Revert "fix: respect inexact flags in row group metadata (apache#16412)"

This reverts commit afc90f7.

Author: LiaCastaneda

datafusion/core/src/datasource/file_format/parquet.rs

@@ -523,23 +523,11 @@ mod tests {
         let dic_array = DictionaryArray::<Int32Type>::try_new(keys, Arc::new(values))?;
         let c_dic: ArrayRef = Arc::new(dic_array);
 
-        // Data for column string_truncation: ["a".repeat(128), null, "b".repeat(128), null]
-        let string_truncation: ArrayRef = Arc::new(StringArray::from(vec![
-            Some("a".repeat(128)),
-            None,
-            Some("b".repeat(128)),
-            None,
-        ]));
-
-        let batch1 = RecordBatch::try_from_iter(vec![
-            ("c_dic", c_dic),
-            ("string_truncation", string_truncation),
-        ])?;
+        let batch1 = RecordBatch::try_from_iter(vec![("c_dic", c_dic)])?;
 
         // Use store_parquet to write each batch to its own file
         // . batch1 written into first file and includes:
         //    - column c_dic that has 4 rows with no null. Stats min and max of dictionary column is available.
-        //    - column string_truncation that has 4 rows with 2 nulls. Stats min and max of string column is available but not exact.
         let store = Arc::new(RequestCountingObjectStore::new(Arc::new(
             LocalFileSystem::new(),
         )));
@@ -575,19 +563,6 @@ mod tests {
             Precision::Exact(Utf8(Some("a".into())))
         );
 
-        // column string_truncation
-        let string_truncation_stats = &stats.column_statistics[1];
-
-        assert_eq!(string_truncation_stats.null_count, Precision::Exact(2));
-        assert_eq!(
-            string_truncation_stats.max_value,
-            Precision::Inexact(ScalarValue::Utf8View(Some("b".repeat(63) + "c")))
-        );
-        assert_eq!(
-            string_truncation_stats.min_value,
-            Precision::Inexact(ScalarValue::Utf8View(Some("a".repeat(64))))
-        );
-
         Ok(())
     }
 

datafusion/datasource-parquet/src/file_format.rs

@@ -19,11 +19,11 @@
 
 use std::any::Any;
 use std::cell::RefCell;
+use std::fmt;
 use std::fmt::Debug;
 use std::ops::Range;
 use std::rc::Rc;
 use std::sync::Arc;
-use std::{fmt, vec};
 
 use arrow::array::RecordBatch;
 use arrow::datatypes::{Fields, Schema, SchemaRef, TimeUnit};
@@ -36,15 +36,16 @@ use datafusion_datasource::write::{
 use datafusion_datasource::file_format::{FileFormat, FileFormatFactory};
 use datafusion_datasource::write::demux::DemuxedStreamReceiver;
 
+use arrow::compute::sum;
 use arrow::datatypes::{DataType, Field, FieldRef};
 use datafusion_common::config::{ConfigField, ConfigFileType, TableParquetOptions};
 #[cfg(feature = "parquet_encryption")]
 use datafusion_common::encryption::map_config_decryption_to_decryption;
 use datafusion_common::encryption::FileDecryptionProperties;
 use datafusion_common::parsers::CompressionTypeVariant;
 use datafusion_common::{
-    internal_datafusion_err, internal_err, not_impl_err, DataFusionError, GetExt,
-    HashSet, Result, DEFAULT_PARQUET_EXTENSION,
+    internal_datafusion_err, internal_err, not_impl_err, ColumnStatistics,
+    DataFusionError, GetExt, HashSet, Result, DEFAULT_PARQUET_EXTENSION,
 };
 use datafusion_common::{HashMap, Statistics};
 use datafusion_common_runtime::{JoinSet, SpawnedTask};

datafusion/functions-aggregate/src/min_max.rs

@@ -736,7 +736,7 @@ macro_rules! min_max {
 }
 
 /// An accumulator to compute the maximum value
-#[derive(Debug, Clone)]
+#[derive(Debug)]
 pub struct MaxAccumulator {
     max: ScalarValue,
 }
@@ -1057,7 +1057,7 @@ impl AggregateUDFImpl for Min {
 }
 
 /// An accumulator to compute the minimum value
-#[derive(Debug, Clone)]
+#[derive(Debug)]
 pub struct MinAccumulator {
     min: ScalarValue,
 }