refactor: Zero-Field Structs and DataFrame with Height Property (#19123)

crates/polars-arrow/src/array/growable/structure.rs

@@ -10,6 +10,7 @@ use crate::bitmap::MutableBitmap;
 /// Concrete [`Growable`] for the [`StructArray`].
 pub struct GrowableStruct<'a> {
     arrays: Vec<&'a StructArray>,
+    length: usize,
     validity: Option<MutableBitmap>,
     values: Vec<Box<dyn Growable<'a> + 'a>>,
 }
@@ -48,6 +49,7 @@ impl<'a> GrowableStruct<'a> {
 
         Self {
             arrays,
+            length: 0,
             values,
             validity: prepare_validity(use_validity, capacity),
         }
@@ -60,6 +62,7 @@ impl<'a> GrowableStruct<'a> {
 
         StructArray::new(
             self.arrays[0].dtype().clone(),
+            self.length,
             values,
             validity.map(|v| v.into()),
         )
@@ -71,6 +74,8 @@ impl<'a> Growable<'a> for GrowableStruct<'a> {
         let array = *self.arrays.get_unchecked_release(index);
         extend_validity(&mut self.validity, array, start, len);
 
+        self.length += len;
+
         if array.null_count() == 0 {
             self.values
                 .iter_mut()
@@ -97,6 +102,7 @@ impl<'a> Growable<'a> for GrowableStruct<'a> {
         if let Some(validity) = &mut self.validity {
             validity.extend_constant(additional, false);
         }
+        self.length += additional;
     }
 
     #[inline]
@@ -123,6 +129,7 @@ impl<'a> From<GrowableStruct<'a>> for StructArray {
 
         StructArray::new(
             val.arrays[0].dtype().clone(),
+            val.length,
             values,
             val.validity.map(|v| v.into()),
         )

crates/polars-arrow/src/array/struct_/data.rs

@@ -21,6 +21,7 @@ impl Arrow2Arrow for StructArray {
 
         Self {
             dtype,
+            length: data.len(),
             values: data.child_data().iter().map(from_data).collect(),
             validity: data.nulls().map(|n| Bitmap::from_null_buffer(n.clone())),
         }

crates/polars-arrow/src/array/struct_/ffi.rs

@@ -68,6 +68,6 @@ impl<A: ffi::ArrowArrayRef> FromFfi<A> for StructArray {
             })
             .collect::<PolarsResult<Vec<Box<dyn Array>>>>()?;
 
-        Self::try_new(dtype, values, validity)
+        Self::try_new(dtype, len, values, validity)
     }
 }

crates/polars-arrow/src/array/struct_/mod.rs

@@ -9,7 +9,7 @@ pub(super) mod fmt;
 mod iterator;
 mod mutable;
 pub use mutable::*;
-use polars_error::{polars_bail, PolarsResult};
+use polars_error::{polars_bail, polars_ensure, PolarsResult};
 
 use crate::compute::utils::combine_validities_and;
 
@@ -27,13 +27,15 @@ use crate::compute::utils::combine_validities_and;
 ///     Field::new("c".into(), ArrowDataType::Int32, false),
 /// ];
 ///
-/// let array = StructArray::new(ArrowDataType::Struct(fields), vec![boolean, int], None);
+/// let array = StructArray::new(ArrowDataType::Struct(fields), 4, vec![boolean, int], None);
 /// ```
 #[derive(Clone)]
 pub struct StructArray {
     dtype: ArrowDataType,
     // invariant: each array has the same length
     values: Vec<Box<dyn Array>>,
+    // invariant: for each v in values: length == v.len()
+    length: usize,
     validity: Option<Bitmap>,
 }
 
@@ -49,22 +51,17 @@ impl StructArray {
     /// * the validity's length is not equal to the length of the first element
     pub fn try_new(
         dtype: ArrowDataType,
+        length: usize,
         values: Vec<Box<dyn Array>>,
         validity: Option<Bitmap>,
     ) -> PolarsResult<Self> {
         let fields = Self::try_get_fields(&dtype)?;
-        if fields.is_empty() {
-            assert!(values.is_empty(), "invalid struct");
-            assert_eq!(validity.map(|v| v.len()).unwrap_or(0), 0, "invalid struct");
-            return Ok(Self {
-                dtype,
-                values,
-                validity: None,
-            });
-        }
-        if fields.len() != values.len() {
-            polars_bail!(ComputeError:"a StructArray must have a number of fields in its DataType equal to the number of child values")
-        }
+
+        polars_ensure!(
+            fields.len() == values.len(),
+            ComputeError:
+                "a StructArray must have a number of fields in its DataType equal to the number of child values"
+        );
 
         fields
             .iter().map(|a| &a.dtype)
@@ -81,29 +78,29 @@ impl StructArray {
                 }
             })?;
 
-        let len = values[0].len();
         values
             .iter()
-            .map(|a| a.len())
+            .map(|f| f.len())
             .enumerate()
-            .try_for_each(|(index, a_len)| {
-                if a_len != len {
-                    polars_bail!(ComputeError: "The children must have an equal number of values.
-                         However, the values at index {index} have a length of {a_len}, which is different from values at index 0, {len}.")
+            .try_for_each(|(index, f_length)| {
+                if f_length != length {
+                    polars_bail!(ComputeError: "The children must have the given number of values.
+                         However, the values at index {index} have a length of {f_length}, which is different from given length {length}.")
                 } else {
                     Ok(())
                 }
             })?;
 
         if validity
             .as_ref()
-            .map_or(false, |validity| validity.len() != len)
+            .map_or(false, |validity| validity.len() != length)
         {
             polars_bail!(ComputeError:"The validity length of a StructArray must match its number of elements")
         }
 
         Ok(Self {
             dtype,
+            length,
             values,
             validity,
         })
@@ -120,10 +117,11 @@ impl StructArray {
     /// * the validity's length is not equal to the length of the first element
     pub fn new(
         dtype: ArrowDataType,
+        length: usize,
         values: Vec<Box<dyn Array>>,
         validity: Option<Bitmap>,
     ) -> Self {
-        Self::try_new(dtype, values, validity).unwrap()
+        Self::try_new(dtype, length, values, validity).unwrap()
     }
 
     /// Creates an empty [`StructArray`].
@@ -133,7 +131,7 @@ impl StructArray {
                 .iter()
                 .map(|field| new_empty_array(field.dtype().clone()))
                 .collect();
-            Self::new(dtype, values, None)
+            Self::new(dtype, 0, values, None)
         } else {
             panic!("StructArray must be initialized with DataType::Struct");
         }
@@ -146,7 +144,7 @@ impl StructArray {
                 .iter()
                 .map(|field| new_null_array(field.dtype().clone(), length))
                 .collect();
-            Self::new(dtype, values, Some(Bitmap::new_zeroed(length)))
+            Self::new(dtype, length, values, Some(Bitmap::new_zeroed(length)))
         } else {
             panic!("StructArray must be initialized with DataType::Struct");
         }
@@ -157,9 +155,10 @@ impl StructArray {
 impl StructArray {
     /// Deconstructs the [`StructArray`] into its individual components.
     #[must_use]
-    pub fn into_data(self) -> (Vec<Field>, Vec<Box<dyn Array>>, Option<Bitmap>) {
+    pub fn into_data(self) -> (Vec<Field>, usize, Vec<Box<dyn Array>>, Option<Bitmap>) {
         let Self {
             dtype,
+            length,
             values,
             validity,
         } = self;
@@ -168,7 +167,7 @@ impl StructArray {
         } else {
             unreachable!()
         };
-        (fields, values, validity)
+        (fields, length, values, validity)
     }
 
     /// Slices this [`StructArray`].
@@ -199,6 +198,7 @@ impl StructArray {
         self.values
             .iter_mut()
             .for_each(|x| x.slice_unchecked(offset, length));
+        self.length = length;
     }
 
     /// Set the outer nulls into the inner arrays.
@@ -227,18 +227,17 @@ impl StructArray {
 impl StructArray {
     #[inline]
     fn len(&self) -> usize {
-        #[cfg(debug_assertions)]
-        if let Some(fst) = self.values.first() {
-            for arr in self.values.iter().skip(1) {
+        if cfg!(debug_assertions) {
+            for arr in self.values.iter() {
                 assert_eq!(
                     arr.len(),
-                    fst.len(),
+                    self.length,
                     "StructArray invariant: each array has same length"
                 );
             }
         }
 
-        self.values.first().map(|arr| arr.len()).unwrap_or(0)
+        self.length
     }
 
     /// The optional validity.
@@ -310,11 +309,13 @@ impl Splitable for StructArray {
         (
             Self {
                 dtype: self.dtype.clone(),
+                length: offset,
                 values: lhs_values,
                 validity: lhs_validity,
             },
             Self {
                 dtype: self.dtype.clone(),
+                length: self.length - offset,
                 values: rhs_values,
                 validity: rhs_validity,
             },