From a7bd5ec927485e0ee27846fe7c182d362b11658b Mon Sep 17 00:00:00 2001
From: Sammy Sidhu <samster25@users.noreply.github.com>
Date: Thu, 16 Nov 2023 21:38:01 -0800
Subject: [PATCH] [BUG] raise error if non valid parquet file (less than
 parquet footer size) (#1628)

---
 src/daft-parquet/src/lib.rs           |  6 ++++++
 src/daft-parquet/src/metadata.rs      |  6 ++++++
 src/daft-parquet/src/stream_reader.rs | 14 ++++++++++++++
 tests/table/table_io/test_parquet.py  | 12 ++++++++++++
 4 files changed, 38 insertions(+)

diff --git a/src/daft-parquet/src/lib.rs b/src/daft-parquet/src/lib.rs
index c2e078afa4..ff0dc57691 100644
--- a/src/daft-parquet/src/lib.rs
+++ b/src/daft-parquet/src/lib.rs
@@ -87,6 +87,12 @@ pub enum Error {
         footer
     ))]
     InvalidParquetFile { path: String, footer: Vec<u8> },
+    #[snafu(display(
+        "File: {} is not a valid parquet file and is only {} bytes, smaller than the minimum size of 12 bytes",
+        path,
+        file_size
+    ))]
+    FileTooSmall { path: String, file_size: usize },
     #[snafu(display(
         "File: {} has a footer size: {} greater than the file size: {}",
         path,
diff --git a/src/daft-parquet/src/metadata.rs b/src/daft-parquet/src/metadata.rs
index b2afc79d27..f0a940b147 100644
--- a/src/daft-parquet/src/metadata.rs
+++ b/src/daft-parquet/src/metadata.rs
@@ -20,6 +20,12 @@ pub(crate) async fn read_parquet_metadata(
 ) -> super::Result<FileMetaData> {
     const FOOTER_SIZE: usize = 8;
     const PARQUET_MAGIC: [u8; 4] = [b'P', b'A', b'R', b'1'];
+    if size < 12 {
+        return Err(Error::FileTooSmall {
+            path: uri.into(),
+            file_size: size,
+        });
+    }
 
     /// The number of bytes read at the end of the parquet file on first read
     const DEFAULT_FOOTER_READ_SIZE: usize = 128 * 1024;
diff --git a/src/daft-parquet/src/stream_reader.rs b/src/daft-parquet/src/stream_reader.rs
index 9f65d19210..e0e7261d4c 100644
--- a/src/daft-parquet/src/stream_reader.rs
+++ b/src/daft-parquet/src/stream_reader.rs
@@ -64,6 +64,20 @@ pub(crate) fn local_parquet_read_into_arrow(
     let mut reader = File::open(uri).with_context(|_| super::InternalIOSnafu {
         path: uri.to_string(),
     })?;
+    let size = reader
+        .metadata()
+        .with_context(|_| super::InternalIOSnafu {
+            path: uri.to_string(),
+        })?
+        .len();
+
+    if size < 12 {
+        return Err(super::Error::FileTooSmall {
+            path: uri.into(),
+            file_size: size as usize,
+        });
+    }
+
     let metadata = read::read_metadata(&mut reader).with_context(|_| {
         super::UnableToParseMetadataFromLocalFileSnafu {
             path: uri.to_string(),
diff --git a/tests/table/table_io/test_parquet.py b/tests/table/table_io/test_parquet.py
index 84dc5c12c2..2becf7bd76 100644
--- a/tests/table/table_io/test_parquet.py
+++ b/tests/table/table_io/test_parquet.py
@@ -314,3 +314,15 @@ def test_parquet_read_int96_timestamps_schema_inference(coerce_to, store_schema)
     ) as f:
         schema = Schema.from_parquet(f, coerce_int96_timestamp_unit=coerce_to)
         assert schema == expected, f"Expected:\n{expected}\n\nReceived:\n{schema}"
+
+
+@pytest.mark.parametrize("n_bytes", [0, 1, 2, 7])
+def test_read_empty_parquet_file(tmpdir, n_bytes):
+
+    tmpdir = pathlib.Path(tmpdir)
+    file_path = tmpdir / "file.parquet"
+    with open(file_path, "wb") as f:
+        for _ in range(n_bytes):
+            f.write(b"0")
+    with pytest.raises(ValueError, match="smaller than the minimum size of 12 bytes"):
+        Table.read_parquet(file_path.as_posix())