Commits on May 4, 2012
  1. Enable Tire to load in Rails 2.3.

    committed May 4, 2012
  2. Release 0.4.2

    karmi committed May 4, 2012
  3. [ALIASES] Added `Tire::Alias`, interface and DSL to manage aliases as…

    … resources
    * Added `Alias.all`, `Alias.all("some_index")` and `Alias.find("my_alias")`
    * Added rich interface for managing aliases as resources
    * Changed Index#add_alias and Index#remove_alias to use Tire::Alias interface
    * Changed the `Tire.aliases` method to use `Alias.all`
    * Added unit and integration tests
    See <> for feature overview.
    The `Tire::Alias` class provides a rich interface for managing aliases.
    To create an alias pointing to multiple indices:
        Tire::Alias.create name: 'my_alias', indices: ['index_1', 'index_2']
    You may pass the routing and other configurations:
        a = name:    'index_anne',
                            indices: ['index_2012_04', 'index_2012_03', 'index_2012_02'],
                            routing: 1,
                            filter:  { :terms => { :user => 'anne' } }
    Of course, you may use the Ruby API directly:
        a ='index_anne')
        # ...
    But you may prefer the DSL-ish block syntax:
    name: 'index_anne' do |a|
          a.index 'index_2012_04'
          a.index 'index_2012_03'
          a.filter :terms, username: 'anne'
    To update an existing alias, find it by name, update its configuration and save it:
        a = Tire::Alias.find('my_alias')
        a.indices.delete 'index_A'
        a.indices.add    'index_B'
        a.indices.add    'index_C'
    Or do it with a block, again:
        Tire::Alias.find('articles_aliased') do |a|
          a.indices.remove 'articles_2'
          puts '---', "#{} says: /me as JSON >", a.as_json, '---'

    To remove indices from alias, you may want to use the `Alias#indices.delete_if` method:
        require 'active_support/core_ext/numeric'
        require 'active_support/core_ext/date/calculations'
        a = Tire::Alias.find('articles_aliased')
        a.indices.delete_if do |i|
          Time.parse( i.gsub(/articles_/, '') ) < 4.weeks.ago rescue false
    To get all aliases, use the `Tire::Alias.all` method:
        Tire::Alias.all.each do |a|
          puts "#{} points to: #{a.indices}"
    To get aliases for a specific index, pass its name to `Alias.all`:
        Tire::Alias.all('articles').each do |a|
          puts "Alias '#{}': #{a.inspect}."
    See the 'sliding window' scenario in the corresponding integration test:
    Fixes #62, fixes #92, fixes #209.
    karmi committed May 2, 2012
  4. [FIX] Fixed, that blocks passed to `` or `Tire.index` …

    …losed the scope
    Without this patch, blocks being passed to `` losed the scope, because
    only `instance_eval` was called.
    This was inconsistent with the `` behaviour, as documented eg. in
    Thanks to @ahfeel for reporting.
    karmi committed May 4, 2012
Commits on May 3, 2012
  1. Release 0.4.1

    karmi committed May 3, 2012
  2. Added Index#url and simplified the Tire::Index class and tests

    Closes #328, fixes #327
    nz committed with karmi Apr 30, 2012
  3. [BULK] Changed, that Index#bulk_store runs against an index endpoint,…

    … not against `/_bulk`
    Closes #327, fixes #328.
    karmi committed May 3, 2012
Commits on Apr 28, 2012
Commits on Apr 27, 2012
  1. [#209] Added support for creating index aliases with configuration, c…

    …leaned up
    * You can do `@index.add_alias 'index-aliased', :routing => '1'` now
    * Fixed incorrect curl-formatted logs from alias operations
    * More consistent naming
    * Fixed expectations for Ruby 1.8
    * Added integration test
    Fixes #209.
    karmi committed Apr 27, 2012
  2. Added support for index aliases

    This commit adds support for:
    * Tire::aliases
    * Tire::Index#aliases
    * Tire::Index#add_alias
    * Tire::Index#remove_alias
    So, to add an alias for an index, you can do:
        Tire.index('articles') do
          add_alias 'articles_aliased
        Tire.index('articles').add_alias 'articles_aliased'
    See the index unit tests for documentation.
    See <>
    for more documentation on the feature in ElasticSearch.
    Closes #209, fixes #92.
    woahdae committed with karmi Jan 18, 2012
Commits on Mar 30, 2012
  1. Added support for reindexing the index documents into new index

    Added a `Tire::Index#reindex` method which uses the "scan/scroll" API to retrieve
    documents from index A and import them into index B.
    If the new index does not exists, it is created: you may pass settings as an options hash.
    You may pass a Tire DSL query as a block to the method.
        require 'tire'
        Tire.configure { logger STDERR }
        Tire.index('scantest-new') { delete }
        documents = (1..100).map do |i|
          { id: i, type: 'test', title: "Document #{i}" }
        puts "", "Importing data into ElasticSearch..."
        Tire.index 'scantest' do
          delete and create :settings => { :number_of_shards => 1, :number_of_replicas => 0 }
          import documents
        Tire.index('scantest').reindex 'scantest-new', settings: { number_of_shards: 3 } do
          query { string '10*' }
    karmi committed Mar 30, 2012
  2. Added support for the "scan" search in the Ruby API

    Given the support for search types (since karmi/retire@d764c0d), we can now
    expose the "scan" search type in the Ruby API and DSL.
        # Let's index some data
        documents = (1..100).map { |i| { id: i, type: 'test', title: "Document #{i}" } }
        Tire.index 'scantest' do
          create :settings => { :number_of_shards => 1, :number_of_replicas => 0 }
          import documents
        # Let's initiate a scan search request
        s = Tire.scan 'scantest'
        # The #each method returns the document batches as received from ElasticSearch,
        # where invididual hits are wrapped as Results::Item
        s.each do |results|
        # The #each_document method returns each individual documents, as received from ElasticSearch,
        # wrapped as a Results::Item
        s.each_document do |document|
          puts "#{count}: #{document.title}"
          count += 1
    The object being returned is fully Enumerable-compatible.
    See the unit and integration tests for details on usage.
    Inspired by Michael Grassotti's and Erick Tryzelaar's work at:
    * grasscode/tire@63fefb9
    See the ElasticSearch documentation:
    Closes #196.
    karmi committed Mar 30, 2012
Commits on Mar 28, 2012
  1. Release 0.4.0

    karmi committed Mar 28, 2012
Commits on Mar 27, 2012
  1. Release 0.4.0.rc

    karmi committed Mar 27, 2012
  2. [#218] Fixed the incorrect serialization of `document_type` in Index#…

    Previously, when storing individual namespaced records, it correctly stored them as "my_namespace/my_model",
    but bulk_store was saving _type as "my_namespace%2Fmy_model".
    This resulted in two different document types where there should only be one, incorrect mapping, incorrect searches.
    Closes #290, closes #291, fixes #218.
    sarmiena committed with karmi Mar 25, 2012
  3. [FIX] Fixed displaying of Rake task usage

    (Let's all forget Rake actually works and makes sense as a Ruby library, shall we?)
    karmi committed Mar 27, 2012
Commits on Mar 26, 2012
  1. [#289] Update README and documentation, use the `prefix` query as the…

    … example of "unsupported" query
    karmi committed Mar 26, 2012
  2. [#289] Added an example of fuzzy query in the Text query integration …

    This is what you probably want to use when exposing the search interface to your users.
    See <>
    karmi committed Mar 26, 2012
  3. [TEST] Improved code in the Text Query integration test

    Use regular Tire API, not custom RestClient calls...
    karmi committed Mar 26, 2012
  4. [#289] Cleaned up the code for "term" and "fuzzy" queries and restruc…

    …tured the search query test suite
    karmi committed Mar 26, 2012
  5. [#289] Improvements to the "term" query type

    This is a combination of 3 commits:
    * little improvement of term api
    * fixing term tests after api changing
    * unit tests for term query with an option hash
    plentz committed with karmi Mar 25, 2012
  6. [#289] Added the "fuzzy" query type

    See <>
    This is a combination of 4 commits:
    * initial implementation of fuzzy api
    * just a bit improvement to the docs to make fuzzy example easier to read
    * unit tests for fuzzy query
    * integration test for fuzzy query
    plentz committed with karmi Mar 25, 2012