Tweak Sitemap generation

Include all public, published items in sitemap Have index point at api/sitemap/ route
nulib · Jul 30, 2024 · 6ba4b8d · 6ba4b8d
1 parent 44fa64d
commit 6ba4b8d
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 5 deletions.
diff --git a/app/config/dev.exs b/app/config/dev.exs
@@ -124,6 +124,7 @@ if prefix = System.get_env("DEV_PREFIX") do
 end
 
 config :meadow, :sitemaps,
+  base_url: "https://dc.library.northwestern.edu/",
   gzip: false,
   store: Sitemapper.FileStore,
   sitemap_url: "https://devbox.library.northwestern.edu:3333/",

diff --git a/app/config/releases.exs b/app/config/releases.exs
@@ -105,7 +105,9 @@ config :meadow,
       bucket: aws_secret("meadow", dig: ["buckets", "sitemap"]),
       path: "/"
     ],
-    sitemap_url: aws_secret("meadow", dig: ["dc", "base_url"])
+    base_url: aws_secret("meadow", dig: ["dc", "base_url"]),
+    sitemap_url:
+      aws_secret("meadow", dig: ["dc", "base_url"], apply: &{:ok, Path.join(&1, "api/sitemap")})
   ],
   validation_ping_interval: environment_secret("VALIDATION_PING_INTERVAL", default: "1000")
 
@@ -125,7 +127,12 @@ config :meadow, Meadow.Scheduler,
   overlap: false,
   timezone: "America/Chicago",
   jobs: [
-    # Runs daily at the configured time (default: 2AM Central)
+    # Sitemap generation runs daily at the configured time (default: 1AM Central)
+    {
+      aws_secret("meadow", dig: ["scheduler", "sitemap"], default: "0 1 * * *"),
+      {Meadow.Utils.Sitemap, :generate, []}
+    },
+    # Preservation check runs daily at the configured time (default: 2AM Central)
     {
       aws_secret("meadow", dig: ["scheduler", "preservation_check"], default: "0 2 * * *"),
       {Meadow.Data.PreservationChecks, :start_job, []}

diff --git a/app/config/test.exs b/app/config/test.exs
@@ -119,6 +119,7 @@ config :honeybadger,
 config :meadow, :sitemaps,
   gzip: true,
   store: Sitemapper.S3Store,
+  base_url: "http://localhost:3333/",
   sitemap_url: "http://localhost:3333/",
   store_config: [bucket: prefix("uploads"), path: ""]
 

diff --git a/app/lib/meadow/utils/sitemap.ex b/app/lib/meadow/utils/sitemap.ex
@@ -2,9 +2,12 @@ defmodule Meadow.Utils.Sitemap do
   @moduledoc """
   Generate and upload Digital Collection sitemaps
   """
-  alias Meadow.Data.{Collections, Works}
+  alias Meadow.Data.Collections
+  alias Meadow.Data.Schemas.Work
   alias Meadow.Repo
 
+  import Ecto.Query
+
   require Logger
 
   @doc """
@@ -74,7 +77,7 @@ defmodule Meadow.Utils.Sitemap do
   end
 
   defp work_urls do
-    Works.work_query(visibility: "OPEN", work_type: "IMAGE")
+    from(w in Work, where: w.visibility["id"] == ^"OPEN" and w.published)
     |> Repo.stream()
     |> Stream.map(fn %{id: id, updated_at: updated_at} ->
       %Sitemapper.URL{
@@ -88,7 +91,7 @@ defmodule Meadow.Utils.Sitemap do
 
   defp expand_url(path) do
     config()
-    |> Keyword.get(:sitemap_url)
+    |> Keyword.get(:base_url)
     |> URI.parse()
     |> URI.merge(path)
     |> URI.to_string()