chore: update datafusion to 50

Signed-off-by: luofucong <[email protected]>

Author: MichaelScofield

Cargo.lock

@@ -99,12 +99,12 @@ rust.unexpected_cfgs = { level = "warn", check-cfg = ['cfg(tokio_unstable)'] }
 # See for more detaiils: https://github.com/rust-lang/cargo/issues/11329
 ahash = { version = "0.8", features = ["compile-time-rng"] }
 aquamarine = "0.6"
-arrow = { version = "56.0", features = ["prettyprint"] }
-arrow-array = { version = "56.0", default-features = false, features = ["chrono-tz"] }
-arrow-buffer = "56.0"
-arrow-flight = "56.0"
-arrow-ipc = { version = "56.0", default-features = false, features = ["lz4", "zstd"] }
-arrow-schema = { version = "56.0", features = ["serde"] }
+arrow = { version = "56.2", features = ["prettyprint"] }
+arrow-array = { version = "56.2", default-features = false, features = ["chrono-tz"] }
+arrow-buffer = "56.2"
+arrow-flight = "56.2"
+arrow-ipc = { version = "56.2", default-features = false, features = ["lz4", "zstd"] }
+arrow-schema = { version = "56.2", features = ["serde"] }
 async-stream = "0.3"
 async-trait = "0.1"
 # Remember to update axum-extra, axum-macros when updating axum
@@ -123,18 +123,18 @@ clap = { version = "4.4", features = ["derive"] }
 config = "0.13.0"
 crossbeam-utils = "0.8"
 dashmap = "6.1"
-datafusion = "49"
-datafusion-common = "49"
-datafusion-expr = "49"
-datafusion-functions = "49"
-datafusion-functions-aggregate-common = "49"
-datafusion-optimizer = "49"
-datafusion-orc = { git = "https://github.com/GreptimeTeam/datafusion-orc", rev = "a0a5f902158f153119316eaeec868cff3fc8a99d" }
-datafusion-pg-catalog = { git = "https://github.com/datafusion-contrib/datafusion-postgres", rev = "3d1b7c7d5b82dd49bafc2803259365e633f654fa" }
-datafusion-physical-expr = "49"
-datafusion-physical-plan = "49"
-datafusion-sql = "49"
-datafusion-substrait = "49"
+datafusion = "50"
+datafusion-common = "50"
+datafusion-expr = "50"
+datafusion-functions = "50"
+datafusion-functions-aggregate-common = "50"
+datafusion-optimizer = "50"
+datafusion-orc = { git = "https://github.com/GreptimeTeam/datafusion-orc", rev = "2971b041ce1c94e315ac5b327589da6d61f449c0" }
+datafusion-postgres = { git = "https://github.com/datafusion-contrib/datafusion-postgres", rev = "7c1db5efdbf25924246f49b061acbbdf2d4af649" }
+datafusion-physical-expr = "50"
+datafusion-physical-plan = "50"
+datafusion-sql = "50"
+datafusion-substrait = "50"
 deadpool = "0.12"
 deadpool-postgres = "0.14"
 derive_builder = "0.20"
@@ -180,7 +180,7 @@ otel-arrow-rust = { git = "https://github.com/GreptimeTeam/otel-arrow", rev = "2
     "server",
 ] }
 parking_lot = "0.12"
-parquet = { version = "56.0", default-features = false, features = ["arrow", "async", "object_store"] }
+parquet = { version = "56.2", default-features = false, features = ["arrow", "async", "object_store"] }
 paste = "1.0"
 pin-project = "1.0"
 pretty_assertions = "1.4.0"
@@ -216,10 +216,7 @@ simd-json = "0.15"
 similar-asserts = "1.6.0"
 smallvec = { version = "1", features = ["serde"] }
 snafu = "0.8"
-sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "39e4fc94c3c741981f77e9d63b5ce8c02e0a27ea", features = [
-    "visitor",
-    "serde",
-] } # branch = "v0.55.x"
+sqlparser = { version = "0.58.0", default-features = false, features = ["std", "visitor", "serde"] }
 sqlx = { version = "0.8", features = [
     "runtime-tokio-rustls",
     "mysql",
@@ -321,16 +318,19 @@ git = "https://github.com/GreptimeTeam/greptime-meter.git"
 rev = "5618e779cf2bb4755b499c630fba4c35e91898cb"
 
 [patch.crates-io]
-datafusion = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-common = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-expr = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-functions = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-functions-aggregate-common = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-optimizer = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-physical-expr = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-physical-plan = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-sql = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
-datafusion-substrait = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "7d5214512740b4dfb742b6b3d91ed9affcc2c9d0" }
+datafusion = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-common = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-expr = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-functions = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-functions-aggregate-common = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-optimizer = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-physical-expr = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-physical-expr-common = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-physical-plan = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-datasource = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-sql = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+datafusion-substrait = { git = "https://github.com/GreptimeTeam/datafusion.git", rev = "fd4b2abcf3c3e43e94951bda452c9fd35243aab0" }
+sqlparser = { git = "https://github.com/GreptimeTeam/sqlparser-rs.git", rev = "4b519a5caa95472cc3988f5556813a583dd35af1" } # branch = "v0.58.x"
 
 [profile.release]
 debug = 1

Cargo.toml

@@ -35,7 +35,7 @@ common-version.workspace = true
 common-workload.workspace = true
 dashmap.workspace = true
 datafusion.workspace = true
-datafusion-pg-catalog.workspace = true
+datafusion-postgres.workspace = true
 datatypes.workspace = true
 futures.workspace = true
 futures-util.workspace = true

src/catalog/Cargo.toml

@@ -26,8 +26,9 @@ use datafusion::datasource::TableType;
 use datafusion::error::DataFusionError;
 use datafusion::execution::TaskContext;
 use datafusion::physical_plan::stream::RecordBatchStreamAdapter as DfRecordBatchStreamAdapter;
-use datafusion_pg_catalog::pg_catalog::catalog_info::CatalogInfo;
-use datafusion_pg_catalog::pg_catalog::{
+use datafusion_postgres::pg_catalog::catalog_info::CatalogInfo;
+use datafusion_postgres::pg_catalog::context::EmptyContextProvider;
+use datafusion_postgres::pg_catalog::{
     PG_CATALOG_TABLES, PgCatalogSchemaProvider, PgCatalogStaticTables, PgCatalogTable,
 };
 use snafu::ResultExt;
@@ -44,7 +45,7 @@ use crate::system_schema::{
 /// [`PGCatalogProvider`] is the provider for a schema named `pg_catalog`, it is not a catalog.
 pub struct PGCatalogProvider {
     catalog_name: String,
-    inner: PgCatalogSchemaProvider<CatalogManagerWrapper>,
+    inner: PgCatalogSchemaProvider<CatalogManagerWrapper, EmptyContextProvider>,
     tables: HashMap<String, TableRef>,
     table_ids: HashMap<&'static str, u32>,
 }
@@ -69,6 +70,7 @@ impl PGCatalogProvider {
                 catalog_manager,
             },
             Arc::new(static_tables),
+            EmptyContextProvider,
         )
         .expect("Failed to initialize PgCatalogSchemaProvider");
 

src/catalog/src/system_schema/pg_catalog.rs

@@ -33,7 +33,7 @@ use bytes::{Buf, Bytes};
 use datafusion::datasource::physical_plan::FileOpenFuture;
 use datafusion::error::{DataFusionError, Result as DataFusionResult};
 use datafusion::physical_plan::SendableRecordBatchStream;
-use futures::StreamExt;
+use futures::{StreamExt, TryStreamExt};
 use object_store::ObjectStore;
 use snafu::ResultExt;
 use tokio_util::compat::FuturesAsyncWriteCompatExt;
@@ -179,7 +179,7 @@ pub fn open_with_decoder<T: ArrowDecoder, F: Fn() -> DataFusionResult<T>>(
             Poll::Ready(decoder.flush().transpose())
         });
 
-        Ok(stream.boxed())
+        Ok(stream.map_err(Into::into).boxed())
     }))
 }
 

src/common/datasource/src/file_format.rs

@@ -36,7 +36,7 @@ datafusion.workspace = true
 datafusion-common.workspace = true
 datafusion-expr.workspace = true
 datafusion-functions-aggregate-common.workspace = true
-datafusion-pg-catalog.workspace = true
+datafusion-postgres.workspace = true
 datafusion-physical-expr.workspace = true
 datatypes.workspace = true
 derive_more = { version = "1", default-features = false, features = ["display"] }

src/common/function/Cargo.toml

@@ -22,6 +22,7 @@
 //! `foo_merge`'s input arg is the same as `foo_state`'s output, and its output is the same as `foo`'s input.
 //!
 
+use std::hash::{Hash, Hasher};
 use std::sync::Arc;
 
 use arrow::array::StructArray;
@@ -272,7 +273,7 @@ impl StateMergeHelper {
 }
 
 /// Wrapper to make an aggregate function out of a state function.
-#[derive(Debug, Clone, PartialEq, Eq)]
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
 pub struct StateWrapper {
     inner: AggregateUDF,
     name: String,
@@ -616,6 +617,20 @@ impl AggregateUDFImpl for MergeWrapper {
     }
 }
 
+impl PartialEq for MergeWrapper {
+    fn eq(&self, other: &Self) -> bool {
+        self.inner == other.inner
+    }
+}
+
+impl Eq for MergeWrapper {}
+
+impl Hash for MergeWrapper {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.inner.hash(state);
+    }
+}
+
 /// The merge accumulator, which modify `update_batch`'s behavior to accept one struct array which
 /// include the state fields of original aggregate function, and merge said states into original accumulator
 /// the output is the same as original aggregate function

src/common/function/src/aggrs/aggr_wrapper.rs

@@ -39,8 +39,7 @@ use datafusion::prelude::SessionContext;
 use datafusion_common::arrow::array::AsArray;
 use datafusion_common::arrow::datatypes::{Float64Type, UInt64Type};
 use datafusion_common::{Column, TableReference};
-use datafusion_expr::expr::AggregateFunction;
-use datafusion_expr::sqlparser::ast::NullTreatment;
+use datafusion_expr::expr::{AggregateFunction, NullTreatment};
 use datafusion_expr::{
     Aggregate, ColumnarValue, Expr, LogicalPlan, ScalarFunctionArgs, SortExpr, TableScan, lit,
 };

src/common/function/src/aggrs/aggr_wrapper/tests.rs

@@ -68,7 +68,7 @@ impl CountHash {
     }
 }
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, Eq, PartialEq, Hash)]
 pub struct CountHash {
     signature: Signature,
 }

src/common/function/src/aggrs/count_hash.rs

@@ -14,6 +14,7 @@
 
 use std::any::Any;
 use std::fmt::{Debug, Formatter};
+use std::hash::{Hash, Hasher};
 
 use datafusion::arrow::datatypes::DataType;
 use datafusion::logical_expr::{ScalarFunctionArgs, ScalarUDFImpl};
@@ -33,6 +34,20 @@ impl Debug for ScalarUdf {
     }
 }
 
+impl PartialEq for ScalarUdf {
+    fn eq(&self, other: &Self) -> bool {
+        self.function.signature() == other.function.signature()
+    }
+}
+
+impl Eq for ScalarUdf {}
+
+impl Hash for ScalarUdf {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.function.signature().hash(state)
+    }
+}
+
 impl ScalarUDFImpl for ScalarUdf {
     fn as_any(&self) -> &dyn Any {
         self