This is an automated email from the ASF dual-hosted git repository.
mbrobbel pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-rs.git
The following commit(s) were added to refs/heads/main by this push:
new 7c6a883302 Migrate `arrow-csv` to Rust 2024 (#8454)
7c6a883302 is described below
commit 7c6a883302551dde7e89bfed1779c74dac677a0a
Author: Matthijs Brobbel <[email protected]>
AuthorDate: Fri Sep 26 15:53:31 2025 +0200
Migrate `arrow-csv` to Rust 2024 (#8454)
# Which issue does this PR close?
- Contribute to #6827
# Rationale for this change
Splitting up #8227.
# What changes are included in this PR?
Migrate `arrow-csv` to Rust 2024
# Are these changes tested?
CI
# Are there any user-facing changes?
Yes
---
arrow-csv/Cargo.toml | 2 +-
arrow-csv/src/lib.rs | 2 +-
arrow-csv/src/reader/mod.rs | 22 +++++++++++++++++-----
arrow-csv/src/writer.rs | 7 +++++--
4 files changed, 24 insertions(+), 9 deletions(-)
diff --git a/arrow-csv/Cargo.toml b/arrow-csv/Cargo.toml
index c44ec01ce3..f823226c21 100644
--- a/arrow-csv/Cargo.toml
+++ b/arrow-csv/Cargo.toml
@@ -25,7 +25,7 @@ authors = { workspace = true }
license = { workspace = true }
keywords = { workspace = true }
include = { workspace = true }
-edition = { workspace = true }
+edition = "2024"
rust-version = { workspace = true }
[lib]
diff --git a/arrow-csv/src/lib.rs b/arrow-csv/src/lib.rs
index a3552eda8a..fd2f0d3d98 100644
--- a/arrow-csv/src/lib.rs
+++ b/arrow-csv/src/lib.rs
@@ -27,9 +27,9 @@
pub mod reader;
pub mod writer;
-pub use self::reader::infer_schema_from_files;
pub use self::reader::Reader;
pub use self::reader::ReaderBuilder;
+pub use self::reader::infer_schema_from_files;
pub use self::writer::Writer;
pub use self::writer::WriterBuilder;
use arrow_schema::ArrowError;
diff --git a/arrow-csv/src/reader/mod.rs b/arrow-csv/src/reader/mod.rs
index d1fc4eb350..0a72b57e85 100644
--- a/arrow-csv/src/reader/mod.rs
+++ b/arrow-csv/src/reader/mod.rs
@@ -128,7 +128,7 @@ mod records;
use arrow_array::builder::{NullBuilder, PrimitiveBuilder};
use arrow_array::types::*;
use arrow_array::*;
-use arrow_cast::parse::{parse_decimal, string_to_datetime, Parser};
+use arrow_cast::parse::{Parser, parse_decimal, string_to_datetime};
use arrow_schema::*;
use chrono::{TimeZone, Utc};
use csv::StringRecord;
@@ -1853,7 +1853,10 @@ mod tests {
let file_name = "test/data/various_invalid_types/invalid_float.csv";
let error = invalid_csv_helper(file_name);
- assert_eq!("Parser error: Error while parsing value '4.x4' as type
'Float32' for column 1 at line 4. Row data: '[4,4.x4,,false]'", error);
+ assert_eq!(
+ "Parser error: Error while parsing value '4.x4' as type 'Float32'
for column 1 at line 4. Row data: '[4,4.x4,,false]'",
+ error
+ );
}
#[test]
@@ -1861,7 +1864,10 @@ mod tests {
let file_name = "test/data/various_invalid_types/invalid_int.csv";
let error = invalid_csv_helper(file_name);
- assert_eq!("Parser error: Error while parsing value '2.3' as type
'UInt64' for column 0 at line 2. Row data: '[2.3,2.2,2.22,false]'", error);
+ assert_eq!(
+ "Parser error: Error while parsing value '2.3' as type 'UInt64'
for column 0 at line 2. Row data: '[2.3,2.2,2.22,false]'",
+ error
+ );
}
#[test]
@@ -1869,7 +1875,10 @@ mod tests {
let file_name = "test/data/various_invalid_types/invalid_bool.csv";
let error = invalid_csv_helper(file_name);
- assert_eq!("Parser error: Error while parsing value 'none' as type
'Boolean' for column 3 at line 2. Row data: '[2,2.2,2.22,none]'", error);
+ assert_eq!(
+ "Parser error: Error while parsing value 'none' as type 'Boolean'
for column 3 at line 2. Row data: '[2,2.2,2.22,none]'",
+ error
+ );
}
/// Infer the data type of a record
@@ -2697,7 +2706,10 @@ mod tests {
.infer_schema(&mut read, None);
assert!(result.is_err());
// Include line number in the error message to help locate and fix the
issue
- assert_eq!(result.err().unwrap().to_string(), "Csv error: Encountered
unequal lengths between records on CSV file. Expected 3 records, found 2
records at line 3");
+ assert_eq!(
+ result.err().unwrap().to_string(),
+ "Csv error: Encountered unequal lengths between records on CSV
file. Expected 3 records, found 2 records at line 3"
+ );
}
#[test]
diff --git a/arrow-csv/src/writer.rs b/arrow-csv/src/writer.rs
index e10943a6a9..3088c12c20 100644
--- a/arrow-csv/src/writer.rs
+++ b/arrow-csv/src/writer.rs
@@ -418,7 +418,7 @@ mod tests {
use crate::ReaderBuilder;
use arrow_array::builder::{
- BinaryBuilder, Decimal128Builder, Decimal256Builder, Decimal32Builder,
Decimal64Builder,
+ BinaryBuilder, Decimal32Builder, Decimal64Builder, Decimal128Builder,
Decimal256Builder,
FixedSizeBinaryBuilder, LargeBinaryBuilder,
};
use arrow_array::types::*;
@@ -717,7 +717,10 @@ sed do eiusmod
tempor,-556132.25,1,,2019-04-18T02:45:55.555,23:46:03,foo
for batch in batches {
let err = writer.write(batch).unwrap_err().to_string();
- assert_eq!(err, "Csv error: Error processing row 2, col 2: Cast
error: Failed to convert 1926632005177685347 to temporal for Date64")
+ assert_eq!(
+ err,
+ "Csv error: Error processing row 2, col 2: Cast error: Failed
to convert 1926632005177685347 to temporal for Date64"
+ )
}
drop(writer);
}
