add dead link monitoring for outgoing content links

Scans page blocks and nav items for broken URLs (internal via DB
lookup, external via HTTP HEAD). Daily Oban cron at 03:30, plus
on-demand checks when pages are saved. Admin UI tab on redirects
page with re-check, ignore, and clickable source links.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

test/berrypod/redirects/link_checker_test.exs

@@ -0,0 +1,111 @@
+defmodule Berrypod.Redirects.LinkCheckerTest do
+  use Berrypod.DataCase, async: true
+
+  alias Berrypod.Redirects.LinkChecker
+
+  describe "check_internal/1" do
+    test "static routes are valid" do
+      assert :ok = LinkChecker.check_internal("/")
+      assert :ok = LinkChecker.check_internal("/about")
+      assert :ok = LinkChecker.check_internal("/contact")
+      assert :ok = LinkChecker.check_internal("/delivery")
+      assert :ok = LinkChecker.check_internal("/privacy")
+      assert :ok = LinkChecker.check_internal("/terms")
+      assert :ok = LinkChecker.check_internal("/cart")
+      assert :ok = LinkChecker.check_internal("/search")
+      assert :ok = LinkChecker.check_internal("/collections/all")
+    end
+
+    test "collection routes are always valid" do
+      assert :ok = LinkChecker.check_internal("/collections/nonexistent")
+    end
+
+    test "existing product slug is valid" do
+      conn = Berrypod.ProductsFixtures.provider_connection_fixture()
+
+      Berrypod.Products.create_product(%{
+        provider_connection_id: conn.id,
+        provider_product_id: "checker_test_1",
+        title: "Test Product",
+        slug: "test-checker-product",
+        status: "active",
+        visible: true,
+        in_stock: true,
+        cheapest_price: 1000
+      })
+
+      assert :ok = LinkChecker.check_internal("/products/test-checker-product")
+    end
+
+    test "nonexistent product slug is broken" do
+      assert {:error, "product not found"} =
+               LinkChecker.check_internal("/products/nonexistent-product")
+    end
+
+    test "existing custom page is valid" do
+      {:ok, _page} =
+        Berrypod.Pages.create_custom_page(%{
+          slug: "checker-test-page",
+          title: "Test",
+          blocks: []
+        })
+
+      assert :ok = LinkChecker.check_internal("/checker-test-page")
+    end
+
+    test "nonexistent custom page is broken" do
+      assert {:error, "page not found"} =
+               LinkChecker.check_internal("/totally-nonexistent-page")
+    end
+  end
+
+  describe "check_external/1" do
+    setup do
+      # Use a function plug to stub HTTP responses in tests
+      Application.put_env(:berrypod, :link_checker_plug, &__MODULE__.test_plug/1)
+      on_exit(fn -> Application.delete_env(:berrypod, :link_checker_plug) end)
+      :ok
+    end
+
+    test "healthy URL returns :ok" do
+      assert :ok = LinkChecker.check_external("http://test/healthy")
+    end
+
+    test "404 URL returns error with status" do
+      assert {:error, 404, "not found"} = LinkChecker.check_external("http://test/not-found")
+    end
+
+    test "500 URL returns error with status" do
+      assert {:error, 500, "server error"} =
+               LinkChecker.check_external("http://test/server-error")
+    end
+
+    test "405 falls back to GET" do
+      assert :ok = LinkChecker.check_external("http://test/head-not-allowed")
+    end
+
+    # Test plug that simulates various responses
+    def test_plug(conn) do
+      case conn.request_path do
+        "/healthy" ->
+          Plug.Conn.send_resp(conn, 200, "OK")
+
+        "/not-found" ->
+          Plug.Conn.send_resp(conn, 404, "Not Found")
+
+        "/server-error" ->
+          Plug.Conn.send_resp(conn, 500, "Error")
+
+        "/head-not-allowed" ->
+          if conn.method == "HEAD" do
+            Plug.Conn.send_resp(conn, 405, "Method Not Allowed")
+          else
+            Plug.Conn.send_resp(conn, 200, "OK")
+          end
+
+        _ ->
+          Plug.Conn.send_resp(conn, 404, "Not Found")
+      end
+    end
+  end
+end

test/berrypod/redirects/link_scanner_test.exs

@@ -0,0 +1,207 @@
+defmodule Berrypod.Redirects.LinkScannerTest do
+  use Berrypod.DataCase, async: true
+
+  alias Berrypod.Redirects.LinkScanner
+  alias Berrypod.Pages
+
+  describe "scan_page/1" do
+    test "extracts URLs from hero block" do
+      page = %{
+        slug: "home",
+        title: "Home",
+        blocks: [
+          %{
+            "type" => "hero",
+            "settings" => %{
+              "cta_href" => "/collections/all",
+              "secondary_cta_href" => "https://example.com"
+            }
+          }
+        ]
+      }
+
+      links = LinkScanner.scan_page(page)
+      urls = Enum.map(links, & &1.url) |> Enum.sort()
+
+      assert "/collections/all" in urls
+      assert "https://example.com" in urls
+    end
+
+    test "extracts URLs from button block" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{"type" => "button", "settings" => %{"href" => "/about"}}
+        ]
+      }
+
+      [link] = LinkScanner.scan_page(page)
+      assert link.url == "/about"
+      assert link.type == :internal
+    end
+
+    test "extracts URLs from image_text block" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{"type" => "image_text", "settings" => %{"link_href" => "/products/mug"}}
+        ]
+      }
+
+      [link] = LinkScanner.scan_page(page)
+      assert link.url == "/products/mug"
+    end
+
+    test "extracts URLs from video_embed block" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{
+            "type" => "video_embed",
+            "settings" => %{"url" => "https://youtube.com/watch?v=abc"}
+          }
+        ]
+      }
+
+      [link] = LinkScanner.scan_page(page)
+      assert link.url == "https://youtube.com/watch?v=abc"
+      assert link.type == :external
+    end
+
+    test "skips empty strings, anchors, mailto, tel, and template vars" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{"type" => "hero", "settings" => %{"cta_href" => ""}},
+          %{"type" => "button", "settings" => %{"href" => "#section"}},
+          %{"type" => "button", "settings" => %{"href" => "mailto:hi@example.com"}},
+          %{"type" => "button", "settings" => %{"href" => "tel:+441234567890"}},
+          %{"type" => "button", "settings" => %{"href" => "{{unsubscribe_url}}"}}
+        ]
+      }
+
+      assert LinkScanner.scan_page(page) == []
+    end
+
+    test "deduplicates URLs and groups sources" do
+      page = %{
+        slug: "home",
+        title: "Home",
+        blocks: [
+          %{"type" => "hero", "settings" => %{"cta_href" => "/about"}},
+          %{"type" => "button", "settings" => %{"href" => "/about"}}
+        ]
+      }
+
+      links = LinkScanner.scan_page(page)
+      assert length(links) == 1
+
+      [link] = links
+      assert link.url == "/about"
+      assert length(link.sources) == 2
+    end
+
+    test "classifies internal and external URLs" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{"type" => "button", "settings" => %{"href" => "/about"}},
+          %{"type" => "button", "settings" => %{"href" => "https://example.com"}}
+        ]
+      }
+
+      links = LinkScanner.scan_page(page) |> Enum.sort_by(& &1.url)
+
+      assert Enum.at(links, 0).type == :internal
+      assert Enum.at(links, 1).type == :external
+    end
+
+    test "handles blocks with no settings" do
+      page = %{
+        slug: "test",
+        title: "Test",
+        blocks: [
+          %{"type" => "spacer"},
+          %{"type" => "divider", "settings" => %{}}
+        ]
+      }
+
+      assert LinkScanner.scan_page(page) == []
+    end
+  end
+
+  describe "scan_nav/0" do
+    test "extracts URLs from header and footer nav" do
+      Berrypod.Settings.put_setting(
+        "header_nav",
+        [%{"label" => "Shop", "href" => "/collections/all"}],
+        "json"
+      )
+
+      Berrypod.Settings.put_setting(
+        "footer_nav",
+        [%{"label" => "Privacy", "href" => "/privacy"}],
+        "json"
+      )
+
+      links = LinkScanner.scan_nav()
+      urls = Enum.map(links, & &1.url)
+
+      assert "/collections/all" in urls
+      assert "/privacy" in urls
+    end
+
+    test "returns empty when no nav configured" do
+      assert LinkScanner.scan_nav() == []
+    end
+  end
+
+  describe "find_sources/1" do
+    test "finds page blocks using a URL" do
+      # Save a custom page (not system page, avoids ETS cache issues)
+      {:ok, _page} =
+        Pages.create_custom_page(%{
+          slug: "test-link-page",
+          title: "Test page",
+          blocks: [
+            %{"type" => "hero", "settings" => %{"cta_href" => "/test-unique-link"}},
+            %{"type" => "button", "settings" => %{"href" => "/about"}}
+          ]
+        })
+
+      sources = LinkScanner.find_sources("/test-unique-link")
+
+      assert length(sources) == 1
+      [source] = sources
+      assert source.type == "page_block"
+      assert source.id == "test-link-page"
+      assert source.label =~ "hero"
+      assert source.edit_path == "/admin/pages/test-link-page"
+    end
+
+    test "finds nav items using a URL" do
+      Berrypod.Settings.put_setting(
+        "header_nav",
+        [%{"label" => "Shop", "href" => "/test-nav-link"}],
+        "json"
+      )
+
+      sources = LinkScanner.find_sources("/test-nav-link")
+
+      assert length(sources) == 1
+      [source] = sources
+      assert source.type == "nav_item"
+      assert source.label =~ "Header nav"
+      assert source.edit_path == "/admin/navigation"
+    end
+
+    test "returns empty for unused URL" do
+      assert LinkScanner.find_sources("/nonexistent") == []
+    end
+  end
+end