Don't return empty completions for single char prefix

Related: #400

VS Code in particular handles empty completions lists very poorly,
refusing to make further completion requests. This commit removes the
logic that was preventing completions from being returned when the
prefix is a single character.

apps/server/lib/lexical/server/code_intelligence/completion.ex

@@ -4,7 +4,6 @@ defmodule Lexical.Server.CodeIntelligence.Completion do
   alias Lexical.Completion.Translatable
   alias Lexical.Document
   alias Lexical.Document.Position
-  alias Lexical.Math
   alias Lexical.Project
   alias Lexical.Protocol.Types.Completion
   alias Lexical.Protocol.Types.InsertTextFormat
@@ -38,7 +37,7 @@ defmodule Lexical.Server.CodeIntelligence.Completion do
     case Env.new(project, document, position) do
       {:ok, env} ->
         completions = completions(project, env, context)
-        Logger.warning("Emitting completions: #{inspect(completions)}")
+        Logger.info("Emitting completions: #{inspect(completions)}")
         completion_list(completions)
 
       {:error, _} = error ->
@@ -51,7 +50,7 @@ defmodule Lexical.Server.CodeIntelligence.Completion do
     prefix_tokens = Env.prefix_tokens(env, 1)
 
     cond do
-      prefix_tokens == [] ->
+      prefix_tokens == [] or not context_will_give_meaningful_completions?(env) ->
         []
 
       match?([{:operator, :do, _}], prefix_tokens) and Env.empty?(env.suffix) ->
@@ -61,9 +60,6 @@ defmodule Lexical.Server.CodeIntelligence.Completion do
         |> Builder.snippet(do_end_snippet, label: "do/end block")
         |> List.wrap()
 
-      Enum.empty?(prefix_tokens) or not context_will_give_meaningful_completions?(env) ->
-        []
-
       Env.in_context?(env, :struct_arguments) and not Env.in_context?(env, :struct_field_value) and
           not prefix_is_trigger?(env) ->
         project
@@ -107,30 +103,7 @@ defmodule Lexical.Server.CodeIntelligence.Completion do
   end
 
   defp context_will_give_meaningful_completions?(%Env{} = env) do
-    case Code.Fragment.cursor_context(env.prefix) do
-      {:local_or_var, name} ->
-        local_length = length(name)
-
-        surround_begin =
-          Math.clamp(env.position.character - local_length - 1, 1, env.position.character)
-
-        case Code.Fragment.surround_context(env.prefix, {1, surround_begin}) do
-          :none ->
-            local_length > 1
-
-          _other ->
-            true
-        end
-
-      :none ->
-        false
-
-      {:unquoted_atom, name} ->
-        length(name) > 1
-
-      _ ->
-        true
-    end
+    Code.Fragment.cursor_context(env.prefix) != :none
   end
 
   defp displayable?(%Project{} = project, result) do

apps/server/test/lexical/server/code_intelligence/completion_test.exs

@@ -52,20 +52,16 @@ defmodule Lexical.Server.CodeIntelligence.CompletionTest do
       end
     end
 
-    test "ignores erlang modules", %{project: project} do
-      assert %Completion.List{is_incomplete: true, items: []} =
-               complete(project, ":e|", as_list: false)
+    test "completes erlang modules", %{project: project} do
+      assert [_ | _] = completions = complete(project, ":e|")
+
+      for completion <- completions do
+        assert completion.kind == :module
+      end
     end
   end
 
   describe "ignoring things" do
-    test "return empty items and mark is_incomplete when single character contexts", %{
-      project: project
-    } do
-      assert %Completion.List{is_incomplete: true, items: []} =
-               complete(project, "def my_thing() d|", as_list: false)
-    end
-
     test "returns an incomplete completion list when the context is empty", %{project: project} do
       assert %Completion.List{is_incomplete: true, items: []} =
                complete(project, " ", as_list: false)