add tests

test/feed_reader/recovery/recovery_test.exs

@@ -0,0 +1,54 @@
+defmodule FeedReader.RecoveryTest do
+  use ExUnit.Case, async: true
+
+  alias FeedReader.Recovery
+
+  describe "declared_encoding/1" do
+    test "defaults to UTF-8 when no declaration is present" do
+      assert Recovery.declared_encoding("<rss/>") == :utf8
+    end
+
+    test "recognizes common Latin-1 aliases case-insensitively" do
+      assert Recovery.declared_encoding(~s(<?xml encoding="ISO-8859-1"?>)) == :latin1
+      assert Recovery.declared_encoding(~s(<?xml encoding='LaTiN-1'?>)) == :latin1
+    end
+
+    test "recognizes US-ASCII and leaves other declarations UTF-8" do
+      assert Recovery.declared_encoding(~s(<?xml encoding="US-ASCII"?>)) == :ascii
+      assert Recovery.declared_encoding(~s(<?xml encoding="UTF-8"?>)) == :utf8
+      assert Recovery.declared_encoding(~s(<?xml encoding="UTF-16"?>)) == :utf8
+    end
+
+    test "finds declarations after XML declaration attributes" do
+      prefix = ~s(<?xml version="1.0" standalone="yes" encoding = 'latin-1'?>)
+
+      assert Recovery.declared_encoding(prefix) == :latin1
+    end
+  end
+
+  test "normalizes all recovery stages across chunk boundaries" do
+    chunks = [
+      <<0xEF>>,
+      <<0xBB, 0xBF>>,
+      " \n<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>",
+      ~s(<item title="A "quoted"">caf),
+      <<0xE9>>,
+      " & B</item>"
+    ]
+
+    assert normalize(chunks, :latin1) ==
+             ~s(<?xml version="1.0" encoding="UTF-8"?><item title="A &quot;quoted&quot;">café &amp; B</item>)
+  end
+
+  test "preserves CDATA and comments while repairing surrounding text" do
+    chunks = [
+      "<root><![C",
+      "DATA[A & B]]><!-- C & D -->E & F</root>"
+    ]
+
+    assert normalize(chunks, :utf8) ==
+             "<root><![CDATA[A & B]]><!-- C & D -->E &amp; F</root>"
+  end
+
+  defp normalize(chunks, encoding), do: chunks |> Recovery.normalize(encoding) |> Enum.join()
+end