Prefer ParquetError::oos to ParquetError::OutOfSpec (#17314)

pola-rs · Jul 1, 2024 · 71e43b6 · 71e43b6
1 parent 227b350
commit 71e43b6
Show file tree

Hide file tree

Showing 7 changed files with 13 additions and 15 deletions.
diff --git a/crates/polars-parquet/src/arrow/write/row_group.rs b/crates/polars-parquet/src/arrow/write/row_group.rs
@@ -42,7 +42,7 @@ pub fn row_group_iter<A: AsRef<dyn Array> + 'static + Send + Sync>(
                         let pages = DynIter::new(
                             pages
                                 .into_iter()
-                                .map(|x| x.map_err(|e| ParquetError::OutOfSpec(e.to_string()))),
+                                .map(|x| x.map_err(|e| ParquetError::oos(e.to_string()))),
                         );
 
                         let compressed_pages = Compressor::new(pages, options.compression, vec![])

diff --git a/crates/polars-parquet/src/parquet/compression.rs b/crates/polars-parquet/src/parquet/compression.rs
@@ -168,9 +168,7 @@ pub fn decompress(
 
             let len = decompress_len(input_buf)?;
             if len > output_buf.len() {
-                return Err(ParquetError::OutOfSpec(String::from(
-                    "snappy header out of spec",
-                )));
+                return Err(ParquetError::oos("snappy header out of spec"));
             }
             Decoder::new()
                 .decompress(input_buf, output_buf)

diff --git a/crates/polars-parquet/src/parquet/error.rs b/crates/polars-parquet/src/parquet/error.rs
@@ -34,6 +34,7 @@ pub enum ParquetError {
 }
 
 impl ParquetError {
+    /// Create an OutOfSpec error from any Into<String>
     pub(crate) fn oos<I: Into<String>>(message: I) -> Self {
         Self::OutOfSpec(message.into())
     }

diff --git a/crates/polars-parquet/src/parquet/read/compression.rs b/crates/polars-parquet/src/parquet/read/compression.rs
@@ -33,8 +33,8 @@ fn decompress_v2(
 
     if can_decompress {
         if offset > buffer.len() || offset > compressed.len() {
-            return Err(ParquetError::OutOfSpec(
-                "V2 Page Header reported incorrect offset to compressed data".to_string(),
+            return Err(ParquetError::oos(
+                "V2 Page Header reported incorrect offset to compressed data",
             ));
         }
 
@@ -43,8 +43,8 @@ fn decompress_v2(
         compression::decompress(compression, &compressed[offset..], &mut buffer[offset..])?;
     } else {
         if buffer.len() != compressed.len() {
-            return Err(ParquetError::OutOfSpec(
-                "V2 Page Header reported incorrect decompressed size".to_string(),
+            return Err(ParquetError::oos(
+                "V2 Page Header reported incorrect decompressed size",
             ));
         }
         buffer.copy_from_slice(compressed);

diff --git a/crates/polars-parquet/src/parquet/read/page/reader.rs b/crates/polars-parquet/src/parquet/read/page/reader.rs
@@ -206,7 +206,7 @@ pub(super) fn build_page<R: Read>(
 
     if bytes_read != read_size {
         return Err(ParquetError::oos(
-            "The page header reported the wrong page size".to_string(),
+            "The page header reported the wrong page size",
         ));
     }
 

diff --git a/crates/polars-parquet/src/parquet/read/page/stream.rs b/crates/polars-parquet/src/parquet/read/page/stream.rs
@@ -112,7 +112,7 @@ fn _get_page_stream<R: AsyncRead + Unpin + Send>(
 
             if bytes_read != read_size {
                 Err(ParquetError::oos(
-                    "The page header reported the wrong page size".to_string(),
+                    "The page header reported the wrong page size",
                 ))?
             }
 

diff --git a/crates/polars-parquet/src/parquet/schema/types/spec.rs b/crates/polars-parquet/src/parquet/schema/types/spec.rs
@@ -62,8 +62,7 @@ fn check_decimal_invariants(
         PhysicalType::ByteArray => {},
         _ => {
             return Err(ParquetError::oos(
-                "DECIMAL can only annotate INT32, INT64, BYTE_ARRAY and FIXED_LEN_BYTE_ARRAY"
-                    .to_string(),
+                "DECIMAL can only annotate INT32, INT64, BYTE_ARRAY and FIXED_LEN_BYTE_ARRAY",
             ))
         },
     };
@@ -111,14 +110,14 @@ pub fn check_converted_invariants(
         Interval => {
             if physical_type != &PhysicalType::FixedLenByteArray(12) {
                 return Err(ParquetError::oos(
-                    "INTERVAL can only annotate FIXED_LEN_BYTE_ARRAY(12)".to_string(),
+                    "INTERVAL can only annotate FIXED_LEN_BYTE_ARRAY(12)",
                 ));
             }
         },
         Enum => {
             if physical_type != &PhysicalType::ByteArray {
                 return Err(ParquetError::oos(
-                    "ENUM can only annotate BYTE_ARRAY fields".to_string(),
+                    "ENUM can only annotate BYTE_ARRAY fields",
                 ));
             }
         },
@@ -153,7 +152,7 @@ pub fn check_logical_invariants(
         (Time { unit, .. }, PhysicalType::Int64) => {
             if unit == TimeUnit::Milliseconds {
                 return Err(ParquetError::oos(
-                    "Cannot use millisecond unit on INT64 type".to_string(),
+                    "Cannot use millisecond unit on INT64 type",
                 ));
             }
         },