Use readers created upfront, add MetadataReaders interface

Signed-off-by: Annanay Agarwal <annanay.agarwal@grafana.com>

config.go

@@ -20,11 +20,11 @@ const (
 	DefaultSkipBloomFilters     = false
 )
 
-type MetadataCacher interface {
-	GetReader() io.ReaderAt
-	ShouldCacheBloomFilters() bool
-	ShouldCacheHeader() bool
-	ShouldCacheFooter() bool
+type MetadataReaders interface {
+	HeaderReader() io.ReaderAt
+	FooterReader() io.ReaderAt
+	PageIndexReader() io.ReaderAt
+	BloomFilterReader() io.ReaderAt
 }
 
 // The FileConfig type carries configuration options for parquet files.
@@ -40,7 +40,7 @@ type MetadataCacher interface {
 type FileConfig struct {
 	SkipPageIndex    bool
 	SkipBloomFilters bool
-	MetadataCacher   MetadataCacher
+	MetadataReaders  MetadataReaders
 }
 
 // DefaultFileConfig returns a new FileConfig value initialized with the
@@ -326,8 +326,8 @@ func SkipBloomFilters(skip bool) FileOption {
 	return fileOption(func(config *FileConfig) { config.SkipBloomFilters = skip })
 }
 
-func AddMetadataCacher(m MetadataCacher) FileOption {
-	return fileOption(func(config *FileConfig) { config.MetadataCacher = m })
+func AddMetadataCacher(m MetadataReaders) FileOption {
+	return fileOption(func(config *FileConfig) { config.MetadataReaders = m })
 }
 
 // PageBufferSize configures the size of column page buffers on parquet writers.

file.go

@@ -22,16 +22,16 @@ const (
 // File represents a parquet file. The layout of a Parquet file can be found
 // here: https://github.com/apache/parquet-format#file-format
 type File struct {
-	metadata       format.FileMetaData
-	protocol       thrift.CompactProtocol
-	reader         io.ReaderAt
-	size           int64
-	schema         *Schema
-	root           *Column
-	columnIndexes  []format.ColumnIndex
-	offsetIndexes  []format.OffsetIndex
-	rowGroups      []RowGroup
-	MetadataCacher MetadataCacher
+	metadata        format.FileMetaData
+	protocol        thrift.CompactProtocol
+	reader          io.ReaderAt
+	size            int64
+	schema          *Schema
+	root            *Column
+	columnIndexes   []format.ColumnIndex
+	offsetIndexes   []format.OffsetIndex
+	rowGroups       []RowGroup
+	MetadataReaders MetadataReaders
 }
 
 // OpenFile opens a parquet file and reads the content between offset 0 and the given
@@ -48,12 +48,12 @@ func OpenFile(r io.ReaderAt, size int64, options ...FileOption) (*File, error) {
 		return nil, err
 	}
 
-	f.MetadataCacher = c.MetadataCacher
+	f.MetadataReaders = c.MetadataReaders
 
 	// check if we should use cached reader for header
 	reader := r
-	if f.MetadataCacher != nil && f.MetadataCacher.ShouldCacheHeader() {
-		r = f.MetadataCacher.GetReader()
+	if f.MetadataReaders != nil {
+		r = f.MetadataReaders.HeaderReader()
 	}
 	if _, err := reader.ReadAt(b[:4], 0); err != nil {
 		return nil, fmt.Errorf("reading magic header of parquet file: %w", err)
@@ -64,8 +64,8 @@ func OpenFile(r io.ReaderAt, size int64, options ...FileOption) (*File, error) {
 
 	// check if we should use cached reader for footer
 	reader = r
-	if f.MetadataCacher != nil && f.MetadataCacher.ShouldCacheFooter() {
-		r = f.MetadataCacher.GetReader()
+	if f.MetadataReaders != nil {
+		r = f.MetadataReaders.FooterReader()
 	}
 	if _, err := reader.ReadAt(b[:8], size-8); err != nil {
 		return nil, fmt.Errorf("reading magic footer of parquet file: %w", err)
@@ -116,8 +116,8 @@ func OpenFile(r io.ReaderAt, size int64, options ...FileOption) (*File, error) {
 		p := thrift.CompactProtocol{}
 
 		reader := f.reader
-		if f.MetadataCacher != nil && f.MetadataCacher.ShouldCacheBloomFilters() {
-			reader = f.MetadataCacher.GetReader()
+		if f.MetadataReaders != nil {
+			reader = f.MetadataReaders.BloomFilterReader()
 		}
 		s := io.NewSectionReader(reader, 0, size)
 		d := thrift.NewDecoder(p.NewReader(s))
@@ -211,8 +211,8 @@ func (f *File) ReadPageIndex() ([]format.ColumnIndex, []format.OffsetIndex, erro
 		columnIndexData := indexBuffer[:columnIndexLength]
 
 		reader := f.reader
-		if f.MetadataCacher != nil && f.MetadataCacher.ShouldCachePageIndex() {
-			reader = f.MetadataCacher.GetReader()
+		if f.MetadataReaders != nil {
+			reader = f.MetadataReaders.PageIndexReader()
 		}
 		if _, err := reader.ReadAt(columnIndexData, columnIndexOffset); err != nil {
 			return nil, nil, fmt.Errorf("reading %d bytes column index at offset %d: %w", columnIndexLength, columnIndexOffset, err)
@@ -236,8 +236,8 @@ func (f *File) ReadPageIndex() ([]format.ColumnIndex, []format.OffsetIndex, erro
 		offsetIndexData := indexBuffer[:offsetIndexLength]
 
 		reader := f.reader
-		if f.MetadataCacher != nil && f.MetadataCacher.ShouldCachePageIndex() {
-			reader = f.MetadataCacher.GetReader()
+		if f.MetadataReaders != nil {
+			reader = f.MetadataReaders.PageIndexReader()
 		}
 		if _, err := reader.ReadAt(offsetIndexData, offsetIndexOffset); err != nil {
 			return nil, nil, fmt.Errorf("reading %d bytes offset index at offset %d: %w", offsetIndexLength, offsetIndexOffset, err)