Merge branch 'fix-search-html-tokenized' into 'master'

Insert spaces before stripping HTML when inserting search data

See merge request framasoft/mobilizon!520

lib/service/formatter/html.ex

@@ -24,5 +24,17 @@ defmodule Mobilizon.Service.Formatter.HTML do
     end
   end
 
+  @doc """
+  Inserts a space before tags closing so that words are not attached once tags stripped
+
+  `<h1>test</h1>next` thing becomes `test next` instead of `testnext`
+  """
+  @spec strip_tags_and_insert_spaces(String.t()) :: String.t()
+  def strip_tags_and_insert_spaces(html) do
+    html
+    |> String.replace("</", " </")
+    |> strip_tags()
+  end
+
   def filter_tags_for_oembed(html), do: Sanitizer.scrub(html, OEmbed)
 end

lib/service/workers/build_search.ex

@@ -45,7 +45,7 @@ defmodule Mobilizon.Service.Workers.BuildSearch do
       [
         event.id,
         event.title,
-        HTML.strip_tags(event.description),
+        HTML.strip_tags_and_insert_spaces(event.description),
         get_tags_string(event)
       ]
     )