fixed MSB selector

jorgecarleitao · Dec 14, 2021 · b5c4184 · b5c4184
1 parent fdb8c68
commit b5c4184
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 3 deletions.
diff --git a/parquet_integration/write_parquet.py b/parquet_integration/write_parquet.py
@@ -50,7 +50,7 @@ def case_basic_nullable(size=1):
 
 
 def case_basic_required(size=1):
-    int64 = [-1, 0, 1, 2, 3, 4, 5, 6, 7, 8]
+    int64 = [-256, -1, 0, 1, 2, 3, 4, 5, 6, 7]
     uint32 = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
     float64 = [0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0]
     string = ["Hello", "bbb", "aa", "", "bbb", "abc", "bbb", "bbb", "def", "aaa"]

diff --git a/src/io/parquet/read/mod.rs b/src/io/parquet/read/mod.rs
@@ -287,7 +287,7 @@ fn page_iter_to_array<I: FallibleStreamingIterator<Item = DataPage, Error = Parq
                                 .map(|v| {
                                     v.and_then(|v1| {
                                         // Pad with the value of the MSB to correctly handle (two's complement) negative integers.
-                                        let msb_set = v1.last().unwrap_or(&0) >> 7 == 1;
+                                        let msb_set = v1.first().unwrap_or(&0) >> 7 == 1;
                                         let padding = if msb_set {
                                             &ones_padding
                                         } else {

diff --git a/tests/it/io/parquet/mod.rs b/tests/it/io/parquet/mod.rs
@@ -399,6 +399,7 @@ pub fn pyarrow_nullable_statistics(column: usize) -> Option<Box<dyn Statistics>>
 // these values match the values in `integration`
 pub fn pyarrow_required(column: usize) -> Box<dyn Array> {
     let i64_values = &[
+        Some(-256),
         Some(-1),
         Some(0),
         Some(1),
@@ -408,7 +409,6 @@ pub fn pyarrow_required(column: usize) -> Box<dyn Array> {
         Some(5),
         Some(6),
         Some(7),
-        Some(8),
     ];
 
     match column {