wgtmac commented on code in PR #14964:
URL: https://github.com/apache/arrow/pull/14964#discussion_r1064140451
##########
cpp/src/parquet/page_index.h:
##########
@@ -126,4 +130,75 @@ class PARQUET_EXPORT OffsetIndex {
virtual const std::vector<PageLocation>& page_locations() const = 0;
};
+/// \brief Interface for reading the page index from a Parquet row group.
+class PARQUET_EXPORT RowGroupPageIndexReader {
+ public:
+ virtual ~RowGroupPageIndexReader() = default;
+
+ /// \brief Read column index of a column chunk.
+ ///
+ /// \param[in] i column id of the column chunk.
+ /// \returns error Result if either column id is invalid or column index
does not exist.
+ virtual ::arrow::Result<std::shared_ptr<ColumnIndex>> GetColumnIndex(int32_t
i) = 0;
+
+ /// \brief Read offset index of a column chunk.
+ ///
+ /// \param[in] i column id of the column chunk.
+ /// \returns error Result if either column id is invalid or offset index
does not exist.
+ virtual ::arrow::Result<std::shared_ptr<OffsetIndex>> GetOffsetIndex(int32_t
i) = 0;
+};
+
+/// \brief Interface for reading the page index from a Parquet file.
+class PARQUET_EXPORT PageIndexReader {
+ public:
+ virtual ~PageIndexReader() = default;
+
+ /// \brief Create a PageIndexReader instance.
+ static std::shared_ptr<PageIndexReader> Make(
+ ::arrow::io::RandomAccessFile* input, std::shared_ptr<FileMetaData>
file_metadata,
Review Comment:
The return value must not outlive any of the input parameters. Added comment
to explain it.
##########
cpp/src/parquet/page_index.h:
##########
@@ -126,4 +130,75 @@ class PARQUET_EXPORT OffsetIndex {
virtual const std::vector<PageLocation>& page_locations() const = 0;
};
+/// \brief Interface for reading the page index from a Parquet row group.
+class PARQUET_EXPORT RowGroupPageIndexReader {
+ public:
+ virtual ~RowGroupPageIndexReader() = default;
+
+ /// \brief Read column index of a column chunk.
+ ///
+ /// \param[in] i column id of the column chunk.
+ /// \returns error Result if either column id is invalid or column index
does not exist.
+ virtual ::arrow::Result<std::shared_ptr<ColumnIndex>> GetColumnIndex(int32_t
i) = 0;
+
+ /// \brief Read offset index of a column chunk.
+ ///
+ /// \param[in] i column id of the column chunk.
+ /// \returns error Result if either column id is invalid or offset index
does not exist.
+ virtual ::arrow::Result<std::shared_ptr<OffsetIndex>> GetOffsetIndex(int32_t
i) = 0;
+};
+
+/// \brief Interface for reading the page index from a Parquet file.
+class PARQUET_EXPORT PageIndexReader {
+ public:
+ virtual ~PageIndexReader() = default;
+
+ /// \brief Create a PageIndexReader instance.
+ static std::shared_ptr<PageIndexReader> Make(
+ ::arrow::io::RandomAccessFile* input, std::shared_ptr<FileMetaData>
file_metadata,
+ const ReaderProperties& properties,
+ std::shared_ptr<InternalFileDecryptor> file_decryptor = NULLPTR);
+
+ /// \brief Get the page index reader of a specific row group.
+ /// \param[in] i row group ordinal to get page index reader.
+ /// \returns RowGroupPageIndexReader of the specified row group.
+ virtual ::arrow::Result<std::shared_ptr<RowGroupPageIndexReader>>
RowGroup(int i) = 0;
+
+ /// \brief Advise the reader which part of page index will be read later.
+ ///
+ /// The PageIndexReader implementation can optionally prefetch and cache
page index
+ /// that may be read later. Follow-up read should not fail even if
WillNeed() is not
+ /// called, or the requested page index is out of range from WillNeed() call.
+ ///
+ /// \param[in] row_group_indices list of row group ordinal to read page
index later.
+ /// \param[in] need_column_index tell if column index is required later.
+ /// \param[in] need_offset_index tell if offset index is required later.
+ virtual void WillNeed(const std::vector<int32_t>& row_group_indices,
+ bool need_column_index, bool need_offset_index) = 0;
Review Comment:
Fixed
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
