Add TunitModifications with first modifier unite_with_next

.iex-local.exs

@@ -1,6 +1,6 @@
 alias Outlook.HtmlPreparations
 alias Outlook.HtmlPreparations.HtmlPreparation
-alias Outlook.InternalTree.{Html,InternalNode,TranslationUnit}
+alias Outlook.InternalTree.{Html,InternalNode,TranslationUnit,TunitModifications}
 alias Outlook.InternalTree
 alias Outlook.Articles
 alias Outlook.Accounts

lib/outlook/internal_tree.ex

@@ -1,6 +1,6 @@
 defmodule Outlook.InternalTree do
 
-  alias Outlook.InternalTree.{Html,Modifiers,RawInternalBasic,InternalTree,Translation}
+  alias Outlook.InternalTree.{Html,Modifiers,RawInternalBasic,InternalTree,Translation,TunitModifications}
   alias Outlook.HtmlPreparations.HtmlPreparation
   alias Outlook.{Hyphenation, Translations}
 
@@ -71,4 +71,8 @@ defmodule Outlook.InternalTree do
     InternalTree.collect_tunit_ids(tree)
     # |> List.flatten()
   end
+
+  def modify_tunits(tree, modifier, tu_ids) do
+    TunitModifications.apply_modifier(tree, modifier, tu_ids)
+  end
 end

lib/outlook/internal_tree/tunit_modifications.ex

@@ -0,0 +1,83 @@
+defmodule Outlook.InternalTree.TunitModifications do
+
+  alias Outlook.InternalTree.{InternalNode,TranslationUnit}
+
+  require Logger
+
+  def modifiers do
+    [
+      %{
+        name: "unite_with_next",
+        fn: &unite_with_next/2,
+        label: "Unite with next",
+        description: "unite translation unit with next"
+      },
+      %{
+        name: "split_tunit",
+        fn: &split_tunit/2,
+        label: "Split Translation unit",
+        description: "split translation unit into two"
+      }
+    ]
+  end
+
+  # Modifier functions
+
+  defp unite_with_next(nodelist, tu_ids) when is_list(tu_ids) do
+    ids_to_process = Enum.reverse(tu_ids)
+    Logger.info "nu: #{inspect tu_ids}"
+    Enum.reduce(ids_to_process, nodelist, fn id, nodes -> unite_with_next(nodes, id) end)
+  end
+  defp unite_with_next(nodelist, tu_id) do
+    ind = Enum.find_index(nodelist, fn n -> n.nid == tu_id end)
+    case Enum.slice(nodelist, ind, 2) do
+      [unit, next] ->
+        nunit = %TranslationUnit{unit | content: unit.content <> next.content}
+        nodelist
+        |> List.replace_at(ind, nunit)
+        |> List.delete_at(ind + 1)
+      [_] ->
+        nodelist
+    end
+  end
+
+  defp split_tunit(_nodelist, _tu_ids) do
+  end
+
+  # Function and helpers to apply modifiers
+
+  def apply_modifier([ %InternalNode{} = node | rest ], modifier, tu_ids) when node.type == :element do
+    content = case List.first(node.content) do
+      %TranslationUnit{} -> process_tunit_list(node.content, modifier, tu_ids)
+      _ -> apply_modifier(node.content, modifier, tu_ids)
+    end
+    [%InternalNode{node| content: content} | apply_modifier(rest, modifier, tu_ids)]
+  end
+  def apply_modifier([node, rest], modifier, tu_ids), do: [node | apply_modifier(rest, modifier, tu_ids)]
+  def apply_modifier([],_, _), do: []
+
+  def process_tunit_list(tunits, modifier, tu_ids) do
+    modi_fun = get_modi_fun(modifier)
+    ids_to_process = get_ids_to_process(tunits, tu_ids)
+    case length(ids_to_process) do
+      0 -> tunits
+      _ -> modi_fun.(tunits, ids_to_process)
+    end
+  end
+
+  @doc false
+  def get_ids_to_process(tunits, tu_ids) do
+    present_ids = Enum.map(tunits, fn u -> u.nid end)
+    found_ids = MapSet.new(present_ids)
+    |> MapSet.intersection(MapSet.new(tu_ids))
+    |> MapSet.to_list()
+    # make sure to return ids in the order they occur in tunits
+    Enum.filter(present_ids, fn pres -> pres in found_ids end)
+  end
+
+  defp get_modi_fun(modifier) do
+    modifiers()
+    |> Enum.find(fn m -> m.name == modifier end)
+    |> Map.get(:fn)
+  end
+end

test/outlook/internaltree_tunitmodifications_test.exs

@@ -0,0 +1,83 @@
+defmodule Outlook.InternalTreeTest do
+  use Outlook.DataCase
+
+  # import Outlook.InternalTreeTestHelpers
+
+  describe "internal_tree" do
+    alias Outlook.InternalTree
+    alias Outlook.InternalTree.{InternalNode,TranslationUnit}
+
+    def tree() do
+     [
+        %InternalNode{
+          name: "p",
+          attributes: %{},
+          type: :element,
+          nid: "rRIib2h8tyix",
+          content: [
+            %TranslationUnit{status: :untranslated, nid: "GuU9v6xeSS7e", content: "Joe Biden a.", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "bzCLsYGNe2PG", content: "k.", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "GyRUrzwH9LcP", content: "a. ", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "y2yb38U4hkya", content: "Crash Test Dummy.", eph: %{}}
+          ],
+          eph: %{sibling_with: :block}
+        }
+      ]
+    end
+
+    test "unite_with_next unites with next in simple case" do
+      assert InternalTree.modify_tunits(tree(), "unite_with_next", ["bzCLsYGNe2PG"]) == [
+        %InternalNode{
+          name: "p",
+          attributes: %{},
+          type: :element,
+          nid: "rRIib2h8tyix",
+          content: [
+            %TranslationUnit{status: :untranslated, nid: "GuU9v6xeSS7e", content: "Joe Biden a.", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "bzCLsYGNe2PG", content: "k.a. ", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "y2yb38U4hkya", content: "Crash Test Dummy.", eph: %{}}
+          ],
+          eph: %{sibling_with: :block}
+        }
+      ]
+    end
+
+    test "unite_with_next unites all with next in complex case" do
+      assert InternalTree.modify_tunits(tree(), "unite_with_next", ["GuU9v6xeSS7e","bzCLsYGNe2PG","GyRUrzwH9LcP"]) == [
+        %InternalNode{
+          name: "p",
+          attributes: %{},
+          type: :element,
+          nid: "rRIib2h8tyix",
+          content: [
+            %TranslationUnit{
+              status: :untranslated,
+              nid: "GuU9v6xeSS7e",
+              content: "Joe Biden a.k.a. Crash Test Dummy.",
+              eph: %{}
+            }
+          ],
+          eph: %{sibling_with: :block}
+        }
+      ]
+    end
+
+    test "unite_with_next ignores id if there is no next tunit" do
+      assert InternalTree.modify_tunits(tree(), "unite_with_next", ["y2yb38U4hkya"]) == [
+        %InternalNode{
+          name: "p",
+          attributes: %{},
+          type: :element,
+          nid: "rRIib2h8tyix",
+          content: [
+            %TranslationUnit{status: :untranslated, nid: "GuU9v6xeSS7e", content: "Joe Biden a.", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "bzCLsYGNe2PG", content: "k.", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "GyRUrzwH9LcP", content: "a. ", eph: %{}},
+            %TranslationUnit{status: :untranslated, nid: "y2yb38U4hkya", content: "Crash Test Dummy.", eph: %{}}
+          ],
+          eph: %{sibling_with: :block}
+        }
+      ]
+    end
+  end
+end