Merge pull request #502 from slovensko-digital/GO-491/substring-search

Allow prefix fulltext search using an asterisk

app/models/filter_subscription.rb

@@ -70,7 +70,7 @@ def scope_searchable(searchable)
     end
 
     scope = scope.where.not("tag_ids && ARRAY[?]", query[:filter_out_tag_ids]) if query[:filter_out_tag_ids].present?
-    scope = scope.fulltext_search(query[:fulltext]) if query[:fulltext].present?
+    scope = scope.fulltext_search(query[:fulltext], prefix_search: query[:prefix_search]) if query[:fulltext].present?
 
     scope
   end

app/models/searchable/message_thread.rb

@@ -22,30 +22,39 @@ class Searchable::MessageThread < ApplicationRecord
   scope :with_tag_id, ->(tag_id) { where("tag_ids && ARRAY[?]", [tag_id]) }
 
   include PgSearch::Model
-  pg_search_scope :pg_search_all,
-                  against: [:title, :content, :note, :tag_names],
-                  using: {
-                    tsearch: {
-                      highlight: {
-                        StartSel: '<span class="bg-yellow-200 text-gray-950">', # if you change classed aad them to view in comment
-                        StopSel: '</span>',
-                        MaxWords: 15,
-                        MinWords: 0,
-                        ShortWord: 1,
-                        HighlightAll: true,
-                        MaxFragments: 2,
-                        FragmentDelimiter: '&hellip;'
-                      }
-                    }
-                  }
+
+
+  pg_search_scope :pg_search_all, lambda { |query, is_prefix = false|
+    {
+      query: query,
+      against: [:title, :content, :note, :tag_names],
+      using: {
+        tsearch: {
+          prefix: is_prefix,
+          highlight: {
+            StartSel: '<span class="bg-yellow-200 text-gray-950">', # if you change classed aad them to view in comment
+            StopSel: '</span>',
+            MaxWords: 15,
+            MinWords: 0,
+            ShortWord: 1,
+            HighlightAll: true,
+            MaxFragments: 2,
+            FragmentDelimiter: '&hellip;'
+          }
+        }
+      }
+    }
+  }
 
   def self.matching(scopeable)
     scopeable.scope_searchable(self) # double dispatch
   end
 
-  def self.fulltext_search(query)
+
+  def self.fulltext_search(query, prefix_search: false)
     pg_search_all(
-      Searchable::IndexHelpers.searchable_string(query)
+      Searchable::IndexHelpers.searchable_string(query),
+      prefix_search
     )
   end
 
@@ -71,7 +80,7 @@ def self.search_ids(query_filter, search_permissions:, cursor:, per_page:, direc
       end
     end
     scope = scope.where.not("tag_ids && ARRAY[?]", query_filter[:filter_out_tag_ids]) if query_filter[:filter_out_tag_ids].present?
-    scope = scope.fulltext_search(query_filter[:fulltext]).with_pg_search_highlight if query_filter[:fulltext].present?
+    scope = scope.fulltext_search(query_filter[:fulltext], prefix_search: query_filter[:prefix_search]).with_pg_search_highlight if query_filter[:fulltext].present?
     scope = scope.select(:message_thread_id, :last_message_delivered_at)
 
     # remove default order rule given by pg_search

app/models/searchable/message_thread_query.rb

@@ -1,6 +1,8 @@
 # frozen_string_literal: true
 
 class Searchable::MessageThreadQuery
+  PREFIX_SEARCH_REGEXP = /^\S{4,}\*$/
+
   def self.parse(query)
     filter_labels = []
     filter_out_labels = []
@@ -21,16 +23,19 @@ def self.parse(query)
       with_text = with_text.gsub("#{match[0]}:#{match[1]}", "")
     end
 
+    with_text = with_text.gsub(/\s+/, ' ').strip
+
     {
-      fulltext: with_text.gsub(/\s+/, ' ').strip,
+      fulltext: with_text,
+      prefix_search: with_text.match?(PREFIX_SEARCH_REGEXP),
       filter_labels: filter_labels,
       filter_out_labels: filter_out_labels,
     }
   end
 
   def self.labels_to_ids(parsed_query, tenant:)
-    fulltext, filter_labels, filter_out_labels =
-      parsed_query.fetch_values(:fulltext, :filter_labels, :filter_out_labels)
+    fulltext, prefix_search, filter_labels, filter_out_labels =
+      parsed_query.fetch_values(:fulltext, :prefix_search, :filter_labels, :filter_out_labels)
 
     # TODO maybe with one query
     found_all, filter_tag_ids = label_names_to_tag_ids(tenant, filter_labels)
@@ -48,6 +53,7 @@ def self.labels_to_ids(parsed_query, tenant:)
 
     result[:filter_out_tag_ids] = filter_out_tag_ids if filter_out_tag_ids.present?
     result[:fulltext] = fulltext if fulltext.present?
+    result[:prefix_search] = prefix_search
 
     result
   end

test/models/searchable/message_thread_query_test.rb

@@ -4,12 +4,14 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
   test "parser empty" do
     assert_equal Searchable::MessageThreadQuery.parse(''), {
       fulltext: '',
+      prefix_search: false,
       filter_labels: [],
       filter_out_labels: []
     }
 
     assert_equal Searchable::MessageThreadQuery.parse(nil), {
       fulltext: '',
+      prefix_search: false,
       filter_labels: [],
       filter_out_labels: []
     }
@@ -18,6 +20,7 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
   test "parser only include tag" do
     assert_equal Searchable::MessageThreadQuery.parse('label:(tag one/with something)'), {
       fulltext: '',
+      prefix_search: false,
       filter_labels: ['tag one/with something'],
       filter_out_labels: []
     }
@@ -26,6 +29,7 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
   test "parser only exclude tag" do
     assert_equal Searchable::MessageThreadQuery.parse('-label:(without)'), {
       fulltext: '',
+      prefix_search: false,
       filter_labels: [],
       filter_out_labels: ['without']
     }
@@ -35,6 +39,7 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
     query = 'label:(tag one/with something) hello label:(tag two) world -label:(without this tag) ending'
     assert_equal Searchable::MessageThreadQuery.parse(query), {
       fulltext: 'hello world ending',
+      prefix_search: false,
       filter_labels: ['tag one/with something', 'tag two'],
       filter_out_labels: ['without this tag']
     }
@@ -44,6 +49,7 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
     query = 'something -label:* else'
     assert_equal Searchable::MessageThreadQuery.parse(query), {
       fulltext: 'something else',
+      prefix_search: false,
       filter_labels: [],
       filter_out_labels: ["*"]
     }
@@ -53,8 +59,29 @@ class Searchable::MessageThreadQueryTest < ActiveSupport::TestCase
     query = 'something -label:* else -label:two'
     assert_equal Searchable::MessageThreadQuery.parse(query), {
       fulltext: 'something else',
+      prefix_search: false,
       filter_labels: [],
       filter_out_labels: ["*", "two"]
     }
   end
+
+  test "parser with prefix search" do
+    query = 'someth* -label:*'
+    assert_equal Searchable::MessageThreadQuery.parse(query), {
+      fulltext: 'someth*',
+      prefix_search: true,
+      filter_labels: [],
+      filter_out_labels: ["*"]
+    }
+  end
+
+  test "parser without prefix search for multiple words" do
+    query = 'tell me someth*'
+    assert_equal Searchable::MessageThreadQuery.parse(query), {
+      fulltext: 'tell me someth*',
+      prefix_search: false,
+      filter_labels: [],
+      filter_out_labels: []
+    }
+  end
 end