From ab5e464901a622e5bfec63e2e9c8a8fe195259a4 Mon Sep 17 00:00:00 2001
From: Gijs Burghoorn <me@gburghoorn.com>
Date: Mon, 21 Oct 2024 10:48:50 +0200
Subject: [PATCH] fix: Ignore Parquet `is_{min,max}_value_exact` when set to
 `true` (#19344)

---
 .../src/parquet/statistics/mod.rs             | 53 ++++++++++++++-----
 1 file changed, 39 insertions(+), 14 deletions(-)

diff --git a/crates/polars-parquet/src/parquet/statistics/mod.rs b/crates/polars-parquet/src/parquet/statistics/mod.rs
index b72e2a7c94b1..03335c27817b 100644
--- a/crates/polars-parquet/src/parquet/statistics/mod.rs
+++ b/crates/polars-parquet/src/parquet/statistics/mod.rs
@@ -11,7 +11,6 @@ pub use primitive::PrimitiveStatistics;
 use crate::parquet::error::ParquetResult;
 use crate::parquet::schema::types::{PhysicalType, PrimitiveType};
 pub use crate::parquet::thrift_format::Statistics as ParquetStatistics;
-use crate::read::ParquetError;
 
 #[derive(Debug, PartialEq)]
 pub enum Statistics {
@@ -42,6 +41,34 @@ impl Statistics {
         }
     }
 
+    pub fn clear_min(&mut self) {
+        use Statistics as S;
+        match self {
+            S::Binary(s) => _ = s.min_value.take(),
+            S::Boolean(s) => _ = s.min_value.take(),
+            S::FixedLen(s) => _ = s.min_value.take(),
+            S::Int32(s) => _ = s.min_value.take(),
+            S::Int64(s) => _ = s.min_value.take(),
+            S::Int96(s) => _ = s.min_value.take(),
+            S::Float(s) => _ = s.min_value.take(),
+            S::Double(s) => _ = s.min_value.take(),
+        };
+    }
+
+    pub fn clear_max(&mut self) {
+        use Statistics as S;
+        match self {
+            S::Binary(s) => _ = s.max_value.take(),
+            S::Boolean(s) => _ = s.max_value.take(),
+            S::FixedLen(s) => _ = s.max_value.take(),
+            S::Int32(s) => _ = s.max_value.take(),
+            S::Int64(s) => _ = s.max_value.take(),
+            S::Int96(s) => _ = s.max_value.take(),
+            S::Float(s) => _ = s.max_value.take(),
+            S::Double(s) => _ = s.max_value.take(),
+        };
+    }
+
     /// Deserializes a raw parquet statistics into [`Statistics`].
     /// # Error
     /// This function errors if it is not possible to read the statistics to the
@@ -51,19 +78,8 @@ impl Statistics {
         statistics: &ParquetStatistics,
         primitive_type: PrimitiveType,
     ) -> ParquetResult<Self> {
-        if statistics.is_min_value_exact.is_some() {
-            return Err(ParquetError::not_supported(
-                "is_min_value_exact in statistics",
-            ));
-        }
-        if statistics.is_max_value_exact.is_some() {
-            return Err(ParquetError::not_supported(
-                "is_max_value_exact in statistics",
-            ));
-        }
-
         use {PhysicalType as T, PrimitiveStatistics as PrimStat};
-        Ok(match primitive_type.physical_type {
+        let mut stats: Self = match primitive_type.physical_type {
             T::ByteArray => BinaryStatistics::deserialize(statistics, primitive_type)?.into(),
             T::Boolean => BooleanStatistics::deserialize(statistics)?.into(),
             T::Int32 => PrimStat::<i32>::deserialize(statistics, primitive_type)?.into(),
@@ -74,7 +90,16 @@ impl Statistics {
             T::FixedLenByteArray(size) => {
                 FixedLenStatistics::deserialize(statistics, size, primitive_type)?.into()
             },
-        })
+        };
+
+        if statistics.is_min_value_exact.is_some_and(|v| !v) {
+            stats.clear_min();
+        }
+        if statistics.is_max_value_exact.is_some_and(|v| !v) {
+            stats.clear_max();
+        }
+
+        Ok(stats)
     }
 }