chore: bump to datafusion 39, arrow 52, pyo3 0.21 (#2581)

# Description
Updates the arrow and datafusion dependencies to 52 and 39(-rc1)
respectively. This is necessary for updating pyo3.

While most changes with trivial, some required big rewrites. Namely, the
logic for the Updates operation had to be rewritten (and simplified) to
accommodate some new sanity checks inside datafusion:
(apache/datafusion#10088).

Depends on delta-kernel having its arrow and object-store version bumped
as well. This PR doesn't include any major changes for pyo3, I'll open a
separate PR depending on this PR.

# Related Issue(s)
<!---
For example:

- closes #106
--->

# Documentation

<!---
Share links to useful documentation
--->

---------

Co-authored-by: R. Tyler Croy <[email protected]>

Cargo.toml

@@ -26,33 +26,33 @@ debug = true
 debug = "line-tables-only"
 
 [workspace.dependencies]
-delta_kernel = { version = "0.1" }
+delta_kernel = { version = "0.1.1" }
 # delta_kernel = { path = "../delta-kernel-rs/kernel" }
 
 # arrow
-arrow = { version = "51" }
-arrow-arith = { version = "51" }
-arrow-array = { version = "51", features = ["chrono-tz"] }
-arrow-buffer = { version = "51" }
-arrow-cast = { version = "51" }
-arrow-ipc = { version = "51" }
-arrow-json = { version = "51" }
-arrow-ord = { version = "51" }
-arrow-row = { version = "51" }
-arrow-schema = { version = "51" }
-arrow-select = { version = "51" }
-object_store = { version = "0.9" }
-parquet = { version = "51" }
+arrow = { version = "52" }
+arrow-arith = { version = "52" }
+arrow-array = { version = "52", features = ["chrono-tz"] }
+arrow-buffer = { version = "52" }
+arrow-cast = { version = "52" }
+arrow-ipc = { version = "52" }
+arrow-json = { version = "52" }
+arrow-ord = { version = "52" }
+arrow-row = { version = "52" }
+arrow-schema = { version = "52" }
+arrow-select = { version = "52" }
+object_store = { version = "0.10.1" }
+parquet = { version = "52" }
 
 # datafusion
-datafusion = { version = "37.1" }
-datafusion-expr = { version = "37.1" }
-datafusion-common = { version = "37.1" }
-datafusion-proto = { version = "37.1" }
-datafusion-sql = { version = "37.1" }
-datafusion-physical-expr = { version = "37.1" }
-datafusion-functions = { version = "37.1" }
-datafusion-functions-array = { version = "37.1" }
+datafusion = { version = "39" }
+datafusion-expr = { version = "39" }
+datafusion-common = { version = "39" }
+datafusion-proto = { version = "39" }
+datafusion-sql = { version = "39" }
+datafusion-physical-expr = { version = "39" }
+datafusion-functions = { version = "39" }
+datafusion-functions-array = { version = "39" }
 
 # serde
 serde = { version = "1.0.194", features = ["derive"] }

crates/aws/src/lib.rs

@@ -189,15 +189,15 @@ impl DynamoDbLockClient {
         if dynamodb_override_endpoint exists/AWS_ENDPOINT_URL_DYNAMODB is specified by user
         use dynamodb_override_endpoint to create dynamodb client
         */
-        let dynamodb_sdk_config = match dynamodb_override_endpoint {
+
+        match dynamodb_override_endpoint {
             Some(dynamodb_endpoint_url) => sdk_config
                 .to_owned()
                 .to_builder()
                 .endpoint_url(dynamodb_endpoint_url)
                 .build(),
             None => sdk_config.to_owned(),
-        };
-        dynamodb_sdk_config
+        }
     }
 
     /// Create the lock table where DynamoDb stores the commit information for all delta tables.

crates/aws/src/storage.rs

@@ -5,22 +5,22 @@ use aws_config::provider_config::ProviderConfig;
 use aws_config::{Region, SdkConfig};
 use bytes::Bytes;
 use deltalake_core::storage::object_store::{
-    aws::AmazonS3ConfigKey, parse_url_opts, GetOptions, GetResult, ListResult, MultipartId,
-    ObjectMeta, ObjectStore, PutOptions, PutResult, Result as ObjectStoreResult,
+    aws::AmazonS3ConfigKey, parse_url_opts, GetOptions, GetResult, ListResult, ObjectMeta,
+    ObjectStore, PutOptions, PutResult, Result as ObjectStoreResult,
 };
 use deltalake_core::storage::{
     limit_store_handler, str_is_truthy, ObjectStoreFactory, ObjectStoreRef, StorageOptions,
 };
 use deltalake_core::{DeltaResult, ObjectStoreError, Path};
 use futures::stream::BoxStream;
 use futures::Future;
+use object_store::{MultipartUpload, PutMultipartOpts, PutPayload};
 use std::collections::HashMap;
 use std::fmt::Debug;
 use std::ops::Range;
 use std::str::FromStr;
 use std::sync::Arc;
 use std::time::Duration;
-use tokio::io::AsyncWrite;
 use url::Url;
 
 use crate::errors::DynamoDbConfigError;
@@ -334,14 +334,14 @@ impl std::fmt::Debug for S3StorageBackend {
 
 #[async_trait::async_trait]
 impl ObjectStore for S3StorageBackend {
-    async fn put(&self, location: &Path, bytes: Bytes) -> ObjectStoreResult<PutResult> {
+    async fn put(&self, location: &Path, bytes: PutPayload) -> ObjectStoreResult<PutResult> {
         self.inner.put(location, bytes).await
     }
 
     async fn put_opts(
         &self,
         location: &Path,
-        bytes: Bytes,
+        bytes: PutPayload,
         options: PutOptions,
     ) -> ObjectStoreResult<PutResult> {
         self.inner.put_opts(location, bytes, options).await
@@ -402,19 +402,16 @@ impl ObjectStore for S3StorageBackend {
         }
     }
 
-    async fn put_multipart(
-        &self,
-        location: &Path,
-    ) -> ObjectStoreResult<(MultipartId, Box<dyn AsyncWrite + Unpin + Send>)> {
+    async fn put_multipart(&self, location: &Path) -> ObjectStoreResult<Box<dyn MultipartUpload>> {
         self.inner.put_multipart(location).await
     }
 
-    async fn abort_multipart(
+    async fn put_multipart_opts(
         &self,
         location: &Path,
-        multipart_id: &MultipartId,
-    ) -> ObjectStoreResult<()> {
-        self.inner.abort_multipart(location, multipart_id).await
+        options: PutMultipartOpts,
+    ) -> ObjectStoreResult<Box<dyn MultipartUpload>> {
+        self.inner.put_multipart_opts(location, options).await
     }
 }
 

crates/aws/tests/common.rs

@@ -87,7 +87,7 @@ impl S3Integration {
             "dynamodb",
             "create-table",
             "--table-name",
-            &table_name,
+            table_name,
             "--provisioned-throughput",
             "ReadCapacityUnits=1,WriteCapacityUnits=1",
             "--attribute-definitions",
@@ -112,7 +112,7 @@ impl S3Integration {
     }
 
     fn wait_for_table(table_name: &str) -> std::io::Result<()> {
-        let args = ["dynamodb", "describe-table", "--table-name", &table_name];
+        let args = ["dynamodb", "describe-table", "--table-name", table_name];
         loop {
             let output = Command::new("aws")
                 .args(args)
@@ -145,7 +145,7 @@ impl S3Integration {
 
     fn delete_dynamodb_table(table_name: &str) -> std::io::Result<ExitStatus> {
         let mut child = Command::new("aws")
-            .args(["dynamodb", "delete-table", "--table-name", &table_name])
+            .args(["dynamodb", "delete-table", "--table-name", table_name])
             .stdout(Stdio::null())
             .spawn()
             .expect("aws command is installed");

crates/aws/tests/repair_s3_rename_test.rs

@@ -9,6 +9,7 @@ use deltalake_core::storage::object_store::{
 use deltalake_core::{DeltaTableBuilder, ObjectStore, Path};
 use deltalake_test::utils::IntegrationContext;
 use futures::stream::BoxStream;
+use object_store::{MultipartUpload, PutMultipartOpts, PutPayload};
 use serial_test::serial;
 use std::ops::Range;
 use std::sync::{Arc, Mutex};
@@ -60,8 +61,8 @@ async fn run_repair_test_case(path: &str, pause_copy: bool) -> Result<(), Object
     };
     let (s3_2, _) = create_s3_backend(&context, "w2", None, None);
 
-    s3_1.put(&src1, Bytes::from("test1")).await.unwrap();
-    s3_2.put(&src2, Bytes::from("test2")).await.unwrap();
+    s3_1.put(&src1, Bytes::from("test1").into()).await.unwrap();
+    s3_2.put(&src2, Bytes::from("test2").into()).await.unwrap();
 
     let rename1 = rename(s3_1, &src1, &dst1);
     // to ensure that first one is started actually first
@@ -166,14 +167,14 @@ impl ObjectStore for DelayedObjectStore {
         self.delete(from).await
     }
 
-    async fn put(&self, location: &Path, bytes: Bytes) -> ObjectStoreResult<PutResult> {
+    async fn put(&self, location: &Path, bytes: PutPayload) -> ObjectStoreResult<PutResult> {
         self.inner.put(location, bytes).await
     }
 
     async fn put_opts(
         &self,
         location: &Path,
-        bytes: Bytes,
+        bytes: PutPayload,
         options: PutOptions,
     ) -> ObjectStoreResult<PutResult> {
         self.inner.put_opts(location, bytes, options).await
@@ -227,19 +228,16 @@ impl ObjectStore for DelayedObjectStore {
         self.inner.rename_if_not_exists(from, to).await
     }
 
-    async fn put_multipart(
-        &self,
-        location: &Path,
-    ) -> ObjectStoreResult<(MultipartId, Box<dyn AsyncWrite + Unpin + Send>)> {
+    async fn put_multipart(&self, location: &Path) -> ObjectStoreResult<Box<dyn MultipartUpload>> {
         self.inner.put_multipart(location).await
     }
 
-    async fn abort_multipart(
+    async fn put_multipart_opts(
         &self,
         location: &Path,
-        multipart_id: &MultipartId,
-    ) -> ObjectStoreResult<()> {
-        self.inner.abort_multipart(location, multipart_id).await
+        options: PutMultipartOpts,
+    ) -> ObjectStoreResult<Box<dyn MultipartUpload>> {
+        self.inner.put_multipart_opts(location, options).await
     }
 }
 

crates/azure/tests/integration.rs

@@ -75,7 +75,10 @@ async fn read_write_test_onelake(context: &IntegrationContext, path: &Path) -> T
 
     let expected = Bytes::from_static(b"test world from delta-rs on friday");
 
-    delta_store.put(path, expected.clone()).await.unwrap();
+    delta_store
+        .put(path, expected.clone().into())
+        .await
+        .unwrap();
     let fetched = delta_store.get(path).await.unwrap().bytes().await.unwrap();
     assert_eq!(expected, fetched);
 

crates/benchmarks/src/bin/merge.rs

@@ -7,9 +7,10 @@ use arrow::datatypes::Schema as ArrowSchema;
 use arrow_array::{RecordBatch, StringArray, UInt32Array};
 use chrono::Duration;
 use clap::{command, Args, Parser, Subcommand};
+use datafusion::functions::expr_fn::random;
 use datafusion::{datasource::MemTable, prelude::DataFrame};
 use datafusion_common::DataFusionError;
-use datafusion_expr::{cast, col, lit, random};
+use datafusion_expr::{cast, col, lit};
 use deltalake_core::protocol::SaveMode;
 use deltalake_core::{
     arrow::{

crates/core/src/delta_datafusion/cdf/scan.rs

@@ -38,7 +38,7 @@ impl ExecutionPlan for DeltaCdfScan {
         self.plan.properties()
     }
 
-    fn children(&self) -> Vec<Arc<dyn ExecutionPlan>> {
+    fn children(&self) -> Vec<&Arc<dyn ExecutionPlan>> {
         vec![]
     }
 

crates/core/src/delta_datafusion/cdf/scan_utils.rs

@@ -83,6 +83,7 @@ pub fn create_partition_values<F: FileAction>(
                 partition_values: new_part_values.clone(),
                 extensions: None,
                 range: None,
+                statistics: None,
             };
 
             file_groups.entry(new_part_values).or_default().push(part);