created_at)"),
		"expected DuckDB date_trunc('month', ...); got:\n{rendered}"
		);
		assert!(
		!rendered.contains("toStartOfMonth"),
		"ClickHouse-only toStartOfMonth must not leak into DuckDB SQL:\n{rendered}"
		);
		}

		#[test]
		fn group_by_truncate_all_units_emit_duckdb_date_trunc() {
		for unit in ["minute", "hour", "day", "week", "month", "quarter", "year"] {
		let json = format!(
		r#"{{
		"query_type": "aggregation",
		"nodes": [
		{{"id": "u", "entity": "Note", "node_ids": [1]}}
		],
		"aggregations": [{{"function": "count", "target": "u", "alias": "n"}}],
		"group_by": [
		{{"kind": "property", "node": "u", "property": "created_at", "transform": {{"kind": "truncate", "unit": "{unit}"}}}}
		],
		"limit": 10
		}}"#
		);
		let result = compile_local(&json, &test_ontology())
		.unwrap_or_else(\|e\| panic!("compile_local failed for unit {unit}: {e:?}"));
		let rendered = result.base.render();
		let expected = format!("date_trunc('{unit}', u.created_at)");
		assert!(
		rendered.contains(&expected),
		"unit {unit}: expected `{expected}` in DuckDB SQL; got:\n{rendered}"
		);
		}
		}

		#[test]
		fn node_ids_expand_params() {
		let sql = parse_duckdb(