fix: Prevent early EOF error in reader.read (#593)

* fix: Python reader should return truncated result when requesting beyond EOF

* test: Add comparison to stdlib BytesIO

Author: nvictus

obstore/src/buffered.rs

@@ -192,7 +192,10 @@ async fn read(reader: Arc<Mutex<BufReader>>, size: Option<usize>) -> PyResult<Py
     let mut reader = reader.lock().await;
     if let Some(size) = size {
         let mut buf = vec![0; size as _];
-        reader.read_exact(&mut buf).await?;
+        let n = reader.read(&mut buf).await?;
+        if n < buf.len() {
+            buf.truncate(n);
+        }
         Ok(Bytes::from(buf).into())
     } else {
         let mut buf = Vec::new();

tests/test_buffered.py

@@ -1,3 +1,5 @@
+from io import BytesIO
+
 import pytest
 
 import obstore as obs
@@ -77,3 +79,36 @@ async def test_writable_file_async():
     resp = await obs.get_async(store, path)
     retour = await resp.bytes_async()
     assert retour == line * 50
+
+
+def test_read_past_eof_sync():
+    store = MemoryStore()
+
+    data = b"Hello, World!"
+    path = "greeting.txt"
+    obs.put(store, path, data)
+
+    file = obs.open_reader(store, path)
+    buffer = file.read(20)
+    assert memoryview(data) == memoryview(buffer)
+
+    buf = BytesIO(data)
+    expected = buf.read(20)
+    assert memoryview(expected) == memoryview(buffer)
+
+
+@pytest.mark.asyncio
+async def test_read_past_eof_async():
+    store = MemoryStore()
+
+    data = b"Hello, World!"
+    path = "greeting.txt"
+    await obs.put_async(store, path, data)
+
+    file = await obs.open_reader_async(store, path)
+    buffer = await file.read(20)
+    assert memoryview(data) == memoryview(buffer)
+
+    buf = BytesIO(data)
+    expected = buf.read(20)
+    assert memoryview(expected) == memoryview(buffer)