Add treatWordAsLuceneQuery to wordsSocialVolume (#4402)

* Add treatWordAsLuceneQuery to wordsSocialVolume

* Add API tests for wordsSocialVolume

lib/sanbase/social_data/social_helper.ex

@@ -16,17 +16,12 @@ defmodule Sanbase.SocialData.SocialHelper do
 
   def sources_total_string(), do: @sources |> Enum.join(",")
 
-  def replace_words_with_original_casing({:error, error}, _words), do: {:error, error}
+  def replace_words_with_original_casing(result, words) do
+    Enum.map(result, fn %{word: lowercase_word} = map ->
+      original_word = Enum.find(words, fn w -> String.downcase(w) == lowercase_word end)
 
-  def replace_words_with_original_casing({:ok, result}, words) do
-    result =
-      Enum.map(result, fn %{word: lowercase_word} = map ->
-        original_word = Enum.find(words, fn w -> String.downcase(w) == lowercase_word end)
-
-        Map.put(map, :word, original_word)
-      end)
-
-    {:ok, result}
+      Map.put(map, :word, original_word)
+    end)
   end
 
   def social_metrics_selector_handler(%{slug: slugs}) when is_list(slugs) do

lib/sanbase/social_data/social_volume.ex

@@ -9,6 +9,8 @@ defmodule Sanbase.SocialData.SocialVolume do
   alias Sanbase.Project
   alias Sanbase.Clickhouse.NftTrade
 
+  import Sanbase.Utils.Transform, only: [maybe_apply_function: 2]
+
   defp http_client, do: Mockery.Macro.mockable(HTTPoison)
 
   @recv_timeout 25_000
@@ -37,12 +39,30 @@ defmodule Sanbase.SocialData.SocialVolume do
 
   def social_volume(%{words: words} = selector, from, to, interval, source, opts)
       when is_list(words) do
-    transformed_words = Enum.reject(words, &(&1 == "***")) |> Enum.map(&String.downcase/1)
+    transformed_words = Enum.reject(words, &(&1 == "***"))
+    treat_word_as_lucene_query = Keyword.get(opts, :treat_word_as_lucene_query, false)
+
+    transformed_words =
+      case Keyword.get(opts, :treat_word_as_lucene_query, false) do
+        false -> transformed_words |> Enum.map(&String.downcase/1)
+        true -> transformed_words
+      end
+
     selector = %{selector | words: transformed_words}
 
     social_volume_list_request(selector, from, to, interval, source, opts)
     |> handle_words_social_volume_response(selector)
-    |> Sanbase.SocialData.SocialHelper.replace_words_with_original_casing(words)
+    |> maybe_apply_function(fn result ->
+      if treat_word_as_lucene_query do
+        result
+      else
+        # If the words are **not** treated as lucene syntax (the default behaviour)
+        # they are lowercased before they are sent to metricshub. Because of that
+        # we need to properly map the result of metricshub back to the original casing
+        # of the word that came from the API call
+        SocialHelper.replace_words_with_original_casing(result, words)
+      end
+    end)
   end
 
   def social_volume(selector, from, to, interval, source, opts) do
@@ -51,8 +71,7 @@ defmodule Sanbase.SocialData.SocialVolume do
   end
 
   defp handle_response(response, selector) do
-    response
-    |> case do
+    case response do
       {:ok, %HTTPoison.Response{status_code: 200, body: body}} ->
         body
         |> Jason.decode!()

lib/sanbase_web/graphql/resolvers/social_data_resolver.ex

@@ -78,10 +78,14 @@ defmodule SanbaseWeb.Graphql.Resolvers.SocialDataResolver do
 
   def words_social_volume(
         _root,
-        %{selector: %{words: _words} = selector, from: from, to: to, interval: interval},
+        %{selector: %{words: _words} = selector, from: from, to: to, interval: interval} = args,
         _resolution
       ) do
-    SocialData.social_volume(selector, from, to, interval, :total)
+    treat_as_lucene = Map.get(args, :treat_word_as_lucene_query, false)
+
+    SocialData.social_volume(selector, from, to, interval, :total,
+      treat_word_as_lucene_query: treat_as_lucene
+    )
   end
 
   def words_social_dominance(

lib/sanbase_web/graphql/schema/queries/social_data_queries.ex

@@ -170,6 +170,13 @@ defmodule SanbaseWeb.Graphql.Schema.SocialDataQueries do
       arg(:to, non_null(:datetime))
       arg(:interval, :interval, default_value: "1d")
 
+      @desc ~s"""
+      If true, the word will be treated as a Lucene query. Default is false.
+      Lucene queries allows the word to be not a single word, but a query like: eth AND nft.
+      If false, the words are all lowercased and the AND/NOT/OR keywords meaning is lost.
+      """
+      arg(:treat_word_as_lucene_query, :boolean, default_value: false)
+
       complexity(&Complexity.from_to_interval/3)
       middleware(AccessControl, %{allow_realtime_data: true})
       cache_resolve(&SocialDataResolver.words_social_volume/3, ttl: 600, max_ttl_offset: 240)

test/sanbase_web/graphql/social_data/words_social_volume_api_test.exs

@@ -0,0 +1,174 @@
+defmodule SanbaseWeb.Graphql.WordsSocialVolumeApiTest do
+  use SanbaseWeb.ConnCase, async: false
+
+  import Sanbase.Factory
+  import SanbaseWeb.Graphql.TestHelpers
+
+  setup do
+    %{user: user} = insert(:subscription_pro_sanbase, user: insert(:user))
+    conn = setup_jwt_auth(build_conn(), user)
+
+    %{conn: conn}
+  end
+
+  test "successfully fetch words social volume", context do
+    # This test does not include treatWordAsLuceneQuery: true, so the
+    # words are lowercased before being send and in the response
+    body =
+      %{
+        "data" => %{
+          "btc" => %{
+            "2024-09-28T00:00:00Z" => 373,
+            "2024-09-29T00:00:00Z" => 487,
+            "2024-09-30T00:00:00Z" => 323
+          },
+          "eth or nft" => %{
+            "2024-09-28T00:00:00Z" => 1681,
+            "2024-09-29T00:00:00Z" => 3246,
+            "2024-09-30T00:00:00Z" => 1577
+          }
+        }
+      }
+      |> Jason.encode!()
+
+    resp = %HTTPoison.Response{status_code: 200, body: body}
+
+    Sanbase.Mock.prepare_mock(HTTPoison, :get, fn _url, _headers, options ->
+      search_texts =
+        Map.new(options[:params])
+        |> Map.get("search_texts")
+
+      # Assert that the words are lowercased before they are sent
+      assert search_texts == "eth or nft,btc"
+
+      {:ok, resp}
+    end)
+    |> Sanbase.Mock.run_with_mocks(fn ->
+      query = """
+      {
+        wordsSocialVolume(
+          selector: { words: ["eth OR nft", "btc"] }
+          from: "2024-09-28T00:00:00Z"
+          to: "2024-09-30T00:00:00Z"
+          interval: "1d"
+        ){
+          word
+          timeseriesData{
+            datetime
+            mentionsCount
+          }
+        }
+      }
+      """
+
+      result =
+        context.conn
+        |> post("/graphql", query_skeleton(query))
+        |> json_response(200)
+
+      assert result == %{
+               "data" => %{
+                 "wordsSocialVolume" => [
+                   %{
+                     "timeseriesData" => [
+                       %{"datetime" => "2024-09-28T00:00:00Z", "mentionsCount" => 373},
+                       %{"datetime" => "2024-09-29T00:00:00Z", "mentionsCount" => 487},
+                       %{"datetime" => "2024-09-30T00:00:00Z", "mentionsCount" => 323}
+                     ],
+                     "word" => "btc"
+                   },
+                   %{
+                     "timeseriesData" => [
+                       %{"datetime" => "2024-09-28T00:00:00Z", "mentionsCount" => 1681},
+                       %{"datetime" => "2024-09-29T00:00:00Z", "mentionsCount" => 3246},
+                       %{"datetime" => "2024-09-30T00:00:00Z", "mentionsCount" => 1577}
+                     ],
+                     "word" => "eth OR nft"
+                   }
+                 ]
+               }
+             }
+    end)
+  end
+
+  test "successfully fetch words social volume - treatWordAsLuceneQuery: true", context do
+    # This test does not include treatWordAsLuceneQuery: true, so the
+    # words are lowercased before being send and in the response
+    body =
+      %{
+        "data" => %{
+          "BTC" => %{
+            "2024-09-28T00:00:00Z" => 373,
+            "2024-09-29T00:00:00Z" => 487,
+            "2024-09-30T00:00:00Z" => 323
+          },
+          "eth OR nft" => %{
+            "2024-09-28T00:00:00Z" => 1681,
+            "2024-09-29T00:00:00Z" => 3246,
+            "2024-09-30T00:00:00Z" => 1577
+          }
+        }
+      }
+      |> Jason.encode!()
+
+    resp = %HTTPoison.Response{status_code: 200, body: body}
+
+    Sanbase.Mock.prepare_mock(HTTPoison, :get, fn _url, _headers, options ->
+      search_texts =
+        Map.new(options[:params])
+        |> Map.get("search_texts")
+
+      # Assert that the words are **not** lowercased before they are sent
+      assert search_texts == "eth OR nft,BTC"
+
+      {:ok, resp}
+    end)
+    |> Sanbase.Mock.run_with_mocks(fn ->
+      query = """
+      {
+        wordsSocialVolume(
+          selector: { words: ["eth OR nft", "BTC"] }
+          from: "2024-09-28T00:00:00Z"
+          to: "2024-09-30T00:00:00Z"
+          interval: "1d"
+          treatWordAsLuceneQuery: true
+        ){
+          word
+          timeseriesData{
+            datetime
+            mentionsCount
+          }
+        }
+      }
+      """
+
+      result =
+        context.conn
+        |> post("/graphql", query_skeleton(query))
+        |> json_response(200)
+
+      assert result == %{
+               "data" => %{
+                 "wordsSocialVolume" => [
+                   %{
+                     "timeseriesData" => [
+                       %{"datetime" => "2024-09-28T00:00:00Z", "mentionsCount" => 373},
+                       %{"datetime" => "2024-09-29T00:00:00Z", "mentionsCount" => 487},
+                       %{"datetime" => "2024-09-30T00:00:00Z", "mentionsCount" => 323}
+                     ],
+                     "word" => "BTC"
+                   },
+                   %{
+                     "timeseriesData" => [
+                       %{"datetime" => "2024-09-28T00:00:00Z", "mentionsCount" => 1681},
+                       %{"datetime" => "2024-09-29T00:00:00Z", "mentionsCount" => 3246},
+                       %{"datetime" => "2024-09-30T00:00:00Z", "mentionsCount" => 1577}
+                     ],
+                     "word" => "eth OR nft"
+                   }
+                 ]
+               }
+             }
+    end)
+  end
+end