Revert "BackPort [ARROW-13572]: ORC support , [ARROW-13797]: column p…

…rojection pushdow, and refactor some classes (#36)" (#41)

This reverts commit a2c70e8.

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

@@ -276,15 +276,6 @@ class ORCFileReader::Impl {
     return ReadTable(opts, schema, out);
   }
 
-  Status Read(const std::vector<std::string>& include_names,
-              std::shared_ptr<Table>* out) {
-    liborc::RowReaderOptions opts;
-    RETURN_NOT_OK(SelectNames(&opts, include_names));
-    std::shared_ptr<Schema> schema;
-    RETURN_NOT_OK(ReadSchema(opts, &schema));
-    return ReadTable(opts, schema, out);
-  }
-
   Status Read(const std::shared_ptr<Schema>& schema,
               const std::vector<int>& include_indices, std::shared_ptr<Table>* out) {
     liborc::RowReaderOptions opts;
@@ -310,16 +301,6 @@ class ORCFileReader::Impl {
     return ReadBatch(opts, schema, stripes_[stripe].num_rows, out);
   }
 
-  Status ReadStripe(int64_t stripe, const std::vector<std::string>& include_names,
-                    std::shared_ptr<RecordBatch>* out) {
-    liborc::RowReaderOptions opts;
-    RETURN_NOT_OK(SelectNames(&opts, include_names));
-    RETURN_NOT_OK(SelectStripe(&opts, stripe));
-    std::shared_ptr<Schema> schema;
-    RETURN_NOT_OK(ReadSchema(opts, &schema));
-    return ReadBatch(opts, schema, stripes_[stripe].num_rows, out);
-  }
-
   Status SelectStripe(liborc::RowReaderOptions* opts, int64_t stripe) {
     ARROW_RETURN_IF(stripe < 0 || stripe >= NumberOfStripes(),
                     Status::Invalid("Out of bounds stripe: ", stripe));
@@ -356,13 +337,6 @@ class ORCFileReader::Impl {
     return Status::OK();
   }
 
-  Status SelectNames(liborc::RowReaderOptions* opts,
-                     const std::vector<std::string>& include_names) {
-    std::list<std::string> include_names_list(include_names.begin(), include_names.end());
-    opts->include(include_names_list);
-    return Status::OK();
-  }
-
   Status ReadTable(const liborc::RowReaderOptions& row_opts,
                    const std::shared_ptr<Schema>& schema, std::shared_ptr<Table>* out) {
     liborc::RowReaderOptions opts(row_opts);
@@ -461,23 +435,10 @@ Status ORCFileReader::Open(const std::shared_ptr<io::RandomAccessFile>& file,
   return Status::OK();
 }
 
-Result<std::unique_ptr<ORCFileReader>> ORCFileReader::Open(
-    const std::shared_ptr<io::RandomAccessFile>& file, MemoryPool* pool) {
-  auto result = std::unique_ptr<ORCFileReader>(new ORCFileReader());
-  RETURN_NOT_OK(result->impl_->Open(file, pool));
-  return std::move(result);
-}
-
 Status ORCFileReader::ReadSchema(std::shared_ptr<Schema>* out) {
   return impl_->ReadSchema(out);
 }
 
-Result<std::shared_ptr<Schema>> ORCFileReader::ReadSchema() {
-  std::shared_ptr<Schema> schema;
-  RETURN_NOT_OK(impl_->ReadSchema(&schema));
-  return schema;
-}
-
 Status ORCFileReader::Read(std::shared_ptr<Table>* out) { return impl_->Read(out); }
 
 Status ORCFileReader::Read(const std::shared_ptr<Schema>& schema,
@@ -490,13 +451,6 @@ Status ORCFileReader::Read(const std::vector<int>& include_indices,
   return impl_->Read(include_indices, out);
 }
 
-Result<std::shared_ptr<Table>> ORCFileReader::Read(
-    const std::vector<std::string>& include_names) {
-  std::shared_ptr<Table> table;
-  RETURN_NOT_OK(impl_->Read(include_names, &table));
-  return table;
-}
-
 Status ORCFileReader::Read(const std::shared_ptr<Schema>& schema,
                            const std::vector<int>& include_indices,
                            std::shared_ptr<Table>* out) {
@@ -507,24 +461,11 @@ Status ORCFileReader::ReadStripe(int64_t stripe, std::shared_ptr<RecordBatch>* o
   return impl_->ReadStripe(stripe, out);
 }
 
-Result<std::shared_ptr<RecordBatch>> ORCFileReader::ReadStripe(int64_t stripe) {
-  std::shared_ptr<RecordBatch> recordBatch;
-  RETURN_NOT_OK(impl_->ReadStripe(stripe, &recordBatch));
-  return recordBatch;
-}
-
 Status ORCFileReader::ReadStripe(int64_t stripe, const std::vector<int>& include_indices,
                                  std::shared_ptr<RecordBatch>* out) {
   return impl_->ReadStripe(stripe, include_indices, out);
 }
 
-Result<std::shared_ptr<RecordBatch>> ORCFileReader::ReadStripe(
-    int64_t stripe, const std::vector<std::string>& include_names) {
-  std::shared_ptr<RecordBatch> recordBatch;
-  RETURN_NOT_OK(impl_->ReadStripe(stripe, include_names, &recordBatch));
-  return recordBatch;
-}
-
 Status ORCFileReader::Seek(int64_t row_number) { return impl_->Seek(row_number); }
 
 Status ORCFileReader::NextStripeReader(int64_t batch_sizes,

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

@@ -48,24 +48,11 @@ class ARROW_EXPORT ORCFileReader {
   static Status Open(const std::shared_ptr<io::RandomAccessFile>& file, MemoryPool* pool,
                      std::unique_ptr<ORCFileReader>* reader);
 
-  /// \brief Creates a new ORC reader
-  ///
-  /// \param[in] file the data source
-  /// \param[in] pool a MemoryPool to use for buffer allocations
-  /// \return the returned reader object
-  static Result<std::unique_ptr<ORCFileReader>> Open(
-      const std::shared_ptr<io::RandomAccessFile>& file, MemoryPool* pool);
-
   /// \brief Return the schema read from the ORC file
   ///
   /// \param[out] out the returned Schema object
   Status ReadSchema(std::shared_ptr<Schema>* out);
 
-  /// \brief Return the schema read from the ORC file
-  ///
-  /// \return the returned Schema object
-  Result<std::shared_ptr<Schema>> ReadSchema();
-
   /// \brief Read the file as a Table
   ///
   /// The table will be composed of one record batch per stripe.
@@ -89,14 +76,6 @@ class ARROW_EXPORT ORCFileReader {
   /// \param[out] out the returned Table
   Status Read(const std::vector<int>& include_indices, std::shared_ptr<Table>* out);
 
-  /// \brief Read the file as a Table
-  ///
-  /// The table will be composed of one record batch per stripe.
-  ///
-  /// \param[in] include_names the selected field names to read
-  /// \return the returned Table
-  Result<std::shared_ptr<Table>> Read(const std::vector<std::string>& include_names);
-
   /// \brief Read the file as a Table
   ///
   /// The table will be composed of one record batch per stripe.
@@ -113,12 +92,6 @@ class ARROW_EXPORT ORCFileReader {
   /// \param[out] out the returned RecordBatch
   Status ReadStripe(int64_t stripe, std::shared_ptr<RecordBatch>* out);
 
-  /// \brief Read a single stripe as a RecordBatch
-  ///
-  /// \param[in] stripe the stripe index
-  /// \return the returned RecordBatch
-  Result<std::shared_ptr<RecordBatch>> ReadStripe(int64_t stripe);
-
   /// \brief Read a single stripe as a RecordBatch
   ///
   /// \param[in] stripe the stripe index
@@ -127,14 +100,6 @@ class ARROW_EXPORT ORCFileReader {
   Status ReadStripe(int64_t stripe, const std::vector<int>& include_indices,
                     std::shared_ptr<RecordBatch>* out);
 
-  /// \brief Read a single stripe as a RecordBatch
-  ///
-  /// \param[in] stripe the stripe index
-  /// \param[in] include_names the selected field names to read
-  /// \return the returned RecordBatch
-  Result<std::shared_ptr<RecordBatch>> ReadStripe(
-      int64_t stripe, const std::vector<std::string>& include_names);
-
   /// \brief Seek to designated row. Invoke NextStripeReader() after seek
   ///        will return stripe reader starting from designated row.
   ///

cpp/src/arrow/dataset/CMakeLists.txt

@@ -36,10 +36,6 @@ if(ARROW_CSV)
   set(ARROW_DATASET_SRCS ${ARROW_DATASET_SRCS} file_csv.cc)
 endif()
 
-if(ARROW_ORC)
-  set(ARROW_DATASET_SRCS ${ARROW_DATASET_SRCS} file_orc.cc)
-endif()
-
 if(ARROW_PARQUET)
   set(ARROW_DATASET_LINK_STATIC ${ARROW_DATASET_LINK_STATIC} parquet_static)
   set(ARROW_DATASET_LINK_SHARED ${ARROW_DATASET_LINK_SHARED} parquet_shared)
@@ -120,10 +116,6 @@ if(ARROW_CSV)
   add_arrow_dataset_test(file_csv_test)
 endif()
 
-if(ARROW_ORC)
-  add_arrow_dataset_test(file_orc_test)
-endif()
-
 if(ARROW_PARQUET)
   add_arrow_dataset_test(file_parquet_test)
 endif()

cpp/src/arrow/dataset/api.h

@@ -25,6 +25,5 @@
 #include "arrow/dataset/file_base.h"
 #include "arrow/dataset/file_csv.h"
 #include "arrow/dataset/file_ipc.h"
-#include "arrow/dataset/file_orc.h"
 #include "arrow/dataset/file_parquet.h"
 #include "arrow/dataset/scanner.h"

cpp/src/arrow/dataset/expression.cc

@@ -612,17 +612,6 @@ std::vector<FieldRef> FieldsInExpression(const Expression& expr) {
   return fields;
 }
 
-bool ExpressionHasFieldRefs(const Expression& expr) {
-  if (expr.literal()) return false;
-
-  if (expr.field_ref()) return true;
-
-  for (const Expression& arg : CallNotNull(expr)->arguments) {
-    if (ExpressionHasFieldRefs(arg)) return true;
-  }
-  return false;
-}
-
 Result<Expression> FoldConstants(Expression expr) {
   return Modify(
       std::move(expr), [](Expression expr) { return expr; },

cpp/src/arrow/dataset/expression.h

@@ -159,10 +159,6 @@ Expression call(std::string function, std::vector<Expression> arguments,
 ARROW_DS_EXPORT
 std::vector<FieldRef> FieldsInExpression(const Expression&);
 
-/// Check if the expression references any fields.
-ARROW_EXPORT
-bool ExpressionHasFieldRefs(const Expression&);
-
 /// Assemble a mapping from field references to known values.
 ARROW_DS_EXPORT
 Result<std::unordered_map<FieldRef, Datum, FieldRef::Hash>> ExtractKnownFieldValues(

cpp/src/arrow/dataset/file_base.cc

@@ -84,13 +84,6 @@ Result<std::shared_ptr<io::InputStream>> FileSource::OpenCompressed(
   return io::CompressedInputStream::Make(codec.get(), std::move(file));
 }
 
-Future<util::optional<int64_t>> FileFormat::CountRows(
-   const std::shared_ptr<FileFragment>& file, Expression predicate,
-   const std::shared_ptr<ScanOptions>& options) {
- // Just adapt to ORC interface
- return Future<util::optional<int64_t>>::MakeFinished(util::nullopt);
-}
-
 Result<std::shared_ptr<FileFragment>> FileFormat::MakeFragment(
     FileSource source, std::shared_ptr<Schema> physical_schema) {
   return MakeFragment(std::move(source), literal(true), std::move(physical_schema));

cpp/src/arrow/dataset/file_base.h

@@ -180,10 +180,6 @@ class ARROW_DS_EXPORT FileFormat : public std::enable_shared_from_this<FileForma
       const std::shared_ptr<ScanOptions>& options,
       const std::shared_ptr<FileFragment>& file);
 
-  virtual Future<util::optional<int64_t>> CountRows(
-      const std::shared_ptr<FileFragment>& file, Expression predicate,
-      const std::shared_ptr<ScanOptions>& options);
-
   /// \brief Open a fragment
   virtual Result<std::shared_ptr<FileFragment>> MakeFragment(
       FileSource source, Expression partition_expression,