From 9619ba4086668b04a14ad07e1a35cb8fbba88529 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=C3=96zg=C3=BCr=20Akkurt?= Date: Fri, 24 Feb 2023 20:32:54 +0300 Subject: [PATCH 1/3] Re-exported parquet bloom filter module --- src/io/parquet/read/mod.rs | 1 + src/io/parquet/write/mod.rs | 1 + 2 files changed, 2 insertions(+) diff --git a/src/io/parquet/read/mod.rs b/src/io/parquet/read/mod.rs index c41d60e4c1e..52e5dffe5f5 100644 --- a/src/io/parquet/read/mod.rs +++ b/src/io/parquet/read/mod.rs @@ -14,6 +14,7 @@ use futures::{AsyncRead, AsyncSeek}; // re-exports of parquet2's relevant APIs pub use parquet2::{ + bloom_filter, error::Error as ParquetError, fallible_streaming_iterator, metadata::{ColumnChunkMetaData, ColumnDescriptor, RowGroupMetaData}, diff --git a/src/io/parquet/write/mod.rs b/src/io/parquet/write/mod.rs index a87dfd54975..68c5d4209bf 100644 --- a/src/io/parquet/write/mod.rs +++ b/src/io/parquet/write/mod.rs @@ -36,6 +36,7 @@ pub use nested::write_rep_and_def; pub use pages::{to_leaves, to_nested, to_parquet_leaves}; use parquet2::schema::types::PrimitiveType as ParquetPrimitiveType; pub use parquet2::{ + bloom_filter, compression::{BrotliLevel, CompressionOptions, GzipLevel, ZstdLevel}, encoding::Encoding, fallible_streaming_iterator, From 40c103d122eebe654376e5921e895cf513815ac7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=C3=96zg=C3=BCr=20Akkurt?= Date: Fri, 24 Feb 2023 20:34:28 +0300 Subject: [PATCH 2/3] re-export from parquet module --- src/io/parquet/mod.rs | 1 + src/io/parquet/read/mod.rs | 1 - src/io/parquet/write/mod.rs | 1 - 3 files changed, 1 insertion(+), 2 deletions(-) diff --git a/src/io/parquet/mod.rs b/src/io/parquet/mod.rs index 04684369110..3a111fac570 100644 --- a/src/io/parquet/mod.rs +++ b/src/io/parquet/mod.rs @@ -3,6 +3,7 @@ use crate::error::Error; pub mod read; pub mod write; +pub use parquet2::bloom_filter; const ARROW_SCHEMA_META_KEY: &str = "ARROW:schema"; diff --git a/src/io/parquet/read/mod.rs b/src/io/parquet/read/mod.rs index 52e5dffe5f5..c41d60e4c1e 100644 --- a/src/io/parquet/read/mod.rs +++ b/src/io/parquet/read/mod.rs @@ -14,7 +14,6 @@ use futures::{AsyncRead, AsyncSeek}; // re-exports of parquet2's relevant APIs pub use parquet2::{ - bloom_filter, error::Error as ParquetError, fallible_streaming_iterator, metadata::{ColumnChunkMetaData, ColumnDescriptor, RowGroupMetaData}, diff --git a/src/io/parquet/write/mod.rs b/src/io/parquet/write/mod.rs index 68c5d4209bf..a87dfd54975 100644 --- a/src/io/parquet/write/mod.rs +++ b/src/io/parquet/write/mod.rs @@ -36,7 +36,6 @@ pub use nested::write_rep_and_def; pub use pages::{to_leaves, to_nested, to_parquet_leaves}; use parquet2::schema::types::PrimitiveType as ParquetPrimitiveType; pub use parquet2::{ - bloom_filter, compression::{BrotliLevel, CompressionOptions, GzipLevel, ZstdLevel}, encoding::Encoding, fallible_streaming_iterator, From d6c8aa8214bc4b1caa2188792881301aedd9fc62 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=C3=96zg=C3=BCr=20Akkurt?= Date: Fri, 24 Feb 2023 20:40:33 +0300 Subject: [PATCH 3/3] feature gate --- Cargo.toml | 3 +++ src/io/parquet/mod.rs | 3 +++ 2 files changed, 6 insertions(+) diff --git a/Cargo.toml b/Cargo.toml index eb1c1017ba6..80d1fa95092 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -186,6 +186,9 @@ io_parquet_lz4_flex = ["parquet2/lz4_flex"] io_parquet_lz4 = ["parquet2/lz4"] io_parquet_brotli = ["parquet2/brotli"] +# parquet bloom filter functions +io_parquet_bloom_filter = ["parquet2/bloom_filter"] + io_avro = ["avro-schema", "streaming-iterator"] io_avro_compression = [ "avro-schema/compression", diff --git a/src/io/parquet/mod.rs b/src/io/parquet/mod.rs index 3a111fac570..cf50980a918 100644 --- a/src/io/parquet/mod.rs +++ b/src/io/parquet/mod.rs @@ -3,6 +3,9 @@ use crate::error::Error; pub mod read; pub mod write; + +#[cfg(feature = "io_parquet_bloom_filter")] +#[cfg_attr(docsrs, doc(cfg(feature = "io_parquet_bloom_filter")))] pub use parquet2::bloom_filter; const ARROW_SCHEMA_META_KEY: &str = "ARROW:schema";