Trino client library for Ruby

Trino is a distributed SQL query engine for big data: https://trino.io/

This is a client library for Ruby to run queries on Trino.

Example

require 'trino-client'

# create a client object:
client = Trino::Client.new(
  server: "localhost:8880",   # required option
  ssl: {verify: false},
  catalog: "native",
  schema: "default",
  user: "frsyuki",
  password: "********",
  time_zone: "US/Pacific",
  language: "English",
  properties: {
    "hive.force_local_scheduling": true,
    "raptor.reader_stream_buffer_size": "32MB"
  },
  http_proxy: "proxy.example.com:8080",
  http_debug: true
)

# run a query and get results as an array of arrays:
columns, rows = client.run("select * from sys.node")
rows.each {|row|
  p row  # row is an array
}

# run a query and get results as an array of hashes:
results = client.run_with_names("select alpha, 1 AS beta from tablename")
results.each {|row|
  p row['alpha']   # access by name
  p row['beta']
  p row.values[0]  # access by index
  p row.values[1]
}

# run a query and fetch results streamingly:
client.query("select * from sys.node") do |q|
  # get columns:
  q.columns.each {|column|
    puts "column: #{column.name}.#{column.type}"
  }

  # get query results. it feeds more rows until
  # query execution finishes:
  q.each_row {|row|
    p row  # row is an array
  }
end

# killing a query
query = client.query("select * from sys.node")
query_id = query.query_info.query_id
query.each_row {|row| ... }  # when a thread is processing the query,
client.kill(query_id)  # another thread / process can kill the query.

# Use Query#transform_row to parse Trino ROW types into Ruby Hashes.
# You can also set a scalar_parser to parse scalars how you'd like them.
scalar_parser = -> (data, type) { (type === 'json') ? JSON.parse(data) : data }
client.query("select * from sys.node") do |q|
  q.scalar_parser = scalar_parser

  # get query results. it feeds more rows until
  # query execution finishes:
  q.each_row {|row|
    p q.transform_row(row)
  }
end

Build models

$ bundle exec rake modelgen:latest

Options

• server sets address (and port) of a Trino coordinator server.
• ssl enables https.
  • Setting true enables SSL and verifies server certificate using system's built-in certificates.
  • Setting {verify: false} enables SSL but doesn't verify server certificate.
  • Setting a Hash object enables SSL and verify server certificate with options:
    • ca_file: path of a CA certification file in PEM format
    • ca_path: path of a CA certification directory containing certifications in PEM format
    • cert_store: a OpenSSL::X509::Store object used for verification
    • client_cert: a OpenSSL::X509::Certificate object as client certificate
    • client_key: a OpenSSL::PKey::RSA or OpenSSL::PKey::DSA object used for client certificate
• catalog sets catalog (connector) name of Trino such as hive-cdh4, hive-hadoop1, etc.
• schema sets default schema name of Trino. You need to use qualified name like FROM myschema.table1 to use non-default schemas.
• source sets source name to connect to a Trino. This name is shown on Trino web interface.
• client_info sets client info to queries. It can be a string to pass a raw string, or an object that can be encoded to JSON.
• client_tags sets client tags to queries. It needs to be an array of strings. The tags are shown on web interface.
• user sets user name to connect to a Trino.
• password sets a password to connect to Trino using basic auth.
• time_zone sets time zone of queries. Time zone affects some functions such as format_datetime.
• language sets language of queries. Language affects some functions such as format_datetime.
• properties set session properties. Session properties affect internal behavior such as hive.force_local_scheduling: true, raptor.reader_stream_buffer_size: "32MB", etc.
• query_timeout sets timeout in seconds for the entire query execution (from the first API call until there're no more output data). If timeout happens, client raises TrinoQueryTimeoutError. Default is nil (disabled).
• plan_timeout sets timeout in seconds for query planning execution (from the first API call until result columns become available). If timeout happens, client raises TrinoQueryTimeoutError. Default is nil (disabled).
• http_headers sets custom HTTP headers. It must be a Hash of string to string.
• http_proxy sets host:port of a HTTP proxy server.
• http_debug enables debug message to STDOUT for each HTTP requests.
• http_debug_logger sets a custom Logger instance for HTTP debug logs. Requires http_debug to be true.
• http_open_timeout sets timeout in seconds to open new HTTP connection.
• http_timeout sets timeout in seconds to read data from a server.
• gzip enables gzip compression.
• follow_redirect enables HTTP redirection support.
• model_version set the Trino version to which a job is submitted. Supported versions are 351, 316, 303, 0.205, 0.178, 0.173, 0.153 and 0.149. Default is 351.

See RDoc for the full documentation.

Development

Releasing a new version

1. Run the following command which will prompt for the new version, update lib/trino/client/version.rb, and generate/let you edit ChangeLog.md.

$ ruby release.rb

2. Commit and tag

$ git commit -am "vX.Y.Z"
$ git tag "vX.Y.Z"
$ git push origin master
$ git push --tags

Pushing a v* tag triggers the GitHub Actions publisher workflow, which automatically builds and publishes the gem to RubyGems.org.