Unsupported table named argument type: FunctionCall(FunctionCall

[BohuTANG]

Summary

SQL: SELECT * FROM INFER_SCHEMA(location => @'https://huggingface.co/datasets/vivym/midjourney-messages/resolve/main/data/000016.parquet'); 

1002: Unsupported table named argument type: FunctionCall(FunctionCall { span: Some(41..42), func_name: "abs", params: [], arguments: [ConstantExpr(ConstantExpr { span: Some(42..134), value: String([104, 116, 116, 112, 115, 58, 47, 47, 104, 117, 103, 103, 105, 110, 103, 102, 97, 99, 101, 46, 99, 111, 47, 100, 97, 116, 97, 115, 101, 116, 115, 47, 118, 105, 118, 121, 109, 47, 109, 105, 100, 106, 111, 117, 114, 110, 101, 121, 45, 109, 101, 115, 115, 97, 103, 101, 115, 47, 114, 101, 115, 111, 108, 118, 101, 47, 109, 97, 105, 110, 47, 100, 97, 116, 97, 47, 48, 48, 48, 48, 49, 54, 46, 112, 97, 114, 113, 117, 101, 116]) })] })

[BohuTANG]

How to infer a remote parquet file schema? The error confused.

[youngsofun]

@BohuTANG not supported yet

[BohuTANG]

Ok, we should give a readable error hints.

[youngsofun]

@BohuTANG INFER_SCHEMA is a normal table function.
this error happend before knowing the table function is INFER_SCHEMA.
cc @andylokandy

[youngsofun]

we did not support uri because it need connection

for now, user can work around by create a stage first.
since we now support used defined connenction, we can use INFER_SCHEMA(location ='<uri>', connection='<conn_name>'), it save user from creating a object for URIs that not need connection (http, fs)

[sundy-li]

🐳 :) SELECT * FROM INFER_SCHEMA(location => 'http://127.0.0.1:9999/hits_0.parquet') limit 3;
┌──────────────────────────────────────────────┐
│ column_name │   type   │ nullable │ order_id │
│    String   │  String  │  Boolean │  UInt64  │
├─────────────┼──────────┼──────────┼──────────┤
│ WatchID     │ BIGINT   │ true     │        0 │
│ JavaEnable  │ SMALLINT │ true     │        1 │
│ Title       │ BINARY   │ true     │        2 │
└──────────────────────────────────────────────┘
3 rows read in 0.058 sec. Processed 105 rows, 4.30 KiB (1.8 thousand rows/s, 73.55 KiB/s)

works now.