www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.announce - Release: avro-d v0.1.0

reply Vijay Nayar <madric gmail.com> writes:
Greetings D Community,

Do you have need for an efficient binary format for encoding and 
decoding data, but are hesitant to use things like [Google 
Protocol Buffers](https://developers.google.com/protocol-buffers) 
due to the need to re-compile code in order to change formats?

[Apache Avro](https://avro.apache.org/docs/current/) addresses 
many of these problems by pairing a binary (and other) format 
with a JSON formatted schema which is part of the protocol.

Sadly, this protocol is not available in the D Programming 
language... until now.

The initial v0.1.0 of the Avro library in D can be found here: 

A set of data classes exist to represent schemas for generation 
in code or the processing of data.

For example:
import avro.schema;
auto schema = new UnionSchema([

Schemas may be parsed from files, text, or JSON.

For example:
import avro.parser;
auto parser = new Parser();
Schema schema =  parser.parseText(q"EOS
{"namespace": "example.avro",
  "type": "record",
  "name": "User",
  "fields": [
      {"name": "name", "type": "string"},
      {"name": "favorite_number", "type": ["int", "null"]},
      {"name": "favorite_color", "type": ["string", "null"]}

Errors in the JSON format of a schema will lead to descriptive 

Generic data objects may be created according to schemas with 
their values set to schema-appropriate
defaults and validation logic when setting values. Most 
`GenericDatum` objects make use of
`.getValue!T()` and `.setValue(T)(T val)` methods, however, many 
convenience functions also exist.

For example:
import avro.generic.genericdata;

// Initializes the GenericDatum according to the schema with 
default values.
GenericDatum datum = new GenericDatum(schema);
assert(datum.getType == Type.RECORD);

// Primitive values can be set and retrieved.

// Convenience shortcut using opIndex() and opAssign() for 
primitive types.
datum["name"] = "bob";

assert(datum["name"].getValue!string == "bob");

// Enums have convenience functions directly on GenericData.
assert(datum["favorite_number"].getUnionIndex() == 0);

// Arrays also have convenience functions.
datum["scores"] ~= 1.23f;
datum["scores"] ~= 4.56f;
assert(datum["scores"].length == 2);
// Maps do as well.
datum["m"]["m1"] = 10L;
datum["m"]["m2"] = 20L;
assert(datum["m"]["m1"].getValue!long == 10L);

`GenericData` objects can be written using an encoder.

For example:
import avro.codec.binaryencoder;
import avro.generic.genericwriter;

ubyte[] data;
auto encoder = binaryEncoder(appender(&data));
GenericWriter writer = new GenericWriter(schema, encoder);

assert(data == [
// Field: name
// len=3     b     o     b
     0x06, 0x62, 0x6F, 0x62,
// Field: favorite_number
// idx=0     8
     0x00, 0x10,
// Field: favorite_color
// idx=0 len=4     b     l     u     e
     0x00, 0x08, 0x62, 0x6C, 0x75, 0x65

They may also be read using a decoder.

For example:
import avro.codec.binarydecoder;
import avro.generic.genericreader;

auto decoder = binaryDecoder(data);
GenericReader reader = new GenericReader(schema, decoder);
GenericDatum datum;

assert(datum["name"].getValue!string() == "bob");
assert(datum["favorite_number"].getValue!int() == 8);
assert(datum["favorite_color"].getValue!string() == "blue");

- Logical Type support
- Specific Data types generated for schemas
- JSON Serialization/Deserialization
- Codex compression support
- Object Container Files
- Protocol wire format
- Schema Resolution
Mar 09
parent Vijay Nayar <madric gmail.com> writes:
On Wednesday, 9 March 2022 at 14:26:10 UTC, Vijay Nayar wrote:
 The initial v0.1.0 of the Avro library in D can be found here: 
Version 0.2.2 has been released, which includes the ability to read and write generic objects in JSON format in addition to the aforementioned binary format.
Mar 16