Project

solrb

0.01
Low commit activity in last 3 years
No release in over a year
Solr Ruby client with a nice object-oriented API
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

>= 0
~> 10.0
~> 3.0
~> 0.80.1

Runtime

 Project Readme

CircleCI Maintainability Gem Version

Solrb

Object-Oriented approach to Solr in Ruby.

Table of contents

  • Installation
  • Configuration
    • Setting Solr URL via environment variable
    • Single core configuration
    • Multiple core configuration
    • Solr Cloud
    • Master-slave
      • Gray list
    • Basic Authentication
  • Indexing
  • Querying
    • Simple Query
    • Querying multiple cores
    • Query with field boost
    • Query with filtering
    • Query with sorting
    • Query with grouping
    • Query with facets
    • Query with boosting functions
      • Dictionary boosting function
    • Field list
  • Deleting documents
  • Active Support instrumentation
  • Testing
  • Running specs

Installation

Add solrb to your Gemfile:

gem 'solrb'

If you are going to use solrb with solr cloud:

gem 'zk' # required for solrb solr-cloud integration
gem 'solrb'

Configuration

Setting Solr URL via environment variable

The simplest way to use Solrb is SORL_URL environment variable (that has a core name in it):

  ENV['SOLR_URL'] = 'http://localhost:8983/solr/demo'

You can also use Solr.configure to specify the solr URL explicitly:

Solr.configure do |config|
  config.url = 'http://localhost:8983/solr/demo'
end

It's important to note that those fields that are not configured, will be passed as-is to solr. So you only need to specify fields in configuration if you want Solrb to modify them at runtime.

Single core configuration

Use Solr.configure for an additional configuration:

Solr.configure do |config|
  config.url = 'http://localhost:8983/solr/demo'

  # This gem uses faraday to make requests to Solr. You can specify additional faraday
  # options here.
  config.faraday_options = {}

  # Core's URL is 'http://localhost:8983/solr/demo'
  # Adding fields to work with
  config.define_core do |f|
    f.field :title, dynamic_field: :text
    f.dynamic_field :text, solr_name: '*_text'
  end
end

Multiple core configuration

Solr.configure do |config|
  config.url = 'http://localhost:8983/solr'

  # Define a core with fields that will be used with Solr.
  # Core URL is 'http://localhost:8983/solr/listings'
  config.define_core(name: :listings) do |f|
    # When a dynamic_field is present, the field name will be mapped to match the dynamic field.
    # Here, "title" will be mapped to "title_text"
    # You must define a dynamic field to be able to use the dynamic_field option
    f.field :title, dynamic_field: :text

    # When solr_name is present, the field name will be mapped to the solr_name at runtime
    f.field :tags, solr_name: :tags_array

    # define a dynamic field
    f.dynamic_field :text, solr_name: '*_text'
  end

  # Pass `default: true` to use one core as a default.
  # Core's URL is 'http://localhost:8983/solr/cars'
  config.define_core(name: :cars, default: true) do |f|
    f.field :manufacturer, solr_name: :manuf_s
    f.field :model, solr_name: :model_s
  end
end

Warning: Solrb doesn't support fields with the same name. If you have two fields with the same name mapping to a single solr field, you'll have to rename one of the fields.

...
config.define_core do |f|
  ...
  # Not allowed: Two fields with same name 'title'
  f.field :title, solr_name: :article_title
  f.field :title, solr_name: :page_title
end
...

Solr Cloud

To enable solr cloud mode you must define a zookeeper url on solr config block. In solr cloud mode you don't need to provide a solr url (config.url or ENV['SOLR_URL']). Solrb will watch the zookeeper state to receive up-to-date information about active solr nodes including the solr urls.

You can also specify the ACL credentials for Zookeeper. More Information

Solr.configure do |config|
  config.zookeeper_urls = ['localhost:2181', 'localhost:2182', 'localhost:2183']
  config.zookeeper_auth_user = 'zk_acl_user'
  config.zookeeper_auth_password = 'zk_acl_password'
end

If you are using puma web server in clustered mode you must call enable_solr_cloud! on on_worker_boot callback to make each puma worker connect with zookeeper.

on_worker_boot do
  Solr.enable_solr_cloud!
end

Master-slave

To enable master-slave mode you must define a master url and slave url on solr config block. In solr master-slave mode you don't need to provide a solr url (config.url or ENV['SOLR_URL']).

Solr.configure do |config|
  config.master_url = 'localhost:8983'
  config.slave_url = 'localhost:8984'
  # Disable select queries from master:
  config.disable_read_from_master = true
  # Specify Gray-list service
  config.nodes_gray_list = Solr::MasterSlave::NodesGrayList::InMemory.new
end

If you are using puma web server in clustered mode you must call enable_master_slave! on on_worker_boot callback to make each puma worker connect with zookeeper.

on_worker_boot do
  Solr.enable_master_slave!
end

Gray list

Solrb provides two built-in services:

  • Solr::MasterSlave::NodesGrayList::Disabled — Disabled service (default). Just does nothing.
  • Solr::MasterSlave::NodesGrayList::InMemory — In memory service. It stores failed URLs in an instance variable, so it's not shared across threads/servers. URLs will be marked as "gray" for 5 minutes, but if all URLs are gray, the policy will try to send requests to these URLs earlier.

You are able to implement your own services with corresponding API.

Force node URL

You can force solrb to use a specific node URL with the with_node_url method:

Solr.with_node_url('http://localhost:9000') do
  Solr::Query::Request.new(search_term: 'example', query_fields: query_fields).run
end

Basic Authentication

Basic authentication is supported by solrb. You can enable it by providing auth_user and auth_password on the config block.

Solr.configure do |config|
  config.auth_user = 'user'
  config.auth_password = 'password'
end

Indexing

# creates a single document and commits it to index
doc = Solr::Update::Commands::Add.new
doc.add_field(:id, 1)
doc.add_field(:name, 'Solrb!!!')

commit = Solr::Update::Commands::Commit.new

request = Solr::Update::Request.new([doc, commit])
request.run

You can also create indexing document directly from attributes:

doc = Solr::Update::Commands::Add.new(doc: { id: 5, name: 'John' })

Querying

Simple Query

  query_field = Solr::Query::Request::QueryField.new(field: :name)

  request = Solr::Query::Request.new(search_term: 'term', query_fields: [query_field])
  request.run(page: 1, page_size: 10)

Querying multiple cores

For multi-core configuration use Solr.with_core block:

Solr.with_core(:models) do
  Solr.delete_by_id(3242343)
  Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  Solr::Update::Request.new([doc])
end

Query with field boost

  query_fields = [
    # Use boost_magnitude argument to apply boost to a specific field that you query
    Solr::Query::Request::QueryField.new(field: :name, boost_magnitude: 16),
    Solr::Query::Request::QueryField.new(field: :title)
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.run(page: 1, page_size: 10)

Query with filtering

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :title)
  ]
  filters = [Solr::Query::Request::Filter.new(type: :equal, field: :title, value: 'A title')]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields, filters: filters)
  request.run(page: 1, page_size: 10)

AND and OR filters

  usa_filter =
    Solr::Query::Request::AndFilter.new(
      Solr::Query::Request::Filter.new(type: :equal, field: :contry, value: 'USA'),
      Solr::Query::Request::Filter.new(type: :equal, field: :region, value: 'Idaho')
    )
  canada_filter =
    Solr::Query::Request::AndFilter.new(
      Solr::Query::Request::Filter.new(type: :equal, field: :contry, value: 'Canada'),
      Solr::Query::Request::Filter.new(type: :equal, field: :region, value: 'Alberta')
    )

  location_filters = Solr::Query::Request::OrFilter.new(usa_filter, canada_filter)
  request = Solr::Query::Request.new(search_term: 'term', filters: location_filters)
  request.run(page: 1, page_size: 10)

Filtering by a Geofilt

  spatial_point = Solr::SpatialPoint.new(lat: 40.0, lon: -120.0)

  filters = [
    Solr::Query::Request::Geofilt.new(field: :location, spatial_point: spatial_point, spatial_radius: 100)
  ]

  request = Solr::Query::Request.new(search_term: 'term', filters: filters)
  request.run(page: 1, page_size: 10)

Filtering by an Arbitrary Rectangle

  spatial_rectangle = Solr::SpatialRectangle.new(
    top_left: Solr::SpatialPoint.new(lat: 40.0, lon: -120.0),
    bottom_right: Solr::SpatialPoint.new(lat: 30.0, lon: -110.0)
  )

  filters = [
    Solr::Query::Request::Filter.new(type: :equal, field: :location, value: spatial_rectangle)
  ]

  request = Solr::Query::Request.new(search_term: 'term', filters: filters)
  request.run(page: 1, page_size: 10)

Query with sorting

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :title)
  ]
  sort_fields = [Solr::Query::Request::Sorting::Field.new(name: :name, direction: :asc)]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.sorting = Solr::Query::Request::Sorting.new(fields: sort_fields)
  request.run(page: 1, page_size: 10)

Default sorting logic is following: nulls last, not-nulls first.

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name)
  ]
  sort_fields = [
    Solr::Query::Request::Sorting::Field.new(name: :is_featured, direction: :desc),
    Solr::Query::Request::Sorting::Function.new(function: "score desc")
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.sorting = Solr::Query::Request::Sorting.new(fields: sort_fields)
  request.run(page: 1, page_size: 10)

Query with grouping

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :category)
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.grouping = Solr::Query::Request::Grouping.new(field: :category, limit: 10)
  request.run(page: 1, page_size: 10)

Query with facets

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :category)
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.facets = [Solr::Query::Request::Facet.new(type: :terms, field: :category, options: { limit: 10 })]
  request.run(page: 1, page_size: 10)

Query with boosting functions

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :category)
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  request.boosting = Solr::Query::Request::Boosting.new(
    multiplicative_boost_functions: [Solr::Query::Request::Boosting::RankingFieldBoostFunction.new(field: :name)],
    phrase_boosts: [Solr::Query::Request::Boosting::PhraseProximityBoost.new(field: :category, boost_magnitude: 4)]
  )
  request.run(page: 1, page_size: 10)

Dictionary boosting function

Sometimes you want to do a dictionary-style boosting example: given a hash (dictionary)

{3025 => 2.0, 3024 => 1.5, 3023 => 1.2}

and a field of category_id the resulting boosting function will be:

if(eq(category_id_it, 3025), 2.0, if(eq(category_id_it, 3024), 1.5, if(eq(category_id_it, 3023), 1.2, 1)))

note that I added spaces for readability, real Solr query functions must always be w/out spaces

Example of usage:

  category_id_boosts = {3025 => 2.0, 3024 => 1.5, 3023 => 1.2}
  request.boosting = Solr::Query::Request::Boosting.new(
    multiplicative_boost_functions: [
      Solr::Query::Request::Boosting::DictionaryBoostFunction.new(field: :category_id,
        dictionary: category_id_boosts)
    ]
  )

Query with shards.preference

  shards_preference = Solr::Query::Request::ShardsPreference.new(
    properties: [
      Solr::Query::Request::ShardsPreferences::Property.new(name: 'replica.type', value: 'PULL')
    ]
  )
  request = Solr::Query::Request.new(search_term: 'term', shards_preference: shards_preference)
  request.run(page: 1, page_size: 10)

Field list

  query_fields = [
    Solr::Query::Request::QueryField.new(field: :name),
    Solr::Query::Request::QueryField.new(field: :category)
  ]
  request = Solr::Query::Request.new(search_term: 'term', query_fields: query_fields)
  # Solr::Query::Request will return only :id field by default.
  # Specify additional return fields (fl param) by setting the request field_list
  request.field_list = [:name, :category]
  request.run(page: 1, page_size: 10)

Deleting documents

# Delete by document ID
Solr.delete_by_id(3242343)
Solr.delete_by_id(3242343, commit: true)

# Delete by query
Solr.delete_by_query('*:*')
Solr.delete_by_query('*:*', commit: true)

# Delete by filters
filters = [Solr::Query::Request::Filter.new(type: :equal, field: :contry, value: 'Canada')]
commands = [Solr::Update::Commands::Delete.new(filters: filters)]
commands << Solr::Update::Commands::Commit.new if commit?
request = Solr::Update::Request.new(commands)
request.run

Active Support instrumentation

This gem publishes events via Active Support Instrumentation

To subscribe to solrb events, you can add this code to initializer:

ActiveSupport::Notifications.subscribe('request.solrb')  do |*args|
  event = ActiveSupport::Notifications::Event.new(*args)
  if Logger::INFO == Rails.logger.level
    Rails.logger.info("Solrb #{event.duration.round(1)}ms")
  elsif Logger::DEBUG == Rails.logger.level && Rails.env.development?
    Pry::ColorPrinter.pp(event.payload)
  end
end

Testing

It's possible to inspect the parameters for each solr query request done using Solrb by requiring solr/testing file in your test suite. The query parameters will be accessible by reading Solr::Testing.last_solr_request after each request.

require 'solr/testing'

RSpec.describe MyTest do
  let(:query) { Solr::Query::Request.new(search_term: 'Solrb') }
  it 'returns the last solr request params' do
    query.run(page: 1, page_size: 10)
    expect(Solr::Testing.last_solr_request.body[:params]).to eq({ ... })
  end
end

Running specs

This project is setup to use CI to run all specs agains a real solr.

If you want to run it locally, you can either use CircleCI CLI or do a completely manual setup (for up-to-date steps see circleci config)

docker run -it --name test-solr -p 8983:8983/tcp -t solr:8.11.2-slim

# Copy default configset
docker exec -u 0 $(docker ps | grep test-solr | cut -d ' ' -f 1) sh -c "mkdir /var/solr/data/configsets \
  && cp -R /opt/solr/server/solr/configsets/_default /var/solr/data/configsets/ \
  && chown -R solr:solr /var/solr/data/configsets"

# create a core
curl 'http://localhost:8983/solr/admin/cores?action=CREATE&name=test-core&configSet=_default'

# disable field guessing
curl http://localhost:8983/solr/test-core/config -d '{"set-user-property": {"update.autoCreateFields":"false"}}'
SOLR_URL=http://localhost:8983/solr/test-core rspec