From 369f49e11b96a0b1706f87b3e01e8726d822c196 Mon Sep 17 00:00:00 2001
From: ritchie <ritchie46@gmail.com>
Date: Thu, 27 Jul 2023 17:22:38 +0200
Subject: [PATCH] fix(rust, python): always sort projection by schema

---
 .../polars-plan/src/logical_plan/file_scan.rs       | 13 -------------
 .../optimizer/projection_pushdown/mod.rs            | 10 ++--------
 2 files changed, 2 insertions(+), 21 deletions(-)

diff --git a/polars/polars-lazy/polars-plan/src/logical_plan/file_scan.rs b/polars/polars-lazy/polars-plan/src/logical_plan/file_scan.rs
index f3d52c115f930..6aa19b79c6287 100644
--- a/polars/polars-lazy/polars-plan/src/logical_plan/file_scan.rs
+++ b/polars/polars-lazy/polars-plan/src/logical_plan/file_scan.rs
@@ -24,19 +24,6 @@ impl FileScan {
         }
     }
 
-    pub(crate) fn sort_projection(&self, _file_options: &FileScanOptions) -> bool {
-        match self {
-            #[cfg(feature = "csv")]
-            Self::Csv { .. } => true,
-            #[cfg(feature = "ipc")]
-            Self::Ipc { .. } => _file_options.row_count.is_some(),
-            #[cfg(feature = "parquet")]
-            Self::Parquet { .. } => _file_options.row_count.is_some(),
-            #[allow(unreachable_patterns)]
-            _ => false,
-        }
-    }
-
     pub fn streamable(&self) -> bool {
         match self {
             #[cfg(feature = "csv")]
diff --git a/polars/polars-lazy/polars-plan/src/logical_plan/optimizer/projection_pushdown/mod.rs b/polars/polars-lazy/polars-plan/src/logical_plan/optimizer/projection_pushdown/mod.rs
index 91c812b29309b..eed4df0011433 100644
--- a/polars/polars-lazy/polars-plan/src/logical_plan/optimizer/projection_pushdown/mod.rs
+++ b/polars/polars-lazy/polars-plan/src/logical_plan/optimizer/projection_pushdown/mod.rs
@@ -127,7 +127,6 @@ fn update_scan_schema(
     acc_projections: &[Node],
     expr_arena: &Arena<AExpr>,
     schema: &Schema,
-    sort_projections: bool,
 ) -> PolarsResult<Schema> {
     let mut new_schema = Schema::with_capacity(acc_projections.len());
     let mut new_cols = Vec::with_capacity(acc_projections.len());
@@ -140,9 +139,8 @@ fn update_scan_schema(
         }
     }
     // make sure that the projections are sorted by the schema.
-    if sort_projections {
-        new_cols.sort_unstable_by_key(|item| item.0);
-    }
+    // otherwise the upstream nodes will have an incorrect schema.
+    new_cols.sort_unstable_by_key(|item| item.0);
     for item in new_cols {
         new_schema.with_column(item.1.clone(), item.2.clone());
     }
@@ -375,7 +373,6 @@ impl ProjectionPushDown {
                             &acc_projections,
                             expr_arena,
                             &file_info.schema,
-                            true,
                         )?))
                     };
                     mut_options.output_schema = output_schema.clone();
@@ -412,7 +409,6 @@ impl ProjectionPushDown {
                         &acc_projections,
                         expr_arena,
                         &schema,
-                        false,
                     )?));
                     projection = get_scan_columns(&mut acc_projections, expr_arena, None);
                 }
@@ -439,7 +435,6 @@ impl ProjectionPushDown {
                         &acc_projections,
                         expr_arena,
                         &options.schema,
-                        true,
                     )?))
                 };
                 Ok(PythonScan { options, predicate })
@@ -465,7 +460,6 @@ impl ProjectionPushDown {
                         &acc_projections,
                         expr_arena,
                         &file_info.schema,
-                        scan_type.sort_projection(&file_options),
                     )?))
                 };