adriangb commented on issue #6221:
URL: https://github.com/apache/arrow-rs/issues/6221#issuecomment-2278595540
Using the file IPC format I get a write error:
```rust
use std::sync::Arc;
use arrow::array::{DictionaryArray, Int32Array, Int8Array, LargeStringArray,
RecordBatch, StringArray};
use arrow_ipc::{reader::FileReader, writer::{FileWriter, IpcWriteOptions}};
use arrow_schema::{DataType, Field, Schema};
use bytes::{BytesMut, BufMut};
#[tokio::main]
async fn main() {
let schema = Arc::new(
Schema::new(
vec![
Field::new("a",
DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::LargeUtf8)),
false),
Field::new("b",
DataType::Dictionary(Box::new(DataType::Int8), Box::new(DataType::Utf8)),
false),
]
)
);
let a_array = DictionaryArray::new(
Int32Array::from(vec![Some(0)]),
Arc::new(LargeStringArray::from(vec![Some("hello")])),
);
let b_array = DictionaryArray::new(
Int8Array::from(vec![Some(0)]),
Arc::new(StringArray::from(vec![Some("world")])),
);
let batch = RecordBatch::try_new(
schema.clone(),
vec![
Arc::new(a_array),
Arc::new(b_array),
],
).unwrap();
for indices in vec![vec![0], vec![1], vec![0, 1]] {
println!("indices: {:?}", indices);
let batch = batch.project(&indices).unwrap();
let options = IpcWriteOptions::default();
let schema = Arc::unwrap_or_clone(batch.schema());
let mut written_bytes: bytes::buf::Writer<BytesMut> =
BytesMut::new().writer();
{
let mut writer = FileWriter::try_new_with_options(&mut
written_bytes, &schema, options).unwrap();
writer.write(&batch).unwrap();
writer.finish().unwrap();
}
let written_bytes = written_bytes.into_inner().freeze();
let cursor = std::io::Cursor::new(written_bytes);
let reader = FileReader::try_new(cursor, None).unwrap();
reader.collect::<Result<Vec<_>, _>>().unwrap();
}
}
```
```
indices: [0]
indices: [1]
indices: [0, 1]
thread 'main' panicked at src/bin/demo.rs:45:34:
called `Result::unwrap()` on an `Err` value:
InvalidArgumentError("Dictionary replacement detected when writing IPC file
format. Arrow IPC files only support a single dictionary for a given field
across all batches.")
```
Reading / writing the same data to Parquet works without issues:
```rust
use std::sync::Arc;
use arrow::array::{DictionaryArray, Int32Array, Int8Array, LargeStringArray,
RecordBatch, StringArray};
use arrow_schema::{DataType, Field, Schema};
use bytes::{BytesMut, BufMut};
use parquet::arrow::{arrow_reader::ArrowReaderBuilder,
arrow_writer::ArrowWriterOptions, ArrowWriter};
#[tokio::main]
async fn main() {
let schema = Arc::new(
Schema::new(
vec![
Field::new("a",
DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::LargeUtf8)),
false),
Field::new("b",
DataType::Dictionary(Box::new(DataType::Int8), Box::new(DataType::Utf8)),
false),
]
)
);
let a_array = DictionaryArray::new(
Int32Array::from(vec![Some(0)]),
Arc::new(LargeStringArray::from(vec![Some("hello")])),
);
let b_array = DictionaryArray::new(
Int8Array::from(vec![Some(0)]),
Arc::new(StringArray::from(vec![Some("world")])),
);
let batch = RecordBatch::try_new(
schema.clone(),
vec![
Arc::new(a_array),
Arc::new(b_array),
],
).unwrap();
for indices in vec![vec![0], vec![1], vec![0, 1]] {
println!("indices: {:?}", indices);
let batch = batch.project(&indices).unwrap();
let options = ArrowWriterOptions::default();
let schema = batch.schema();
let mut written_bytes: bytes::buf::Writer<BytesMut> =
BytesMut::new().writer();
{
let mut writer = ArrowWriter::try_new_with_options(&mut
written_bytes, schema, options).unwrap();
writer.write(&batch).unwrap();
writer.finish().unwrap();
}
let written_bytes = written_bytes.into_inner().freeze();
let reader =
ArrowReaderBuilder::try_new(written_bytes).unwrap().build().unwrap();
reader.collect::<Result<Vec<_>, _>>().unwrap();
}
}
```
Let me know if I'm doing something wrong...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
