🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.
TypeScript analyticsautogenevaluationlangchainlarge-language-modelsllama-indexllmllm-evaluationllm-observabilityllmopsmonitoringobservabilityopen-sourceopenaiplaygroundprompt-engineeringprompt-managementself-hostedycombinator
fix(scores): update SQL query to filter scores by experiment events (#16011)
marliessophie7e6918ec6730
diff --git a/packages/shared/src/server/repositories/scores.ts b/packages/shared/src/server/repositories/scores.tsindex 31345e0cf6..8e0d6c7031 100644--- a/packages/shared/src/server/repositories/scores.ts+++ b/packages/shared/src/server/repositories/scores.ts@@ -963,8 +963,8 @@ export const getScoresGroupedByNameSourceType = async ({s.data_type as data_typeFROM scores s${performDatasetRunItemsJoin ? `JOIN dataset_run_items_rmt dri ON s.trace_id = dri.trace_id AND s.project_id = dri.project_id` : ""}- ${hasEventsFilters ? `ANY JOIN experiment_events e ON s.trace_id = e.trace_id AND s.project_id = e.project_id` : ""}WHERE s.project_id = {projectId: String}+ ${hasEventsFilters ? `AND (s.project_id, s.trace_id) IN (SELECT project_id, trace_id FROM experiment_events)` : ""}${scoresFilterRes?.query ? `AND ${scoresFilterRes.query}` : ""}${fromTimestamp ? `AND s.timestamp >= {fromTimestamp: DateTime64(3)}` : ""}${toTimestamp ? `AND s.timestamp <= {toTimestamp: DateTime64(3)}` : ""}diff --git a/web/src/__tests__/server/repositories/score-repository.servertest.ts b/web/src/__tests__/server/repositories/score-repository.servertest.tsindex 687ccb943c..f2764ff097 100644--- a/web/src/__tests__/server/repositories/score-repository.servertest.ts+++ b/web/src/__tests__/server/repositories/score-repository.servertest.ts@@ -22,6 +22,8 @@ import {createDatasetRunItem,createDatasetRunItemsCh,createDatasetRunScore,+ createEvent,+ createEventsCh,createSessionScore,createOrgProjectAndApiKey,getScoreStringValues,@@ -407,6 +409,64 @@ describe("Clickhouse Scores Repository Test", () => {dataType: "CATEGORICAL",});});++ it("should return every score name on an experiment trace when filtering by experiment ids", async () => {+ const { projectId: isolatedProjectId } =+ await createOrgProjectAndApiKey();+ const experimentId = `exp-${v4()}`;+ const traceIds = [v4(), v4()];++ await createEventsCh(+ traceIds.map((traceId) =>+ createEvent({+ project_id: isolatedProjectId,+ trace_id: traceId,+ experiment_id: experimentId,+ }),+ ),+ );++ // Multiple scores per experiment trace: the shape a composite evaluator+ // produces (github.com/langfuse/langfuse/issues/14454).+ const scoreNames = ["metric-one", "metric-two", "metric-three"];+ await createScoresCh([+ ...traceIds.flatMap((traceId) =>+ scoreNames.map((name) =>+ createTraceScore({+ project_id: isolatedProjectId,+ trace_id: traceId,+ observation_id: v4(),+ name,+ source: "EVAL",+ data_type: "NUMERIC",+ }),+ ),+ ),+ createTraceScore({+ project_id: isolatedProjectId,+ trace_id: v4(),+ name: "unrelated-score",+ source: "EVAL",+ data_type: "NUMERIC",+ }),+ ]);++ const result = await getScoresGroupedByNameSourceType({+ projectId: isolatedProjectId,+ filter: [+ {+ column: "experimentIds",+ operator: "any of",+ value: [experimentId],+ type: "stringOptions",+ },+ ],+ });++ expect(result.map((score) => score.name).sort()).toEqual(+ [...scoreNames].sort(),+ );+ });});describe("filter option score grouping limits", () => {