Merge branch 'function_score' of https://github.com/monkbroc/searchkick into monkbroc-function_score

Andrew Kane
2 parents 49940253 79bcd5a8
Showing 3 changed files with 47 additions and 0 deletions Show diff stats
README.md
lib/searchkick/query.rb
test/sql_test.rb
@@ -616,6 +616,17 @@ Bounded by a box
 City.search "san", where: {location: {top_left: [38, -123], bottom_right: [37, -122]}}
 ```
  
+Modify search score by proximity to a point:
+
+```ruby
+City.search "san", proximity_factor: { field: :location, origin: [37, -122] } # field and origin mandatory
+City.search "san", proximity_factor: { field: :location, origin: [37, -122], function: :linear, scale: "30mi", decay: 0.5 } # at 30mi, decrease score by 50%
+```
+
+The difference between where with a geo_point and proximity_factor is that the former is a hard cutoff that simply includes or excludes results but has no impact on the order of results, whereas the latter smoothly modifies the search score so that the results with the best match for the query text and the distance are returned.
+
+See [more details on modifying the score by proximity](http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html#_decay_functions).
+
 ## Inheritance
  
 Searchkick supports single table inheritance.
@@ -219,6 +219,33 @@ module Searchkick
           }
         end
  
+        # Modify the query scoring by proximity to a geo point
+        if options[:proximity_factor]
+          proximity_factor = { function: :gauss, score_mode: :multiply, scale: "5mi" }.merge(options[:proximity_factor])
+          (proximity_factor[:field] and proximity_factor[:origin]) or raise ArgumentError, "query() option :proximity_factor must contain at least :field and :origin"
+
+          field = proximity_factor[:field]
+          function = proximity_factor[:function]
+          function_params = proximity_factor.select { |k,v| [:origin, :scale, :offset, :decay].include? k }
+          # Conform to GeoJSON convention of [longitude, latitude]
+          function_params[:origin] = function_params[:origin].reverse
+          score_mode = proximity_factor[:score_mode]
+
+          payload = {
+            function_score: {
+              functions: [
+                {
+                  function => {
+                    field => function_params
+                  }
+                }
+              ],
+              query: payload,
+              score_mode: score_mode
+            }
+          }
+        end
+
         payload = {
           query: payload,
           size: per_page,
@@ -261,6 +261,15 @@ class TestSql &lt; Minitest::Test
     assert_search "product", ["Product"], where: {latitude: {gt: 99}}
   end
  
+  def test_proximity
+    store [
+      {name: "San Francisco", latitude: 37.7833, longitude: -122.4167},
+      {name: "San Antonio", latitude: 29.4167, longitude: -98.5000},
+      {name: "San Marino", latitude: 43.9333, longitude: 12.4667}
+    ]
+    assert_order "san", ["San Francisco", "San Antonio", "San Marino"], proximity_factor: {field: :location, origin: [37, -122]}
+  end
+
   # load
  
   def test_load_default
...	...	@@ -616,6 +616,17 @@ Bounded by a box
616	616	City.search "san", where: {location: {top_left: [38, -123], bottom_right: [37, -122]}}
617	617	```
618	618
	619	+Modify search score by proximity to a point:
	620	+
	621	+```ruby
	622	+City.search "san", proximity_factor: { field: :location, origin: [37, -122] } # field and origin mandatory
	623	+City.search "san", proximity_factor: { field: :location, origin: [37, -122], function: :linear, scale: "30mi", decay: 0.5 } # at 30mi, decrease score by 50%
	624	+```
	625	+
	626	+The difference between where with a geo_point and proximity_factor is that the former is a hard cutoff that simply includes or excludes results but has no impact on the order of results, whereas the latter smoothly modifies the search score so that the results with the best match for the query text and the distance are returned.
	627	+
	628	+See [more details on modifying the score by proximity](http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html#_decay_functions).
	629	+
619	630	## Inheritance
620	631
621	632	Searchkick supports single table inheritance.
...	...
...	...	@@ -219,6 +219,33 @@ module Searchkick
219	219	}
220	220	end
221	221
	222	+ # Modify the query scoring by proximity to a geo point
	223	+ if options[:proximity_factor]
	224	+ proximity_factor = { function: :gauss, score_mode: :multiply, scale: "5mi" }.merge(options[:proximity_factor])
	225	+ (proximity_factor[:field] and proximity_factor[:origin]) or raise ArgumentError, "query() option :proximity_factor must contain at least :field and :origin"
	226	+
	227	+ field = proximity_factor[:field]
	228	+ function = proximity_factor[:function]
	229	+ function_params = proximity_factor.select { \|k,v\| [:origin, :scale, :offset, :decay].include? k }
	230	+ # Conform to GeoJSON convention of [longitude, latitude]
	231	+ function_params[:origin] = function_params[:origin].reverse
	232	+ score_mode = proximity_factor[:score_mode]
	233	+
	234	+ payload = {
	235	+ function_score: {
	236	+ functions: [
	237	+ {
	238	+ function => {
	239	+ field => function_params
	240	+ }
	241	+ }
	242	+ ],
	243	+ query: payload,
	244	+ score_mode: score_mode
	245	+ }
	246	+ }
	247	+ end
	248	+
222	249	payload = {
223	250	query: payload,
224	251	size: per_page,
...	...
...	...	@@ -261,6 +261,15 @@ class TestSql < Minitest::Test
261	261	assert_search "product", ["Product"], where: {latitude: {gt: 99}}
262	262	end
263	263
	264	+ def test_proximity
	265	+ store [
	266	+ {name: "San Francisco", latitude: 37.7833, longitude: -122.4167},
	267	+ {name: "San Antonio", latitude: 29.4167, longitude: -98.5000},
	268	+ {name: "San Marino", latitude: 43.9333, longitude: 12.4667}
	269	+ ]
	270	+ assert_order "san", ["San Francisco", "San Antonio", "San Marino"], proximity_factor: {field: :location, origin: [37, -122]}
	271	+ end
	272	+
264	273	# load
265	274
266	275	def test_load_default
...	...