fix

Signed-off-by: Joe Isaacs <[email protected]>

Author: joseph-isaacs

bench-vortex/src/bin/compress.rs

@@ -6,9 +6,7 @@ use std::io::{Write, stdout};
 use std::path::PathBuf;
 use std::time::Duration;
 
-use bench_vortex::compress::bench::{
-    CompressMeasurements, CompressOp, {self as compress},
-};
+use bench_vortex::compress::bench::{self as compress, CompressMeasurements, CompressOp};
 use bench_vortex::datasets::Dataset;
 use bench_vortex::datasets::struct_list_of_ints::StructListOfInts;
 use bench_vortex::datasets::taxi_data::TaxiData;

bench-vortex/src/compress/bench.rs

@@ -3,43 +3,38 @@
 
 use std::borrow::Cow;
 use std::fmt;
-#[cfg(feature = "lance")]
-use std::fs;
-#[cfg(feature = "lance")]
-use std::path::PathBuf;
-#[cfg(feature = "lance")]
-use std::sync::Arc;
 use std::sync::atomic::{AtomicU64, Ordering};
 use std::time::Duration;
 
 use anyhow::Result;
-#[cfg(feature = "lance")]
-use arrow_array::RecordBatch;
 use bytes::Bytes;
 use clap::ValueEnum;
-#[cfg(feature = "lance")]
-use parking_lot::Mutex;
 use parquet::basic::{Compression, ZstdLevel};
 use serde::Serialize;
 use tokio::runtime::Runtime;
 use vortex::Array;
 use vortex::arrays::ChunkedVTable;
 use vortex::utils::aliases::hash_map::HashMap;
-
 #[cfg(feature = "lance")]
-use super::lance::*;
+use {
+    super::lance::*,
+    crate::{
+        bench_run::run_with_setup,
+        utils::{convert_utf8view_batch, convert_utf8view_schema},
+    },
+    arrow_array::RecordBatch,
+    parking_lot::Mutex,
+    std::fs,
+    std::path::PathBuf,
+    std::sync::Arc,
+};
+
 use crate::Format;
 use crate::bench_run::run;
-#[cfg(feature = "lance")]
-use crate::bench_run::run_with_setup;
 use crate::compress::chunked_to_vec_record_batch;
 use crate::compress::parquet::{parquet_compress_write, parquet_decompress_read};
 use crate::compress::vortex::{vortex_compress_write, vortex_decompress_read};
 use crate::measurements::{CompressionTimingMeasurement, CustomUnitMeasurement};
-#[cfg(feature = "lance")]
-use crate::utils::convert_utf8view_batch;
-#[cfg(feature = "lance")]
-use crate::utils::convert_utf8view_schema;
 
 #[derive(Default)]
 pub struct CompressMeasurements {

bench-vortex/src/datasets/file.rs

@@ -11,16 +11,12 @@ use datafusion::datasource::listing::{
 };
 use datafusion::prelude::SessionContext;
 use glob::Pattern;
-#[cfg(feature = "lance")]
-use lance::datafusion::LanceTableProvider;
-#[cfg(feature = "lance")]
-use lance::dataset::Dataset;
 use tracing::info;
 use url::Url;
 use vortex_datafusion::VortexFormat;
-
 #[cfg(feature = "lance")]
-use crate::Format;
+use {crate::Format, lance::datafusion::LanceTableProvider, lance::dataset::Dataset};
+
 use crate::SESSION;
 use crate::datasets::BenchmarkDataset;
 

bench-vortex/src/datasets/taxi_data.rs

@@ -1,25 +1,22 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright the Vortex contributors
 
-#[cfg(feature = "lance")]
-use std::fs::File;
 use std::path::PathBuf;
 
 use anyhow::Result;
 use async_trait::async_trait;
-#[cfg(feature = "lance")]
-use lance::dataset::Dataset as LanceDataset;
-#[cfg(feature = "lance")]
-use lance::dataset::WriteParams;
-#[cfg(feature = "lance")]
-use lance_encoding::version::LanceFileVersion;
-#[cfg(feature = "lance")]
-use parquet::arrow::arrow_reader::ParquetRecordBatchReaderBuilder;
 use tokio::fs::File as TokioFile;
 use tokio::io::AsyncWriteExt;
 use vortex::ArrayRef;
 use vortex::file::{OpenOptionsSessionExt, WriteOptionsSessionExt};
 use vortex::stream::ArrayStreamExt;
+#[cfg(feature = "lance")]
+use {
+    lance::dataset::{Dataset as LanceDataset, WriteParams},
+    lance_encoding::version::LanceFileVersion,
+    parquet::arrow::arrow_reader::ParquetRecordBatchReaderBuilder,
+    std::fs::File,
+};
 
 use crate::conversions::parquet_to_vortex;
 use crate::datasets::Dataset;

bench-vortex/src/public_bi.rs

@@ -2,9 +2,7 @@
 // SPDX-FileCopyrightText: Copyright the Vortex contributors
 
 use std::borrow::Cow;
-use std::fmt::{
-    Display, {self},
-};
+use std::fmt::{self, Display};
 use std::fs;
 use std::os::unix::fs::MetadataExt;
 use std::path::{Path, PathBuf};

bench-vortex/src/random_access/take.rs

@@ -11,9 +11,7 @@ use arrow_select::take::take_record_batch;
 use futures::stream;
 use itertools::Itertools;
 #[cfg(feature = "lance")]
-use lance::dataset::Dataset;
-#[cfg(feature = "lance")]
-use lance::dataset::ProjectionRequest;
+use lance::dataset::{Dataset, ProjectionRequest};
 use parquet::arrow::ParquetRecordBatchStreamBuilder;
 use parquet::arrow::arrow_reader::ArrowReaderOptions;
 use parquet::arrow::async_reader::AsyncFileReader;

bench-vortex/src/tpch/dbgen.rs

@@ -6,13 +6,7 @@
 /// This is a simple wrapper around the `dbgen` tool.
 ///
 /// For more information, see the [dbgen](https://github.com/vortex-data/tpch-dbgen) project.
-use std::fmt::Display;
-/// Download TPC-H data via Docker.
-///
-/// This is a simple wrapper around the `dbgen` tool.
-///
-/// For more information, see the [dbgen](https://github.com/vortex-data/tpch-dbgen) project.
-use std::fmt::Formatter;
+use std::fmt::{Display, Formatter};
 use std::fs::File;
 use std::io::copy;
 use std::path::{Path, PathBuf};

bench-vortex/src/tpch/schema.rs

@@ -6,15 +6,7 @@ use std::sync::LazyLock;
 /// Arrow schemas for TPC-H tables.
 ///
 /// Adapted from the SQL definitions in https://github.com/dimitri/tpch-citus/blob/master/schema/tpch-schema.sql
-use arrow_schema::DataType;
-/// Arrow schemas for TPC-H tables.
-///
-/// Adapted from the SQL definitions in https://github.com/dimitri/tpch-citus/blob/master/schema/tpch-schema.sql
-use arrow_schema::Field;
-/// Arrow schemas for TPC-H tables.
-///
-/// Adapted from the SQL definitions in https://github.com/dimitri/tpch-citus/blob/master/schema/tpch-schema.sql
-use arrow_schema::Schema;
+use arrow_schema::{DataType, Field, Schema};
 
 pub static NATION: LazyLock<Schema> = LazyLock::new(|| {
     Schema::new(vec![

bench-vortex/src/tpch/tpch_benchmark.rs

@@ -14,19 +14,17 @@ use log::{info, warn};
 use similar::{ChangeTag, TextDiff};
 use tokio::runtime::Runtime;
 use url::Url;
+#[cfg(feature = "lance")]
+use {crate::file::register_lance_files, crate::utils};
 
 use crate::benchmark_trait::Benchmark;
 use crate::engines::{EngineCtx, ddb};
-#[cfg(feature = "lance")]
-use crate::file::register_lance_files;
 use crate::tpch::schema::{CUSTOMER, LINEITEM, NATION, ORDERS, PART, PARTSUPP, REGION, SUPPLIER};
 use crate::tpch::tpchgen::TpchGenOptions;
 use crate::tpch::{
     EXPECTED_ROW_COUNTS_SF1, EXPECTED_ROW_COUNTS_SF10, register_arrow, register_parquet,
     register_vortex_compact_file, register_vortex_file, tpch_queries, tpchgen,
 };
-#[cfg(feature = "lance")]
-use crate::utils;
 use crate::{BenchmarkDataset, Format, IdempotentPath, Target};
 
 /// TPCH benchmark implementation

bench-vortex/src/utils/parquet_utils.rs

@@ -1,11 +1,7 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright the Vortex contributors
 
-#[cfg(feature = "lance")]
-use std::fs;
 use std::fs::File;
-#[cfg(feature = "lance")]
-use std::path::Path;
 #[cfg(not(feature = "lance"))]
 use std::path::PathBuf;
 use std::sync::Arc;
@@ -14,20 +10,17 @@ use anyhow::anyhow;
 use arrow_array::{RecordBatch, RecordBatchReader};
 use arrow_cast::cast;
 use arrow_schema::{ArrowError, DataType, Field, Schema, SchemaRef};
-#[cfg(feature = "lance")]
-use lance::dataset::Dataset as LanceDataset;
-#[cfg(feature = "lance")]
-use lance::dataset::WriteParams;
-#[cfg(feature = "lance")]
-use lance_encoding::version::LanceFileVersion;
-#[cfg(feature = "lance")]
-use log::info;
 use parquet::arrow::arrow_reader::ParquetRecordBatchReaderBuilder;
 #[cfg(feature = "lance")]
-use tokio::fs::create_dir_all;
-
-#[cfg(feature = "lance")]
-use crate::utils::idempotent_async;
+use {
+    crate::utils::idempotent_async,
+    lance::dataset::{Dataset as LanceDataset, WriteParams},
+    lance_encoding::version::LanceFileVersion,
+    log::info,
+    std::fs,
+    std::path::{Path, PathBuf},
+    tokio::fs::create_dir_all,
+};
 
 /// A streaming iterator that reads RecordBatches from multiple Parquet files sequentially.
 /// Works equally well for single files and multiple files.