GH-40633: [C++][Python] Fix ORC crash when file contains unknown timezone

cpp/src/arrow/adapters/orc/adapter.cc

@@ -145,28 +145,29 @@ class OrcStripeReader : public RecordBatchReader {
 
   Status ReadNext(std::shared_ptr<RecordBatch>* out) override {
     std::unique_ptr<liborc::ColumnVectorBatch> batch;
-    ORC_CATCH_NOT_OK(batch = row_reader_->createRowBatch(batch_size_));
+    std::unique_ptr<RecordBatchBuilder> builder;
+
+    ORC_BEGIN_CATCH_NOT_OK
+    batch = row_reader_->createRowBatch(batch_size_);
 
     const liborc::Type& type = row_reader_->getSelectedType();
     if (!row_reader_->next(*batch)) {
       out->reset();
       return Status::OK();
     }
 
-    std::unique_ptr<RecordBatchBuilder> builder;
     ARROW_ASSIGN_OR_RAISE(builder,
                           RecordBatchBuilder::Make(schema_, pool_, batch->numElements));
-
     // The top-level type must be a struct to read into an arrow table
     const auto& struct_batch = checked_cast<liborc::StructVectorBatch&>(*batch);
 
     for (int i = 0; i < builder->num_fields(); i++) {
       RETURN_NOT_OK(AppendBatch(type.getSubtype(i), struct_batch.fields[i], 0,
                                 batch->numElements, builder->GetField(i)));
     }
+    ORC_END_CATCH_NOT_OK
 
-    ARROW_ASSIGN_OR_RAISE(*out, builder->Flush());
-    return Status::OK();
+    return builder->Flush().Value(out);
   }
 
  private:
@@ -470,15 +471,13 @@ class ORCFileReader::Impl {
                                                  int64_t nrows) {
     std::unique_ptr<liborc::RowReader> row_reader;
     std::unique_ptr<liborc::ColumnVectorBatch> batch;
+    std::unique_ptr<RecordBatchBuilder> builder;
 
     ORC_BEGIN_CATCH_NOT_OK
     row_reader = reader_->createRowReader(opts);
     batch = row_reader->createRowBatch(std::min(nrows, kReadRowsBatch));
-    ORC_END_CATCH_NOT_OK
 
-    std::unique_ptr<RecordBatchBuilder> builder;
     ARROW_ASSIGN_OR_RAISE(builder, RecordBatchBuilder::Make(schema, pool_, nrows));
-
     // The top-level type must be a struct to read into an arrow table
     const auto& struct_batch = checked_cast<liborc::StructVectorBatch&>(*batch);
 
@@ -489,6 +488,7 @@ class ORCFileReader::Impl {
                                   batch->numElements, builder->GetField(i)));
       }
     }
+    ORC_END_CATCH_NOT_OK
 
     return builder->Flush();
   }

python/pyarrow/tests/test_orc.py

@@ -15,9 +15,14 @@
 # specific language governing permissions and limitations
 # under the License.
 
-import pytest
 import decimal
 import datetime
+from pathlib import Path
+import shutil
+import subprocess
+import sys
+
+import pytest
 
 import pyarrow as pa
 from pyarrow import fs
@@ -140,6 +145,57 @@ def test_example_using_json(filename, datadir):
     check_example_file(path, table, need_fix=True)
 
 
+def test_timezone_database_absent(datadir):
+    # Example file relies on the timezone "US/Pacific". It should gracefully
+    # fail, not crash, if the timezone database is not found.
+    path = datadir / 'TestOrcFile.testDate1900.orc'
+    code = f"""if 1:
+        import os
+        os.environ['TZDIR'] = '/tmp/non_existent'
+
+        from pyarrow import orc
+        try:
+            orc_file = orc.ORCFile({str(path)!r})
+            orc_file.read()
+        except Exception as e:
+            assert "time zone database" in str(e).lower(), e
+        else:
+            assert False, "Should have raised exception"
+    """
+    subprocess.run([sys.executable, "-c", code], check=True)
+
+
+def test_timezone_absent(datadir, tmpdir):
+    # Example file relies on the timezone "US/Pacific". It should gracefully
+    # fail, not crash, if the timezone database is present but the timezone
+    # is not found (GH-40633).
+    source_tzdir = Path('/usr/share/zoneinfo')
+    if not source_tzdir.exists():
+        pytest.skip(f"Test needs timezone database in {source_tzdir}")
+    tzdir = Path(tmpdir / 'zoneinfo')
+    try:
+        shutil.copytree(source_tzdir, tzdir, symlinks=True)
+    except OSError as e:
+        pytest.skip(f"Failed to copy timezone database: {e}")
+    (tzdir / 'US' / 'Pacific').unlink(missing_ok=True)
+
+    path = datadir / 'TestOrcFile.testDate1900.orc'
+    code = f"""if 1:
+        import os
+        os.environ['TZDIR'] = {str(tzdir)!r}
+
+        from pyarrow import orc
+        orc_file = orc.ORCFile({str(path)!r})
+        try:
+            orc_file.read()
+        except Exception as e:
+            assert "zoneinfo/US/Pacific" in str(e), e
+        else:
+            assert False, "Should have raised exception"
+    """
+    subprocess.run([sys.executable, "-c", code], check=True)
+
+
 def test_orcfile_empty(datadir):
     from pyarrow import orc