Add reprs for py classes and dataclass-esque constructor for WriteOptions

Author: Kimahriman

python/conftest.py

@@ -3,6 +3,7 @@
 import urllib.parse
 
 import fsspec
+from hdfs_native import Client
 import pytest
 
 from hdfs_native.fsspec import HdfsFileSystem
@@ -39,6 +40,11 @@ def minidfs():
         child.kill()
 
 
+@pytest.fixture(scope="module")
+def client(minidfs: str) -> Client:
+    return Client(minidfs)
+
+
 @pytest.fixture(scope="module")
 def fs(minidfs: str) -> HdfsFileSystem:
     url = urllib.parse.urlparse(minidfs)

python/hdfs_native/_internal.pyi

@@ -30,6 +30,15 @@ class WriteOptions:
     overwrite: bool
     create_parent: bool
 
+    def __init__(
+        self,
+        block_size: Optional[int] = None,
+        replication: Optional[int] = None,
+        permission: Optional[int] = None,
+        overwrite: Optional[bool] = None,
+        create_parent: Optional[bool] = None,
+    ): ...
+
 class RawFileReader:
     def file_length(self) -> int:
         """Returns the size of the file"""

python/src/lib.rs

@@ -50,6 +50,25 @@ impl From<FileStatus> for PyFileStatus {
     }
 }
 
+#[pymethods]
+impl PyFileStatus {
+    /// Return a dataclass-esque format for the repr
+    fn __repr__(&self) -> String {
+        format!("FileStatus(path='{}', length={}, isdir={}, permission={}, owner={}, group={}, modification_time={}, access_time={}, replication={}, blocksize={})",
+        self.path,
+        self.length,
+        self.isdir,
+        self.permission,
+        self.owner,
+        self.group,
+        self.modification_time,
+        self.access_time,
+        self.replication.map(|r| r.to_string()).unwrap_or("None".to_string()),
+        self.blocksize.map(|r| r.to_string()).unwrap_or("None".to_string())
+    )
+    }
+}
+
 #[pyclass(name = "FileStatusIter")]
 struct PyFileStatusIter {
     inner: ListStatusIterator,
@@ -96,6 +115,21 @@ impl From<ContentSummary> for PyContentSummary {
     }
 }
 
+#[pymethods]
+impl PyContentSummary {
+    /// Return a dataclass-esque format for the repr
+    fn __repr__(&self) -> String {
+        format!("ContentSummary(length={}, file_count={}, directory_count={}, quota={}, space_consumed={}, space_quota={})",
+            self.length,
+            self.file_count,
+            self.directory_count,
+            self.quota,
+            self.space_consumed,
+            self.space_quota,
+        )
+    }
+}
+
 #[pyclass]
 struct RawFileReader {
     inner: FileReader,
@@ -173,9 +207,42 @@ impl From<WriteOptions> for PyWriteOptions {
 #[pymethods]
 impl PyWriteOptions {
     #[new]
-    #[pyo3(signature = ())]
-    pub fn new() -> Self {
-        Self::from(WriteOptions::default())
+    pub fn new(
+        block_size: Option<u64>,
+        replication: Option<u32>,
+        permission: Option<u32>,
+        overwrite: Option<bool>,
+        create_parent: Option<bool>,
+    ) -> Self {
+        let mut write_options = WriteOptions::default();
+        if let Some(block_size) = block_size {
+            write_options = write_options.block_size(block_size);
+        }
+        if let Some(replication) = replication {
+            write_options = write_options.replication(replication);
+        }
+        if let Some(permission) = permission {
+            write_options = write_options.permission(permission);
+        }
+        if let Some(overwrite) = overwrite {
+            write_options = write_options.overwrite(overwrite);
+        }
+        if let Some(create_parent) = create_parent {
+            write_options = write_options.create_parent(create_parent);
+        }
+
+        PyWriteOptions::from(write_options)
+    }
+
+    /// Return a dataclass-esque format for the repr
+    fn __repr__(&self) -> String {
+        format!("WriteOptions(block_size={}, replication={}, permission={}, overwrite={}, create_parent={})",
+            self.block_size.map(|x| x.to_string()).unwrap_or("None".to_string()),
+            self.replication.map(|x| x.to_string()).unwrap_or("None".to_string()),
+            self.permission,
+            self.overwrite,
+            self.create_parent
+        )
     }
 }
 

python/tests/test_integration.py

@@ -3,9 +3,8 @@
 from hdfs_native import Client, WriteOptions
 
 
-def test_integration(minidfs: str):
-    client = Client(minidfs)
-    client.create("/testfile", WriteOptions()).close()
+def test_integration(client: Client):
+    client.create("/testfile").close()
     file_info = client.get_file_info("/testfile")
 
     assert file_info.path == "/testfile"
@@ -25,7 +24,7 @@ def test_integration(minidfs: str):
     file_list = list(client.list_status("/", False))
     assert len(file_list) == 0
 
-    with client.create("/testfile", WriteOptions()) as file:
+    with client.create("/testfile") as file:
         data = io.BytesIO()
 
         for i in range(0, 32 * 1024 * 1024):
@@ -101,3 +100,18 @@ def test_integration(minidfs: str):
     assert content_summary.length == 33 * 1024 * 1024 * 4
 
     client.delete("/testfile", False)
+
+
+def test_write_options(client: Client):
+    with client.create("/testfile") as file:
+        file.write(b"abcd")
+
+    client.create(
+        "/testfile",
+        WriteOptions(overwrite=True, permission=0o700, block_size=1024 * 1024),
+    ).close()
+
+    file_info = client.get_file_info("/testfile")
+    assert file_info.length == 0
+    assert file_info.permission == 0o700
+    assert file_info.blocksize == 1024 * 1024