fix(query-engine): keep list inputs from nesting SQL past max_parser_depth

packages/backend/src/services/warehouse/ci-runs.clickhouse.e2e.test.ts

@@ -361,6 +361,20 @@ describe.skipIf(!clickhouseE2eEnabled)("ci_runs / ci_jobs materialization", () =
 		assert.deepStrictEqual(traceIds(await read(CH.ciRunsQuery({ pullRequest: 5 }))), ["r2", "r1"])
 	})
 
+	it("counts actors against an org with hundreds of repositories", async () => {
+		// One clause per default branch: a nested OR chain here passed ClickHouse's max_parser_depth.
+		const defaultBranches = Array.from({ length: 500 }, (_, i) => ({
+			owner: "acme",
+			repository: `repo-${i}`,
+			branch: "main",
+		}))
+		const rows = await read(CH.ciActorStatsQuery({ defaultBranches }))
+		assert.deepStrictEqual(
+			rows.map(({ actor, runs, defaultBranchRuns }) => ({ actor, runs, defaultBranchRuns })),
+			[{ actor: "octocat", runs: 3, defaultBranchRuns: 0 }],
+		)
+	})
+
 	it("finds a workflow's red streak and the failed jobs of its runs", async () => {
 		const ran = await read(CH.ciRunsQuery({ pipeline: "CI", ranOnly: true }))
 		assert.deepStrictEqual(traceIds(ran), ["r5", "r2", "r1"])

packages/query-engine/package.json

@@ -35,6 +35,7 @@
 		"effect": "catalog:effect"
 	},
 	"devDependencies": {
+		"@effect/vitest": "catalog:effect",
 		"@types/node": "catalog:tooling",
 		"typescript": "catalog:tooling",
 		"vitest": "catalog:"

packages/query-engine/src/__sql_baseline__/catalog.sql

@@ -407,7 +407,7 @@ SELECT
         LIMIT 200
         FORMAT JSON
 
--- builder:ci-runs:ciActorStatsQuery:allWorkflows  [d856dc60]
+-- builder:ci-runs:ciActorStatsQuery:allWorkflows  [890fdba4]
 SELECT
           ci_runs.Actor AS actor,
           count() AS runs,
@@ -416,8 +416,8 @@ SELECT
           countIf(ci_runs.Attempt > 1) AS reruns,
           countIf(ci_runs.Result = 'cancellation') AS cancelledRuns,
           sum(jobs.runnerMs) AS runnerMs,
-          countIf(((ci_runs.TriggerEvent = 'push' AND ((ci_runs.Owner = 'acme' AND ci_runs.Repository = 'web') AND ci_runs.Branch = 'main')) AND (ci_runs.Result = 'success' OR ci_runs.Result IN ('failure', 'timeout', 'error')))) AS defaultBranchRuns,
-          countIf(((ci_runs.TriggerEvent = 'push' AND ((ci_runs.Owner = 'acme' AND ci_runs.Repository = 'web') AND ci_runs.Branch = 'main')) AND ci_runs.Result IN ('failure', 'timeout', 'error'))) AS defaultBranchFailed
+          countIf(((ci_runs.TriggerEvent = 'push' AND (ci_runs.Owner = 'acme' AND ci_runs.Repository = 'web' AND ci_runs.Branch = 'main')) AND (ci_runs.Result = 'success' OR ci_runs.Result IN ('failure', 'timeout', 'error')))) AS defaultBranchRuns,
+          countIf(((ci_runs.TriggerEvent = 'push' AND (ci_runs.Owner = 'acme' AND ci_runs.Repository = 'web' AND ci_runs.Branch = 'main')) AND ci_runs.Result IN ('failure', 'timeout', 'error'))) AS defaultBranchFailed
         FROM ci_runs
         INNER JOIN (SELECT
           ci_jobs.TraceId AS TraceId,
@@ -3229,7 +3229,7 @@ SELECT
         LIMIT 50
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelBreakdownQuery:attribute-session-step  [b3496b8e]
+-- builder:product-events:productEventsFunnelBreakdownQuery:attribute-session-step  [c28317af]
 SELECT
           groups.group AS group,
           arrayJoin([1, 2, 3, 4]) AS step,
@@ -3272,7 +3272,7 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND e.UserId != ''
 ) AS funnel_events
         GROUP BY key) AS levels
@@ -3282,7 +3282,7 @@ SELECT
         ORDER BY group ASC, step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelBreakdownQuery:session-dimension  [bb00cfab]
+-- builder:product-events:productEventsFunnelBreakdownQuery:session-dimension  [f210b7b0]
 SELECT
           groups.group AS group,
           arrayJoin([1, 2, 3]) AS step,
@@ -3326,7 +3326,7 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND multiIf(e.UserId != '', e.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), e.VisitorId) != '') AS funnel_events
         GROUP BY key) AS levels
         GROUP BY group
@@ -3335,7 +3335,7 @@ SELECT
         ORDER BY group ASC, step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelLeaversQuery:visitor-filtered  [ad6d68b1]
+-- builder:product-events:productEventsFunnelLeaversQuery:visitor-filtered  [95bb693c]
 SELECT
           hops.step AS step,
           tupleElement(hop, 1) AS next,
@@ -3424,7 +3424,7 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND e.VisitorId != ''
           AND e.VisitorId IN (SELECT
           session_replays.VisitorId AS key
@@ -3472,7 +3472,7 @@ SELECT
         ORDER BY step ASC, count DESC, next ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelQuery:person  [0ee5feae]
+-- builder:product-events:productEventsFunnelQuery:person  [877c57db]
 SELECT
           arrayJoin([1, 2, 3]) AS step,
           arrayElement(totals.counts, step) AS count
@@ -3503,13 +3503,13 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND multiIf(e.UserId != '', e.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), e.VisitorId) != '') AS funnel_events
         GROUP BY key) AS levels) AS totals
         ORDER BY step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelQuery:session-key  [76d377e9]
+-- builder:product-events:productEventsFunnelQuery:session-key  [118d45ba]
 SELECT
           arrayJoin([1, 2, 3]) AS step,
           arrayElement(totals.counts, step) AS count
@@ -3529,13 +3529,13 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND e.SessionId != '') AS funnel_events
         GROUP BY key) AS levels) AS totals
         ORDER BY step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelQuery:session-step-filtered  [dbec8c07]
+-- builder:product-events:productEventsFunnelQuery:session-step-filtered  [7bae7ef8]
 SELECT
           arrayJoin([1, 2, 3, 4]) AS step,
           arrayElement(totals.counts, step) AS count
@@ -3641,7 +3641,7 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND multiIf(e.UserId != '', e.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), e.VisitorId) != ''
           AND multiIf(e.UserId != '', e.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), e.VisitorId) IN (SELECT
           multiIf(s.UserId != '', s.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), s.VisitorId) AS key
@@ -3695,7 +3695,7 @@ SELECT
         ORDER BY step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelQuery:visitor-session-step  [c63a28ff]
+-- builder:product-events:productEventsFunnelQuery:visitor-session-step  [59b281c2]
 SELECT
           arrayJoin([1, 2, 3, 4]) AS step,
           arrayElement(totals.counts, step) AS count
@@ -3732,14 +3732,14 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND e.VisitorId != ''
 ) AS funnel_events
         GROUP BY key) AS levels) AS totals
         ORDER BY step ASC
         FORMAT JSON
 
--- builder:product-events:productEventsFunnelTimingQuery:person  [78104129]
+-- builder:product-events:productEventsFunnelTimingQuery:person  [82b59764]
 SELECT
           arrayJoin([2, 3]) AS step,
           arrayElement(totals.p50s, step) AS p50Ms,
@@ -3779,7 +3779,7 @@ SELECT
         WHERE e.OrgId = 'org_sql_catalog'
           AND e.Timestamp >= '2026-01-01 10:30:00'
           AND e.Timestamp <= '2026-01-03 14:15:00'
-          AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
+          AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))
           AND multiIf(e.UserId != '', e.UserId, coalesce(link.UserId, '') != '', coalesce(link.UserId, ''), e.VisitorId) != '') AS funnel_events
         GROUP BY key) AS chain_events) AS chain) AS totals
         ORDER BY step ASC
@@ -11795,7 +11795,7 @@ SELECT
         LIMIT 50
         FORMAT JSON
 
--- pipe:list_logs:searched:bloom  [74bedc54]
+-- pipe:list_logs:searched:bloom  [226ed830]
 SELECT
           logs.Timestamp AS timestamp,
           toString(logs.Timestamp) AS exactTimestamp,
@@ -11817,7 +11817,7 @@ SELECT
           AND logs.ServiceName = 'api'
           AND logs.SeverityText IN ('ERROR', 'Error', 'error')
           AND logs.TraceId = '0af7651916cd43dd8448eb211c80319c'
-          AND (((hasToken(lower(logs.Body), 'connection') AND hasToken(lower(logs.Body), 'refused')) AND hasToken(lower(logs.Body), 'by')) AND logs.Body ILIKE '%upstream connection refused by peer%')
+          AND (hasToken(lower(logs.Body), 'connection') AND hasToken(lower(logs.Body), 'refused') AND hasToken(lower(logs.Body), 'by') AND logs.Body ILIKE '%upstream connection refused by peer%')
           AND logs.Timestamp >= (SELECT min(ts) FROM (SELECT
           logs.Timestamp AS ts
         FROM logs
@@ -11829,7 +11829,7 @@ SELECT
           AND logs.ServiceName = 'api'
           AND logs.SeverityText IN ('ERROR', 'Error', 'error')
           AND logs.TraceId = '0af7651916cd43dd8448eb211c80319c'
-          AND (((hasToken(lower(logs.Body), 'connection') AND hasToken(lower(logs.Body), 'refused')) AND hasToken(lower(logs.Body), 'by')) AND logs.Body ILIKE '%upstream connection refused by peer%')
+          AND (hasToken(lower(logs.Body), 'connection') AND hasToken(lower(logs.Body), 'refused') AND hasToken(lower(logs.Body), 'by') AND logs.Body ILIKE '%upstream connection refused by peer%')
         ORDER BY ts DESC
         LIMIT 50))
         ORDER BY timestamp DESC, serviceName ASC, traceId ASC, spanId ASC, recordIdentity ASC

packages/query-engine/src/ch/queries/ci-runs.ts

@@ -297,16 +297,17 @@ export function ciActorStatsQuery(opts: CiActorStatsOpts) {
 				const firstTry = $.Attempt.lte(1)
 				const passed = $.Result.eq("success")
 				const failed = CH.inList($.Result, CI_FAILED_RESULTS)
-				const isDefaultBranch = opts.defaultBranches
-					.map((repo) =>
-						$.Owner.eq(repo.owner)
-							.and($.Repository.eq(repo.repository))
-							.and($.Branch.eq(repo.branch)),
-					)
-					.reduce<CH.Condition | undefined>(
-						(all, next) => (all === undefined ? next : all.or(next)),
-						undefined,
-					)
+				// One flat OR: chaining `.or()` nests a paren per repository and an org with a few
+				// hundred repositories passes ClickHouse's max_parser_depth (1000).
+				const isDefaultBranch = CH.or(
+					...opts.defaultBranches.map((repo) =>
+						CH.and(
+							$.Owner.eq(repo.owner),
+							$.Repository.eq(repo.repository),
+							$.Branch.eq(repo.branch),
+						),
+					),
+				)
 				// No repository: an empty `inList` compiles to `1 = 0`, so nothing counts as the default branch.
 				const onDefaultBranch = $.TriggerEvent.eq("push").and(
 					isDefaultBranch ?? CH.inList($.Branch, []),

packages/query-engine/src/ch/queries/list-input-nesting.property.test.ts

@@ -0,0 +1,90 @@
+// ClickHouse refuses a query nested deeper than max_parser_depth (1000). A builder that chains
+// `.or()` / `.and()` once per list item nests one level per item and reaches it at a few hundred,
+// so a list input must never make the SQL deeper. Fixtures use one or two items and cannot see it.
+import { describe, expect, it } from "@effect/vitest"
+import { Schema } from "effect"
+import { compileUnsafe } from "@maple-dev/effect-orm/clickhouse"
+import { ciActorStatsQuery } from "./ci-runs"
+import { logsCountQuery } from "./logs"
+import { tracesListQuery } from "./traces"
+
+const params = { orgId: "org_nesting", startTime: "2026-10-01 00:00:00", endTime: "2026-10-02 00:00:00" }
+
+/** Deepest parenthesis nesting outside string literals. */
+const nestingDepth = (sql: string): number => {
+	let depth = 0
+	let deepest = 0
+	let quoted = false
+	for (let i = 0; i < sql.length; i++) {
+		const char = sql[i]
+		if (char === "\\" && quoted) i++
+		else if (char === "'") quoted = !quoted
+		else if (!quoted && char === "(") deepest = Math.max(deepest, ++depth)
+		else if (!quoted && char === ")") depth--
+	}
+	return deepest
+}
+
+const listSize = Schema.Number.check(Schema.isInt(), Schema.isBetween({ minimum: 1, maximum: 2_000 }))
+
+const items = <A>(n: number, make: (i: number) => A): ReadonlyArray<A> =>
+	Array.from({ length: n }, (_, i) => make(i))
+const words = (n: number) => items(n + 2, (i) => `word${i}`).join(" ")
+const orGroup = (n: number) => ({
+	key: "http.route",
+	value: "/0",
+	mode: "equals" as const,
+	or: items(n, (i) => ({ key: "http.route", value: `/${i + 1}`, mode: "equals" as const })),
+})
+const filters = (n: number) => items(n, (i) => ({ key: `attr.${i}`, value: "v", mode: "equals" as const }))
+
+const cases: ReadonlyArray<{ readonly name: string; readonly sql: (n: number) => string }> = [
+	{
+		name: "CI actor stats over many repositories' default branches",
+		sql: (n) =>
+			compileUnsafe(
+				ciActorStatsQuery({
+					defaultBranches: items(n, (i) => ({
+						owner: "acme",
+						repository: `repo-${i}`,
+						branch: "main",
+					})),
+				}),
+				params,
+			).sql,
+	},
+	{
+		name: "log search with many words (token index)",
+		sql: (n) =>
+			compileUnsafe(logsCountQuery({ search: words(n), bodySearchMode: "tokenbf" }), params).sql,
+	},
+	{
+		name: "log search with many words (text index)",
+		sql: (n) => compileUnsafe(logsCountQuery({ search: words(n), bodySearchMode: "text" }), params).sql,
+	},
+	{
+		name: "log attribute OR group",
+		sql: (n) => compileUnsafe(logsCountQuery({ attributeFilters: [orGroup(n)] }), params).sql,
+	},
+	{
+		name: "many log attribute filters",
+		sql: (n) => compileUnsafe(logsCountQuery({ attributeFilters: filters(n) }), params).sql,
+	},
+	{
+		name: "span attribute OR group",
+		sql: (n) => compileUnsafe(tracesListQuery({ attributeFilters: [orGroup(n)] }), params).sql,
+	},
+	{
+		name: "many span attribute filters",
+		sql: (n) => compileUnsafe(tracesListQuery({ attributeFilters: filters(n) }), params).sql,
+	},
+]
+
+describe("list inputs never deepen the SQL", () => {
+	for (const { name, sql } of cases) {
+		const twoItems = nestingDepth(sql(2))
+		it.prop(name, [listSize], ([n]) => {
+			expect(nestingDepth(sql(n))).toBeLessThanOrEqual(twoItems)
+		})
+	}
+})

packages/query-engine/src/ch/queries/logs.ts

@@ -93,13 +93,11 @@ function logBodySearchCondition(body: CH.Expr<string>, opts: LogsQueryOpts): CH.
 		for (let offset = 0; offset < tokens.length; offset += 50) {
 			batches.push(CH.hasAllTokens(normalizedBody, tokens.slice(offset, offset + 50).join(" ")))
 		}
-		return batches.reduce((condition, next) => condition.and(next)).and(exact)
+		return CH.and(...batches, exact)
 	}
 
-	return tokens
-		.map((token) => CH.hasToken(normalizedBody, token))
-		.reduce((condition, next) => condition.and(next))
-		.and(exact)
+	// One flat AND: a chained `.and()` per word passes ClickHouse's max_parser_depth on a pasted stack trace.
+	return CH.and(...tokens.map((token) => CH.hasToken(normalizedBody, token)), exact)
 }
 
 /**

packages/query-engine/src/ch/queries/product-events.test.ts

@@ -89,7 +89,7 @@ describe("productEventsFunnelQuery", () => {
 			"toUInt8((e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup')) AS s3",
 		)
 		expect(oneLine(sql)).toContain(
-			"AND ((((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed') OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))",
+			"AND (((e.Kind = 'navigation' AND e.PagePath = '/pricing') AND e.Host = 'maple.dev') OR e.EventName = 'signup_completed' OR (e.EventName = 'plan_started' AND e.Attributes['plan'] = 'startup'))",
 		)
 	})
 

packages/query-engine/src/ch/queries/product-events.ts

@@ -453,10 +453,7 @@ function eventsBranch(plan: FunnelPlan): FunnelBranch {
 			const stepConditions = opts.steps
 				.map((step) => eventStepCondition($, step))
 				.filter((cond): cond is CH.Condition => cond !== undefined)
-			const anyStep = stepConditions.reduce<CH.Condition | undefined>(
-				(acc, cond) => (acc ? acc.or(cond) : cond),
-				undefined,
-			)
+			const anyStep = stepConditions.length === 0 ? undefined : CH.or(...stepConditions)
 			return [
 				$.OrgId.eq(orgIdParam),
 				$.Timestamp.gte(param.dateTime("startTime")),