|
| 1 | +use std::sync::Arc; |
| 2 | + |
| 3 | +use arrow::{ |
| 4 | + array::{Int32Array, RecordBatch, StringArray}, |
| 5 | + datatypes::{DataType, Field, Schema}, |
| 6 | +}; |
| 7 | +use arrow_flight::sql::client::FlightSqlServiceClient; |
| 8 | +use datafusion::{ |
| 9 | + datasource::MemTable, |
| 10 | + execution::context::{SessionContext, SessionState}, |
| 11 | +}; |
| 12 | +use datafusion_flight_sql_server::{config::FlightSqlServiceConfig, service::FlightSqlService}; |
| 13 | +use tokio::time::{sleep, Duration}; |
| 14 | +use tonic::transport::{Channel, Endpoint}; |
| 15 | + |
| 16 | +fn create_test_session() -> SessionState { |
| 17 | + let ctx = SessionContext::new(); |
| 18 | + let schema = Arc::new(Schema::new(vec![ |
| 19 | + Field::new("id", DataType::Int32, false), |
| 20 | + Field::new("name", DataType::Utf8, false), |
| 21 | + ])); |
| 22 | + |
| 23 | + let batch = RecordBatch::try_new( |
| 24 | + schema.clone(), |
| 25 | + vec![ |
| 26 | + Arc::new(Int32Array::from(vec![1, 2, 3])), |
| 27 | + Arc::new(StringArray::from(vec!["Alice", "Bob", "Charlie"])), |
| 28 | + ], |
| 29 | + ) |
| 30 | + .unwrap(); |
| 31 | + |
| 32 | + let table = MemTable::try_new(schema, vec![vec![batch]]).unwrap(); |
| 33 | + ctx.register_table("users", Arc::new(table)).unwrap(); |
| 34 | + |
| 35 | + let orders_schema = Arc::new(Schema::new(vec![ |
| 36 | + Field::new("order_id", DataType::Int32, false), |
| 37 | + Field::new("user_id", DataType::Int32, false), |
| 38 | + Field::new("amount", DataType::Int32, false), |
| 39 | + ])); |
| 40 | + |
| 41 | + let orders_batch = RecordBatch::try_new( |
| 42 | + orders_schema.clone(), |
| 43 | + vec![ |
| 44 | + Arc::new(Int32Array::from(vec![100, 101, 102, 103])), |
| 45 | + Arc::new(Int32Array::from(vec![1, 2, 1, 3])), |
| 46 | + Arc::new(Int32Array::from(vec![50, 75, 100, 25])), |
| 47 | + ], |
| 48 | + ) |
| 49 | + .unwrap(); |
| 50 | + |
| 51 | + let orders_table = MemTable::try_new(orders_schema, vec![vec![orders_batch]]).unwrap(); |
| 52 | + ctx.register_table("orders", Arc::new(orders_table)) |
| 53 | + .unwrap(); |
| 54 | + |
| 55 | + ctx.state() |
| 56 | +} |
| 57 | + |
| 58 | +async fn start_test_server(addr: String, state: SessionState) { |
| 59 | + let config = FlightSqlServiceConfig { |
| 60 | + schema_with_metadata: true, |
| 61 | + }; |
| 62 | + |
| 63 | + let service = FlightSqlService::new(state).with_config(config); |
| 64 | + |
| 65 | + tokio::spawn(async move { |
| 66 | + service |
| 67 | + .serve(addr) |
| 68 | + .await |
| 69 | + .expect("Server should start successfully") |
| 70 | + }); |
| 71 | + sleep(Duration::from_millis(500)).await; |
| 72 | +} |
| 73 | + |
| 74 | +async fn create_test_client(addr: &str) -> FlightSqlServiceClient<Channel> { |
| 75 | + let endpoint = Endpoint::new(addr.to_string()).expect("Valid endpoint"); |
| 76 | + let channel = endpoint.connect().await.expect("Connection successful"); |
| 77 | + FlightSqlServiceClient::new(channel) |
| 78 | +} |
| 79 | + |
| 80 | +#[tokio::test] |
| 81 | +async fn test_schema_contains_table_name_metadata() { |
| 82 | + let addr = "0.0.0.0:50071"; |
| 83 | + let state = create_test_session(); |
| 84 | + start_test_server(addr.to_string(), state).await; |
| 85 | + |
| 86 | + let mut client = create_test_client(&format!("http://{}", addr)).await; |
| 87 | + |
| 88 | + let flight_info = client |
| 89 | + .execute("SELECT id, name FROM users".to_string(), None) |
| 90 | + .await |
| 91 | + .expect("Query should succeed"); |
| 92 | + |
| 93 | + let schema = flight_info |
| 94 | + .try_decode_schema() |
| 95 | + .expect("Should decode schema"); |
| 96 | + |
| 97 | + for field in schema.fields() { |
| 98 | + assert!( |
| 99 | + field.metadata().contains_key("table_name"), |
| 100 | + "Field {} should have table_name metadata", |
| 101 | + field.name() |
| 102 | + ); |
| 103 | + |
| 104 | + assert_eq!( |
| 105 | + field.metadata().get("table_name").unwrap(), |
| 106 | + "users", |
| 107 | + "Field {} should have table_name='users'", |
| 108 | + field.name() |
| 109 | + ); |
| 110 | + } |
| 111 | +} |
| 112 | + |
| 113 | +#[tokio::test] |
| 114 | +async fn test_schema_metadata_with_subquery_and_join() { |
| 115 | + let addr = "0.0.0.0:50072"; |
| 116 | + let state = create_test_session(); |
| 117 | + start_test_server(addr.to_string(), state).await; |
| 118 | + |
| 119 | + let mut client = create_test_client(&format!("http://{}", addr)).await; |
| 120 | + |
| 121 | + let query = r#" |
| 122 | + SELECT u.id, u.name, o.amount |
| 123 | + FROM users u |
| 124 | + JOIN (SELECT * FROM orders WHERE AMOUNT > 25 ) o |
| 125 | + ON u.id = o.user_id |
| 126 | + "#; |
| 127 | + |
| 128 | + let flight_info = client |
| 129 | + .execute(query.to_string(), None) |
| 130 | + .await |
| 131 | + .expect("Query with subquery and join should succeed"); |
| 132 | + |
| 133 | + let schema = flight_info |
| 134 | + .try_decode_schema() |
| 135 | + .expect("Should decode schema"); |
| 136 | + |
| 137 | + assert_eq!(schema.fields().len(), 3, "Should have 3 fields"); |
| 138 | + |
| 139 | + // Both fields should have table_name metadata pointing to 'u' |
| 140 | + let id_field = schema.field(0); |
| 141 | + let name_field = schema.field(1); |
| 142 | + |
| 143 | + // This field should have table_name metadata pointing to 'o' |
| 144 | + let amount_field = schema.field(2); |
| 145 | + |
| 146 | + assert_eq!(id_field.name(), "id"); |
| 147 | + assert_eq!(name_field.name(), "name"); |
| 148 | + assert_eq!(amount_field.name(), "amount"); |
| 149 | + |
| 150 | + assert!( |
| 151 | + id_field.metadata().contains_key("table_name"), |
| 152 | + "id field should have table_name metadata" |
| 153 | + ); |
| 154 | + assert_eq!( |
| 155 | + id_field.metadata().get("table_name").unwrap(), |
| 156 | + "u", |
| 157 | + "id field should have table_name='u'" |
| 158 | + ); |
| 159 | + |
| 160 | + assert!( |
| 161 | + name_field.metadata().contains_key("table_name"), |
| 162 | + "name field should have table_name metadata" |
| 163 | + ); |
| 164 | + assert_eq!( |
| 165 | + name_field.metadata().get("table_name").unwrap(), |
| 166 | + "u", |
| 167 | + "name field should have table_name='u'" |
| 168 | + ); |
| 169 | + |
| 170 | + assert!( |
| 171 | + amount_field.metadata().contains_key("table_name"), |
| 172 | + "amount field should have table_name metadata" |
| 173 | + ); |
| 174 | + assert_eq!( |
| 175 | + amount_field.metadata().get("table_name").unwrap(), |
| 176 | + "o", |
| 177 | + "amount field should have table_name='o'" |
| 178 | + ); |
| 179 | +} |
0 commit comments