perf(rust): improve json_extract

Cargo.toml

@@ -14,6 +14,7 @@ members = [
   "polars/polars-sql",
   "polars/polars-error",
   "polars/polars-row",
+  "polars/polars-json",
   "examples/read_csv",
   "examples/read_parquet",
   "examples/read_parquet_cloud",
@@ -44,7 +45,7 @@ package = "arrow2"
 git = "https://github.com/ritchie46/arrow2"
 # rev = "1491c6e8f4fd100f53c358e4f3ef1536d9e75090"
 # path = "../arrow2"
-branch = "polars_2023-05-09"
+branch = "polars_2023-05-16"
 version = "0.17"
 default-features = false
 features = [

polars/Makefile

@@ -21,7 +21,8 @@ check:
 		-p polars-time \
 		-p polars-error \
 		-p polars-ops \
-		-p polars-sql
+		-p polars-sql \
+		-p polars-json
 
 clippy:
 	cargo clippy --all-features \
@@ -34,7 +35,8 @@ clippy:
 		-p polars-error \
 		-p polars-row \
 		-p polars-time \
-		-p polars-sql
+		-p polars-sql \
+		-p polars-json
 
 clippy-default:
 	cargo clippy

polars/polars-core/src/datatypes/aliases.rs

@@ -18,17 +18,11 @@ pub type IdxType = UInt32Type;
 #[cfg(feature = "bigidx")]
 pub type IdxType = UInt64Type;
 
-#[cfg(feature = "private")]
 pub type PlHashMap<K, V> = hashbrown::HashMap<K, V, RandomState>;
-#[cfg(feature = "private")]
-
 /// This hashmap has the uses an IdHasher
 pub type PlIdHashMap<K, V> = hashbrown::HashMap<K, V, IdBuildHasher>;
-#[cfg(feature = "private")]
 pub type PlHashSet<V> = hashbrown::HashSet<V, RandomState>;
-#[cfg(feature = "private")]
 pub type PlIndexMap<K, V> = indexmap::IndexMap<K, V, RandomState>;
-#[cfg(feature = "private")]
 pub type PlIndexSet<K> = indexmap::IndexSet<K, RandomState>;
 
 pub trait InitHashMaps {

polars/polars-io/src/lib.rs

@@ -15,7 +15,7 @@ pub mod ipc;
 #[cfg(feature = "json")]
 pub mod json;
 #[cfg(feature = "json")]
-pub mod ndjson_core;
+pub mod ndjson;
 #[cfg(feature = "cloud")]
 pub use crate::cloud::glob as async_glob;
 

polars/polars-io/src/ndjson_core/ndjson.rs → polars/polars-io/src/ndjson/core.rs

@@ -13,7 +13,7 @@ use rayon::prelude::*;
 
 use crate::csv::utils::*;
 use crate::mmap::{MmapBytesReader, ReaderBytes};
-use crate::ndjson_core::buffer::*;
+use crate::ndjson::buffer::*;
 use crate::prelude::*;
 const NEWLINE: u8 = b'\n';
 const RETURN: u8 = b'\r';

polars/polars-io/src/ndjson_core/mod.rs → polars/polars-io/src/ndjson/mod.rs

@@ -1,2 +1,2 @@
 pub(crate) mod buffer;
-pub mod ndjson;
+pub mod core;

polars/polars-io/src/prelude.rs

@@ -8,7 +8,7 @@ pub use crate::ipc::*;
 #[cfg(feature = "json")]
 pub use crate::json::*;
 #[cfg(feature = "json")]
-pub use crate::ndjson_core::ndjson::*;
+pub use crate::ndjson::core::*;
 #[cfg(feature = "parquet")]
 pub use crate::parquet::*;
 pub use crate::utils::*;

polars/polars-json/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+name = "polars-json"
+version = "0.29.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+ahash.workspace = true
+arrow.workspace = true
+hashbrown.workspace = true
+num-traits.workspace = true
+polars-arrow = { version = "0.29.0", path = "../polars-arrow", default-features = false }
+polars-error = { version = "0.29.0", path = "../polars-error" }
+polars-utils = { version = "0.29.0", path = "../polars-utils" }
+simd-json = { version = "0.9", features = ["allow-non-simd", "known-key"] }