Simplified ArrowError

src/array/binary/mutable.rs

@@ -410,8 +410,8 @@ impl<O: Offset, T: AsRef<[u8]>> TryPush<Option<T>> for MutableBinaryArray<O> {
             Some(value) => {
                 let bytes = value.as_ref();
 
-                let size = O::from_usize(self.values.len() + bytes.len())
-                    .ok_or(ArrowError::KeyOverflowError)?;
+                let size =
+                    O::from_usize(self.values.len() + bytes.len()).ok_or(ArrowError::Overflow)?;
 
                 self.values.extend_from_slice(bytes);
 

src/array/dictionary/mutable.rs

@@ -70,7 +70,7 @@ impl<K: DictionaryKey, M: MutableArray> MutableDictionaryArray<K, M> {
                 Ok(false)
             }
             None => {
-                let key = K::from_usize(self.map.len()).ok_or(ArrowError::KeyOverflowError)?;
+                let key = K::from_usize(self.map.len()).ok_or(ArrowError::Overflow)?;
                 self.map.insert(hash, key);
                 self.keys.push(Some(key));
                 Ok(true)

src/array/fixed_size_list/mutable.rs

@@ -58,7 +58,7 @@ impl<M: MutableArray> MutableFixedSizeListArray<M> {
     #[inline]
     fn try_push_valid(&mut self) -> Result<()> {
         if self.values.len() % self.size != 0 {
-            return Err(ArrowError::KeyOverflowError);
+            return Err(ArrowError::Overflow);
         };
         if let Some(validity) = &mut self.validity {
             validity.push(true)

src/array/list/mutable.rs

@@ -130,7 +130,7 @@ impl<O: Offset, M: MutableArray> MutableListArray<O, M> {
     /// This is a relatively low level function, prefer `try_push` when you can.
     pub fn try_push_valid(&mut self) -> Result<()> {
         let size = self.values.len();
-        let size = O::from_usize(size).ok_or(ArrowError::KeyOverflowError)?; // todo: make this error
+        let size = O::from_usize(size).ok_or(ArrowError::Overflow)?;
         assert!(size >= *self.offsets.last().unwrap());
 
         self.offsets.push(size);

src/array/utf8/mutable.rs

@@ -448,7 +448,7 @@ impl<O: Offset, T: AsRef<str>> TryPush<Option<T>> for MutableUtf8Array<O> {
                 let bytes = value.as_ref().as_bytes();
                 self.values.extend_from_slice(bytes);
 
-                let size = O::from_usize(self.values.len()).ok_or(ArrowError::KeyOverflowError)?;
+                let size = O::from_usize(self.values.len()).ok_or(ArrowError::Overflow)?;
 
                 self.offsets.push(size);
 

src/compute/cast/dictionary_to.rs

@@ -13,7 +13,7 @@ macro_rules! key_cast {
         // Failure to cast keys (because they don't fit in the
         // target type) results in NULL values;
         if cast_keys.null_count() > $keys.null_count() {
-            return Err(ArrowError::KeyOverflowError);
+            return Err(ArrowError::Overflow);
         }
         Ok(Box::new(DictionaryArray::<$to_type>::from_data(
             cast_keys, $values,
@@ -74,7 +74,7 @@ where
     let casted_keys = primitive_to_primitive::<K1, K2>(keys, &K2::DATA_TYPE);
 
     if casted_keys.null_count() > keys.null_count() {
-        Err(ArrowError::KeyOverflowError)
+        Err(ArrowError::Overflow)
     } else {
         Ok(DictionaryArray::from_data(casted_keys, values.clone()))
     }
@@ -94,7 +94,7 @@ where
     let casted_keys = primitive_as_primitive::<K1, K2>(keys, &K2::DATA_TYPE);
 
     if casted_keys.null_count() > keys.null_count() {
-        Err(ArrowError::KeyOverflowError)
+        Err(ArrowError::Overflow)
     } else {
         Ok(DictionaryArray::from_data(casted_keys, values.clone()))
     }

src/datatypes/field.rs

@@ -177,7 +177,7 @@ impl Field {
                 for (key, from_value) in from_metadata {
                     if let Some(self_value) = self_metadata.get(key) {
                         if self_value != from_value {
-                            return Err(ArrowError::Schema(format!(
+                            return Err(ArrowError::InvalidArgumentError(format!(
                                 "Fail to merge field due to conflicting metadata data value for key {}", key),
                             ));
                         }
@@ -193,12 +193,12 @@ impl Field {
             _ => {}
         }
         if from.dict_id != self.dict_id {
-            return Err(ArrowError::Schema(
+            return Err(ArrowError::InvalidArgumentError(
                 "Fail to merge schema Field due to conflicting dict_id".to_string(),
             ));
         }
         if from.dict_is_ordered != self.dict_is_ordered {
-            return Err(ArrowError::Schema(
+            return Err(ArrowError::InvalidArgumentError(
                 "Fail to merge schema Field due to conflicting dict_is_ordered".to_string(),
             ));
         }
@@ -220,7 +220,7 @@ impl Field {
                     }
                 }
                 _ => {
-                    return Err(ArrowError::Schema(
+                    return Err(ArrowError::InvalidArgumentError(
                         "Fail to merge schema Field due to conflicting datatype".to_string(),
                     ));
                 }
@@ -241,7 +241,7 @@ impl Field {
                     }
                 }
                 _ => {
-                    return Err(ArrowError::Schema(
+                    return Err(ArrowError::InvalidArgumentError(
                         "Fail to merge schema Field due to conflicting datatype".to_string(),
                     ));
                 }
@@ -279,7 +279,7 @@ impl Field {
             | DataType::Map(_, _)
             | DataType::Decimal(_, _) => {
                 if self.data_type != from.data_type {
-                    return Err(ArrowError::Schema(
+                    return Err(ArrowError::InvalidArgumentError(
                         "Fail to merge schema Field due to conflicting datatype".to_string(),
                     ));
                 }

src/datatypes/schema.rs

@@ -120,7 +120,7 @@ impl Schema {
                     // merge metadata
                     if let Some(old_val) = merged.metadata.get(&key) {
                         if old_val != &value {
-                            return Err(ArrowError::Schema(
+                            return Err(ArrowError::InvalidArgumentError(
                                 "Fail to merge schema due to conflicting metadata.".to_string(),
                             ));
                         }

src/error.rs

@@ -5,31 +5,23 @@ use std::error::Error;
 
 /// Enum with all errors in this crate.
 #[derive(Debug)]
+#[non_exhaustive]
 pub enum ArrowError {
     /// Returned when functionality is not yet available.
     NotYetImplemented(String),
-    /// Triggered by an external error, such as CSV, serde, chrono.
+    /// Wrapper for an error triggered by a dependency
     External(String, Box<dyn Error + Send + Sync>),
-    /// Error associated with incompatible schemas.
-    Schema(String),
-    /// Errors associated with IO
+    /// Wrapper for IO errors
     Io(std::io::Error),
     /// When an invalid argument is passed to a function.
     InvalidArgumentError(String),
-    /// Error during import or export to/from C Data Interface
-    Ffi(String),
-    /// Error during import or export to/from IPC
-    Ipc(String),
     /// Error during import or export to/from a format
     ExternalFormat(String),
     /// Whenever pushing to a container fails because it does not support more entries.
-    /// (e.g. maximum size of the keys of a dictionary overflowed)
-    KeyOverflowError,
-    /// Error during arithmetic operation. Normally returned
-    /// during checked operations
-    ArithmeticError(String),
-    /// Any other error.
-    Other(String),
+    /// The solution is usually to use a higher-capacity container-backing type.
+    Overflow,
+    /// Whenever incoming data from the C data interface, IPC or Flight does not fulfil the Arrow specification.
+    OutOfSpec(String),
 }
 
 impl ArrowError {
@@ -66,27 +58,17 @@ impl Display for ArrowError {
             ArrowError::External(message, source) => {
                 write!(f, "External error{}: {}", message, &source)
             }
-            ArrowError::Schema(desc) => write!(f, "Schema error: {}", desc),
             ArrowError::Io(desc) => write!(f, "Io error: {}", desc),
             ArrowError::InvalidArgumentError(desc) => {
                 write!(f, "Invalid argument error: {}", desc)
             }
-            ArrowError::Ffi(desc) => {
-                write!(f, "FFI error: {}", desc)
-            }
-            ArrowError::Ipc(desc) => {
-                write!(f, "IPC error: {}", desc)
-            }
             ArrowError::ExternalFormat(desc) => {
                 write!(f, "External format error: {}", desc)
             }
-            ArrowError::KeyOverflowError => {
-                write!(f, "Dictionary key bigger than the key type")
-            }
-            ArrowError::ArithmeticError(desc) => {
-                write!(f, "Arithmetic error: {}", desc)
+            ArrowError::Overflow => {
+                write!(f, "Operation overflew the backing container.")
             }
-            ArrowError::Other(message) => {
+            ArrowError::OutOfSpec(message) => {
                 write!(f, "{}", message)
             }
         }

src/ffi/ffi.rs

@@ -184,7 +184,9 @@ unsafe fn create_buffer<T: NativeType>(
     index: usize,
 ) -> Result<Buffer<T>> {
     if array.buffers.is_null() {
-        return Err(ArrowError::Ffi("The array buffers are null".to_string()));
+        return Err(ArrowError::OutOfSpec(
+            "The array buffers are null".to_string(),
+        ));
     }
 
     let buffers = array.buffers as *mut *const u8;
@@ -197,7 +199,9 @@ unsafe fn create_buffer<T: NativeType>(
     let offset = buffer_offset(array, data_type, index);
     let bytes = ptr
         .map(|ptr| Bytes::new(ptr, len, deallocation))
-        .ok_or_else(|| ArrowError::Ffi(format!("The buffer at position {} is null", index)))?;
+        .ok_or_else(|| {
+            ArrowError::OutOfSpec(format!("The buffer at position {} is null", index))
+        })?;
 
     Ok(Buffer::from_bytes(bytes).slice(offset, len - offset))
 }
@@ -215,7 +219,9 @@ unsafe fn create_bitmap(
     index: usize,
 ) -> Result<Bitmap> {
     if array.buffers.is_null() {
-        return Err(ArrowError::Ffi("The array buffers are null".to_string()));
+        return Err(ArrowError::OutOfSpec(
+            "The array buffers are null".to_string(),
+        ));
     }
     let len = array.length as usize;
     let offset = array.offset as usize;
@@ -229,7 +235,7 @@ unsafe fn create_bitmap(
     let bytes = ptr
         .map(|ptr| Bytes::new(ptr, bytes_len, deallocation))
         .ok_or_else(|| {
-            ArrowError::Ffi(format!(
+            ArrowError::OutOfSpec(format!(
                 "The buffer {} is a null pointer and cannot be interpreted as a bitmap",
                 index
             ))

src/ffi/schema.rs

@@ -243,7 +243,7 @@ fn to_integer_type(format: &str) -> Result<IntegerType> {
         "l" => Int64,
         "L" => UInt64,
         _ => {
-            return Err(ArrowError::Ffi(
+            return Err(ArrowError::OutOfSpec(
                 "Dictionary indices can only be integers".to_string(),
             ))
         }
@@ -312,36 +312,40 @@ unsafe fn to_data_type(schema: &Ffi_ArrowSchema) -> Result<DataType> {
             } else if parts.len() == 2 && parts[0] == "tsn" {
                 DataType::Timestamp(TimeUnit::Nanosecond, Some(parts[1].to_string()))
             } else if parts.len() == 2 && parts[0] == "w" {
-                let size = parts[1]
-                    .parse::<usize>()
-                    .map_err(|_| ArrowError::Ffi("size is not a valid integer".to_string()))?;
+                let size = parts[1].parse::<usize>().map_err(|_| {
+                    ArrowError::OutOfSpec("size is not a valid integer".to_string())
+                })?;
                 DataType::FixedSizeBinary(size)
             } else if parts.len() == 2 && parts[0] == "+w" {
-                let size = parts[1]
-                    .parse::<usize>()
-                    .map_err(|_| ArrowError::Ffi("size is not a valid integer".to_string()))?;
+                let size = parts[1].parse::<usize>().map_err(|_| {
+                    ArrowError::OutOfSpec("size is not a valid integer".to_string())
+                })?;
                 let child = to_field(schema.child(0))?;
                 DataType::FixedSizeList(Box::new(child), size)
             } else if parts.len() == 2 && parts[0] == "d" {
                 let parts = parts[1].split(',').collect::<Vec<_>>();
                 if parts.len() < 2 || parts.len() > 3 {
-                    return Err(ArrowError::Ffi(
+                    return Err(ArrowError::OutOfSpec(
                         "Decimal must contain 2 or 3 comma-separated values".to_string(),
                     ));
                 };
                 if parts.len() == 3 {
                     let bit_width = parts[0].parse::<usize>().map_err(|_| {
-                        ArrowError::Ffi("Decimal bit width is not a valid integer".to_string())
+                        ArrowError::OutOfSpec(
+                            "Decimal bit width is not a valid integer".to_string(),
+                        )
                     })?;
                     if bit_width != 128 {
-                        return Err(ArrowError::Ffi("Decimal256 is not supported".to_string()));
+                        return Err(ArrowError::OutOfSpec(
+                            "Decimal256 is not supported".to_string(),
+                        ));
                     }
                 }
                 let precision = parts[0].parse::<usize>().map_err(|_| {
-                    ArrowError::Ffi("Decimal precision is not a valid integer".to_string())
+                    ArrowError::OutOfSpec("Decimal precision is not a valid integer".to_string())
                 })?;
                 let scale = parts[1].parse::<usize>().map_err(|_| {
-                    ArrowError::Ffi("Decimal scale is not a valid integer".to_string())
+                    ArrowError::OutOfSpec("Decimal scale is not a valid integer".to_string())
                 })?;
                 DataType::Decimal(precision, scale)
             } else if !parts.is_empty() && ((parts[0] == "+us") || (parts[0] == "+ud")) {
@@ -351,7 +355,9 @@ unsafe fn to_data_type(schema: &Ffi_ArrowSchema) -> Result<DataType> {
                     .split(',')
                     .map(|x| {
                         x.parse::<i32>().map_err(|_| {
-                            ArrowError::Ffi("Union type id is not a valid integer".to_string())
+                            ArrowError::OutOfSpec(
+                                "Union type id is not a valid integer".to_string(),
+                            )
                         })
                     })
                     .collect::<Result<Vec<_>>>()?;
@@ -360,7 +366,7 @@ unsafe fn to_data_type(schema: &Ffi_ArrowSchema) -> Result<DataType> {
                     .collect::<Result<Vec<_>>>()?;
                 DataType::Union(fields, Some(type_ids), mode)
             } else {
-                return Err(ArrowError::Ffi(format!(
+                return Err(ArrowError::OutOfSpec(format!(
                     "The datatype \"{}\" is still not supported in Rust implementation",
                     other
                 )));
@@ -456,7 +462,7 @@ pub(super) fn get_field_child(field: &Field, index: usize) -> Result<Field> {
         (0, DataType::Map(field, _)) => Ok(field.as_ref().clone()),
         (index, DataType::Struct(fields)) => Ok(fields[index].clone()),
         (index, DataType::Union(fields, _, _)) => Ok(fields[index].clone()),
-        (child, data_type) => Err(ArrowError::Ffi(format!(
+        (child, data_type) => Err(ArrowError::OutOfSpec(format!(
             "Requested child {} to type {:?} that has no such child",
             child, data_type
         ))),

src/io/avro/read/decompress.rs

@@ -30,12 +30,12 @@ fn decompress_block(
         #[cfg(feature = "io_avro_compression")]
         Some(Compression::Snappy) => {
             let len = snap::raw::decompress_len(&block[..block.len() - 4])
-                .map_err(|_| ArrowError::Other("Failed to decompress snap".to_string()))?;
+                .map_err(|_| ArrowError::ExternalFormat("Failed to decompress snap".to_string()))?;
             decompress.clear();
             decompress.resize(len, 0);
             snap::raw::Decoder::new()
                 .decompress(&block[..block.len() - 4], decompress)
-                .map_err(|_| ArrowError::Other("Failed to decompress snap".to_string()))?;
+                .map_err(|_| ArrowError::ExternalFormat("Failed to decompress snap".to_string()))?;
             Ok(false)
         }
         #[cfg(not(feature = "io_avro_compression"))]

src/io/avro/read/nested.rs

@@ -43,7 +43,7 @@ impl<O: Offset> DynMutableListArray<O> {
     #[inline]
     pub fn try_push_valid(&mut self) -> Result<()> {
         let size = self.values.len();
-        let size = O::from_usize(size).ok_or(ArrowError::KeyOverflowError)?; // todo: make this error
+        let size = O::from_usize(size).ok_or(ArrowError::Overflow)?;
         assert!(size >= *self.offsets.last().unwrap());
 
         self.offsets.push(size);