refactor: Move path logic to from utils to path_utils in polars-io (

#17635)

crates/polars-io/src/cloud/options.rs

@@ -37,9 +37,9 @@ use url::Url;
 #[cfg(feature = "file_cache")]
 use crate::file_cache::get_env_file_cache_ttl;
 #[cfg(feature = "aws")]
-use crate::pl_async::with_concurrency_budget;
+use crate::path_utils::resolve_homedir;
 #[cfg(feature = "aws")]
-use crate::utils::resolve_homedir;
+use crate::pl_async::with_concurrency_budget;
 
 #[cfg(feature = "aws")]
 static BUCKET_REGION: Lazy<std::sync::Mutex<FastFixedCache<SmartString, SmartString>>> =

crates/polars-io/src/csv/read/parser.rs

@@ -12,7 +12,7 @@ use super::buffer::Buffer;
 use super::options::{CommentPrefix, NullValuesCompiled};
 use super::splitfields::SplitFields;
 use super::utils::get_file_chunks;
-use crate::prelude::is_cloud_url;
+use crate::path_utils::is_cloud_url;
 use crate::utils::get_reader_bytes;
 
 /// Read the number of rows without parsing columns

crates/polars-io/src/csv/read/reader.rs

@@ -12,9 +12,10 @@ use super::read_impl::batched::to_batched_owned;
 use super::read_impl::CoreReader;
 use super::{infer_file_schema, BatchedCsvReader, OwnedBatchedCsvReader};
 use crate::mmap::MmapBytesReader;
+use crate::path_utils::resolve_homedir;
 use crate::predicates::PhysicalIoExpr;
 use crate::shared::SerReader;
-use crate::utils::{get_reader_bytes, resolve_homedir};
+use crate::utils::get_reader_bytes;
 
 /// Create a new DataFrame by reading a csv file.
 ///

crates/polars-io/src/file_cache/cache.rs

@@ -11,8 +11,7 @@ use super::entry::{FileCacheEntry, DATA_PREFIX, METADATA_PREFIX};
 use super::eviction::EvictionManager;
 use super::file_fetcher::FileFetcher;
 use super::utils::FILE_CACHE_PREFIX;
-use crate::prelude::is_cloud_url;
-use crate::utils::ensure_directory_init;
+use crate::path_utils::{ensure_directory_init, is_cloud_url};
 
 pub static FILE_CACHE: Lazy<FileCache> = Lazy::new(|| {
     let prefix = FILE_CACHE_PREFIX.as_ref();
@@ -101,7 +100,7 @@ impl FileCache {
         #[cfg(debug_assertions)]
         {
             // Local paths must be absolute or else the cache would be wrong.
-            if !crate::utils::is_cloud_url(uri.as_ref()) {
+            if !crate::path_utils::is_cloud_url(uri.as_ref()) {
                 let path = Path::new(uri.as_ref());
                 assert_eq!(path, std::fs::canonicalize(path).unwrap().as_path());
             }

crates/polars-io/src/file_cache/utils.rs

@@ -11,9 +11,8 @@ use super::file_fetcher::{CloudFileFetcher, LocalFileFetcher};
 use crate::cloud::{
     build_object_store, object_path_from_string, CloudLocation, CloudOptions, PolarsObjectStore,
 };
+use crate::path_utils::{ensure_directory_init, is_cloud_url, POLARS_TEMP_DIR_BASE_PATH};
 use crate::pl_async;
-use crate::prelude::{is_cloud_url, POLARS_TEMP_DIR_BASE_PATH};
-use crate::utils::ensure_directory_init;
 
 pub static FILE_CACHE_PREFIX: Lazy<Box<Path>> = Lazy::new(|| {
     let path = POLARS_TEMP_DIR_BASE_PATH

crates/polars-io/src/lib.rs

@@ -21,6 +21,7 @@ mod options;
 pub mod parquet;
 #[cfg(feature = "parquet")]
 pub mod partition;
+pub mod path_utils;
 #[cfg(feature = "async")]
 pub mod pl_async;
 pub mod predicates;
@@ -31,6 +32,7 @@ pub mod utils;
 #[cfg(feature = "cloud")]
 pub use cloud::glob as async_glob;
 pub use options::*;
+pub use path_utils::*;
 pub use shared::*;
 
 pub mod hive;

crates/polars-io/src/ndjson/core.rs

@@ -137,7 +137,7 @@ where
 impl<'a> JsonLineReader<'a, File> {
     /// This is the recommended way to create a json reader as this allows for fastest parsing.
     pub fn from_path<P: Into<PathBuf>>(path: P) -> PolarsResult<Self> {
-        let path = resolve_homedir(&path.into());
+        let path = crate::resolve_homedir(&path.into());
         let f = polars_utils::open_file(&path)?;
         Ok(Self::new(f).with_path(Some(path)))
     }

crates/polars-io/src/prelude.rs

@@ -11,5 +11,6 @@ pub use crate::ndjson::core::*;
 pub use crate::parquet::{metadata::*, read::*, write::*};
 #[cfg(feature = "parquet")]
 pub use crate::partition::write_partitioned_dataset;
+pub use crate::path_utils::*;
 pub use crate::shared::{SerReader, SerWriter};
 pub use crate::utils::*;

crates/polars-io/src/utils/mod.rs

@@ -1,5 +1,3 @@
 mod other;
-mod path;
 
 pub use other::*;
-pub use path::*;

crates/polars-lazy/src/scan/csv.rs

@@ -5,7 +5,8 @@ use polars_io::cloud::CloudOptions;
 use polars_io::csv::read::{
     infer_file_schema, CommentPrefix, CsvEncoding, CsvParseOptions, CsvReadOptions, NullValues,
 };
-use polars_io::utils::{expand_paths, get_reader_bytes};
+use polars_io::path_utils::expand_paths;
+use polars_io::utils::get_reader_bytes;
 use polars_io::RowIndex;
 
 use crate::prelude::*;

crates/polars-mem-engine/src/executors/scan/ipc.rs

@@ -4,8 +4,8 @@ use hive::HivePartitions;
 use polars_core::config;
 use polars_core::utils::accumulate_dataframes_vertical;
 use polars_io::cloud::CloudOptions;
+use polars_io::path_utils::is_cloud_url;
 use polars_io::predicates::apply_predicate;
-use polars_io::utils::is_cloud_url;
 use rayon::prelude::*;
 
 use super::*;

crates/polars-mem-engine/src/executors/scan/parquet.rs

@@ -7,7 +7,7 @@ use polars_core::config::{get_file_prefetch_size, verbose};
 use polars_core::utils::accumulate_dataframes_vertical;
 use polars_io::cloud::CloudOptions;
 use polars_io::parquet::metadata::FileMetaDataRef;
-use polars_io::utils::is_cloud_url;
+use polars_io::path_utils::is_cloud_url;
 use polars_io::RowIndex;
 
 use super::*;

crates/polars-pipe/src/executors/sources/csv.rs

@@ -3,7 +3,7 @@ use std::path::PathBuf;
 
 use polars_core::{config, POOL};
 use polars_io::csv::read::{BatchedCsvReader, CsvReadOptions, CsvReader};
-use polars_io::utils::is_cloud_url;
+use polars_io::path_utils::is_cloud_url;
 use polars_plan::global::_set_n_rows_for_scan;
 use polars_plan::prelude::FileScanOptions;
 use polars_utils::iter::EnumerateIdxTrait;

crates/polars-pipe/src/executors/sources/parquet.rs

@@ -10,12 +10,12 @@ use polars_core::POOL;
 use polars_io::cloud::CloudOptions;
 use polars_io::parquet::metadata::FileMetaDataRef;
 use polars_io::parquet::read::{BatchedParquetReader, ParquetOptions, ParquetReader};
+use polars_io::path_utils::is_cloud_url;
 use polars_io::pl_async::get_runtime;
 use polars_io::predicates::PhysicalIoExpr;
 use polars_io::prelude::materialize_projection;
 #[cfg(feature = "async")]
 use polars_io::prelude::ParquetAsyncReader;
-use polars_io::utils::is_cloud_url;
 use polars_io::SerReader;
 use polars_plan::plans::FileInfo;
 use polars_plan::prelude::hive::HivePartitions;

crates/polars-plan/src/plans/conversion/dsl_to_ir.rs

@@ -8,9 +8,9 @@ use hive::{hive_partitions_from_paths, HivePartitions};
 #[cfg(any(feature = "ipc", feature = "parquet"))]
 use polars_io::cloud::CloudOptions;
 #[cfg(any(feature = "csv", feature = "json"))]
-use polars_io::utils::expand_paths;
+use polars_io::path_utils::expand_paths;
 #[cfg(any(feature = "ipc", feature = "parquet"))]
-use polars_io::utils::{expand_paths_hive, expanded_from_single_directory};
+use polars_io::path_utils::{expand_paths_hive, expanded_from_single_directory};
 
 use super::stack_opt::ConversionOptimizer;
 use super::*;

crates/polars-plan/src/plans/conversion/scans.rs

@@ -1,10 +1,10 @@
 use std::path::PathBuf;
 
 use either::Either;
+use polars_io::path_utils::is_cloud_url;
 #[cfg(feature = "cloud")]
 use polars_io::pl_async::get_runtime;
 use polars_io::prelude::*;
-use polars_io::utils::is_cloud_url;
 use polars_io::RowIndex;
 
 use super::*;

crates/polars-plan/src/plans/functions/count.rs

@@ -11,7 +11,7 @@ use polars_io::parquet::read::ParquetReader;
 #[cfg(all(feature = "parquet", feature = "async"))]
 use polars_io::pl_async::{get_runtime, with_concurrency_budget};
 #[cfg(any(feature = "parquet", feature = "ipc"))]
-use polars_io::{utils::is_cloud_url, SerReader};
+use polars_io::{path_utils::is_cloud_url, SerReader};
 
 use super::*;
 

crates/polars-plan/src/plans/hive.rs

@@ -230,7 +230,7 @@ pub fn hive_partitions_from_paths(
 /// Determine the path separator for identifying Hive partitions.
 #[cfg(target_os = "windows")]
 fn separator(url: &Path) -> char {
-    if polars_io::utils::is_cloud_url(url) {
+    if polars_io::path_utils::is_cloud_url(url) {
         '/'
     } else {
         '\\'