feat: Ensure unique names in HCOncat

crates/polars-core/src/frame/horizontal.rs

@@ -66,7 +66,7 @@ impl DataFrame {
 }
 /// Concat [`DataFrame`]s horizontally.
 /// Concat horizontally and extend with null values if lengths don't match
-pub fn concat_df_horizontal(dfs: &[DataFrame]) -> PolarsResult<DataFrame> {
+pub fn concat_df_horizontal(dfs: &[DataFrame], check_duplicates: bool) -> PolarsResult<DataFrame> {
     let max_len = dfs
         .iter()
         .map(|df| df.height())
@@ -99,18 +99,26 @@ pub fn concat_df_horizontal(dfs: &[DataFrame]) -> PolarsResult<DataFrame> {
     let height = first_df.height();
     let is_empty = first_df.is_empty();
 
-    let columns = first_df
-        .columns
-        .iter()
-        .map(|s| SmartString::from(s.name()))
-        .collect::<Vec<_>>();
+    let columns;
+    let mut names = if check_duplicates {
+        columns = first_df
+            .columns
+            .iter()
+            .map(|s| SmartString::from(s.name()))
+            .collect::<Vec<_>>();
 
-    let mut names = columns.iter().map(|n| n.as_str()).collect::<PlHashSet<_>>();
+        columns.iter().map(|n| n.as_str()).collect::<PlHashSet<_>>()
+    } else {
+        Default::default()
+    };
 
     for df in &dfs[1..] {
         let cols = df.get_columns();
-        for col in cols {
-            check_hstack(col, &mut names, height, is_empty)?;
+
+        if check_duplicates {
+            for col in cols {
+                check_hstack(col, &mut names, height, is_empty)?;
+            }
         }
 
         unsafe { first_df.hstack_mut_unchecked(cols) };

crates/polars-mem-engine/src/executors/hconcat.rs

@@ -58,6 +58,7 @@ impl Executor for HConcatExec {
             out?.into_iter().flatten().collect()
         };
 
-        concat_df_horizontal(&dfs)
+        // Invariant of IR. Schema is already checked to contain no duplicates.
+        concat_df_horizontal(&dfs, false)
     }
 }

crates/polars-plan/src/plans/ir/mod.rs

@@ -136,6 +136,8 @@ pub enum IR {
         inputs: Vec<Node>,
         options: UnionOptions,
     },
+    /// Horizontal concatenation
+    /// - Invariant: the names will be unique
     HConcat {
         inputs: Vec<Node>,
         schema: SchemaRef,

crates/polars-plan/src/plans/optimizer/simplify_functions.rs

@@ -1,5 +1,3 @@
-use polars_core::chunked_array::cast::CastOptions;
-
 use super::*;
 
 pub(super) fn optimize_functions(
@@ -67,17 +65,6 @@ pub(super) fn optimize_functions(
                 None
             }
         },
-        FunctionExpr::Boolean(BooleanFunction::AllHorizontal | BooleanFunction::AnyHorizontal) => {
-            if input.len() == 1 {
-                Some(AExpr::Cast {
-                    expr: input[0].node(),
-                    data_type: DataType::Boolean,
-                    options: CastOptions::NonStrict,
-                })
-            } else {
-                None
-            }
-        },
         FunctionExpr::Boolean(BooleanFunction::Not) => {
             let y = expr_arena.get(input[0].node());
 

docs/src/rust/user-guide/transformations/concatenation.rs

@@ -30,7 +30,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             "r2"=> &[7, 8],
             "r3"=> &[9, 10],
     )?;
-    let df_horizontal_concat = polars::functions::concat_df_horizontal(&[df_h1, df_h2])?;
+    let df_horizontal_concat = polars::functions::concat_df_horizontal(&[df_h1, df_h2], true)?;
     println!("{}", &df_horizontal_concat);
     // --8<-- [end:horizontal]
     //
@@ -43,7 +43,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             "r1"=> &[5, 6, 7],
             "r2"=> &[8, 9, 10],
     )?;
-    let df_horizontal_concat = polars::functions::concat_df_horizontal(&[df_h1, df_h2])?;
+    let df_horizontal_concat = polars::functions::concat_df_horizontal(&[df_h1, df_h2], true)?;
     println!("{}", &df_horizontal_concat);
     // --8<-- [end:horizontal_different_lengths]
 

py-polars/src/functions/eager.rs

@@ -88,6 +88,6 @@ pub fn concat_df_horizontal(dfs: &Bound<'_, PyAny>) -> PyResult<PyDataFrame> {
         })
         .collect::<PyResult<Vec<_>>>()?;
 
-    let df = functions::concat_df_horizontal(&dfs).map_err(PyPolarsErr::from)?;
+    let df = functions::concat_df_horizontal(&dfs, true).map_err(PyPolarsErr::from)?;
     Ok(df.into())
 }