README
¶
Parquet
Parquet generates a parquet reader and writer for a single struct.
Installation
go get -u github.com/parsyl/parquet/...
Usage
First define a struct for the data to be written to parquet:
type Person struct {
ID int32 `parquet:"id"`
Age *int32 `parquet:"age"`
}
Next, add a go:generate comment somewhere (in this example all code lives in main.go):
//go:generate parquetgen -input main.go -type Person -package main
Generate the code for the reader and writer:
$ go generate
A new file (parquet.go) has now been written that defines ParquetWriter and ParquetReader. Next, make use of the writer and reader:
package main
import (
"bytes"
"encoding/json"
)
func main() {
var buf bytes.Buffer
// MaxPageSize optionally defines the number of rows in each column chunk (default is 1000)
w, err := NewParquetWriter(&buf, MaxPageSize(10000))
if err != nil {
log.Fatal(err)
}
w.Add(Person{ID: 1, Age: getAge(30)})
w.Add(Person{ID: 2})
// Each call to write creates a new parquet row group.
if err := w.Write(); err != nil {
log.Fatal(err)
}
// Close must be called when you are done. It writes
// the parquet metadata at the end of the file.
if err := w.Close(); err != nil {
log.Fatal(err)
}
r, err := NewParquetReader(bytes.NewReader(buf.Bytes()))
if err != nil {
log.Fatal(err)
}
enc := json.NewEncoder(os.Stdout)
for r.Next() {
var p Person
r.Scan(&p)
enc.Encode(p)
}
if err := r.Error(); err != nil {
log.Fatal(err)
}
}
func getAge(a int32) *int32 { return &a }
See this for a complete example of how to use it.
Supported Types
The struct used to define the parquet data can have the following types:
int32
uint32
int64
uint64
float32
float64
string
bool
Each of these types may be a pointer to indicate that the data is optional. The struct can also embed another struct:
type Being struct {
ID int32 `parquet:"id"`
Age *int32 `parquet:"age"`
}
type Person struct {
Being
Username string `parquet:"username"`
}
Nested structs, however, are not supported at this time. If you want a field to be excluded from parquet you can tag it with a dash or make it private like so:
type Being struct {
ID int32 `parquet:"id"`
Password string`parquet:"-"` //will not be written to parquet
age int32 //will not be written to parquet
}
Documentation
¶
Index ¶
- func BoolType(se *sch.SchemaElement)
- func Float32Type(se *sch.SchemaElement)
- func Float64Type(se *sch.SchemaElement)
- func GetBools(r io.Reader, n int) ([]bool, error)
- func Int32Type(se *sch.SchemaElement)
- func Int64Type(se *sch.SchemaElement)
- func ReadLevels(r io.Reader) ([]int64, int, error)
- func RepetitionOptional(se *sch.SchemaElement)
- func RepetitionRequired(se *sch.SchemaElement)
- func StringType(se *sch.SchemaElement)
- func Uint32Type(se *sch.SchemaElement)
- func Uint64Type(se *sch.SchemaElement)
- func WriteLevels(w io.Writer, levels []int64) error
- type Field
- type FieldFunc
- type Metadata
- func (m *Metadata) Footer(w io.Writer) error
- func (m *Metadata) Offsets() (map[string][]Position, error)
- func (m *Metadata) PageHeader(r io.ReadSeeker) (*sch.PageHeader, error)
- func (m *Metadata) ReadFooter(r io.ReadSeeker) error
- func (m *Metadata) RowGroups() []RowGroup
- func (m *Metadata) Rows() int64
- func (m *Metadata) StartRowGroup(fields ...Field)
- func (m *Metadata) WritePageHeader(w io.Writer, col string, dataLen, compressedLen, count int) error
- type OptionalField
- type Position
- type ReadCounter
- type RequiredField
- type RowGroup
- type WriteCounter
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func BoolType ¶
func BoolType(se *sch.SchemaElement)
func Float32Type ¶
func Float32Type(se *sch.SchemaElement)
func Float64Type ¶
func Float64Type(se *sch.SchemaElement)
func Int32Type ¶
func Int32Type(se *sch.SchemaElement)
func Int64Type ¶
func Int64Type(se *sch.SchemaElement)
func ReadLevels ¶
ReadLevels reads the RLE encoded definition levels
func RepetitionOptional ¶
func RepetitionOptional(se *sch.SchemaElement)
func RepetitionRequired ¶
func RepetitionRequired(se *sch.SchemaElement)
func StringType ¶
func StringType(se *sch.SchemaElement)
func Uint32Type ¶
func Uint32Type(se *sch.SchemaElement)
func Uint64Type ¶
func Uint64Type(se *sch.SchemaElement)
Types ¶
type FieldFunc ¶
type FieldFunc func(*sch.SchemaElement)
type Metadata ¶
type Metadata struct {
// contains filtered or unexported fields
}
func (*Metadata) PageHeader ¶
func (m *Metadata) PageHeader(r io.ReadSeeker) (*sch.PageHeader, error)
func (*Metadata) ReadFooter ¶
func (m *Metadata) ReadFooter(r io.ReadSeeker) error
func (*Metadata) StartRowGroup ¶
type OptionalField ¶ added in v0.0.6
type OptionalField struct {
Defs []int64
// contains filtered or unexported fields
}
func NewOptionalField ¶ added in v0.0.6
func NewOptionalField(col string) OptionalField
func (*OptionalField) DoRead ¶ added in v0.0.6
func (f *OptionalField) DoRead(r io.ReadSeeker, meta *Metadata, pos Position) (io.Reader, error)
func (*OptionalField) Name ¶ added in v0.0.6
func (f *OptionalField) Name() string
func (*OptionalField) Values ¶ added in v0.0.6
func (f *OptionalField) Values() int
type ReadCounter ¶ added in v0.0.6
type ReadCounter struct {
// contains filtered or unexported fields
}
type RequiredField ¶ added in v0.0.6
type RequiredField struct {
// contains filtered or unexported fields
}
func NewRequiredField ¶ added in v0.0.6
func NewRequiredField(col string) RequiredField
func (*RequiredField) DoRead ¶ added in v0.0.6
func (f *RequiredField) DoRead(r io.ReadSeeker, meta *Metadata, pos Position) (io.Reader, error)
func (*RequiredField) Name ¶ added in v0.0.6
func (f *RequiredField) Name() string
type RowGroup ¶ added in v0.0.6
type RowGroup struct {
Rows int64
// contains filtered or unexported fields
}
func (*RowGroup) Columns ¶ added in v0.0.6
func (r *RowGroup) Columns() []*sch.ColumnChunk
type WriteCounter ¶ added in v0.0.6
type WriteCounter struct {
// contains filtered or unexported fields
}
func NewWriteCounter ¶ added in v0.0.6
func NewWriteCounter(w io.Writer) *WriteCounter
Source Files
Directories
Click to show internal directories.
Click to hide internal directories.