hanzoai/o11y
2
0
Fork 0
mirror of https://github.com/hanzoai/o11y.git synced 2026-10-15 12:58:33 +00:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
Files
iam-datastore
o11y/pkg/telemetrylogs/json_stmt_builder_test.go
T
zeekayandHanzo Dev ded468300a readers: the query builders speak the envelope, not the fork's dead dialect 
The traces and logs statement/condition builders emitted the OTLP-fork columns
(timestamp, duration_nano, attributes_string/number/bool, resources_string,
resource_string_service$name, ts_bucket_start, parent_span_id, trace_flags,
scope_name) against event.span / event.log, which carry the 15-column envelope.
Every span lookup and log search failed Code 47 UNKNOWN_IDENTIFIER — executed and
proven before this change, not inferred.

TWO COLUMNS ADDED to event.span and event.log because no rewrite can satisfy them:
  resource_fingerprint String — the builders stamp
    'resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter)'
    on the MAIN table, and __resource_filter reads event.span_resource /
    event.log_resource keyed (org, seen_at_ts_bucket_start, fingerprint). The
    fingerprint is a write-time hash of the whole resource label set — derivable
    from NO envelope column. event.log's pre-existing UInt64 'resource' is
    type-incompatible and empty.
  ts_bucket_start UInt64 MATERIALIZED intDiv(toUnixTimestamp(time),1800)*1800 —
    the builders put bucket-range predicates directly on the main table.
    MATERIALIZED-from-time keeps writers 15-column-envelope-only and shares ONE
    bucketing scheme with the four support tables that already speak
    seen_at_ts_bucket_start.
EVERYTHING ELSE REWRITTEN onto the envelope — one naming scheme wins:
timestamp->time, parent_span_id->parent, duration_nano->duration,
kind_string->kind (the live kind IS the string enum, so numeric kind/status_code
predicates translate by enum NAME), status_code_string->status,
has_error->status='error', attributes_string[k]->attributes[k],
attributes_number[k]->toFloat64OrNull(attributes[k]),
attributes_bool[k]->attributes[k]='true', resources_string['service.name'] and
resource_string_service$name->service, http_host->host, the calculated columns
to attribute expressions, and the 28 deprecated camelCase aliases through
oldToNew to the SAME envelope targets — never to table columns.

spantypes TraceDB/TraceSummaryTable move o11y_traces/distributed_trace_summary ->
event/trace; impltracedetail builds its FROM from those constants. Swapping them
ALONE panics TestGetTraceSummary on a nil pointer — the store and its test move
together, which is why they are in this commit and not a smaller one.

Pins moved deliberately, each with the reason in place. live_proof_test.go in
three packages executes the builders' emitted SQL against the real store, so the
next dialect drift fails loudly instead of at runtime.

go build ./... and go test ./... green across every package.

Co-authored-by: Hanzo Dev <[email protected]>
2026-08-01 13:11:33 -07:00

1215 lines
75 KiB
Go
Raw Permalink Blame History

package telemetrylogs
import (
"context"
"slices"
"strings"
"testing"
"time"
"github.com/hanzoai/o11y/pkg/flagger/flaggertest"
"github.com/hanzoai/o11y/pkg/instrumentation/instrumentationtest"
"github.com/hanzoai/o11y/pkg/querybuilder"
qbtypes "github.com/hanzoai/o11y/pkg/types/querybuildertypes/querybuildertypesv5"
"github.com/hanzoai/o11y/pkg/types/telemetrytypes"
"github.com/hanzoai/o11y/pkg/types/telemetrytypes/telemetrytypestest"
schemamigrator "github.com/hanzoai/otel-collector/cmd/o11yschemamigrator/schema_migrator"
"github.com/stretchr/testify/require"
)
const (
testJSONQueryPrefix = "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, trace_id, span_id, toUInt32OrZero(attributes['trace_flags']) AS trace_flags, severity_text, severity_number, attributes['scope.name'] AS scope_name, attributes['scope.version'] AS scope_version, body_v2 as body, attributes AS attributes_string, CAST(map() AS Map(String, Float64)) AS attributes_number, CAST(map() AS Map(String, Bool)) AS attributes_bool, map('service.name', toString(service), 'host', toString(host)) AS resources_string, CAST(map() AS Map(String, String)) AS scope_string FROM event.log WHERE"
testJSONQuerySuffix = "AND time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? LIMIT ?"
)
type TestExpected struct {
WhereClause string
Args []any
Warnings []string
}
func (t TestExpected) GetQuery() string {
return strings.Join([]string{testJSONQueryPrefix, t.WhereClause, testJSONQuerySuffix}, " ")
}
func TestJSONStmtBuilder_TimeSeries(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, false)
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected qbtypes.Statement
expectedErrContains string
}{
{
name: "Group By Simple Path",
requestType: qbtypes.RequestTypeTimeSeries,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Aggregations: []qbtypes.LogAggregation{
{
Expression: "count()",
},
},
Limit: 10,
GroupBy: []qbtypes.GroupByKey{
{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "user.age",
},
},
},
},
expected: qbtypes.Statement{
Query: "WITH __limit_cte AS (SELECT toString(multiIf((dynamicElement(body_v2.`user.age`, 'Int64') IS NOT NULL), toString(dynamicElement(body_v2.`user.age`, 'Int64')), (dynamicElement(body_v2.`user.age`, 'String') IS NOT NULL), dynamicElement(body_v2.`user.age`, 'String'), NULL)) AS `user.age`, count() AS __result_0 FROM event.log WHERE time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? GROUP BY `user.age` ORDER BY __result_0 DESC LIMIT ?) SELECT toStartOfInterval(time, INTERVAL 30 SECOND) AS ts, toString(multiIf((dynamicElement(body_v2.`user.age`, 'Int64') IS NOT NULL), toString(dynamicElement(body_v2.`user.age`, 'Int64')), (dynamicElement(body_v2.`user.age`, 'String') IS NOT NULL), dynamicElement(body_v2.`user.age`, 'String'), NULL)) AS `user.age`, count() AS __result_0 FROM event.log WHERE time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? AND (`user.age`) GLOBAL IN (SELECT `user.age` FROM __limit_cte) GROUP BY ts, `user.age`",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448)},
},
},
{
name: "Group By One Array Join",
requestType: qbtypes.RequestTypeTimeSeries,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Aggregations: []qbtypes.LogAggregation{
{
Expression: "count()",
},
},
Limit: 10,
GroupBy: []qbtypes.GroupByKey{
{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "education[].awards[].type",
},
},
},
},
expectedErrContains: "Group by/Aggregation isn't available for the Array Paths",
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErrContains != "" {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErrContains)
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
/* Promoted path tests commented out — Materialized now means type hint (direct sub-column),
not a body_promoted.* column. These tests assumed the old coalesce(body_promoted.x, body_v2.x) path.
func TestStmtBuilderTimeSeriesBodyGroupByPromoted(t *testing.T) {
statementBuilder := buildJSONTestStatementBuilder(t, "user.age", "user.name")
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected qbtypes.Statement
expectedErrContains string
}{
{
name: "Two simple paths",
requestType: qbtypes.RequestTypeTimeSeries,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Aggregations: []qbtypes.LogAggregation{
{
Expression: "count()",
},
},
Limit: 10,
GroupBy: []qbtypes.GroupByKey{
{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "user.age",
},
},
{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "user.name",
},
},
},
},
expected: qbtypes.Statement{
Args: []any{ "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448)},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErrContains != "" {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErrContains)
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
*/
func TestJSONStmtBuilder_PrimitivePaths(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, false)
cases := []struct {
name string
filter string
expected TestExpected
expectedErr error
}{
// ── intrinsic sub-column (body_v2.message) ──────────────────────────────
{
name: "message Exists",
filter: "message Exists",
expected: TestExpected{
WhereClause: "body_v2.message <> ?",
Args: []any{"", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "message Contains 'Iron Award'",
filter: "message Contains 'Iron Award'",
expected: TestExpected{
WhereClause: "body_v2.message ILIKE ?",
Args: []any{"%Iron Award%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── flat nested primitive (x.y) ─────────────────────────────────────────
{
name: "Simple string filter",
filter: "body.user.name = 'x'",
expected: TestExpected{
WhereClause: "((dynamicElement(body_v2.`user.name`, 'String') = ?) AND has(JSONAllPaths(body_v2), 'user.name'))",
Args: []any{"x", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Simple Int field Contains value",
filter: "body.user.age Contains 25",
expected: TestExpected{
WhereClause: "(((toString(dynamicElement(body_v2.`user.age`, 'Int64')) ILIKE ?) AND has(JSONAllPaths(body_v2), 'user.age')) OR ((dynamicElement(body_v2.`user.age`, 'String') ILIKE ?) AND has(JSONAllPaths(body_v2), 'user.age')))",
Args: []any{"%25%", "%25%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `user.age` is ambiguous, found 2 different combinations of field context / data type: [name=user.age,context=body,datatype=int64 name=user.age,context=body,datatype=string]."},
},
},
{
name: "Simple Float field Contains value",
filter: "body.user.height Contains 5.8",
expected: TestExpected{
WhereClause: "((toString(dynamicElement(body_v2.`user.height`, 'Float64')) ILIKE ?) AND has(JSONAllPaths(body_v2), 'user.height'))",
Args: []any{"%5.8%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Float GT",
filter: "body.user.height > 5.8",
expected: TestExpected{
WhereClause: "((toFloat64(dynamicElement(body_v2.`user.height`, 'Float64')) > ?) AND has(JSONAllPaths(body_v2), 'user.height'))",
Args: []any{5.8, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "String Like",
filter: "body.user.name LIKE '%ali%'",
expected: TestExpected{
WhereClause: "((dynamicElement(body_v2.`user.name`, 'String') LIKE ?) AND has(JSONAllPaths(body_v2), 'user.name'))",
Args: []any{"%ali%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "String Regexp",
filter: "body.user.name REGEXP 'al.*'",
expected: TestExpected{
WhereClause: "((match(dynamicElement(body_v2.`user.name`, 'String'), ?)) AND has(JSONAllPaths(body_v2), 'user.name'))",
Args: []any{"al.*", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive Bool NotEqual",
filter: "body.user.active != true",
expected: TestExpected{
WhereClause: "(assumeNotNull(dynamicElement(body_v2.`user.active`, 'Bool')) <> ?)",
Args: []any{true, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive Bool NotExists",
filter: "body.user.active NOT EXISTS",
expected: TestExpected{
WhereClause: "(dynamicElement(body_v2.`user.active`, 'Bool') IS NULL)",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive String NotContains",
filter: "body.user.name NOT CONTAINS 'IIT'",
expected: TestExpected{
WhereClause: "(assumeNotNull(dynamicElement(body_v2.`user.name`, 'String')) NOT ILIKE ?)",
Args: []any{"%IIT%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive String NotLike",
filter: "body.user.name NOT LIKE '%ali%'",
expected: TestExpected{
WhereClause: "(assumeNotNull(dynamicElement(body_v2.`user.name`, 'String')) NOT LIKE ?)",
Args: []any{"%ali%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive String NotRegexp",
filter: "body.user.name NOT REGEXP 'al.*'",
expected: TestExpected{
WhereClause: "(NOT match(assumeNotNull(dynamicElement(body_v2.`user.name`, 'String')), ?))",
Args: []any{"al.*", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Primitive String NotIn",
filter: "body.user.name NOT IN ['alice', 'bob']",
expected: TestExpected{
WhereClause: "(assumeNotNull(dynamicElement(body_v2.`user.name`, 'String')) NOT IN (?, ?))",
Args: []any{"alice", "bob", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── deeply nested primitive (x.y.z) ─────────────────────────────────────
{
name: "Primitive Int64 NotBetween",
filter: "body.user.address.zip NOT BETWEEN 100000 AND 999999",
expected: TestExpected{
WhereClause: "(toFloat64(assumeNotNull(dynamicElement(body_v2.`user.address.zip`, 'Int64'))) NOT BETWEEN ? AND ?)",
Args: []any{float64(100000), float64(999999), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, qbtypes.RequestTypeRaw,
qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: c.filter},
Limit: 10,
}, nil)
if c.expectedErr != nil {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErr.Error())
return
}
require.NoError(t, err)
require.Equal(t, c.expected.GetQuery(), q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
})
}
}
/* Promoted path list-query tests commented out — Materialized now means type hint
(direct sub-column access), not a body_promoted.* column.
func TestStatementBuilderListQueryBodyPromoted(t *testing.T) {
statementBuilder := buildJSONTestStatementBuilder(t, "education", "tags")
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected TestExpected
expectedErr error
}{
{
name: "Has Array promoted uses body fallback",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "has(body.tags, 'production')"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "production", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Key inside Array(JSON) exists",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].name Exists"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Key inside Array(JSON) -> Array(Dynamic) + Array(JSON) exists",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].awards[].name Exists"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Key inside Array(JSON) -> Array(Dynamic) + Array(JSON) = 'Iron Award'",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].awards[].name = 'Iron Award'"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "Iron Award", "Iron Award", "Iron Award", "Iron Award", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Key inside Array(JSON) contains value",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].parameters Contains 1.65"},
Limit: 10,
},
expected: TestExpected{
Args: []any{uint64(1747945619), uint64(1747983448), "%1.65%", 1.65, "%1.65%", 1.65, "%1.65%", 1.65, "%1.65%", 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
},
expectedErr: nil,
},
{
name: "Key inside Array(JSON) contains String value",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].name Contains 'IIT'"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "%IIT%", "%IIT%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Dynamic array contains boolean",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].parameters Contains true"},
Limit: 10,
},
expected: TestExpected{
Args: []any{uint64(1747945619), uint64(1747983448), "%true%", true, "%true%", true, "%true%", true, "%true%", true, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
},
expectedErr: nil,
},
{
name: "Dynamic array contains String",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].parameters Contains 'passed'"},
Limit: 10,
},
expected: TestExpected{
Args: []any{uint64(1747945619), uint64(1747983448), "%passed%", "passed", "%passed%", "passed", "%passed%", "passed", "%passed%", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64,materialized=true name=education[].parameters,context=body,datatype=[]dynamic,materialized=true]."},
},
expectedErr: nil,
},
{
name: "Array(Dynamic) + Array(JSON) both can contain a value",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].awards[].type = 'sports'"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "sports", "sports", "sports", "sports", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "Multi nested Array(Dynamic) + Array(JSON) both can contain a value",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.education[].awards[].participated[].team[].branch Contains 'Civil'"},
Limit: 10,
},
expected: TestExpected{
Args: []any{ "%Civil%", "%Civil%", "%Civil%", "%Civil%", "%Civil%", "%Civil%", "%Civil%", "%Civil%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
expectedErr: nil,
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErr != nil {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErr.Error())
} else {
require.NoError(t, err)
require.Equal(t, c.expected.GetQuery(), q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
*/
func TestJSONStmtBuilder_ArrayPaths(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, false)
cases := []struct {
name string
filter string
expected TestExpected
expectedErr error
}{
// ── single-hop (x[].y) ──────────────────────────────────────────────────
{
name: "Key inside Array(JSON) exists",
filter: "body.education[].name Exists",
expected: TestExpected{
WhereClause: "(arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') IS NOT NULL, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Key inside Array(JSON) contains String value",
filter: "body.education[].name Contains 'IIT'",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') ILIKE ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{"%IIT%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Int in array Contains value",
filter: "body.education[].year Contains 2020",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> toString(dynamicElement(`body_v2.education`.`year`, 'Int64')) ILIKE ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{"%2020%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Array simple path NotEqual",
filter: "body.education[].type != '10001'",
expected: TestExpected{
WhereClause: "((NOT arrayExists(`body_v2.education`-> toFloat64OrNull(dynamicElement(`body_v2.education`.`type`, 'String')) = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND (NOT arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`type`, 'Int64') = ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))))",
Args: []any{int64(10001), int64(10001), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].type` is ambiguous, found 2 different combinations of field context / data type: [name=education[].type,context=body,datatype=string name=education[].type,context=body,datatype=int64]."},
},
},
{
name: "Array simple path NotExists",
filter: "body.education[].name NOT EXISTS",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') IS NOT NULL, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Array simple path NotContains",
filter: "body.education[].name NOT CONTAINS 'IIT'",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> dynamicElement(`body_v2.education`.`name`, 'String') ILIKE ?, dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"%IIT%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── single-hop terminal array (x[].y[]) ─────────────────────────────────
{
name: "Key inside Array(JSON) contains Float value",
filter: "body.education[].parameters Contains 1.65",
expected: TestExpected{
WhereClause: "(((arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
Args: []any{"%1.65%", 1.65, "%1.65%", 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
},
},
{
name: "Dynamic array element comparison",
filter: "body.education[].parameters = 'passed'",
expected: TestExpected{
WhereClause: "(((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
Args: []any{"passed", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
},
},
{
name: "Dynamic array contains String",
filter: "body.education[].parameters Contains 'passed'",
expected: TestExpected{
WhereClause: "(((arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toString(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> toString(x) = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
Args: []any{"%passed%", "passed", "%passed%", "passed", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
},
},
{
name: "Dynamic array IN Operator",
filter: "body.education[].parameters IN [1.65, 1.99]",
expected: TestExpected{
WhereClause: "(((arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) IN (?, ?), dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')) OR ((arrayExists(`body_v2.education`-> arrayExists(x -> toString(x) IN (?, ?), arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education')))",
Args: []any{1.65, 1.99, "1.65", "1.99", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
},
},
{
name: "Terminal array NotContains float",
filter: "body.education[].parameters NOT CONTAINS 1.65",
expected: TestExpected{
WhereClause: "((NOT arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND (NOT arrayExists(`body_v2.education`-> (arrayExists(x -> toString(x) ILIKE ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'))) OR arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)')))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))))",
Args: []any{"%1.65%", float64(1.65), "%1.65%", float64(1.65), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic]."},
},
},
{
name: "Simple has filter",
filter: "has(body.education[].parameters, 1.65)",
expected: TestExpected{
WhereClause: "(has(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->dynamicElement(`body_v2.education`.`parameters`, 'Array(Nullable(Float64))'), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?) OR has(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->dynamicElement(`body_v2.education`.`parameters`, 'Array(Dynamic)'), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?))",
Args: []any{1.65, 1.65, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{
"Key `education[].parameters` is ambiguous, found 2 different combinations of field context / data type: [name=education[].parameters,context=body,datatype=[]float64 name=education[].parameters,context=body,datatype=[]dynamic].",
},
},
},
{
name: "Flat path hasAll filter",
filter: "hasAll(body.user.permissions, ['read', 'write'])",
expected: TestExpected{
WhereClause: "hasAll(dynamicElement(body_v2.`user.permissions`, 'Array(Nullable(String))'), ?)",
Args: []any{[]any{"read", "write"}, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── non-array intermediate hop (x[].y.z) ────────────────────────────────
{
name: "Non-array intermediate NotEqual",
filter: "body.http-events[].request-info.host != 'example.com'",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.http-events`-> dynamicElement(`body_v2.http-events`.`request-info.host`, 'String') = ?, dynamicElement(body_v2.`http-events`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"example.com", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── double-hop (x[].y[].z) ──────────────────────────────────────────────
{
name: "Key inside Array(JSON) -> Array(Dynamic) + Array(JSON) exists",
filter: "body.education[].awards[].name Exists",
expected: TestExpected{
WhereClause: "(arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') IS NOT NULL, dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') IS NOT NULL, arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Key inside Array(JSON) -> Array(Dynamic) + Array(JSON) = 'Iron Award'",
filter: "body.education[].awards[].name = 'Iron Award'",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') = ?, dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') = ?, arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{"Iron Award", "Iron Award", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Integer BETWEEN Operator",
filter: "education[].awards[].semester BETWEEN 2 AND 4",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> toFloat64(dynamicElement(`body_v2.education[].awards`.`semester`, 'Int64')) BETWEEN ? AND ?, dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> toFloat64(dynamicElement(`body_v2.education[].awards`.`semester`, 'Int64')) BETWEEN ? AND ?, arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{float64(2), float64(4), float64(2), float64(4), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Integer IN Operator",
filter: "education[].awards[].semester IN [2, 4]",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> toFloat64(dynamicElement(`body_v2.education[].awards`.`semester`, 'Int64')) IN (?, ?), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> toFloat64(dynamicElement(`body_v2.education[].awards`.`semester`, 'Int64')) IN (?, ?), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{float64(2), float64(4), float64(2), float64(4), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Nested array NotEqual",
filter: "body.education[].awards[].type != 'sports'",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`type`, 'String') = ?, dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`type`, 'String') = ?, arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"sports", "sports", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Nested array NotExists",
filter: "body.education[].awards[].name NOT EXISTS",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') IS NOT NULL, dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> dynamicElement(`body_v2.education[].awards`.`name`, 'String') IS NOT NULL, arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── deep multi-hop (x[].y[].z[].w...) ──────────────────────────────────
{
name: "Super nested array contains Int64 value",
filter: "body.interests[].entities[].reviews[].entries[].metadata[].positions[].ratings Contains 4",
expected: TestExpected{
WhereClause: "(((arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(`body_v2.interests[].entities[].reviews`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`-> (arrayExists(x -> toString(x) ILIKE ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(Int64))')) OR arrayExists(x -> toFloat64(x) = ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(Int64))'))), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata`.`positions`, 'Array(JSON(max_dynamic_types=0, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews[].entries`.`metadata`, 'Array(JSON(max_dynamic_types=1, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews`.`entries`, 'Array(JSON(max_dynamic_types=2, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities`.`reviews`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'interests')) OR ((arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(`body_v2.interests[].entities[].reviews`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata`-> arrayExists(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`-> (arrayExists(x -> x ILIKE ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(String))')) OR arrayExists(x -> toFloat64OrNull(x) = ?, dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata[].positions`.`ratings`, 'Array(Nullable(String))'))), dynamicElement(`body_v2.interests[].entities[].reviews[].entries[].metadata`.`positions`, 'Array(JSON(max_dynamic_types=0, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews[].entries`.`metadata`, 'Array(JSON(max_dynamic_types=1, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities[].reviews`.`entries`, 'Array(JSON(max_dynamic_types=2, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests[].entities`.`reviews`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'interests')))",
Args: []any{"%4%", float64(4), "%4%", float64(4), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
Warnings: []string{"Key `interests[].entities[].reviews[].entries[].metadata[].positions[].ratings` is ambiguous, found 2 different combinations of field context / data type: [name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]int64 name=interests[].entities[].reviews[].entries[].metadata[].positions[].ratings,context=body,datatype=[]string]."},
},
},
{
name: "Multi nested Array(Dynamic) + Array(JSON) both can contain a value",
filter: "body.education[].awards[].participated[].team[].branch Contains 'Civil'",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> (arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(`body_v2.education[].awards[].participated[].team`-> dynamicElement(`body_v2.education[].awards[].participated[].team`.`branch`, 'String') ILIKE ?, dynamicElement(`body_v2.education[].awards[].participated`.`team`, 'Array(JSON(max_dynamic_types=2, max_dynamic_paths=0))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(`body_v2.education[].awards[].participated[].team`-> dynamicElement(`body_v2.education[].awards[].participated[].team`.`branch`, 'String') ILIKE ?, dynamicElement(`body_v2.education[].awards[].participated`.`team`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')) OR arrayExists(`body_v2.education[].awards`-> (arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(`body_v2.education[].awards[].participated[].team`-> dynamicElement(`body_v2.education[].awards[].participated[].team`.`branch`, 'String') ILIKE ?, dynamicElement(`body_v2.education[].awards[].participated`.`team`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=64))')), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')) OR arrayExists(`body_v2.education[].awards[].participated`-> arrayExists(`body_v2.education[].awards[].participated[].team`-> dynamicElement(`body_v2.education[].awards[].participated[].team`.`branch`, 'String') ILIKE ?, dynamicElement(`body_v2.education[].awards[].participated`.`team`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->dynamicElement(x, 'JSON'), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{"%Civil%", "%Civil%", "%Civil%", "%Civil%", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
// IN on a single-type Int64 array terminal: any element of scores IN {90, 95}
name: "Terminal Int64 array IN operator",
filter: "body.education[].scores IN [90, 95]",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) IN (?, ?), dynamicElement(`body_v2.education`.`scores`, 'Array(Nullable(Int64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{float64(90), float64(95), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
// NOT IN on a single-type Int64 array terminal: no element of scores IN {90, 95}.
// Negative operator → NOT wraps outer arrayExists; no path-index guard added.
name: "Terminal Int64 array NOT IN operator",
filter: "body.education[].scores NOT IN [90, 95]",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) IN (?, ?), dynamicElement(`body_v2.education`.`scores`, 'Array(Nullable(Int64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{float64(90), float64(95), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
// BETWEEN on a single-type Int64 array terminal: any element of scores BETWEEN 80 AND 95.
// BETWEEN is an IsArrayOperator → value arrives as []any{80,95} → c.valueType.IsArray=true.
name: "Terminal Int64 array BETWEEN operator",
filter: "body.education[].scores BETWEEN 80 AND 95",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) BETWEEN ? AND ?, dynamicElement(`body_v2.education`.`scores`, 'Array(Nullable(Int64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'education'))",
Args: []any{float64(80), float64(95), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
// NOT BETWEEN on a single-type Int64 array terminal: no element BETWEEN 80 AND 95.
// Negative → NOT wraps outer arrayExists; no path-index guard.
name: "Terminal Int64 array NOT BETWEEN operator",
filter: "body.education[].scores NOT BETWEEN 80 AND 95",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.education`-> arrayExists(x -> toFloat64(x) BETWEEN ? AND ?, dynamicElement(`body_v2.education`.`scores`, 'Array(Nullable(Int64))')), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{float64(80), float64(95), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Nested path hasAny filter",
filter: "hasAny(education[].awards[].participated[].members, ['Piyush', 'Tushar'])",
expected: TestExpected{
WhereClause: "hasAny(arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->arrayConcat(arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?)",
Args: []any{[]any{"Piyush", "Tushar"}, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "dynamic_array_element_compare_CONTAINS",
filter: "interests[].entities[].product_codes Contains 1",
expected: TestExpected{
WhereClause: "((arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> (arrayExists(x -> toString(x) ILIKE ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'))) OR arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)')))), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))'))) AND has(JSONAllPaths(body_v2), 'interests'))",
Args: []any{"%1%", float64(1), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "dynamic_array_element_compare_HAS_STRING",
filter: "has(interests[].entities[].product_codes, '2002')",
expected: TestExpected{
WhereClause: "has(arrayFlatten(arrayConcat(arrayMap(`body_v2.interests`->arrayMap(`body_v2.interests[].entities`->dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?)",
Args: []any{"2002", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "dynamic_array_element_compare_NOT_EQUAL",
filter: "interests[].entities[].product_codes != '1'",
expected: TestExpected{
WhereClause: "(NOT arrayExists(`body_v2.interests`-> arrayExists(`body_v2.interests[].entities`-> arrayExists(x -> accurateCastOrNull(x, 'Float64') = ?, arrayFilter(x->(dynamicType(x) IN ('String', 'Int64', 'Float64', 'Bool')), dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'))), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))",
Args: []any{int64(1), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "dynamic_array_element_compare_HAS_INT",
filter: "has(interests[].entities[].product_codes, 1001)",
expected: TestExpected{
WhereClause: "has(arrayFlatten(arrayConcat(arrayMap(`body_v2.interests`->arrayMap(`body_v2.interests[].entities`->dynamicElement(`body_v2.interests[].entities`.`product_codes`, 'Array(Dynamic)'), dynamicElement(`body_v2.interests`.`entities`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), dynamicElement(body_v2.`interests`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))), ?)",
Args: []any{float64(1001), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, qbtypes.RequestTypeRaw,
qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: c.filter},
Limit: 10,
}, nil)
if c.expectedErr != nil {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErr.Error())
return
}
require.NoError(t, err)
require.Equal(t, c.expected.GetQuery(), q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
})
}
}
func TestJSONStmtBuilder_IndexedPaths(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, true)
cases := []struct {
name string
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected TestExpected
expectedErr error
}{
// ── indexed equal: emits assumeNotNull(dynamicElement) = ? ───────────
{
name: "Indexed String Equal",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.user.name = 'alice'"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "((lower(assumeNotNull(dynamicElement(body_v2.user.name, 'String'))) = ?) AND has(JSONAllPaths(body_v2), 'user.name'))",
Args: []any{"alice", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Indexed Int64 Equal",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.user.address.zip = 110001"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "((toFloat64(assumeNotNull(dynamicElement(body_v2.user.address.zip, 'Int64'))) = ?) AND has(JSONAllPaths(body_v2), 'user.address.zip'))",
Args: []any{float64(110001), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── indexed exists: index is skipped; emits plain IS NOT NULL ───────────
// EXISTS/NOT EXISTS bypass the index because indexed columns store a default
// empty value for absent fields, making != "" unreliable for existence checks
// (a field with value "" would be incorrectly excluded).
{
name: "Indexed String Exists",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.user.name Exists"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "(dynamicElement(body_v2.`user.name`, 'String') IS NOT NULL)",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── indexed not-equal: assumeNotNull wrapping + no path index ─────────
{
name: "Indexed String NotEqual",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.user.name != 'alice'"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "(lower(assumeNotNull(dynamicElement(body_v2.user.name, 'String'))) <> ?)",
Args: []any{"alice", "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── special-character indexed paths ───────────────────────────────────
{
name: "Indexed hyphen key Equal",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.http-status = 200"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "(((toFloat64(assumeNotNull(dynamicElement(body_v2.`http-status`, 'Int64'))) = ?) AND has(JSONAllPaths(body_v2), 'http-status')) OR ((toFloat64OrNull(lower(assumeNotNull(dynamicElement(body_v2.`http-status`, 'String')))) = ?) AND has(JSONAllPaths(body_v2), 'http-status')))",
Args: []any{float64(200), float64(200), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "Indexed nested hyphen key Equal",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.response.time-taken > 1.5"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "((toFloat64(assumeNotNull(dynamicElement(body_v2.response.`time-taken`, 'Float64'))) > ?) AND has(JSONAllPaths(body_v2), 'response.time-taken'))",
Args: []any{float64(1.5), "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
// ── Bool is NOT IndexSupported — index branch must be skipped ─────────
// user.active is Bool; even with indexed builder it must not use the index.
{
name: "Bool skips index — Equal",
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
Filter: &qbtypes.Filter{Expression: "body.user.active = true"},
Limit: 10,
},
expected: TestExpected{
WhereClause: "((dynamicElement(body_v2.`user.active`, 'Bool') = ?) AND has(JSONAllPaths(body_v2), 'user.active'))",
Args: []any{true, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, qbtypes.RequestTypeRaw, c.query, nil)
if c.expectedErr != nil {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErr.Error())
return
}
require.NoError(t, err)
require.Equal(t, c.expected.GetQuery(), q.Query)
require.Equal(t, c.expected.Args, q.Args)
})
}
}
func TestJSONStmtBuilder_SelectField(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, false)
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected qbtypes.Statement
expectedErrContains string
}{
{
name: "select_x_education[].awards[].participated[].members",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Limit: 10,
SelectFields: []telemetrytypes.TelemetryFieldKey{
{
Name: "education[].awards[].participated[].members",
},
},
Filter: &qbtypes.Filter{
Expression: "user.name exists",
},
},
expected: qbtypes.Statement{
Query: "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->arrayConcat(arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))) AS `education[].awards[].participated[].members` FROM event.log WHERE (dynamicElement(body_v2.`user.name`, 'String') IS NOT NULL) AND time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? LIMIT ?",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "select_x_education[].awards[].type",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Limit: 10,
SelectFields: []telemetrytypes.TelemetryFieldKey{
{
Name: "education[].awards[].type",
},
},
Filter: &qbtypes.Filter{
Expression: "user.name exists",
},
},
expected: qbtypes.Statement{
Query: "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->arrayConcat(arrayMap(`body_v2.education[].awards`->dynamicElement(`body_v2.education[].awards`.`type`, 'String'), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards`->dynamicElement(`body_v2.education[].awards`.`type`, 'String'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))) AS `education[].awards[].type` FROM event.log WHERE (dynamicElement(body_v2.`user.name`, 'String') IS NOT NULL) AND time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? LIMIT ?",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "select_x_user.name",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Limit: 10,
SelectFields: []telemetrytypes.TelemetryFieldKey{
{
Name: "user.name",
},
},
},
expected: qbtypes.Statement{
Query: "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, dynamicElement(body_v2.`user.name`, 'String') AS `user.name` FROM event.log WHERE time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? LIMIT ?",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErrContains != "" {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErrContains)
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
func TestJSONStmtBuilder_OrderBy(t *testing.T) {
statementBuilder, _ := buildJSONTestStatementBuilder(t, false)
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected qbtypes.Statement
expectedErrContains string
}{
{
name: "order_by_education[].awards[].participated[].members",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Limit: 10,
Order: []qbtypes.OrderBy{
{
Key: qbtypes.OrderByKey{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "education[].awards[].participated[].members",
},
},
Direction: qbtypes.OrderDirectionAsc,
},
},
Filter: &qbtypes.Filter{
Expression: "user.name exists",
},
},
expected: qbtypes.Statement{
Query: "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, trace_id, span_id, toUInt32OrZero(attributes['trace_flags']) AS trace_flags, severity_text, severity_number, attributes['scope.name'] AS scope_name, attributes['scope.version'] AS scope_version, body_v2 as body, attributes AS attributes_string, CAST(map() AS Map(String, Float64)) AS attributes_number, CAST(map() AS Map(String, Bool)) AS attributes_bool, map('service.name', toString(service), 'host', toString(host)) AS resources_string, CAST(map() AS Map(String, String)) AS scope_string FROM event.log WHERE (dynamicElement(body_v2.`user.name`, 'String') IS NOT NULL) AND time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? ORDER BY arrayFlatten(arrayConcat(arrayMap(`body_v2.education`->arrayConcat(arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=4, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), dynamicElement(`body_v2.education`.`awards`, 'Array(JSON(max_dynamic_types=8, max_dynamic_paths=0))')), arrayMap(`body_v2.education[].awards`->arrayConcat(arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=256))')), arrayMap(`body_v2.education[].awards[].participated`->dynamicElement(`body_v2.education[].awards[].participated`.`members`, 'Array(Nullable(String))'), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education[].awards`.`participated`, 'Array(Dynamic)'))))), arrayMap(x->assumeNotNull(dynamicElement(x, 'JSON')), arrayFilter(x->(dynamicType(x) = 'JSON'), dynamicElement(`body_v2.education`.`awards`, 'Array(Dynamic)'))))), dynamicElement(body_v2.`education`, 'Array(JSON(max_dynamic_types=16, max_dynamic_paths=0))')))) AS `education[].awards[].participated[].members` asc LIMIT ?",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
{
name: "order_by_user.name",
requestType: qbtypes.RequestTypeRaw,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Limit: 10,
Order: []qbtypes.OrderBy{
{
Key: qbtypes.OrderByKey{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "user.name",
},
},
Direction: qbtypes.OrderDirectionAsc,
},
},
},
expected: qbtypes.Statement{
Query: "SELECT toUnixTimestamp64Nano(time) AS timestamp, id, trace_id, span_id, toUInt32OrZero(attributes['trace_flags']) AS trace_flags, severity_text, severity_number, attributes['scope.name'] AS scope_name, attributes['scope.version'] AS scope_version, body_v2 as body, attributes AS attributes_string, CAST(map() AS Map(String, Float64)) AS attributes_number, CAST(map() AS Map(String, Bool)) AS attributes_bool, map('service.name', toString(service), 'host', toString(host)) AS resources_string, CAST(map() AS Map(String, String)) AS scope_string FROM event.log WHERE time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? ORDER BY dynamicElement(body_v2.`user.name`, 'String') AS `user.name` asc LIMIT ?",
Args: []any{"1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448), 10},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErrContains != "" {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErrContains)
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
func TestResourceAggrAndGroupBy_WithJSONEnabled(t *testing.T) {
statementBuilder, metadataStore := buildJSONTestStatementBuilder(t, false)
releaseTime := time.Date(2024, 1, 15, 10, 0, 0, 0, time.UTC)
keysMap := buildCompleteFieldKeyMap(releaseTime)
for _, keys := range keysMap {
for _, key := range keys {
metadataStore.SetKey(key)
}
}
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]
expected qbtypes.Statement
expectedErrContains string
}{
{
name: "resource_aggregation_and_group_by_with_json_enabled",
requestType: qbtypes.RequestTypeTimeSeries,
query: qbtypes.QueryBuilderQuery[qbtypes.LogAggregation]{
Signal: telemetrytypes.SignalLogs,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
GroupBy: []qbtypes.GroupByKey{
{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "region",
},
},
},
Filter: &qbtypes.Filter{
Expression: "user.name exists",
},
Aggregations: []qbtypes.LogAggregation{
{
Expression: "count_distinct(service.name)",
},
},
},
expected: qbtypes.Statement{
Query: "SELECT toStartOfInterval(time, INTERVAL 30 SECOND) AS ts, toString(multiIf(mapContains(attributes, 'region') = ?, attributes['region'], NULL)) AS `region`, countDistinct(multiIf(notEmpty(service), service, NULL)) AS __result_0 FROM event.log WHERE ((dynamicElement(body_v2.`user.name`, 'String') IS NOT NULL) OR mapContains(attributes, 'user.name') = ?) AND time >= ? AND ts_bucket_start >= ? AND time < ? AND ts_bucket_start <= ? GROUP BY ts, `region`",
Args: []any{true, true, "1747947419000000000", uint64(1747945619), "1747983448000000000", uint64(1747983448)},
Warnings: []string{"Key `user.name` is ambiguous, found 2 different combinations of field context / data type: [name=user.name,context=body,datatype=string name=user.name,context=attribute,datatype=string]."},
},
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErrContains != "" {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErrContains)
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
func buildTestTelemetryMetadataStore(t *testing.T, addIndexes bool) *telemetrytypestest.MockMetadataStore {
mockMetadataStore := telemetrytypestest.NewMockMetadataStore()
mockMetadataStore.SetStaticFields(IntrinsicFields)
types, _ := telemetrytypes.TestJSONTypeSet()
for path, fieldDataTypes := range types {
for _, fdt := range fieldDataTypes {
key := &telemetrytypes.TelemetryFieldKey{
Name: path,
Signal: telemetrytypes.SignalLogs,
FieldContext: telemetrytypes.FieldContextBody,
FieldDataType: fdt,
}
if addIndexes {
jsonType := telemetrytypes.MappingFieldDataTypeToJSONDataType[fdt]
idx := slices.IndexFunc(telemetrytypes.TestIndexedPaths, func(entry telemetrytypes.TestIndexedPathEntry) bool {
return entry.Path == path && entry.Type == jsonType
})
if idx >= 0 {
key.Indexes = append(key.Indexes, telemetrytypes.TelemetryFieldKeySkipIndex{
Name: path,
FieldContext: telemetrytypes.FieldContextBody,
FieldDataType: fdt,
BaseColumn: LogsV2BodyV2Column,
IndexExpression: schemamigrator.JSONSubColumnIndexExpr(LogsV2BodyV2Column, path, jsonType.StringValue()),
})
}
}
err := key.SetJSONAccessPlan(telemetrytypes.JSONColumnMetadata{
BaseColumn: LogsV2BodyV2Column,
PromotedColumn: LogsV2BodyPromotedColumn,
}, types)
require.NoError(t, err)
mockMetadataStore.SetKey(key)
}
}
return mockMetadataStore
}
func buildJSONTestStatementBuilder(t *testing.T, addIndexes bool) (*logQueryStatementBuilder, *telemetrytypestest.MockMetadataStore) {
t.Helper()
mockMetadataStore := buildTestTelemetryMetadataStore(t, addIndexes)
fl := flaggertest.WithUseJSONBody(t, true)
fm := NewFieldMapper(fl)
cb := NewConditionBuilder(fm, fl)
aggExprRewriter := querybuilder.NewAggExprRewriter(instrumentationtest.New().ToProviderSettings(), nil, fm, cb, nil, fl)
statementBuilder := NewLogQueryStatementBuilder(
instrumentationtest.New().ToProviderSettings(),
mockMetadataStore,
fm,
cb,
aggExprRewriter,
DefaultFullTextColumn,
GetBodyJSONKey,
fl,
nil,
false,
100000,
)
return statementBuilder, mockMetadataStore
}
Reference in New Issue View Git Blame Copy Permalink