From d236a44c5f9d864a8f5cbdbade8475ea5c71b55a Mon Sep 17 00:00:00 2001
From: Alexander Beedie <alexander-beedie@users.noreply.github.com>
Date: Mon, 10 Jun 2024 18:20:17 +0400
Subject: [PATCH] feat: Support homedir expansion in lazy/scan read functions

---
 crates/polars-lazy/src/scan/csv.rs     | 11 +++++++++--
 crates/polars-lazy/src/scan/ipc.rs     | 11 +++++++++--
 crates/polars-lazy/src/scan/ndjson.rs  |  9 ++++++++-
 crates/polars-lazy/src/scan/parquet.rs |  9 ++++++++-
 4 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/crates/polars-lazy/src/scan/csv.rs b/crates/polars-lazy/src/scan/csv.rs
index 577441e3dee7..4045ee718aae 100644
--- a/crates/polars-lazy/src/scan/csv.rs
+++ b/crates/polars-lazy/src/scan/csv.rs
@@ -5,6 +5,7 @@ use polars_io::cloud::CloudOptions;
 use polars_io::csv::read::{
     infer_file_schema, CommentPrefix, CsvEncoding, CsvParseOptions, CsvReadOptions, NullValues,
 };
+use polars_io::prelude::resolve_homedir;
 use polars_io::utils::get_reader_bytes;
 use polars_io::RowIndex;
 
@@ -35,7 +36,7 @@ impl LazyCsvReader {
 
     pub fn new(path: impl AsRef<Path>) -> Self {
         LazyCsvReader {
-            path: path.as_ref().to_owned(),
+            path: resolve_homedir(path.as_ref()),
             paths: Arc::new([]),
             glob: true,
             cache: true,
@@ -305,11 +306,17 @@ impl LazyFileListReader for LazyCsvReader {
     }
 
     fn with_path(mut self, path: PathBuf) -> Self {
-        self.path = path;
+        self.path = resolve_homedir(&path);
         self
     }
 
     fn with_paths(mut self, paths: Arc<[PathBuf]>) -> Self {
+        let paths = paths
+            .iter()
+            .map(|p| resolve_homedir(p))
+            .collect::<Vec<_>>()
+            .into();
+
         self.paths = paths;
         self
     }
diff --git a/crates/polars-lazy/src/scan/ipc.rs b/crates/polars-lazy/src/scan/ipc.rs
index 5672802a61f2..498eebf4a04d 100644
--- a/crates/polars-lazy/src/scan/ipc.rs
+++ b/crates/polars-lazy/src/scan/ipc.rs
@@ -3,6 +3,7 @@ use std::path::{Path, PathBuf};
 use polars_core::prelude::*;
 use polars_io::cloud::CloudOptions;
 use polars_io::ipc::IpcScanOptions;
+use polars_io::utils::resolve_homedir;
 use polars_io::RowIndex;
 
 use crate::prelude::*;
@@ -41,7 +42,7 @@ impl LazyIpcReader {
     fn new(path: PathBuf, args: ScanArgsIpc) -> Self {
         Self {
             args,
-            path,
+            path: resolve_homedir(&path),
             paths: Arc::new([]),
         }
     }
@@ -96,11 +97,17 @@ impl LazyFileListReader for LazyIpcReader {
     }
 
     fn with_path(mut self, path: PathBuf) -> Self {
-        self.path = path;
+        self.path = resolve_homedir(&path);
         self
     }
 
     fn with_paths(mut self, paths: Arc<[PathBuf]>) -> Self {
+        let paths = paths
+            .iter()
+            .map(|p| resolve_homedir(p))
+            .collect::<Vec<_>>()
+            .into();
+
         self.paths = paths;
         self
     }
diff --git a/crates/polars-lazy/src/scan/ndjson.rs b/crates/polars-lazy/src/scan/ndjson.rs
index e2b2691e4e08..996d066a01c8 100644
--- a/crates/polars-lazy/src/scan/ndjson.rs
+++ b/crates/polars-lazy/src/scan/ndjson.rs
@@ -3,6 +3,7 @@ use std::path::{Path, PathBuf};
 use std::sync::RwLock;
 
 use polars_core::prelude::*;
+use polars_io::utils::resolve_homedir;
 use polars_io::RowIndex;
 
 use super::*;
@@ -114,11 +115,17 @@ impl LazyFileListReader for LazyJsonLineReader {
     }
 
     fn with_path(mut self, path: PathBuf) -> Self {
-        self.path = path;
+        self.path = resolve_homedir(&path);
         self
     }
 
     fn with_paths(mut self, paths: Arc<[PathBuf]>) -> Self {
+        let paths = paths
+            .iter()
+            .map(|p| resolve_homedir(p))
+            .collect::<Vec<_>>()
+            .into();
+
         self.paths = paths;
         self
     }
diff --git a/crates/polars-lazy/src/scan/parquet.rs b/crates/polars-lazy/src/scan/parquet.rs
index 4e57b25d42bf..e135099b13a8 100644
--- a/crates/polars-lazy/src/scan/parquet.rs
+++ b/crates/polars-lazy/src/scan/parquet.rs
@@ -3,6 +3,7 @@ use std::path::{Path, PathBuf};
 use polars_core::prelude::*;
 use polars_io::cloud::CloudOptions;
 use polars_io::parquet::read::ParallelStrategy;
+use polars_io::prelude::resolve_homedir;
 use polars_io::{HiveOptions, RowIndex};
 
 use crate::prelude::*;
@@ -112,11 +113,17 @@ impl LazyFileListReader for LazyParquetReader {
     }
 
     fn with_path(mut self, path: PathBuf) -> Self {
-        self.path = path;
+        self.path = resolve_homedir(&path);
         self
     }
 
     fn with_paths(mut self, paths: Arc<[PathBuf]>) -> Self {
+        let paths = paths
+            .iter()
+            .map(|p| resolve_homedir(p))
+            .collect::<Vec<_>>()
+            .into();
+
         self.paths = paths;
         self
     }