fix: patched-base Int range start index out of range for slice of length (#77)

fix

Author: cxzl25

src/bin/orc/export.rs

@@ -90,11 +90,10 @@ fn run_export<R: ChunkReader>(
     output: Box<dyn io::Write>,
 ) -> Result<()> {
     // Build projection mask if columns are specified
-    let projection = if args.columns.is_some() {
+    let projection = if let Some(selected) = &args.columns {
         // Need to read metadata to build projection
         let metadata = read_metadata(&mut source)?;
 
-        let selected = args.columns.as_ref().unwrap();
         let root_children = metadata.root_data_type().children();
         let mut missing: Vec<&str> = selected
             .iter()

src/encoding/integer/rle_v2/patched_base.rs

@@ -69,8 +69,11 @@ pub fn read_patched_base<N: NInt, R: Read, S: EncodingSign>(
 
     let patch_list_length = (fourth_byte & 0x1f) as usize;
 
-    let base = N::read_big_endian(reader, base_byte_width)?;
+    // Read base value as i64 directly
+    // This matches Java ORC's implementation which always uses long for base values
+    let base = i64::read_big_endian(reader, base_byte_width)?;
     let base = S::decode_signed_msb(base, base_byte_width);
+    let base = N::from_i64(base);
 
     // Get data values
     // TODO: this should read into Vec<i64>

tests/basic/data/patched_int.orc

@@ -275,6 +275,26 @@ pub fn basic_test_bigint() {
     assert_batches_eq(&[last_3_rows], &expected);
 }
 
+#[test]
+pub fn basic_test_patched_int() {
+    let path = basic_path("patched_int.orc");
+    let reader = new_arrow_reader(&path, &["c1"]);
+    let batch = reader.collect::<Result<Vec<_>, _>>().unwrap();
+
+    let total_rows: usize = batch.iter().map(|b| b.num_rows()).sum();
+    assert_eq!(999596, total_rows);
+
+    let last_batch_idx = batch.len() - 1;
+    let total_rows = batch[last_batch_idx].num_rows();
+    let last_3_rows = batch[last_batch_idx].slice(total_rows - 3, 3);
+
+    let expected = [
+        "+----+", "| c1 |", "+----+", "| 1  |", "| 1  |", "| 1  |", "+----+",
+    ];
+
+    assert_batches_eq(&[last_3_rows], &expected);
+}
+
 #[test]
 pub fn basic_test_nested_struct() {
     let path = basic_path("nested_struct.orc");