fnord

defmodule AI.Embeddings.Migration do
  @moduledoc """
  Detects and performs one-time migration from OpenAI embeddings to the local
  embedding model. Migration is triggered when stored embedding vectors have
  a different dimensionality than the current model expects.

  The migration deletes all stored embeddings so they can be regenerated by
  the background indexers or a manual `fnord index` run.
  """

  @doc """
  Checks whether migration is needed and performs it if so.
  Returns :ok in all cases (migration is best-effort).

  `context` controls the user-facing messaging only:
    * `:ask` - rebuild happens in the background while ask runs
    * `:index` - full foreground reindex is about to start
    * `:read_only` - read-only command (search, conversations, memory);
      no background indexer is running, user must run `fnord index`
  """
  @spec maybe_migrate(:ask | :index | :read_only) :: :ok
  def maybe_migrate(context) do
    case detect_stale_embeddings() do
      :ok -> :ok
      :stale -> do_migrate(context)
    end
  end

  # Sample one embedding file to check dimensions. If no embeddings exist
  # yet, there's nothing to migrate.
  defp detect_stale_embeddings do
    expected = AI.Embeddings.dimensions()

    with {:ok, project} <- Store.get_project(),
         sample when not is_nil(sample) <- sample_embedding(project) do
      if length(sample) == expected, do: :ok, else: :stale
    else
      # No project or no embeddings on disk - nothing to migrate
      _ -> :ok
    end
  end

  # Try to find one existing embedding vector from any index type.
  defp sample_embedding(project) do
    # File embeddings
    sample_file_embedding(project) ||
      sample_commit_embedding(project) ||
      sample_conversation_embedding(project) ||
      sample_memory_embedding(project)
  end

  defp sample_file_embedding(project) do
    project
    |> Store.Project.files_root()
    |> Path.join("*/embeddings.json")
    |> Path.wildcard()
    |> List.first()
    |> read_embedding_file()
  end

  defp sample_commit_embedding(project) do
    project
    |> Store.Project.CommitIndex.root()
    |> Path.join("*/embeddings.json")
    |> Path.wildcard()
    |> List.first()
    |> read_embedding_file()
  end

  defp sample_conversation_embedding(project) do
    project
    |> Store.Project.ConversationIndex.root()
    |> Path.join("*/embeddings.json")
    |> Path.wildcard()
    |> List.first()
    |> read_embedding_file()
  end

  defp sample_memory_embedding(project) do
    project_dir = Path.join(project.store_path, "memory")
    global_dir = Path.join(Settings.fnord_home(), "memory")

    sample_memory_embedding_in_dir(project_dir) || sample_memory_embedding_in_dir(global_dir)
  end

  defp sample_memory_embedding_in_dir(dir) do
    dir
    |> Path.join("*.json")
    |> Path.wildcard()
    |> Enum.find_value(fn path ->
      case File.read(path) do
        {:ok, data} ->
          case SafeJson.decode(data) do
            {:ok, %{"embeddings" => emb}} when is_list(emb) and length(emb) > 0 -> emb
            _ -> nil
          end

        _ ->
          nil
      end
    end)
  end

  defp read_embedding_file(nil), do: nil

  defp read_embedding_file(path) do
    case File.read(path) do
      {:ok, data} ->
        case SafeJson.decode(data) do
          {:ok, list} when is_list(list) -> list
          _ -> nil
        end

      _ ->
        nil
    end
  end

  defp do_migrate(context) do
    UI.warn("""
    [Embeddings Migration] Detected embeddings from a different model (wrong dimensions).
    Wiping the file, commit, and conversation indexes and clearing memory embeddings
    so everything can be regenerated with the local model.
    """)

    with {:ok, project} <- Store.get_project() do
      delete_file_index(project)
      delete_commit_index(project)
      delete_conversation_index(project)
      clear_memory_embeddings(project)
    end

    memory_warning()

    case context do
      :ask ->
        UI.info("""
        File/commit/conversation indexes cleared; they'll rebuild in the background
        as the indexer runs. Run `fnord index` to reindex everything immediately.
        """)

      :index ->
        UI.info("""
        Indexes cleared. Full reindexing will begin now.
        """)

      :read_only ->
        UI.info("""
        File/commit/conversation indexes cleared. This command will return
        limited or empty results until you run `fnord index` to rebuild.
        """)
    end

    :ok
  end

  # Memories are different from the other indexes: their embeddings are
  # cleared in place (the memory text itself is preserved), and they are
  # backfilled by background services + `fnord index`. Until a memory is
  # re-embedded, it is effectively unreachable via semantic search - the
  # user needs to know that.
  defp memory_warning do
    UI.error("""

    ================================================================================
      MEMORIES ARE TEMPORARILY UNAVAILABLE
    ================================================================================
      All saved memories (project + global) had their embeddings cleared as part
      of the model migration. The memory text is intact - only the vectors are
      missing. Until each memory is re-embedded with the new model, semantic
      recall against it will return nothing.

      Re-embedding happens slowly in the background via the memory indexer.
      To force it to finish immediately, run:

          fnord index

    ================================================================================
    """)
  end

  # Nuke the entire files/ tree. Deleting only embeddings.json would leave
  # stale summaries on disk, generated against the previous embedding-input
  # contract and prompt. Reindex regenerates summary + embeddings from
  # scratch.
  #
  # This operation is destructive and non-transactional: if the process dies
  # after wiping some indexes but before others, `detect_stale_embeddings`
  # on the next run will still trigger correctly as long as *any* surviving
  # on-disk embedding (file, commit, conversation, memory) has the old
  # dimension. Memories have an in-place guard (Memory.do_search skips
  # dim-mismatched entries) and are backfilled by the memory indexer.
  defp delete_file_index(project) do
    project
    |> Store.Project.files_root()
    |> File.rm_rf()
  end

  defp delete_commit_index(project) do
    project
    |> Store.Project.CommitIndex.root()
    |> File.rm_rf()
  end

  defp delete_conversation_index(project) do
    project
    |> Store.Project.ConversationIndex.root()
    |> File.rm_rf()
  end

  defp clear_memory_embeddings(project) do
    # Project-scoped memories
    clear_memory_embeddings_in_dir(Path.join(project.store_path, "memory"))

    # Global memories
    clear_memory_embeddings_in_dir(Path.join(Settings.fnord_home(), "memory"))
  end

  # Memory files are rewritten in place with embeddings set to nil. A
  # non-atomic File.write/2 that gets interrupted mid-stream leaves a
  # truncated file on disk. Since detect_stale_embeddings samples only
  # one non-nil embedding, a corrupt file plus an otherwise-fully-migrated
  # on-disk state would skip re-migration AND leave the memory
  # unreadable by the Memory module. Write atomically via tmp + rename so
  # the on-disk state is always either the pre-wipe or post-wipe JSON.
  defp clear_memory_embeddings_in_dir(dir) do
    dir
    |> Path.join("*.json")
    |> Path.wildcard()
    |> Enum.each(fn path ->
      case File.read(path) do
        {:ok, data} ->
          case SafeJson.decode(data) do
            {:ok, %{"embeddings" => emb} = memory} when is_list(emb) ->
              updated = Map.put(memory, "embeddings", nil)

              case SafeJson.encode(updated) do
                {:ok, json} ->
                  try do
                    Settings.write_atomic!(path, json)
                  rescue
                    e ->
                      UI.warn(
                        "[Embeddings Migration] Failed to clear embeddings in #{path}: " <>
                          Exception.message(e)
                      )

                      :ok
                  end

                _ ->
                  :ok
              end

            _ ->
              :ok
          end

        _ ->
          :ok
      end
    end)
  end
end