Made len/len_proxy consistent with Offsets (jorgecarleitao#1434)

src/array/binary/mod.rs

@@ -82,7 +82,7 @@ impl<O: Offset> BinaryArray<O> {
 
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != offsets.len())
+            .map_or(false, |validity| validity.len() != offsets.len_proxy())
         {
             return Err(Error::oos(
                 "validity mask length must match the number of values",
@@ -127,7 +127,7 @@ impl<O: Offset> BinaryArray<O> {
     /// Returns the length of this array
     #[inline]
     pub fn len(&self) -> usize {
-        self.offsets.len()
+        self.offsets.len_proxy()
     }
 
     /// Returns the element at index `i`

src/array/list/mod.rs

@@ -44,7 +44,7 @@ impl<O: Offset> ListArray<O> {
 
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != offsets.len())
+            .map_or(false, |validity| validity.len() != offsets.len_proxy())
         {
             return Err(Error::oos(
                 "validity mask length must match the number of values",
@@ -138,7 +138,7 @@ impl<O: Offset> ListArray<O> {
     /// Returns the length of this array
     #[inline]
     pub fn len(&self) -> usize {
-        self.offsets.len()
+        self.offsets.len_proxy()
     }
 
     /// Returns the element at index `i`

src/array/map/mod.rs

@@ -59,7 +59,7 @@ impl MapArray {
 
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != offsets.len())
+            .map_or(false, |validity| validity.len() != offsets.len_proxy())
         {
             return Err(Error::oos(
                 "validity mask length must match the number of values",
@@ -154,7 +154,7 @@ impl MapArray {
     /// Returns the length of this array
     #[inline]
     pub fn len(&self) -> usize {
-        self.offsets.len()
+        self.offsets.len_proxy()
     }
 
     /// returns the offsets

src/array/utf8/mod.rs

@@ -95,7 +95,7 @@ impl<O: Offset> Utf8Array<O> {
         try_check_utf8(&offsets, &values)?;
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != offsets.len())
+            .map_or(false, |validity| validity.len() != offsets.len_proxy())
         {
             return Err(Error::oos(
                 "validity mask length must match the number of values",
@@ -144,7 +144,7 @@ impl<O: Offset> Utf8Array<O> {
     /// Returns the length of this array
     #[inline]
     pub fn len(&self) -> usize {
-        self.offsets.len()
+        self.offsets.len_proxy()
     }
 
     /// Returns the value of the element at index `i`, ignoring the array's validity.
@@ -365,7 +365,7 @@ impl<O: Offset> Utf8Array<O> {
 
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != offsets.len())
+            .map_or(false, |validity| validity.len() != offsets.len_proxy())
         {
             return Err(Error::oos(
                 "validity mask length must match the number of values",

src/compute/aggregate/memory.rs

@@ -63,7 +63,7 @@ pub fn estimated_bytes_size(array: &dyn Array) -> usize {
         List => {
             let array = array.as_any().downcast_ref::<ListArray<i32>>().unwrap();
             estimated_bytes_size(array.values().as_ref())
-                + array.offsets().len() * std::mem::size_of::<i32>()
+                + array.offsets().len_proxy() * std::mem::size_of::<i32>()
                 + validity_size(array.validity())
         }
         FixedSizeList => {
@@ -73,7 +73,7 @@ pub fn estimated_bytes_size(array: &dyn Array) -> usize {
         LargeList => {
             let array = array.as_any().downcast_ref::<ListArray<i64>>().unwrap();
             estimated_bytes_size(array.values().as_ref())
-                + array.offsets().len() * std::mem::size_of::<i64>()
+                + array.offsets().len_proxy() * std::mem::size_of::<i64>()
                 + validity_size(array.validity())
         }
         Struct => {
@@ -111,7 +111,7 @@ pub fn estimated_bytes_size(array: &dyn Array) -> usize {
         }),
         Map => {
             let array = array.as_any().downcast_ref::<MapArray>().unwrap();
-            let offsets = array.offsets().len() * std::mem::size_of::<i32>();
+            let offsets = array.offsets().len_proxy() * std::mem::size_of::<i32>();
             offsets + estimated_bytes_size(array.field().as_ref()) + validity_size(array.validity())
         }
     }

src/io/parquet/write/pages.rs

@@ -50,8 +50,8 @@ impl Nested {
     pub fn len(&self) -> usize {
         match self {
             Nested::Primitive(_, _, length) => *length,
-            Nested::List(nested) => nested.offsets.len(),
-            Nested::LargeList(nested) => nested.offsets.len(),
+            Nested::List(nested) => nested.offsets.len_proxy(),
+            Nested::LargeList(nested) => nested.offsets.len_proxy(),
             Nested::Struct(_, _, len) => *len,
         }
     }

src/offset.rs

@@ -377,12 +377,18 @@ impl<O: Offset> OffsetsBuffer<O> {
         &self.0
     }
 
-    /// Returns the length of this container
+    /// Returns the length an array with these offsets would be.
     #[inline]
-    pub fn len(&self) -> usize {
+    pub fn len_proxy(&self) -> usize {
         self.0.len() - 1
     }
 
+    /// Returns the number of offsets in this container.
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.0.len()
+    }
+
     /// Returns the byte slice stored in this buffer
     #[inline]
     pub fn as_slice(&self) -> &[O] {
@@ -419,7 +425,7 @@ impl<O: Offset> OffsetsBuffer<O> {
     #[inline]
     pub fn start_end(&self, index: usize) -> (usize, usize) {
         // soundness: the invariant of the function
-        assert!(index < self.len());
+        assert!(index < self.len_proxy());
         unsafe { self.start_end_unchecked(index) }
     }