Documentation ¶
Overview ¶
Package avro encodes/decodes avro schemas to your struct or a map.
Overview ¶
Go-avro parses .avsc schemas from files and then lets you work with them.
schema, err := avro.ParseSchemaFile("person.avsc") // important: handle err!
Struct Mapping ¶
When using SpecificDecoder, the implementation uses struct tags to map avro messages into your struct. This helps because it makes schema evolution easier and avoids having a ton of casting and dictionary checks in your user code.
Say you had the schema:
{ "type": "record", "name": "Person", "fields" : [ {"name": "id", "type": "int"}, {"name": "name", "type": "string"} {"name": "location", "type": { "type": "record", "name": "Location", "fields": [ {"name": "latitude", "type": "double"}, {"name": "longitude", "type": "double"} ] }} ] }
This could be mapped to a SpecificRecord by using structs:
type Person struct { Id int32 `avro:"id"` Name string `avro:"name"` Location *Location `avro:"location"` } type Location struct { Latitude float64 Longitude float64 }
If the `avro:` struct tag is omitted, the default mapping lower-cases the first letter only. It's better to just explicitly define where possible.
Mapped types:
- avro 'int' is always 32-bit, so maps to golang 'int32'
- avro 'long' is always mapped to 'int64'
- avro 'float' -> float32
- avro 'double' -> 'float64'
- most other ones are obvious
Type unions are a bit more tricky. For a complex type union, the only valid mapping is interface{}. However, for a type union with only "null" and one other type (very typical) you can map it as a pointer type and keep type safety.
Index ¶
- Constants
- Variables
- func GetFullName(schema Schema) string
- func LoadSchemas(path string) map[string]Schema
- func NewFieldDoesNotExistError(field string) error
- type ArraySchema
- type AvroRecord
- type BooleanSchema
- type BytesSchema
- type CodeGenerator
- type DataBlock
- type DataFileReader
- type DataFileWriter
- type DatumReader
- type DatumWriter
- type Decoder
- type DoubleSchema
- type Encoder
- type EnumSchema
- type FixedSchema
- type FloatSchema
- type GenericDatumReader
- type GenericDatumWriter
- type GenericEnum
- type GenericRecord
- type IntSchema
- type LongSchema
- type MapSchema
- type Marshaler
- type NullSchema
- type RecordSchema
- type RecursiveSchema
- type Schema
- type SchemaField
- type SpecificDatumReader
- type SpecificDatumWriter
- type StringSchema
- type UnionSchema
- func (*UnionSchema) GetName() string
- func (s *UnionSchema) GetType(v reflect.Value) int
- func (s *UnionSchema) MarshalJSON() ([]byte, error)
- func (*UnionSchema) Prop(key string) (interface{}, bool)
- func (s *UnionSchema) String() string
- func (*UnionSchema) Type() int
- func (s *UnionSchema) Validate(v reflect.Value) bool
- type Unmarshaler
Examples ¶
Constants ¶
const ( // Record schema type constant Record int = iota // Enum schema type constant Enum // Array schema type constant Array // Map schema type constant Map // Union schema type constant Union // Fixed schema type constant Fixed // String schema type constant String // Bytes schema type constant Bytes // Int schema type constant Int // Long schema type constant Long // Float schema type constant Float // Double schema type constant Double // Boolean schema type constant Boolean // Null schema type constant Null // Recursive schema type constant. Recursive is an artificial type that means a Record schema without its definition // that should be looked up in some registry. Recursive )
Variables ¶
var ErrBlockNotFinished = errors.New("Block read is unfinished")
Happens when trying to read next block without finishing the previous one.
var ErrIntOverflow = errors.New("Overflowed an int value")
Happens when the given value to decode overflows maximum int32 value.
var ErrInvalidBool = errors.New("Invalid bool value")
Happens when given value to decode as bool is neither 0x00 nor 0x01.
var ErrInvalidFixedSize = errors.New("Invalid Fixed type size")
Happens when avro schema contains invalid value for fixed size.
var ErrInvalidInt = errors.New("Invalid int value")
Happens when given value to decode as a int is invalid
var ErrInvalidLong = errors.New("Invalid long value")
Happens when given value to decode as a long is invalid
var ErrInvalidSchema = errors.New("Invalid schema")
Happens when avro schema is unparsable or is invalid in any other way.
var ErrInvalidStringLength = errors.New("Invalid string length")
Happens when given value to decode as string has either negative or undecodable length.
var ErrLongOverflow = errors.New("Overflowed a long value")
Happens when the given value to decode overflows maximum int64 value.
var ErrNegativeBytesLength = errors.New("Negative bytes length")
Happens when given value to decode as bytes has negative length.
var ErrNotAvroFile = errors.New("Not an Avro data file")
Indicates the given file to decode does not correspond to Avro data file format.
var ErrSchemaNotSet = errors.New("Schema not set")
Happens when a datum reader has no set schema.
var ErrUnexpectedEOF = io.ErrUnexpectedEOF
Signals that an end of file or stream has been reached unexpectedly.
var ErrUnionTypeOverflow = errors.New("Union type overflow")
UnionTypeOverflow happens when the numeric index of the union type is invalid.
Functions ¶
func GetFullName ¶
GetFullName returns a fully-qualified name for a schema if possible. The format is namespace.name.
func LoadSchemas ¶
LoadSchemas loads and parses a schema file or directory. Directory names MUST end with "/"
func NewFieldDoesNotExistError ¶
Specify a custom error message for indicating which necessary field in the struct is missing.
Types ¶
type ArraySchema ¶
ArraySchema implements Schema and represents Avro array type.
func (*ArraySchema) GetName ¶
func (*ArraySchema) GetName() string
GetName returns a type name for this ArraySchema.
func (*ArraySchema) MarshalJSON ¶
func (s *ArraySchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON.
func (*ArraySchema) Prop ¶
func (s *ArraySchema) Prop(key string) (interface{}, bool)
Prop gets a custom non-reserved property from this schema and a bool representing if it exists.
func (*ArraySchema) String ¶
func (s *ArraySchema) String() string
String returns a JSON representation of ArraySchema.
func (*ArraySchema) Type ¶
func (*ArraySchema) Type() int
Type returns a type constant for this ArraySchema.
type AvroRecord ¶
type AvroRecord interface { // Schema returns an Avro schema for this AvroRecord. Schema() Schema }
AvroRecord is an interface for anything that has an Avro schema and can be serialized/deserialized by this library.
type BooleanSchema ¶
type BooleanSchema struct{}
BooleanSchema implements Schema and represents Avro boolean type.
func (*BooleanSchema) GetName ¶
func (*BooleanSchema) GetName() string
GetName returns a type name for this BooleanSchema.
func (*BooleanSchema) MarshalJSON ¶
func (*BooleanSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*BooleanSchema) Prop ¶
func (*BooleanSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for BooleanSchema.
func (*BooleanSchema) String ¶
func (*BooleanSchema) String() string
String returns a JSON representation of BooleanSchema.
func (*BooleanSchema) Type ¶
func (*BooleanSchema) Type() int
Type returns a type constant for this BooleanSchema.
type BytesSchema ¶
type BytesSchema struct{}
BytesSchema implements Schema and represents Avro bytes type.
func (*BytesSchema) GetName ¶
func (*BytesSchema) GetName() string
GetName returns a type name for this BytesSchema.
func (*BytesSchema) MarshalJSON ¶
func (*BytesSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*BytesSchema) Prop ¶
func (*BytesSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for BytesSchema.
func (*BytesSchema) String ¶
func (*BytesSchema) String() string
String returns a JSON representation of BytesSchema.
func (*BytesSchema) Type ¶
func (*BytesSchema) Type() int
Type returns a type constant for this BytesSchema.
type CodeGenerator ¶
type CodeGenerator struct {
// contains filtered or unexported fields
}
CodeGenerator is a code generation tool for structs from given Avro schemas.
func NewCodeGenerator ¶
func NewCodeGenerator(schemas []string) *CodeGenerator
NewCodeGenerator creates a new CodeGenerator for given Avro schemas.
func (*CodeGenerator) Generate ¶
func (codegen *CodeGenerator) Generate() (string, error)
Generate generates source code for Avro schemas specified on creation. The ouput is Go formatted source code that contains struct definitions for all given schemas. May return an error if code generation fails, e.g. due to unparsable schema.
type DataBlock ¶
type DataBlock struct { // Number of entries encoded in Data. NumEntries int64 // Size of data buffer in bytes. BlockSize int // Number of unread entries in this DataBlock. BlockRemaining int64 // contains filtered or unexported fields }
DataBlock is a structure that holds a certain amount of entries and the actual buffer to read from.
type DataFileReader ¶
type DataFileReader struct {
// contains filtered or unexported fields
}
DataFileReader is a reader for Avro Object Container Files. More here: https://avro.apache.org/docs/current/spec.html#Object+Container+Files
Example ¶
package main import ( "log" avro "gopkg.in/avro.v0" ) type SomeStruct struct { Name string } func main() { // Create a reader open for reading on a data file. reader, err := avro.NewDataFileReader("filename.avro") if err != nil { log.Fatal(err) } defer reader.Close() for reader.HasNext() { var dest SomeStruct // or a *avro.GenericRecord if err := reader.Next(&dest); err != nil { // Error specific to decoding a single record } log.Printf("Decoded record %v", dest) } // If there was any error that stopped the reader loop, this is how we know if err := reader.Err(); err != nil { log.Fatal(err) } }
Output:
func NewDataFileReader ¶
func NewDataFileReader(filename string, ignoreMe ...DatumReader) (*DataFileReader, error)
NewDataFileReader enables reading an object container file from the filesystem. May return an error if the file contains invalid data or is just missing.
The second DatumReader argument is deprecated, only there for source compatibility. Will be removed in an upcoming compatibility break.
func (*DataFileReader) Close ¶
func (reader *DataFileReader) Close() error
Close the underlying file if necessary.
Needed with filesystem files if you want to not leak filehandles. Returns any error in closing.
func (*DataFileReader) Err ¶
func (reader *DataFileReader) Err() error
Err returns the last encountered error.
Will not return io.EOF if that was the last error.
func (*DataFileReader) HasNext ¶
func (reader *DataFileReader) HasNext() bool
HasNext is used in a for loop to know you can continue on.
If there was an I/O or decoding error in decoding a block, then HasNext will be false, even if there might be more data in the data file.
It might be possible to recover from corrupted data by jumping to the next block by using the NextBlock() but this is not guaranteed.
func (*DataFileReader) Next ¶
func (reader *DataFileReader) Next(v interface{}) error
Next reads the next value from file and fills the given value with data.
v can be anything a DatumReader would accept, including a pointer to a struct that is compatible with this file's schema, an allocated *GenericRecord, or a **GenericRecord (library allocates for you)
Will error with io.EOF if you're past the end, loop HasNext() to prevent.
func (*DataFileReader) NextBlock ¶
func (reader *DataFileReader) NextBlock() error
NextBlock tells this DataFileReader to skip current block and move to next one.
This is not typically needed as the Next() loop will automatically advance to the next block for you.
May return an error if the block is malformed or io.EOF if no more blocks left to read.
type DataFileWriter ¶
type DataFileWriter struct {
// contains filtered or unexported fields
}
DataFileWriter lets you write object container files.
func NewDataFileWriter ¶
func NewDataFileWriter(output io.Writer, schema Schema, datumWriter DatumWriter) (writer *DataFileWriter, err error)
NewDataFileWriter creates a new DataFileWriter for given output and schema using the given DatumWriter to write the data to that Writer. May return an error if writing fails.
func (*DataFileWriter) Close ¶
func (w *DataFileWriter) Close() error
Close this DataFileWriter. This is required to finish out the data file format. After Close() is called, this DataFileWriter cannot be used anymore.
func (*DataFileWriter) Flush ¶
func (w *DataFileWriter) Flush() error
Flush out any previously written datums to our underlying io.Writer. Does nothing if no datums had previously been written.
It's up to the library user to decide how often to flush; doing it often will spend a lot of time on tiny I/O but save memory.
func (*DataFileWriter) Write ¶
func (w *DataFileWriter) Write(v interface{}) error
Write out a single datum.
Encoded datums are buffered internally and will not be written to the underlying io.Writer until Flush() is called.
type DatumReader ¶
type DatumReader interface { // Reads a single structured entry using this DatumReader according to provided Schema. // Accepts a value to fill with data and a Decoder to read from. Given value MUST be of pointer type. // May return an error indicating a read failure. Read(v interface{}, dec Decoder) error }
DatumReader is an interface that is responsible for reading structured data according to schema from a decoder
func NewDatumReader ¶
func NewDatumReader(schema Schema) DatumReader
NewDatumReader creates a DatumReader that can handle both GenericRecord and also aribtrary structs.
This is the preferred implementation at this point in time.
type DatumWriter ¶
type DatumWriter interface { // Write writes a single entry using this DatumWriter according to provided Schema. // Accepts a value to write and Encoder to write to. // May return an error indicating a write failure. Write(obj interface{}, enc Encoder) error }
DatumWriter is an interface that is responsible for writing structured data according to schema to an encoder.
func NewDatumWriter ¶
func NewDatumWriter(schema Schema) DatumWriter
NewDatumWriter creates a DatumWriter that can handle both GenericRecord and also aribtrary structs.
This is the preferred implementation at this point in time.
type Decoder ¶
type Decoder interface { // Reads a null value. Returns a decoded value and an error if it occurs. ReadNull() (interface{}, error) // Reads a boolean value. Returns a decoded value and an error if it occurs. ReadBoolean() (bool, error) // Reads an in value. Returns a decoded value and an error if it occurs. ReadInt() (int32, error) // Reads a long value. Returns a decoded value and an error if it occurs. ReadLong() (int64, error) // Reads a float value. Returns a decoded value and an error if it occurs. ReadFloat() (float32, error) // Reads a double value. Returns a decoded value and an error if it occurs. ReadDouble() (float64, error) // Reads a bytes value. Returns a decoded value and an error if it occurs. ReadBytes() ([]byte, error) // Reads a string value. Returns a decoded value and an error if it occurs. ReadString() (string, error) // Reads an enum value (which is an Avro int value). Returns a decoded value and an error if it occurs. ReadEnum() (int32, error) // Reads and returns the size of the first block of an array. If call to this return non-zero, then the caller // should read the indicated number of items and then call ArrayNext() to find out the number of items in the // next block. Returns a decoded value and an error if it occurs. ReadArrayStart() (int64, error) // Processes the next block of an array and returns the number of items in the block. // Returns a decoded value and an error if it occurs. ArrayNext() (int64, error) // Reads and returns the size of the first block of map entries. If call to this return non-zero, then the caller // should read the indicated number of items and then call MapNext() to find out the number of items in the // next block. Usage is similar to ReadArrayStart(). Returns a decoded value and an error if it occurs. ReadMapStart() (int64, error) // Processes the next block of map entries and returns the number of items in the block. // Returns a decoded value and an error if it occurs. MapNext() (int64, error) // Reads fixed sized binary object into the provided buffer. // Returns an error if it occurs. ReadFixed([]byte) error }
Decoder is an interface that provides low-level support for deserializing Avro values.
func NewBinaryDecoder ¶
NewBinaryDecoder creates a new BinaryDecoder to read from a given buffer.
func NewBinaryDecoderReader ¶
NewBinaryDecoderReader creates a new BinaryDecoder to read from a given io.Reader.
This decoder makes a lot of very small reads from the underlying io.Reader. If this is some high-latency object like a network socket or file, consider passing some sort of buffered reader like a bufio.Reader.
type DoubleSchema ¶
type DoubleSchema struct{}
DoubleSchema implements Schema and represents Avro double type.
func (*DoubleSchema) GetName ¶
func (*DoubleSchema) GetName() string
GetName returns a type name for this DoubleSchema.
func (*DoubleSchema) MarshalJSON ¶
func (*DoubleSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*DoubleSchema) Prop ¶
func (*DoubleSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for DoubleSchema.
func (*DoubleSchema) String ¶
func (*DoubleSchema) String() string
Returns a JSON representation of DoubleSchema.
func (*DoubleSchema) Type ¶
func (*DoubleSchema) Type() int
Type returns a type constant for this DoubleSchema.
type Encoder ¶
type Encoder interface { // Writes a null value. Doesn't actually do anything but may advance the state of Encoder implementation if it // is stateful. WriteNull(interface{}) // Writes a boolean value. WriteBoolean(bool) // Writes an int value. WriteInt(int32) // Writes a long value. WriteLong(int64) // Writes a float value. WriteFloat(float32) // Writes a double value. WriteDouble(float64) // Writes a bytes value. WriteBytes([]byte) // Writes a string value. WriteString(string) // WriteArrayStart should be called when starting to serialize an array providing it with a number of items in // array block. WriteArrayStart(int64) // WriteArrayNext should be called after finishing writing an array block either passing it the number of items in // next block or 0 indicating the end of array. WriteArrayNext(int64) // WriteMapStart should be called when starting to serialize a map providing it with a number of items in // map block. WriteMapStart(int64) // WriteMapNext should be called after finishing writing a map block either passing it the number of items in // next block or 0 indicating the end of map. WriteMapNext(int64) // Writes raw bytes to this Encoder. WriteRaw([]byte) }
Encoder is an interface that provides low-level support for serializing Avro values.
func NewBinaryEncoder ¶
NewBinaryEncoder creates a new BinaryEncoder that will write to a given io.Writer.
type EnumSchema ¶
type EnumSchema struct { Name string Namespace string Aliases []string Doc string Symbols []string Properties map[string]interface{} }
EnumSchema implements Schema and represents Avro enum type.
func (*EnumSchema) GetName ¶
func (s *EnumSchema) GetName() string
GetName returns an enum name for this EnumSchema.
func (*EnumSchema) MarshalJSON ¶
func (s *EnumSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON.
func (*EnumSchema) Prop ¶
func (s *EnumSchema) Prop(key string) (interface{}, bool)
Prop gets a custom non-reserved property from this schema and a bool representing if it exists.
func (*EnumSchema) String ¶
func (s *EnumSchema) String() string
String returns a JSON representation of EnumSchema.
func (*EnumSchema) Type ¶
func (*EnumSchema) Type() int
Type returns a type constant for this EnumSchema.
type FixedSchema ¶
FixedSchema implements Schema and represents Avro fixed type.
func (*FixedSchema) GetName ¶
func (s *FixedSchema) GetName() string
GetName returns a fixed name for this FixedSchema.
func (*FixedSchema) MarshalJSON ¶
func (s *FixedSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON.
func (*FixedSchema) Prop ¶
func (s *FixedSchema) Prop(key string) (interface{}, bool)
Prop gets a custom non-reserved property from this schema and a bool representing if it exists.
func (*FixedSchema) String ¶
func (s *FixedSchema) String() string
String returns a JSON representation of FixedSchema.
func (*FixedSchema) Type ¶
func (*FixedSchema) Type() int
Type returns a type constant for this FixedSchema.
type FloatSchema ¶
type FloatSchema struct{}
FloatSchema implements Schema and represents Avro float type.
func (*FloatSchema) GetName ¶
func (*FloatSchema) GetName() string
GetName returns a type name for this FloatSchema.
func (*FloatSchema) MarshalJSON ¶
func (*FloatSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*FloatSchema) Prop ¶
func (*FloatSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for FloatSchema.
func (*FloatSchema) String ¶
func (*FloatSchema) String() string
String returns a JSON representation of FloatSchema.
func (*FloatSchema) Type ¶
func (*FloatSchema) Type() int
Type returns a type constant for this FloatSchema.
type GenericDatumReader ¶
type GenericDatumReader struct {
// contains filtered or unexported fields
}
GenericDatumReader implements DatumReader and is used for filling GenericRecords or other Avro supported types (full list is: interface{}, bool, int32, int64, float32, float64, string, slices of any type, maps with string keys and any values, GenericEnums) with data. Each value passed to Read is expected to be a pointer.
func NewGenericDatumReader ¶
func NewGenericDatumReader() *GenericDatumReader
NewGenericDatumReader creates a new GenericDatumReader.
func (*GenericDatumReader) Read ¶
func (reader *GenericDatumReader) Read(v interface{}, dec Decoder) error
Read reads a single entry using this GenericDatumReader. Accepts a value to fill with data and a Decoder to read from. Given value MUST be of pointer type. May return an error indicating a read failure.
func (*GenericDatumReader) SetSchema ¶
func (reader *GenericDatumReader) SetSchema(schema Schema) DatumReader
SetSchema sets the schema for this GenericDatumReader to know the data structure. Note that it must be called before calling Read.
type GenericDatumWriter ¶
type GenericDatumWriter struct {
// contains filtered or unexported fields
}
GenericDatumWriter implements DatumWriter and is used for writing GenericRecords or other Avro supported types (full list is: interface{}, bool, int32, int64, float32, float64, string, slices of any type, maps with string keys and any values, GenericEnums) to a given Encoder.
func NewGenericDatumWriter ¶
func NewGenericDatumWriter() *GenericDatumWriter
NewGenericDatumWriter creates a new GenericDatumWriter.
func (*GenericDatumWriter) SetSchema ¶
func (writer *GenericDatumWriter) SetSchema(schema Schema) DatumWriter
SetSchema sets the provided schema for this GenericDatumWriter to know the data structure. Note that it must be called before calling Write.
func (*GenericDatumWriter) Write ¶
func (writer *GenericDatumWriter) Write(obj interface{}, enc Encoder) error
Write writes a single entry using this GenericDatumWriter according to provided Schema. Accepts a value to write and Encoder to write to. May return an error indicating a write failure.
type GenericEnum ¶
type GenericEnum struct { // Avro enum symbols. Symbols []string // contains filtered or unexported fields }
GenericEnum is a generic Avro enum representation. This is still subject to change and may be rethought.
func NewGenericEnum ¶
func NewGenericEnum(symbols []string) *GenericEnum
NewGenericEnum returns a new GenericEnum that uses provided enum symbols.
func (*GenericEnum) Get ¶
func (enum *GenericEnum) Get() string
Get gets the string value for this enum (e.g. symbol).
func (*GenericEnum) GetIndex ¶
func (enum *GenericEnum) GetIndex() int32
GetIndex gets the numeric value for this enum.
func (*GenericEnum) Set ¶
func (enum *GenericEnum) Set(symbol string)
Set sets the string value for this enum (e.g. symbol). Panics if the given symbol does not exist in this enum.
func (*GenericEnum) SetIndex ¶
func (enum *GenericEnum) SetIndex(index int32)
SetIndex sets the numeric value for this enum.
type GenericRecord ¶
type GenericRecord struct {
// contains filtered or unexported fields
}
GenericRecord is a generic instance of a record schema. Fields are accessible by their name.
func NewGenericRecord ¶
func NewGenericRecord(schema Schema) *GenericRecord
NewGenericRecord creates a new GenericRecord.
func (*GenericRecord) Get ¶
func (gr *GenericRecord) Get(name string) interface{}
Get gets a value by its name.
func (*GenericRecord) Map ¶
func (gr *GenericRecord) Map() map[string]interface{}
Map returns a map representation of this GenericRecord.
func (*GenericRecord) Schema ¶
func (gr *GenericRecord) Schema() Schema
Schema returns a schema for this GenericRecord.
func (*GenericRecord) Set ¶
func (gr *GenericRecord) Set(name string, value interface{})
Set sets a value for a given name.
func (*GenericRecord) String ¶
func (gr *GenericRecord) String() string
String returns a JSON representation of this GenericRecord.
type IntSchema ¶
type IntSchema struct{}
IntSchema implements Schema and represents Avro int type.
func (*IntSchema) MarshalJSON ¶
MarshalJSON serializes the given schema as JSON. Never returns an error.
type LongSchema ¶
type LongSchema struct{}
LongSchema implements Schema and represents Avro long type.
func (*LongSchema) GetName ¶
func (*LongSchema) GetName() string
GetName returns a type name for this LongSchema.
func (*LongSchema) MarshalJSON ¶
func (*LongSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*LongSchema) Prop ¶
func (*LongSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for LongSchema.
func (*LongSchema) String ¶
func (*LongSchema) String() string
Returns a JSON representation of LongSchema.
func (*LongSchema) Type ¶
func (*LongSchema) Type() int
Type returns a type constant for this LongSchema.
type MapSchema ¶
MapSchema implements Schema and represents Avro map type.
func (*MapSchema) MarshalJSON ¶
MarshalJSON serializes the given schema as JSON.
func (*MapSchema) Prop ¶
Prop gets a custom non-reserved property from this schema and a bool representing if it exists.
type Marshaler ¶
Marshaler is an interface that may be implemented to avoid using runtime reflection during serialization. Implementing it is optional and may be used as an optimization. Falls back to using reflection if not implemented.
type NullSchema ¶
type NullSchema struct{}
NullSchema implements Schema and represents Avro null type.
func (*NullSchema) GetName ¶
func (*NullSchema) GetName() string
GetName returns a type name for this NullSchema.
func (*NullSchema) MarshalJSON ¶
func (*NullSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*NullSchema) Prop ¶
func (*NullSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for NullSchema.
func (*NullSchema) String ¶
func (*NullSchema) String() string
String returns a JSON representation of NullSchema.
func (*NullSchema) Type ¶
func (*NullSchema) Type() int
Type returns a type constant for this NullSchema.
type RecordSchema ¶
type RecordSchema struct { Name string `json:"name,omitempty"` Namespace string `json:"namespace,omitempty"` Doc string `json:"doc,omitempty"` Aliases []string `json:"aliases,omitempty"` Properties map[string]interface{} Fields []*SchemaField `json:"fields"` }
RecordSchema implements Schema and represents Avro record type.
func (*RecordSchema) GetName ¶
func (s *RecordSchema) GetName() string
GetName returns a record name for this RecordSchema.
func (*RecordSchema) MarshalJSON ¶
func (s *RecordSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON.
func (*RecordSchema) Prop ¶
func (s *RecordSchema) Prop(key string) (interface{}, bool)
Prop gets a custom non-reserved property from this schema and a bool representing if it exists.
func (*RecordSchema) String ¶
func (s *RecordSchema) String() string
String returns a JSON representation of RecordSchema.
func (*RecordSchema) Type ¶
func (*RecordSchema) Type() int
Type returns a type constant for this RecordSchema.
type RecursiveSchema ¶
type RecursiveSchema struct {
Actual *RecordSchema
}
RecursiveSchema implements Schema and represents Avro record type without a definition (e.g. that should be looked up).
func (*RecursiveSchema) GetName ¶
func (s *RecursiveSchema) GetName() string
GetName returns a record name for enclosed RecordSchema.
func (*RecursiveSchema) MarshalJSON ¶
func (s *RecursiveSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*RecursiveSchema) Prop ¶
func (*RecursiveSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for RecursiveSchema.
func (*RecursiveSchema) String ¶
func (s *RecursiveSchema) String() string
String returns a JSON representation of RecursiveSchema.
func (*RecursiveSchema) Type ¶
func (*RecursiveSchema) Type() int
Type returns a type constant for this RecursiveSchema.
type Schema ¶
type Schema interface { // Returns an integer constant representing this schema type. Type() int // If this is a record, enum or fixed, returns its name, otherwise the name of the primitive type. GetName() string // Gets a custom non-reserved property from this schema and a bool representing if it exists. Prop(key string) (interface{}, bool) // Converts this schema to its JSON representation. String() string // Checks whether the given value is writeable to this schema. Validate(v reflect.Value) bool }
Schema is an interface representing a single Avro schema (both primitive and complex).
func MustParseSchema ¶
MustParseSchema is like ParseSchema, but panics if the given schema cannot be parsed.
func ParseSchema ¶
ParseSchema parses a given schema without provided schemas to reuse. Equivalent to call ParseSchemaWithResistry(rawSchema, make(map[string]Schema)) May return an error if schema is not parsable or has insufficient information about any type.
func ParseSchemaFile ¶
ParseSchemaFile parses a given file. May return an error if schema is not parsable or file does not exist.
func ParseSchemaWithRegistry ¶
ParseSchemaWithRegistry parses a given schema using the provided registry for type lookup. Registry will be filled up during parsing. May return an error if schema is not parsable or has insufficient information about any type.
type SchemaField ¶
type SchemaField struct { Name string `json:"name,omitempty"` Doc string `json:"doc,omitempty"` Default interface{} `json:"default"` Type Schema `json:"type,omitempty"` Properties map[string]interface{} }
SchemaField represents a schema field for Avro record.
func (*SchemaField) MarshalJSON ¶
func (s *SchemaField) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema field as JSON.
func (*SchemaField) Prop ¶
func (this *SchemaField) Prop(key string) (interface{}, bool)
Gets a custom non-reserved property from this schemafield and a bool representing if it exists.
func (*SchemaField) String ¶
func (s *SchemaField) String() string
String returns a JSON representation of SchemaField.
type SpecificDatumReader ¶
type SpecificDatumReader struct {
// contains filtered or unexported fields
}
SpecificDatumReader implements DatumReader and is used for filling Go structs with data. Each value passed to Read is expected to be a pointer.
func NewSpecificDatumReader ¶
func NewSpecificDatumReader() *SpecificDatumReader
NewSpecificDatumReader creates a new SpecificDatumReader.
func (*SpecificDatumReader) Read ¶
func (reader *SpecificDatumReader) Read(v interface{}, dec Decoder) error
Read reads a single structured entry using this SpecificDatumReader. Accepts a Go struct with exported fields to fill with data and a Decoder to read from. Given value MUST be of pointer type. Field names should match field names in Avro schema but be exported (e.g. "some_value" in Avro schema is expected to be Some_value in struct) or you may provide Go struct tags to explicitly show how to map fields (e.g. if you want to map "some_value" field of type int to SomeValue in Go struct you should define your struct field as follows: SomeValue int32 `avro:"some_field"`). May return an error indicating a read failure.
func (*SpecificDatumReader) SetSchema ¶
func (reader *SpecificDatumReader) SetSchema(schema Schema) DatumReader
SetSchema sets the schema for this SpecificDatumReader to know the data structure. Note that it must be called before calling Read.
type SpecificDatumWriter ¶
type SpecificDatumWriter struct {
// contains filtered or unexported fields
}
SpecificDatumWriter implements DatumWriter and is used for writing Go structs in Avro format.
Example (Basic) ¶
package main import ( "bytes" "log" avro "gopkg.in/avro.v0" ) var someSchema avro.Schema type SomeStruct struct { Name string } func main() { writer := avro.NewSpecificDatumWriter() writer.SetSchema(someSchema) var buf bytes.Buffer v := &SomeStruct{Name: "abc"} if err := writer.Write(v, avro.NewBinaryEncoder(&buf)); err != nil { log.Fatal(err) // i/o errors OR encoding errors } // buf.Bytes() now contains the encoded value of 'v' }
Output:
Example (Full) ¶
package main import ( "bytes" "fmt" "log" avro "gopkg.in/avro.v0" ) func main() { // Parse a schema from JSON to get the schema object. schema, err := avro.ParseSchema(`{ "type": "record", "name": "Person", "fields": [ {"name": "first_name", "type": "string"}, {"name": "age", "type": "int"} ] }`) if err != nil { log.Fatal(err) } // This is a struct which supports the above schema. type Person struct { FirstName string `avro:"first_name"` // the avro: tag specifies which avro field matches this field. Age int32 `avro:"age"` } // Create a SpecificDatumWriter, which you can re-use multiple times. writer := avro.NewSpecificDatumWriter() writer.SetSchema(schema) // Write a person to a byte buffer as avro. person := &Person{ FirstName: "Bob", Age: 48, } var buf bytes.Buffer encoder := avro.NewBinaryEncoder(&buf) err = writer.Write(person, encoder) if err != nil { log.Fatal(err) } fmt.Println(buf.Bytes()) }
Output: [6 66 111 98 96]
func NewSpecificDatumWriter ¶
func NewSpecificDatumWriter() *SpecificDatumWriter
NewSpecificDatumWriter creates a new SpecificDatumWriter.
func (*SpecificDatumWriter) SetSchema ¶
func (writer *SpecificDatumWriter) SetSchema(schema Schema) DatumWriter
SetSchema sets the provided schema for this SpecificDatumWriter to know the data structure. Note that it must be called before calling Write.
func (*SpecificDatumWriter) Write ¶
func (writer *SpecificDatumWriter) Write(obj interface{}, enc Encoder) error
Write writes a single Go struct using this SpecificDatumWriter according to provided Schema. Accepts a value to write and Encoder to write to. Field names should match field names in Avro schema but be exported (e.g. "some_value" in Avro schema is expected to be Some_value in struct) or you may provide Go struct tags to explicitly show how to map fields (e.g. if you want to map "some_value" field of type int to SomeValue in Go struct you should define your struct field as follows: SomeValue int32 `avro:"some_field"`). May return an error indicating a write failure.
type StringSchema ¶
type StringSchema struct{}
StringSchema implements Schema and represents Avro string type.
func (*StringSchema) GetName ¶
func (*StringSchema) GetName() string
GetName returns a type name for this StringSchema.
func (*StringSchema) MarshalJSON ¶
func (*StringSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON. Never returns an error.
func (*StringSchema) Prop ¶
func (*StringSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for StringSchema.
func (*StringSchema) String ¶
func (*StringSchema) String() string
Returns a JSON representation of StringSchema.
func (*StringSchema) Type ¶
func (*StringSchema) Type() int
Type returns a type constant for this StringSchema.
type UnionSchema ¶
type UnionSchema struct {
Types []Schema
}
UnionSchema implements Schema and represents Avro union type.
func (*UnionSchema) GetName ¶
func (*UnionSchema) GetName() string
GetName returns a type name for this UnionSchema.
func (*UnionSchema) GetType ¶
func (s *UnionSchema) GetType(v reflect.Value) int
GetType gets the index of actual union type for a given value.
func (*UnionSchema) MarshalJSON ¶
func (s *UnionSchema) MarshalJSON() ([]byte, error)
MarshalJSON serializes the given schema as JSON.
func (*UnionSchema) Prop ¶
func (*UnionSchema) Prop(key string) (interface{}, bool)
Prop doesn't return anything valuable for UnionSchema.
func (*UnionSchema) String ¶
func (s *UnionSchema) String() string
String returns a JSON representation of UnionSchema.
func (*UnionSchema) Type ¶
func (*UnionSchema) Type() int
Type returns a type constant for this UnionSchema.
type Unmarshaler ¶
Unmarshaler is an interface that may be implemented to avoid using runtime reflection during deserialization. Implementing it is optional and may be used as an optimization. Falls back to using reflection if not implemented.