Refactor types for output

cmd/parquet-cli/cmd_schema.go

@@ -140,6 +140,6 @@ func (r *metadataRow) Cells() []any {
 	}
 }
 
-func (r *metadataRow) Data() any {
+func (r *metadataRow) SerializableData() any {
 	return r.s
 }

pkg/inspect/aggregate.go

@@ -26,7 +26,7 @@ type Aggregate struct {
 	Stats []AggregateCellStats `json:"stats"`
 }
 
-func (rs *Aggregate) Data() any {
+func (rs *Aggregate) SerializableData() any {
 	return rs
 }
 

pkg/inspect/col_stats.go

@@ -51,7 +51,7 @@ type ColumnStats struct {
 	cells []any
 }
 
-func (rs *ColumnStats) Data() any {
+func (rs *ColumnStats) SerializableData() any {
 	return rs
 }
 

pkg/inspect/file_info.go

@@ -68,7 +68,7 @@ func (i *FileInfo) Add(k string, v any) {
 	i.keys = append(i.keys, k)
 }
 
-func (i *FileInfo) Data() any {
+func (i *FileInfo) SerializableData() any {
 	return i.elem
 }
 

pkg/inspect/row_dump.go

@@ -12,7 +12,7 @@ type DumpLine struct {
 	Line      []*parquet.Value
 }
 
-func (d *DumpLine) Data() any {
+func (d *DumpLine) SerializableData() any {
 	return d.Line
 }
 

pkg/inspect/row_stats.go

@@ -23,7 +23,7 @@ type RowStats struct {
 	Stats     []RowCellStats
 }
 
-func (rs *RowStats) Data() any {
+func (rs *RowStats) SerializableData() any {
 	return rs.Stats
 }
 

pkg/output/format.go

@@ -0,0 +1,44 @@
+package output
+
+import "errors"
+
+// Format describes a printable data representation.
+type Format string
+
+const (
+	FormatJSON = "json"
+	FormatCSV  = "csv"
+	FormatTab  = "tab"
+	FormatText = "text"
+)
+
+func (f *Format) Validate() error {
+	switch *f {
+	case FormatJSON, FormatTab, FormatCSV, FormatText:
+		return nil
+	default:
+		return errors.New("output format is expected to be 'json', 'tab', 'text' or 'csv'")
+	}
+}
+
+func formatsFor(data any) []Format {
+	var formats []Format
+	switch data.(type) {
+	case Serializable, SerializableIterator:
+		formats = append(formats, FormatJSON)
+	case Table, TableIterator:
+		formats = append(formats, FormatTab, FormatCSV)
+	case Text:
+		formats = append(formats, FormatText)
+	}
+	return formats
+}
+
+func supportsFormat(data any, f Format) bool {
+	for _, format := range formatsFor(data) {
+		if format == f {
+			return true
+		}
+	}
+	return false
+}

pkg/output/interfaces.go

@@ -0,0 +1,42 @@
+package output
+
+// A Table represents a tabular data that can also be printed as CSV.
+// Suitable for small tables that fit into memory.
+type Table interface {
+	Header() []string
+	Rows() []TableRow
+}
+
+// A TableIterator that can efficiently be printed as large table or CSV.
+// Suitable for larger tables that do not fit into memory.
+type TableIterator interface {
+	// Header returns the header of the table
+	Header() []any
+	// NextRow returns a new TableRow until the error is io.EOF
+	NextRow() (TableRow, error)
+}
+
+// A TableRow represents all data that belongs to a table row.
+type TableRow interface {
+	// Cells returns all table cells for this row. This is used to
+	// print tabular formats such csv. The returned slice has the same
+	// length as the header slice returned by the parent TableIterator.
+	Cells() []any
+}
+
+// Serializable represents data that can be converted to JSON or YAML.
+type Serializable interface {
+	// SerializableData returns arbitrary data that can be converted to formats like JSON or YAML.
+	SerializableData() any
+}
+
+// SerializableIterator represents a stream of data that can be converted to JSON or YAML.
+type SerializableIterator interface {
+	NextSerializable() (any, error)
+}
+
+// Text represents a multi line text that can be printed but is not a table or another
+// structured format such as JSON or YAML.
+type Text interface {
+	Text() (string, error)
+}

pkg/output/output.go

@@ -11,49 +11,8 @@ import (
 	"text/tabwriter"
 )
 
-// Format describes a printable data representation.
-type Format string
-
-const (
-	FormatJSON = "json"
-	FormatCSV  = "csv"
-	FormatTab  = "tab"
-)
-
-func (f *Format) Validate() error {
-	switch *f {
-	case FormatJSON, FormatTab, FormatCSV:
-		return nil
-	default:
-		return errors.New("output format is expected to be 'json', 'tab', or 'csv'")
-	}
-}
-
-// A Table that can be printed / encoded in different output formats.
-type Table interface {
-	// Header returns the header of the table
-	Header() []any
-	// NextRow returns a new TableRow until the error is io.EOF
-	NextRow() (TableRow, error)
-}
-
-// SerializableData represents table data that can be converted to JSON.
-type SerializableData interface {
-	// Data returns the table data suitable for structured data formats
-	// such as json.
-	Data() any
-}
-
-// A TableRow represents all data that belongs to a table row.
-type TableRow interface {
-	// Cells returns all table cells for this row. This is used to
-	// print tabular formats such csv. The returned slice has the same
-	// length as the header slice returned by the parent Table.
-	Cells() []any
-}
-
-// PrintTable writes the Table data to w using the provided format.
-func PrintTable(w io.Writer, f Format, data Table) error {
+// PrintTable writes the TableIterator data to w using the provided format.
+func PrintTable(w io.Writer, f Format, data TableIterator) error {
 	switch f {
 	case FormatJSON:
 		return printJSON(w, data)
@@ -66,7 +25,7 @@ func PrintTable(w io.Writer, f Format, data Table) error {
 	}
 }
 
-func printTab(w io.Writer, data Table) error {
+func printTab(w io.Writer, data TableIterator) error {
 	tw := tabwriter.NewWriter(w, 0, 0, 2, ' ', 0)
 
 	formatBuilder := strings.Builder{}
@@ -97,7 +56,7 @@ func printTab(w io.Writer, data Table) error {
 	return tw.Flush()
 }
 
-func printCSV(w io.Writer, data Table) error {
+func printCSV(w io.Writer, data TableIterator) error {
 	cw := csv.NewWriter(w)
 	cw.Comma = ';'
 
@@ -128,11 +87,11 @@ func printCSV(w io.Writer, data Table) error {
 	return cw.Error()
 }
 
-func printJSON(w io.Writer, data Table) error {
-	if serializable, ok := data.(SerializableData); ok {
+func printJSON(w io.Writer, data TableIterator) error {
+	if serializable, ok := data.(Serializable); ok {
 		enc := json.NewEncoder(w)
 		enc.SetIndent("", "  ")
-		return enc.Encode(serializable.Data())
+		return enc.Encode(serializable.SerializableData())
 	}
 
 	_, err := fmt.Fprintln(w, "[")
@@ -153,13 +112,13 @@ func printJSON(w io.Writer, data Table) error {
 		if err != nil {
 			return err
 		}
-		serializableRow, ok := row.(SerializableData)
+		serializableRow, ok := row.(Serializable)
 		if !ok {
 			return errors.New("JSON not supported for sub command")
 		}
 
 		buf.Reset()
-		err = json.NewEncoder(buf).Encode(serializableRow.Data())
+		err = json.NewEncoder(buf).Encode(serializableRow.SerializableData())
 		if err != nil {
 			return err
 		}