ARROW-11163: [C++] Fix reading of compressed IPC/Feather files written with Arrow 0.17

cpp/src/arrow/ipc/reader.cc

@@ -49,6 +49,7 @@
 #include "arrow/util/key_value_metadata.h"
 #include "arrow/util/logging.h"
 #include "arrow/util/parallel.h"
+#include "arrow/util/string.h"
 #include "arrow/util/ubsan.h"
 #include "arrow/visitor_inline.h"
 
@@ -535,8 +536,9 @@ Status GetCompressionExperimental(const flatbuf::Message* message,
     RETURN_NOT_OK(internal::GetKeyValueMetadata(message->custom_metadata(), &metadata));
     int index = metadata->FindKey("ARROW:experimental_compression");
     if (index != -1) {
-      ARROW_ASSIGN_OR_RAISE(*out,
-                            util::Codec::GetCompressionType(metadata->value(index)));
+      // Arrow 0.17 stored string in upper case, internal utils now require lower case
+      auto name = arrow::internal::AsciiToLower(metadata->value(index));
+      ARROW_ASSIGN_OR_RAISE(*out, util::Codec::GetCompressionType(name));
     }
     return internal::CheckCompressionSupported(*out);
   }

python/pyarrow/tests/test_feather.py

@@ -39,6 +39,11 @@
     pass
 
 
+@pytest.fixture(scope='module')
+def datadir(base_datadir):
+    return base_datadir / 'feather'
+
+
 def random_path(prefix='feather_'):
     return tempfile.mktemp(prefix=prefix)
 
@@ -769,3 +774,19 @@ def test_nested_types(compression):
 @h.given(past.all_tables, st.sampled_from(["uncompressed", "lz4", "zstd"]))
 def test_roundtrip(table, compression):
     _check_arrow_roundtrip(table, compression=compression)
+
+
+def test_feather_v017_experimental_compression_backward_compatibility(datadir):
+    # ARROW-11163 - ensure newer pyarrow versions can read the old feather
+    # files from version 0.17.0 with experimental compression support (before
+    # it was officially added to IPC format in 1.0.0)
+
+    # file generated with:
+    #     table = pa.table({'a': range(5)})
+    #     from pyarrow import feather
+    #     feather.write_feather(
+    #         table, "v0.17.0.version=2-compression=lz4.feather",
+    #         compression="lz4", version=2)
+    expected = pa.table({'a': range(5)})
+    result = read_table(datadir / "v0.17.0.version=2-compression=lz4.feather")
+    assert result.equals(expected)

r/NEWS.md

@@ -43,6 +43,7 @@
 * `write_parquet()` can now write RecordBatches
 * Reading a Table from a RecordBatchStreamReader containing 0 batches no longer crashes
 * `readr`'s `problems` attribute is removed when converting to Arrow RecordBatch and table to prevent large amounts of metadata from accumulating inadvertently [ARROW-10624](https://issues.apache.org/jira/browse/ARROW-10624)
+* Fixed reading of compressed Feather files written with Arrow 0.17 [ARROW-10850](https://issues.apache.org/jira/browse/ARROW-10850)
 
 ## Packaging and installation