chore: include path when read invalid parquet file. (#18343)

Author: youngsofun

src/common/storage/src/parquet.rs

@@ -116,7 +116,7 @@ pub async fn read_metadata_async(
     let buffer_len = buffer.len();
 
     let map_err =
-        |e: ParquetError| ErrorCode::BadBytes(format!("Invalid Parquet File {path}: {e}",));
+        |e: ParquetError| ErrorCode::BadBytes(format!("Invalid Parquet file '{path}': {e}",));
     let footer_tail = ParquetMetaDataReader::decode_footer_tail(
         &buffer[(buffer_len - FOOTER_SIZE as usize)..]
             .try_into()
@@ -160,7 +160,7 @@ pub fn read_metadata_sync(
     check_footer_size(file_size, path)?;
 
     let map_err =
-        |e: ParquetError| ErrorCode::BadBytes(format!("Invalid Parquet File {path}: {e}",));
+        |e: ParquetError| ErrorCode::BadBytes(format!("Invalid Parquet file '{path}': {e}",));
     // read and cache up to DEFAULT_FOOTER_READ_SIZE bytes from the end and process the footer
     let default_end_len = DEFAULT_FOOTER_READ_SIZE.min(file_size);
     let buffer = blocking
@@ -198,7 +198,7 @@ pub fn read_metadata_sync(
 fn check_footer_size(file_size: u64, path: &str) -> Result<()> {
     if file_size < FOOTER_SIZE {
         Err(ErrorCode::BadBytes(format!(
-            "Invalid Parquet file {path}. Size is smaller than footer."
+            "Not a parquet file ({path}): only {file_size} bytes."
         )))
     } else {
         Ok(())
@@ -209,7 +209,7 @@ fn check_footer_size(file_size: u64, path: &str) -> Result<()> {
 fn check_meta_size(file_size: u64, metadata_len: u64, path: &str) -> Result<()> {
     if metadata_len + FOOTER_SIZE > file_size {
         Err(ErrorCode::BadBytes(format!(
-            "Invalid Parquet file {path}. Reported metadata length of {} + {} byte footer, but file is only {} bytes",
+            "Invalid Parquet file '{path}': Reported metadata length of {} + {} byte footer, but file is only {} bytes",
             metadata_len, FOOTER_SIZE, file_size
         )))
     } else {

src/query/storages/parquet/src/parquet_reader/reader/full_reader.rs

@@ -179,12 +179,11 @@ impl ParquetWholeFileReader {
 
     /// Read a [`DataBlock`] from bytes.
     pub fn read_blocks_from_binary(&self, bytes: Bytes, path: &str) -> Result<DataBlockIterator> {
-        let mut builder = ParquetRecordBatchReaderBuilder::try_new_with_options(
-            bytes,
-            ArrowReaderOptions::new(),
-        )?
-        .with_projection(self.projection.clone())
-        .with_batch_size(self.batch_size);
+        let mut builder =
+            ParquetRecordBatchReaderBuilder::try_new_with_options(bytes, ArrowReaderOptions::new())
+                .map_err(|e| format!("Invalid Parquet file '{path}': {e}"))?
+                .with_projection(self.projection.clone())
+                .with_batch_size(self.batch_size);
 
         // Prune row groups.
         let file_meta = builder.metadata().clone();

tests/sqllogictests/suites/stage/formats/parquet/select_parquet.test

@@ -59,3 +59,12 @@ query I
 select count() from @data/parquet/no-stats.parquet (pattern => '') where line_item_usage_start_date < '2024-09-11 06:00:00'
 ----
 0
+
+query error gen
+settings (parquet_fast_read_bytes=0) select * from @data/parquet/diff_schema/ (files=>('f1.parquet', 'gen.py'));
+
+query error gen
+settings select * from @data/parquet/diff_schema/ (files=>('f1.parquet', 'gen.py'));
+
+query error gen
+settings select * from @data/parquet/diff_schema/ (files=>('gen.py', ''f1.parquet'));