@@ -165,7 +165,7 @@ use super::ByteArrayType;
165165pub struct GenericByteViewArray < T : ByteViewType + ?Sized > {
166166 data_type : DataType ,
167167 views : ScalarBuffer < u128 > ,
168- buffers : Vec < Buffer > ,
168+ buffers : Arc < [ Buffer ] > ,
169169 phantom : PhantomData < T > ,
170170 nulls : Option < NullBuffer > ,
171171}
@@ -188,7 +188,10 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
188188/// # Panics
189189///
190190/// Panics if [`GenericByteViewArray::try_new`] returns an error
191- pub fn new ( views : ScalarBuffer < u128 > , buffers : Vec < Buffer > , nulls : Option < NullBuffer > ) -> Self {
191+ pub fn new < U > ( views : ScalarBuffer < u128 > , buffers : U , nulls : Option < NullBuffer > ) -> Self
192+ where
193+ U : Into < Arc < [ Buffer ] > > ,
194+ {
192195 Self :: try_new ( views, buffers, nulls) . unwrap ( )
193196 }
194197
@@ -198,11 +201,16 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
198201///
199202/// * `views.len() != nulls.len()`
200203/// * [ByteViewType::validate] fails
201- pub fn try_new (
204+ pub fn try_new < U > (
202205 views : ScalarBuffer < u128 > ,
203- buffers : Vec < Buffer > ,
206+ buffers : U ,
204207 nulls : Option < NullBuffer > ,
205- ) -> Result < Self , ArrowError > {
208+ ) -> Result < Self , ArrowError >
209+ where
210+ U : Into < Arc < [ Buffer ] > > ,
211+ {
212+ let buffers: Arc < [ Buffer ] > = buffers. into ( ) ;
213+
206214 T :: validate ( & views, & buffers) ?;
207215
208216 if let Some ( n) = nulls. as_ref ( ) {
@@ -230,11 +238,14 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
230238/// # Safety
231239///
232240/// Safe if [`Self::try_new`] would not error
233- pub unsafe fn new_unchecked (
241+ pub unsafe fn new_unchecked < U > (
234242 views : ScalarBuffer < u128 > ,
235- buffers : Vec < Buffer > ,
243+ buffers : U ,
236244 nulls : Option < NullBuffer > ,
237- ) -> Self {
245+ ) -> Self
246+ where
247+ U : Into < Arc < [ Buffer ] > > ,
248+ {
238249 if cfg ! ( feature = "force_validate" ) {
239250 return Self :: new ( views, buffers, nulls) ;
240251 }
@@ -243,7 +254,7 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
243254 data_type : T :: DATA_TYPE ,
244255 phantom : Default :: default ( ) ,
245256 views,
246- buffers,
257+ buffers : buffers . into ( ) ,
247258 nulls,
248259 }
249260 }
@@ -253,7 +264,7 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
253264 Self {
254265 data_type : T :: DATA_TYPE ,
255266 views : vec ! [ 0 ; len] . into ( ) ,
256- buffers : vec ! [ ] ,
267+ buffers : vec ! [ ] . into ( ) ,
257268 nulls : Some ( NullBuffer :: new_null ( len) ) ,
258269 phantom : Default :: default ( ) ,
259270 }
@@ -279,7 +290,7 @@ impl<T: ByteViewType + ?Sized> GenericByteViewArray<T> {
279290 }
280291
281292 /// Deconstruct this array into its constituent parts
282- pub fn into_parts ( self ) -> ( ScalarBuffer < u128 > , Vec < Buffer > , Option < NullBuffer > ) {
293+ pub fn into_parts ( self ) -> ( ScalarBuffer < u128 > , Arc < [ Buffer ] > , Option < NullBuffer > ) {
283294 ( self . views , self . buffers , self . nulls )
284295 }
285296
@@ -854,6 +865,8 @@ impl<T: ByteViewType + ?Sized> Debug for GenericByteViewArray<T> {
854865 }
855866}
856867
868+ impl < T : ByteViewType + ?Sized > super :: private:: Sealed for GenericByteViewArray < T > { }
869+
857870impl < T : ByteViewType + ?Sized > Array for GenericByteViewArray < T > {
858871 fn as_any ( & self ) -> & dyn Any {
859872 self
@@ -885,8 +898,21 @@ impl<T: ByteViewType + ?Sized> Array for GenericByteViewArray<T> {
885898
886899 fn shrink_to_fit ( & mut self ) {
887900 self . views . shrink_to_fit ( ) ;
888- self . buffers . iter_mut ( ) . for_each ( |b| b. shrink_to_fit ( ) ) ;
889- self . buffers . shrink_to_fit ( ) ;
901+
902+ // The goal of `shrink_to_fit` is to minimize the space used by any of
903+ // its allocations. The use of `Arc::get_mut` over `Arc::make_mut` is
904+ // because if the reference count is greater than 1, `Arc::make_mut`
905+ // will first clone its contents. So, any large allocations will first
906+ // be cloned before being shrunk, leaving the pre-cloned allocations
907+ // intact, before adding the extra (used) space of the new clones.
908+ if let Some ( buffers) = Arc :: get_mut ( & mut self . buffers ) {
909+ buffers. iter_mut ( ) . for_each ( |b| b. shrink_to_fit ( ) ) ;
910+ }
911+
912+ // With the assumption that this is a best-effort function, no attempt
913+ // is made to shrink `self.buffers`, which it can't because it's type
914+ // does not expose a `shrink_to_fit` method.
915+
890916 if let Some ( nulls) = & mut self . nulls {
891917 nulls. shrink_to_fit ( ) ;
892918 }
@@ -944,7 +970,7 @@ impl<T: ByteViewType + ?Sized> From<ArrayData> for GenericByteViewArray<T> {
944970 fn from ( value : ArrayData ) -> Self {
945971 let views = value. buffers ( ) [ 0 ] . clone ( ) ;
946972 let views = ScalarBuffer :: new ( views, value. offset ( ) , value. len ( ) ) ;
947- let buffers = value. buffers ( ) [ 1 ..] . to_vec ( ) ;
973+ let buffers = value. buffers ( ) [ 1 ..] . to_vec ( ) . into ( ) ;
948974 Self {
949975 data_type : T :: DATA_TYPE ,
950976 views,
@@ -1012,12 +1038,15 @@ where
10121038}
10131039
10141040impl < T : ByteViewType + ?Sized > From < GenericByteViewArray < T > > for ArrayData {
1015- fn from ( mut array : GenericByteViewArray < T > ) -> Self {
1041+ fn from ( array : GenericByteViewArray < T > ) -> Self {
10161042 let len = array. len ( ) ;
1017- array. buffers . insert ( 0 , array. views . into_inner ( ) ) ;
1043+
1044+ let mut buffers = array. buffers . to_vec ( ) ;
1045+ buffers. insert ( 0 , array. views . into_inner ( ) ) ;
1046+
10181047 let builder = ArrayDataBuilder :: new ( T :: DATA_TYPE )
10191048 . len ( len)
1020- . buffers ( array . buffers )
1049+ . buffers ( buffers)
10211050 . nulls ( array. nulls ) ;
10221051
10231052 unsafe { builder. build_unchecked ( ) }
0 commit comments