S3 Parquet output plugin for Embulk

Embulk output plugin to dump records as Apache Parquet files on S3.

Overview

Plugin type: output
Load all or nothing: no
Resume supported: no
Cleanup supported: yes

Configuration

bucket: s3 bucket name (string, required)
path_prefix: prefix of target keys (string, optional)
sequence_format: format of the sequence number of the output files (string, default: "%03d.%02d.")
- sequence_format formats task index and sequence number in a task.
file_ext: path suffix of the output files (string, default: "parquet")
compression_codec: compression codec for parquet file ("uncompressed","snappy","gzip","lzo","brotli","lz4" or "zstd", default: "uncompressed")
default_timestamp_format: default timestamp format (string, default: "%Y-%m-%d %H:%M:%S.%6N %z")
default_timezone: default timezone (string, default: "UTC")
column_options: a map whose keys are name of columns, and values are configuration with following parameters (optional)
- timezone: timezone if type of this column is timestamp. If not set, default_timezone is used. (string, optional)
- format: timestamp format if type of this column is timestamp. If not set, default_timestamp_format: is used. (string, optional)
- converted_type: a Parquet converted type name (timestamp-millis, timestamp-micros, timestamp-nanos, json, int8, int16, int32, int64, uint8, uint16, uint32, uint64) (string, optional)
- logical_type: [DEPRECATED: Use converted_type instead] a Parquet converted type name (timestamp-millis, timestamp-micros, timestamp-nanos, json, int8, int16, int32, int64, uint8, uint16, uint32, uint64) (string, optional)
- logical_type: configuration for the detailed logical type. See Logical Type Specification (optional)
- name: The name of logical type ("date", "decimal", "int", "json", "time", "timestamp") (string, required)
- bit_width: The bit width for "int" logical type (Allowed bit width values are 8, 16, 32, 64). (int, default: 64)
- is_signed: Signed or not for "int" logical type (boolean, default: true)
- scale: The scale for "decimal" logical type (int, default: 0)
- precision: The precision for "decimal" logical type (int, default: 0)
- is_adjusted_to_utc: (boolean, default: true)
- time_unit: The precision for "time" or "timestamp" logical type (Allowed values are "MILLIS, MICROS, NANOS)
canned_acl: grants one of canned ACLs for created objects (string, default: private)
block_size: The block size is the size of a row group being buffered in memory. This limits the memory usage when writing. Larger values will improve the I/O when reading but consume more memory when writing. (int, default: 134217728 (128MB))
page_size: The page size is for compression. When reading, each page can be decompressed independently. A block is composed of pages. The page is the smallest unit that must be read fully to access a single record. If this value is too small, the compression will deteriorate. (int, default: 1048576 (1MB))
max_padding_size: The max size (bytes) to write as padding and the min size of a row group (int, default: 8388608 (8MB))
enable_dictionary_encoding: The boolean value is to enable/disable dictionary encoding. (boolean, default: true)

auth_method: name of mechanism to authenticate requests ("basic", "env", "instance", "profile", "properties", "anonymous", "session", "web_identity_token", default: "default")

"basic": uses access_key_id and secret_access_key to authenticate.
"env": uses AWS_ACCESS_KEY_ID (or AWS_ACCESS_KEY) and AWS_SECRET_KEY (or AWS_SECRET_ACCESS_KEY) environment variables.
"instance": uses EC2 instance profile or attached ECS task role.
"profile": uses credentials written in a file. Format of the file is as following, where [...] is a name of profile.
```
[default]
aws_access_key_id=YOUR_ACCESS_KEY_ID
aws_secret_access_key=YOUR_SECRET_ACCESS_KEY
```

[profile2] ...


- `"properties"`: uses aws.accessKeyId and aws.secretKey Java system properties.
- `"anonymous"`: uses anonymous access. This auth method can access only public files.
- `"session"`: uses temporary-generated **access_key_id**, **secret_access_key** and **session_token**.
- `"assume_role"`: uses temporary-generated credentials by assuming **role_arn** role.
- `"web_identity_token"`: uses temporary-generated credentials by assuming **role_arn** role with web identity.
- `"default"`: uses AWS SDK's default strategy to look up available credentials from runtime environment. This method behaves like the combination of the following methods.
1. `"env"`
1. `"properties"`
1. `"profile"`
1. `"instance"`

profile_file: path to a profiles file. this is optionally used when auth_method is "profile". (string, default: given by AWS_CREDENTIAL_PROFILES_FILE environment variable, or ~/.aws/credentials).
profile_name: name of a profile. this is optionally used when auth_method is "profile". (string, default: "default")
access_key_id: aws access key id. this is required when auth_method is "basic" or "session". (string, optional)
secret_access_key: aws secret access key. this is required when auth_method is "basic" or "session". (string, optional)
session_token: aws session token. this is required when auth_method is "session". (string, optional)
role_arn: arn of the role to assume. this is required for auth_method is "assume_role" or "web_identity_token". (string, optional)
role_session_name: an identifier for the assumed role session. this is required when auth_method is "assume_role" or "web_identity_token". (string, optional)
role_external_id: a unique identifier that is used by third parties when assuming roles in their customers' accounts. this is optionally used for auth_method: "assume_role". (string, optional)
role_session_duration_seconds: duration, in seconds, of the role session. this is optionally used for auth_method: "assume_role". (int, optional)
web_identity_token_file: the absolute path to the web identity token file. this is required when auth_method is "web_identity_token". (string, optional)
scope_down_policy: an iam policy in json format. this is optionally used for auth_method: "assume_role". (string, optional)

catalog: Register a table if this option is specified (optional)

catalog_id: glue data catalog id if you use a catalog different from account/region default catalog. (string, optional)
database: The name of the database (string, required)
table: The name of the table (string, required)
column_options: a key-value pairs where key is a column name and value is options for the column. (string to options map, default: {})

type: type of column when this plugin creates new tables (e.g. string, bigint) (string, default: depends on the input embulk column type, or the parquet logical type. See the below table)

embulk column type	glue data type
long	bigint
boolean	boolean
double	double
string	string
timestamp	string
json	string

parquet converted type	glue data type	note
timestamp-millis	timestamp
timestamp-micros	long	Glue cannot recognize timestamp-micros.
timestamp-nanos	long	Glue cannot recognize timestamp-nanos.
int8	tinyint
int16	smallint
int32	int
int64	bigint
uint8	smallint	Glue tinyint is a minimum value of -2^7 and a maximum value of 2^7-1
uint16	int	Glue smallint is a minimum value of -2^15 and a maximum value of 2^15-1.
uint32	bigint	Glue int is a minimum value of-2^31 and a maximum value of 2^31-1.
uint64	ConfigException	Glue bigint supports only a 64-bit signed integer.
json	string

operation_if_exists: operation if the table already exist. Available operations are "delete" and "skip" (string, default: "delete")

endpoint: The AWS Service endpoint (string, optional)
region: The AWS region (string, optional)
http_proxy: Indicate whether using when accessing AWS via http proxy. (optional)
- host proxy host (string, required)
- port proxy port (int, optional)
- protocol proxy protocol (string, default: "https")
- user proxy user (string, optional)
- password proxy password (string, optional)
buffer_dir: buffer directory for parquet files to be uploaded on S3 (string, default: Create a Temporary Directory)
type_options: a map whose keys are name of embulk type(boolean, long, double, string, timestamp, json), and values are configuration with following parameters (optional)
- converted_type: a Parquet converted type name (timestamp-millis, timestamp-micros, timestamp-nanos, json, int8, int16, int32, int64, uint8, uint16, uint32, uint64) (string, optional)
- logical_type: [DEPRECATED: Use converted_type instead] a Parquet converted type name (timestamp-millis, timestamp-micros, timestamp-nanos, json, int8, int16, int32, int64, uint8, uint16, uint32, uint64) (string, optional)
- logical_type: configuration for the detailed logical type. See Logical Type Specification (optional)
- name: The name of logical type ("date", "decimal", "int", "json", "time", "timestamp") (string, required)
- bit_width: The bit width for "int" logical type (Allowed bit width values are 8, 16, 32, 64). (int, default: 64)
- is_signed: Signed or not for "int" logical type (boolean, default: true)
- scale: The scale for "decimal" logical type (int, default: 0)
- precision: The precision for "decimal" logical type (int, default: 0)
- is_adjusted_to_utc: (boolean, default: true)
- time_unit: The precision for "time" or "timestamp" logical type (Allowed values are "MILLIS, MICROS, NANOS)

Example

out:
  type: s3_parquet
  bucket: my-bucket
  path_prefix: path/to/my-obj.
  file_ext: snappy.parquet
  compression_codec: snappy
  default_timezone: Asia/Tokyo
  canned_acl: bucket-owner-full-control

Note

This plugin implements the Parquet LogicalTypes as much as possible. But it does not implement all of ones.
Some kind of LogicalTypes are sometimes not supported on your middleware. Be careful to giving logical type name.

Development

Run example:

$ ./run_s3_local.sh
$ ./example/prepare_s3_bucket.sh
$ ./gradlew gem
$ embulk run example/config.yml -Ibuild/gemContents/lib

Run test:

$ ./run_s3_local.sh
$ ./gradlew scalatest

Build

$ ./gradlew gem --write-locks  # -t to watch change of files and rebuild continuously

Release gem:

Fix build.gradle, then

$ ./gradlew gemPush

ChangeLog

CHANGELOG.md