Prune index for <package>-<version> before updating (#61)

Author: ruslandoga

lib/hexdocs/queue.ex

@@ -167,23 +167,20 @@ defmodule Hexdocs.Queue do
   end
 
   defp process_search(key, package, version, body, start) do
-    case Version.parse(version) do
-      {:ok, version} ->
-        case Hexdocs.Tar.unpack(body, package: package, version: version) do
-          {:ok, files} ->
-            update_search_index(key, package, version, files)
-
-            elapsed = System.os_time(:millisecond) - start
-            Logger.info("FINISHED INDEXING DOCS #{key} #{elapsed}ms")
+    version =
+      case Version.parse(version) do
+        {:ok, version} -> version
+        :error when package in @special_package_names -> version
+      end
 
-          {:error, reason} ->
-            Logger.error("Failed unpack #{package} #{version}: #{reason}")
-        end
+    case Hexdocs.Tar.unpack(body, package: package, version: version) do
+      {:ok, files} ->
+        update_search_index(key, package, version, files)
+        elapsed = System.os_time(:millisecond) - start
+        Logger.info("FINISHED INDEXING DOCS #{key} #{elapsed}ms")
 
-      :error when package in @special_package_names ->
-        # Skip for special packages, it's probably a tag push that's not valid semver
-        # and we don't need to index those
-        :ok
+      {:error, reason} ->
+        Logger.error("Failed unpack #{package} #{version}: #{reason}")
     end
   end
 
@@ -342,6 +339,8 @@ defmodule Hexdocs.Queue do
   defp update_search_index(key, package, version, files) do
     case Hexdocs.Search.find_search_items(package, version, files) do
       {proglang, items} ->
+        Logger.info("DELETING SEARCH INDEX #{key}")
+        Hexdocs.Search.delete(package, version)
         Logger.info("UPDATING SEARCH INDEX #{key}")
         Hexdocs.Search.index(package, version, proglang, items)
         Logger.info("UPDATED SEARCH INDEX #{key}")

test/hexdocs/search_test.exs

@@ -117,6 +117,36 @@ defmodule Hexdocs.SearchTest do
              })
   end
 
+  test "prunes previous docs before updating search index", %{package: _package} do
+    package = "elixir"
+    version = "main"
+
+    run_upload(package, version, [
+      {"index.html", "contents"},
+      {"dist/search_data-0F918FFD.js",
+       """
+       searchData={"items":[\
+       {"type":"function","title":"Example.old_test/4","doc":"does old example things","ref":"Example.html#old_test/4"},\
+       {"type":"module","title":"Example","doc":"example text","ref":"Example.html"}\
+       ],"content_type":"text/markdown","producer":{"name":"ex_doc","version":[48,46,51,52,46,50]}}\
+       """}
+    ])
+
+    run_upload(package, version, [
+      {"index.html", "contents"},
+      {"dist/search_data-0F918FFD.js",
+       """
+       searchData={"items":[\
+       {"type":"function","title":"Example.new_test/4","doc":"does new example things","ref":"Example.html#new_test/4"},\
+       {"type":"module","title":"Example","doc":"example text","ref":"Example.html"}\
+       ],"content_type":"text/markdown","producer":{"name":"ex_doc","version":[48,46,51,52,46,50]}}\
+       """}
+    ])
+
+    assert [%{"document" => %{"title" => "Example.new_test/4"}}] =
+             typesense_search(%{"q" => "test", "query_by" => "title"})
+  end
+
   test "logs an info message if search_data is not found", %{package: package} do
     original_log_level = Logger.level()
     Logger.configure(level: :info)