[add] API to Crawly.Manager (#118)

* [add] an API to Crawly.Manager

Co-authored-by: oltarasenko <oltarasenko@gmail.com>

lib/crawly/engine.ex

@@ -21,6 +21,23 @@ defmodule Crawly.Engine do
     GenServer.call(__MODULE__, {:start_spider, spider_name})
   end
 
+  @spec get_manager(module()) ::
+          pid() | {:error, :spider_not_found}
+  def get_manager(spider_name) do
+    case Map.fetch(running_spiders(), spider_name) do
+      :error ->
+        {:error, :spider_not_found}
+
+      {:ok, pid_sup} ->
+        Supervisor.which_children(pid_sup)
+        |> Enum.find(&({Crawly.Manager, _, :worker, [Crawly.Manager]} = &1))
+        |> case do
+          nil -> {:error, :spider_not_found}
+          {_, pid, :worker, _} -> pid
+        end
+    end
+  end
+
   @spec stop_spider(module(), reason) :: result
         when reason: :itemcount_limit | :itemcount_timeout | atom(),
              result: :ok | {:error, :spider_not_running}
@@ -47,6 +64,19 @@ defmodule Crawly.Engine do
     {:ok, %Crawly.Engine{}}
   end
 
+  def handle_call({:get_manager, spider_name}, _, state) do
+    pid =
+      case Map.get(state.started_spiders, spider_name) do
+        nil ->
+          {:error, :spider_not_found}
+
+        pid ->
+          pid
+      end
+
+    {:reply, pid, state}
+  end
+
   def handle_call(:running_spiders, _from, state) do
     {:reply, state.started_spiders, state}
   end

lib/crawly/manager.ex

@@ -31,13 +31,26 @@ defmodule Crawly.Manager do
 
   use GenServer
 
-  alias Crawly.Utils
+  alias Crawly.{Engine, Utils}
+
+  @spec add_workers(module(), non_neg_integer()) ::
+          :ok | {:error, :spider_non_exist}
+  def add_workers(spider_name, num_of_workers) do
+    case Engine.get_manager(spider_name) do
+      {:error, reason} ->
+        {:error, reason}
+
+      pid ->
+        GenServer.cast(pid, {:add_workers, num_of_workers})
+    end
+  end
 
   def start_link(spider_name) do
     Logger.debug("Starting the manager for #{spider_name}")
     GenServer.start_link(__MODULE__, spider_name)
   end
 
+  @impl true
   def init(spider_name) do
     # Getting spider start urls
     [start_urls: urls] = spider_name.init()
@@ -83,6 +96,18 @@ defmodule Crawly.Manager do
      %{name: spider_name, tref: tref, prev_scraped_cnt: 0, workers: worker_pids}}
   end
 
+  @impl true
+  def handle_cast({:add_workers, num_of_workers}, state) do
+    Logger.info("Adding #{num_of_workers} workers for #{state.name}")
+
+    Enum.each(1..num_of_workers, fn _ ->
+      DynamicSupervisor.start_child(state.name, {Crawly.Worker, [state.name]})
+    end)
+
+    {:noreply, state}
+  end
+
+  @impl true
   def handle_info(:operations, state) do
     Process.cancel_timer(state.tref)
 
@@ -109,11 +134,11 @@ defmodule Crawly.Manager do
       |> Utils.get_settings(state.name)
       |> maybe_convert_to_integer()
 
-      maybe_stop_spider_by_timeout(
-        state.name,
+    maybe_stop_spider_by_timeout(
+      state.name,
       delta,
-        closespider_timeout_limit
-      )
+      closespider_timeout_limit
+    )
 
     tref =
       Process.send_after(

test/manager_test.exs

@@ -26,6 +26,30 @@ defmodule ManagerTest do
     end)
   end
 
+  test "it is possible to add more workers to a spider" do
+    spider_name = Manager.TestSpider
+    :ok = Crawly.Engine.start_spider(spider_name)
+    initial_number_of_workers = 1
+
+    assert initial_number_of_workers ==
+             DynamicSupervisor.count_children(spider_name)[:workers]
+
+    workers = 2
+    assert :ok == Crawly.Manager.add_workers(spider_name, workers)
+
+    pid = Crawly.Engine.get_manager(spider_name)
+    state = :sys.get_state(pid)
+    assert spider_name == state.name
+
+    assert initial_number_of_workers + workers ==
+             DynamicSupervisor.count_children(spider_name)[:workers]
+  end
+
+  test "returns error when spider doesn't exist" do
+    assert {:error, :spider_not_found} ==
+             Crawly.Manager.add_workers(Manager.NonExistentSpider, 2)
+  end
+
   test "max request per minute is respected" do
     :ok = Crawly.Engine.start_spider(Manager.TestSpider)
 
@@ -53,7 +77,6 @@ defmodule ManagerTest do
   end
 
   test "Closespider timeout is respected" do
-
     Process.register(self(), :spider_closed_callback_test)
 
     # Ignore closespider_itemcount