fix(rust, python): add file path to io error in scan_csv (pola-rs#10076)

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

@@ -323,7 +323,7 @@ impl<'a> CsvReader<'a, File> {
     /// This is the recommended way to create a csv 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 f = std::fs::File::open(&path)?;
+        let f = polars_utils::open_file(&path)?;
         Ok(Self::new(f).with_path(Some(path)))
     }
 }

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

@@ -92,7 +92,7 @@ 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 f = std::fs::File::open(&path)?;
+        let f = polars_utils::open_file(&path)?;
         Ok(Self::new(f).with_path(Some(path)))
     }
 }

polars/polars-io/src/parquet/mod.rs

@@ -29,7 +29,6 @@ use super::*;
 
 #[cfg(test)]
 mod test {
-    use std::fs::File;
     use std::io::Cursor;
 
     use polars_core::df;
@@ -40,7 +39,7 @@ mod test {
     #[test]
     fn test_parquet() {
         // In CI: This test will be skipped because the file does not exist.
-        if let Ok(r) = File::open("data/simple.parquet") {
+        if let Ok(r) = polars_utils::open_file("data/simple.parquet") {
             let reader = ParquetReader::new(r);
             let df = reader.finish().unwrap();
             assert_eq!(df.get_column_names(), ["a", "b"]);

polars/polars-io/src/partition.rs

@@ -175,7 +175,7 @@ mod test {
                 .collect::<PolarsResult<Vec<_>>>()?;
 
             assert_eq!(ipc_paths.len(), 1);
-            let reader = BufReader::new(std::fs::File::open(&ipc_paths[0])?);
+            let reader = BufReader::new(polars_utils::open_file(&ipc_paths[0])?);
             let df = IpcReader::new(reader).finish()?;
             assert!(expected_df.frame_equal(&df));
         }

polars/polars-lazy/polars-pipe/src/executors/sinks/groupby/generic/source.rs

@@ -64,7 +64,7 @@ impl Source for GroupBySource {
         if partition_dir.exists() {
             for file in std::fs::read_dir(partition_dir).expect("should be there") {
                 let spilled = file.unwrap().path();
-                let file = std::fs::File::open(spilled)?;
+                let file = polars_utils::open_file(spilled)?;
                 let reader = IpcReader::new(file);
                 let spilled = reader.finish().unwrap();
                 if spilled.n_chunks() > 1 {

polars/polars-lazy/polars-pipe/src/executors/sinks/sort/ooc.rs

@@ -16,7 +16,7 @@ use crate::executors::sinks::sort::source::SortSource;
 use crate::operators::FinalizedSink;
 
 pub(super) fn read_df(path: &Path) -> PolarsResult<DataFrame> {
-    let file = std::fs::File::open(path)?;
+    let file = polars_utils::open_file(path)?;
     IpcReader::new(file).set_rechunk(false).finish()
 }
 

polars/polars-lazy/polars-pipe/src/executors/sources/ipc_one_shot.rs

@@ -15,7 +15,7 @@ pub struct IpcSourceOneShot {
 impl IpcSourceOneShot {
     #[allow(unused_variables)]
     pub(crate) fn new(path: &Path) -> PolarsResult<Self> {
-        let file = File::open(path)?;
+        let file = polars_utils::open_file(path)?;
         let reader = Some(IpcReader::new(file));
 
         Ok(IpcSourceOneShot { reader })

polars/polars-lazy/polars-plan/src/logical_plan/builder.rs

@@ -144,7 +144,7 @@ impl LogicalPlanBuilder {
                 ParquetAsyncReader::file_info(&uri, cloud_options.as_ref())?
             }
         } else {
-            let file = std::fs::File::open(&path)?;
+            let file = polars_utils::open_file(&path)?;
             let mut reader = ParquetReader::new(file);
             (reader.schema()?, reader.num_rows()?)
         };
@@ -195,7 +195,7 @@ impl LogicalPlanBuilder {
         use polars_io::SerReader as _;
 
         let path = path.into();
-        let file = std::fs::File::open(&path)?;
+        let file = polars_utils::open_file(&path)?;
         let mut reader = IpcReader::new(file);
 
         let mut schema = reader.schema()?;
@@ -253,7 +253,15 @@ impl LogicalPlanBuilder {
         try_parse_dates: bool,
     ) -> PolarsResult<Self> {
         let path = path.into();
-        let mut file = std::fs::File::open(&path)?;
+        let mut file = polars_utils::open_file(&path).map_err(|e| {
+            let path = path.to_string_lossy();
+            if path.len() > 88 {
+                let path: String = path.chars().skip(path.len() - 88).collect();
+                polars_err!(ComputeError: "error open file: ...{}, {}", path, e)
+            } else {
+                polars_err!(ComputeError: "error open file: {}, {}", path, e)
+            }
+        })?;
         let mut magic_nr = [0u8; 2];
         file.read_exact(&mut magic_nr)
             .map_err(|_| polars_err!(NoData: "empty csv"))?;

polars/polars-lazy/src/frame/csv.rs

@@ -210,9 +210,9 @@ impl<'a> LazyCsvReader<'a> {
                 Some(globresult) => globresult?,
                 None => polars_bail!(ComputeError: "globbing pattern did not match any files"),
             };
-            std::fs::File::open(&path)
+            polars_utils::open_file(&path)
         } else {
-            std::fs::File::open(&self.path)
+            polars_utils::open_file(&self.path)
         }?;
         let reader_bytes = get_reader_bytes(&mut file).expect("could not mmap file");
         let mut skip_rows = self.skip_rows;

polars/polars-lazy/src/physical_plan/executors/scan/ndjson.rs

@@ -20,7 +20,7 @@ impl AnonymousScan for LazyJsonLineReader {
     }
 
     fn schema(&self, infer_schema_length: Option<usize>) -> PolarsResult<Schema> {
-        let f = std::fs::File::open(&self.path)?;
+        let f = polars_utils::open_file(&self.path)?;
         let mut reader = std::io::BufReader::new(f);
 
         let data_type =

polars/polars-utils/Cargo.toml

@@ -13,6 +13,7 @@ ahash.workspace = true
 hashbrown.workspace = true
 num-traits.workspace = true
 once_cell.workspace = true
+polars-error = { version = "0.31.1", path = "../polars-error" }
 rayon.workspace = true
 smartstring.workspace = true
 sysinfo = { version = "0.29", default-features = false, optional = true }

polars/polars-utils/src/io.rs

@@ -0,0 +1,19 @@
+use std::fs::File;
+use std::path::Path;
+
+use polars_error::*;
+
+pub fn open_file<P>(path: P) -> PolarsResult<File>
+where
+    P: AsRef<Path>,
+{
+    std::fs::File::open(&path).map_err(|e| {
+        let path = path.as_ref().to_string_lossy();
+        if path.len() > 88 {
+            let path: String = path.chars().skip(path.len() - 88).collect();
+            polars_err!(ComputeError: "error open file: ...{}, {}", path, e)
+        } else {
+            polars_err!(ComputeError: "error open file: {}, {}", path, e)
+        }
+    })
+}

polars/polars-utils/src/lib.rs

@@ -27,3 +27,6 @@ pub mod macros;
 pub mod vec;
 #[cfg(target_family = "wasm")]
 pub mod wasm;
+
+pub mod io;
+pub use io::open_file;