Added search for similar items

Andrew Kane
1 parent 55cfb407
Showing 6 changed files with 57 additions and 1 deletions Show diff stats
README.md
lib/searchkick.rb
lib/searchkick/model.rb
lib/searchkick/similar.rb
test/similar_test.rb
test/test_helper.rb
@@ -284,6 +284,17 @@ Advanced
 Product.search "2% Milk", facets: {store_id: {where: {in_stock: true}}}
 ```
  
+### Similar Items [master]
+
+**Note:** Subject to change before the next gem release
+
+Find similar items.
+
+```ruby
+product = Product.first
+product.similar(fields: ["name"])
+```
+
 ## Deployment
  
 Searchkick uses `ENV["ELASTICSEARCH_URL"]` for the Elasticsearch server.  This defaults to `http://localhost:9200`.
@@ -3,6 +3,7 @@ require &quot;searchkick/version&quot;
 require "searchkick/reindex"
 require "searchkick/results"
 require "searchkick/search"
+require "searchkick/similar"
 require "searchkick/model"
 require "searchkick/tasks"
 require "searchkick/logger" if defined?(Rails)
@@ -7,6 +7,7 @@ module Searchkick
       class_eval do
         extend Searchkick::Search
         extend Searchkick::Reindex
+        include Searchkick::Similar
         include Tire::Model::Search
         include Tire::Model::Callbacks
         tire do
@@ -0,0 +1,32 @@
+module Searchkick
+  module Similar
+    def similar(options = {})
+      like_text = index.retrieve(document_type, id).to_hash
+        .keep_if{|k,v| k[0] != "_" and (!options[:fields] or options[:fields].map(&:to_sym).include?(k)) }
+        .values.compact.join(" ")
+
+      fields = options[:fields] ? options[:fields].map{|f| "#{f}.analyzed" } : ["_all"]
+
+      payload = {
+        query: {
+          more_like_this: {
+            fields: fields,
+            like_text: like_text,
+            min_doc_freq: 1,
+            min_term_freq: 1
+          }
+        },
+        filter: {
+          not: {
+            term: {
+              _id: id
+            }
+          }
+        }
+      }
+
+      search = Tire::Search::Search.new(index_name, payload: payload)
+      Searchkick::Results.new(search.json, search.options)
+    end
+  end
+end
@@ -0,0 +1,10 @@
+require_relative "test_helper"
+
+class TestSimilar < Minitest::Unit::TestCase
+
+  def test_fields
+    store_names ["1% Organic Milk", "2% Organic Milk", "Popcorn"]
+    assert_equal ["2% Organic Milk"], Product.find_by(name: "1% Organic Milk").similar(fields: ["name"]).map(&:name)
+  end
+
+end
@@ -40,7 +40,8 @@ class Product &lt; ActiveRecord::Base
  
   searchkick \
     settings: {
-      number_of_shards: 1
+      number_of_shards: 1,
+      number_of_replicas: 0
     },
     synonyms: [
       ["clorox", "bleach"],
...	...	@@ -284,6 +284,17 @@ Advanced
284	284	Product.search "2% Milk", facets: {store_id: {where: {in_stock: true}}}
285	285	```
286	286
	287	+### Similar Items [master]
	288	+
	289	+Note: Subject to change before the next gem release
	290	+
	291	+Find similar items.
	292	+
	293	+```ruby
	294	+product = Product.first
	295	+product.similar(fields: ["name"])
	296	+```
	297	+
287	298	## Deployment
288	299
289	300	Searchkick uses `ENV["ELASTICSEARCH_URL"]` for the Elasticsearch server. This defaults to `http://localhost:9200`.
...	...
...	...	@@ -3,6 +3,7 @@ require "searchkick/version"
3	3	require "searchkick/reindex"
4	4	require "searchkick/results"
5	5	require "searchkick/search"
	6	+require "searchkick/similar"
6	7	require "searchkick/model"
7	8	require "searchkick/tasks"
8	9	require "searchkick/logger" if defined?(Rails)
...	...
...	...	@@ -7,6 +7,7 @@ module Searchkick
7	7	class_eval do
8	8	extend Searchkick::Search
9	9	extend Searchkick::Reindex
	10	+ include Searchkick::Similar
10	11	include Tire::Model::Search
11	12	include Tire::Model::Callbacks
12	13	tire do
...	...
...	...	@@ -0,0 +1,32 @@
	1	+module Searchkick
	2	+ module Similar
	3	+ def similar(options = {})
	4	+ like_text = index.retrieve(document_type, id).to_hash
	5	+ .keep_if{\|k,v\| k[0] != "_" and (!options[:fields] or options[:fields].map(&:to_sym).include?(k)) }
	6	+ .values.compact.join(" ")
	7	+
	8	+ fields = options[:fields] ? options[:fields].map{\|f\| "#{f}.analyzed" } : ["_all"]
	9	+
	10	+ payload = {
	11	+ query: {
	12	+ more_like_this: {
	13	+ fields: fields,
	14	+ like_text: like_text,
	15	+ min_doc_freq: 1,
	16	+ min_term_freq: 1
	17	+ }
	18	+ },
	19	+ filter: {
	20	+ not: {
	21	+ term: {
	22	+ _id: id
	23	+ }
	24	+ }
	25	+ }
	26	+ }
	27	+
	28	+ search = Tire::Search::Search.new(index_name, payload: payload)
	29	+ Searchkick::Results.new(search.json, search.options)
	30	+ end
	31	+ end
	32	+end
...	...
...	...	@@ -0,0 +1,10 @@
	1	+require_relative "test_helper"
	2	+
	3	+class TestSimilar < Minitest::Unit::TestCase
	4	+
	5	+ def test_fields
	6	+ store_names ["1% Organic Milk", "2% Organic Milk", "Popcorn"]
	7	+ assert_equal ["2% Organic Milk"], Product.find_by(name: "1% Organic Milk").similar(fields: ["name"]).map(&:name)
	8	+ end
	9	+
	10	+end
...	...
...	...	@@ -40,7 +40,8 @@ class Product < ActiveRecord::Base
40	40
41	41	searchkick \
42	42	settings: {
43		- number_of_shards: 1
	43	+ number_of_shards: 1,
	44	+ number_of_replicas: 0
44	45	},
45	46	synonyms: [
46	47	["clorox", "bleach"],
...	...