feat(influxql): rewrite regex conditions in subqueries (#20962)

Co-authored-by: Tecker.Yu <[email protected]>

CHANGELOG.md

@@ -17,6 +17,7 @@ the endpoint has been removed. Use the `/metrics` endpoint to collect system sta
 1. [20846](https://github.com/influxdata/influxdb/pull/20846): Add `--compression` option to `influx write` to support GZIP inputs.
 1. [20845](https://github.com/influxdata/influxdb/pull/20845): Add `influx task retry-failed` command to rerun failed runs.
 1. [20965](https://github.com/influxdata/influxdb/pull/20965): Add `--metrics-disabled` option to `influxd` to disable exposing Prometheus metrics over HTTP.
+1. [20962](https://github.com/influxdata/influxdb/pull/20962): Rewrite regex conditions in InfluxQL subqueries for performance. Thanks @yujiahaol68!
 
 ### Bug Fixes
 

influxql/query/compile.go

@@ -189,6 +189,7 @@ func (c *compiledStatement) compile(stmt *influxql.SelectStatement) error {
 			if err := c.subquery(source.Statement); err != nil {
 				return err
 			}
+			source.Statement.RewriteRegexConditions()
 		}
 	}
 	return nil

influxql/query/compile_internal_test.go

@@ -0,0 +1,24 @@
+package query
+
+import (
+	"testing"
+
+	"github.com/influxdata/influxql"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCompile_RewriteSubqueryRegex(t *testing.T) {
+	q := `SELECT top(mean, 10), host FROM (SELECT mean(value) FROM cpu WHERE id =~ /^(server-1|server-2|server-3)$/ GROUP BY host)`
+	stmt, err := influxql.ParseStatement(q)
+	require.NoError(t, err)
+	s := stmt.(*influxql.SelectStatement)
+
+	compiled, err := Compile(s, CompileOptions{})
+	require.NoError(t, err)
+
+	c := compiled.(*compiledStatement)
+	require.Len(t, c.stmt.Sources, 1)
+
+	subquery := c.stmt.Sources[0]
+	require.Equal(t, `(SELECT mean(value) FROM cpu WHERE id = 'server-1' OR id = 'server-2' OR id = 'server-3' GROUP BY host)`, subquery.String())
+}

influxql/query/compile_test.go

@@ -77,6 +77,7 @@ func TestCompile_Success(t *testing.T) {
 		`SELECT max(value) FROM cpu WHERE time >= now() - 1m GROUP BY time(10s, '2000-01-01T00:00:05Z')`,
 		`SELECT max(value) FROM cpu WHERE time >= now() - 1m GROUP BY time(10s, now())`,
 		`SELECT max(mean) FROM (SELECT mean(value) FROM cpu GROUP BY host)`,
+		`SELECT top(mean, 10), host FROM (SELECT mean(value) FROM cpu WHERE id =~ /^(server-1|server-2|server-3)$/ GROUP BY host)`,
 		`SELECT max(derivative) FROM (SELECT derivative(mean(value)) FROM cpu) WHERE time >= now() - 1m GROUP BY time(10s)`,
 		`SELECT max(value) FROM (SELECT value + total FROM cpu) WHERE time >= now() - 1m GROUP BY time(10s)`,
 		`SELECT value FROM cpu WHERE time >= '2000-01-01T00:00:00Z' AND time <= '2000-01-01T01:00:00Z'`,