fix: avoid convert special characters to escape characters in windows machine

Author: zhuliquan

.gitignore

@@ -64,9 +64,6 @@ datafusion/sqllogictests/test_files/tpch/data/*
 # Scratch temp dir for sqllogictests
 datafusion/sqllogictest/test_files/scratch*
 
-# temp file for core
-datafusion/core/*.parquet
-
 # rat
 filtered_rat.txt
 rat.txt

datafusion/core/src/datasource/listing/mod.rs

@@ -90,7 +90,9 @@ impl PartitionedFile {
     pub fn new(path: impl Into<String>, size: u64) -> Self {
         Self {
             object_meta: ObjectMeta {
-                location: Path::from(path.into()),
+                // windows will convert special characters to escape characters (i.g. '~' => '%7E'),
+                // so we need to use Path::parse
+                location: Path::parse(path.into()).unwrap(),
                 last_modified: chrono::Utc.timestamp_nanos(0),
                 size: size as usize,
                 e_tag: None,

datafusion/core/tests/parquet/external_access_plan.rs

@@ -33,7 +33,6 @@ use datafusion_physical_plan::ExecutionPlan;
 use parquet::arrow::arrow_reader::{RowSelection, RowSelector};
 use parquet::arrow::ArrowWriter;
 use parquet::file::properties::WriterProperties;
-use std::path::Path;
 use std::sync::{Arc, OnceLock};
 use tempfile::NamedTempFile;
 
@@ -315,19 +314,12 @@ impl TestFull {
 
         let TestData {
             _temp_file: _,
-            ref schema,
-            ref file_name,
-            ref file_size,
+            schema,
+            file_name,
+            file_size,
         } = get_test_data();
 
-        let new_file_name = if cfg!(target_os = "windows") {
-            // Windows path separator is different from Unix
-            file_name.replace("\\", "/")
-        } else {
-            file_name.clone()
-        };
-
-        let mut partitioned_file = PartitionedFile::new(new_file_name, *file_size);
+        let mut partitioned_file = PartitionedFile::new(file_name, *file_size);
 
         // add the access plan, if any, as an extension
         if let Some(access_plan) = access_plan {
@@ -363,8 +355,6 @@ impl TestFull {
             pretty_format_batches(&results).unwrap()
         );
 
-        std::fs::remove_file(file_name).unwrap();
-
         Ok(MetricsFinder::find_metrics(plan.as_ref()).unwrap())
     }
 }
@@ -379,43 +369,45 @@ struct TestData {
     file_size: u64,
 }
 
-static _TEST_DATA: OnceLock<TestData> = OnceLock::new();
+static TEST_DATA: OnceLock<TestData> = OnceLock::new();
 
 /// Return a parquet file with 2 row groups each with 5 rows
-fn get_test_data() -> TestData {
-    let scenario = Scenario::UTF8;
-    let row_per_group = 5;
+fn get_test_data() -> &'static TestData {
+    TEST_DATA.get_or_init(|| {
+        let scenario = Scenario::UTF8;
+        let row_per_group = 5;
 
-    let mut temp_file = tempfile::Builder::new()
-        .prefix("user_access_plan")
-        .suffix(".parquet")
-        .tempfile_in(Path::new(""))
-        .expect("tempfile creation");
+        let mut temp_file = tempfile::Builder::new()
+            .prefix("user_access_plan")
+            .suffix(".parquet")
+            .tempfile()
+            .expect("tempfile creation");
 
-    let props = WriterProperties::builder()
-        .set_max_row_group_size(row_per_group)
-        .build();
+        let props = WriterProperties::builder()
+            .set_max_row_group_size(row_per_group)
+            .build();
 
-    let batches = create_data_batch(scenario);
-    let schema = batches[0].schema();
+        let batches = create_data_batch(scenario);
+        let schema = batches[0].schema();
 
-    let mut writer =
-        ArrowWriter::try_new(&mut temp_file, schema.clone(), Some(props)).unwrap();
+        let mut writer =
+            ArrowWriter::try_new(&mut temp_file, schema.clone(), Some(props)).unwrap();
 
-    for batch in batches {
-        writer.write(&batch).expect("writing batch");
-    }
-    writer.close().unwrap();
+        for batch in batches {
+            writer.write(&batch).expect("writing batch");
+        }
+        writer.close().unwrap();
 
-    let file_name = temp_file.path().to_string_lossy().to_string();
-    let file_size = temp_file.path().metadata().unwrap().len();
+        let file_name = temp_file.path().to_string_lossy().to_string();
+        let file_size = temp_file.path().metadata().unwrap().len();
 
-    TestData {
-        _temp_file: temp_file,
-        schema,
-        file_name,
-        file_size,
-    }
+        TestData {
+            _temp_file: temp_file,
+            schema,
+            file_name,
+            file_size,
+        }
+    })
 }
 
 /// Return the total value of the specified metric name