berrypod/lib/simpleshop_theme/search.ex

defmodule SimpleshopTheme.Search do
  @moduledoc """
  Full-text product search backed by SQLite FTS5.

  Uses a contentless FTS5 index with BM25 ranking. The index is rebuilt
  from the products table after each provider sync.
  """

  import Ecto.Query

  alias SimpleshopTheme.Products.Product
  alias SimpleshopTheme.Repo

  @listing_preloads [images: :image]

  # BM25 column weights: title(10), category(5), variant_info(3), description(1)
  @bm25_weights "10.0, 5.0, 3.0, 1.0"

  @doc """
  Searches products by query string. Returns ranked list of Product structs
  with listing preloads, or empty list for blank/short queries.
  """
  def search(query) when is_binary(query) do
    query = String.trim(query)

    if String.length(query) < 2 do
      []
    else
      fts_query = build_fts_query(query)

      case search_fts(fts_query) do
        [] -> search_like(query)
        results -> results
      end
    end
  end

  def search(_), do: []

  @doc """
  Rebuilds the entire FTS5 index from visible, active products.
  """
  def rebuild_index do
    Repo.transaction(fn ->
      # Clear existing index data
      Repo.query!("DELETE FROM products_search_map")
      Repo.query!("DELETE FROM products_search")

      # Load all visible products with variants for indexing
      products =
        Product
        |> where([p], p.visible == true and p.status == "active")
        |> preload([:variants])
        |> Repo.all()

      Enum.each(products, &insert_into_index/1)
    end)
  end

  @doc """
  Indexes or reindexes a single product.
  Removes existing entry first if present.
  """
  def index_product(%Product{} = product) do
    product = Repo.preload(product, [:variants], force: true)

    Repo.transaction(fn ->
      remove_from_index(product.id)
      insert_into_index(product)
    end)
  end

  @doc """
  Removes a product from the search index.
  """
  def remove_product(product_id) do
    remove_from_index(product_id)
  end

  # Build an FTS5 MATCH query from user input.
  # Strips special chars, splits into tokens, adds * prefix match to last token.
  defp build_fts_query(input) do
    tokens =
      input
      |> String.replace(~r/[^\w\s]/, "")
      |> String.split(~r/\s+/, trim: true)

    case tokens do
      [] ->
        nil

      tokens ->
        {complete, [last]} = Enum.split(tokens, -1)

        parts =
          Enum.map(complete, &~s("#{&1}")) ++
            [~s("#{last}" *)]

        Enum.join(parts, " ")
    end
  end

  defp search_fts(nil), do: []

  defp search_fts(fts_query) do
    result =
      Repo.query!(
        """
        SELECT m.product_id, bm25(products_search, #{@bm25_weights}) AS rank
        FROM products_search
        JOIN products_search_map m ON m.rowid = products_search.rowid
        WHERE products_search MATCH ?1
        ORDER BY rank
        LIMIT 20
        """,
        [fts_query]
      )

    product_ids = Enum.map(result.rows, fn [id, _rank] -> id end)

    if product_ids == [] do
      []
    else
      # Fetch full structs preserving rank order
      products_by_id =
        Product
        |> where([p], p.id in ^product_ids)
        |> where([p], p.visible == true and p.status == "active")
        |> preload(^@listing_preloads)
        |> Repo.all()
        |> Map.new(&{&1.id, &1})

      Enum.flat_map(product_ids, fn id ->
        case Map.get(products_by_id, id) do
          nil -> []
          product -> [product]
        end
      end)
    end
  end

  # Substring fallback when FTS5 prefix matching returns nothing
  defp search_like(query) do
    pattern = "%#{sanitize_like(query)}%"

    Product
    |> where([p], p.visible == true and p.status == "active")
    |> where([p], like(p.title, ^pattern) or like(p.category, ^pattern))
    |> order_by([p], p.title)
    |> limit(20)
    |> preload(^@listing_preloads)
    |> Repo.all()
  end

  defp sanitize_like(input) do
    input
    |> String.replace("\\", "\\\\")
    |> String.replace("%", "\\%")
    |> String.replace("_", "\\_")
  end

  defp insert_into_index(%Product{} = product) do
    Repo.query!(
      "INSERT INTO products_search_map (product_id) VALUES (?1)",
      [product.id]
    )

    %{rows: [[rowid]]} = Repo.query!("SELECT last_insert_rowid()")

    variant_info = build_variant_info(product.variants || [])
    description = strip_html(product.description || "")

    Repo.query!(
      """
      INSERT INTO products_search (rowid, title, category, variant_info, description)
      VALUES (?1, ?2, ?3, ?4, ?5)
      """,
      [rowid, product.title || "", product.category || "", variant_info, description]
    )
  end

  defp remove_from_index(product_id) do
    case Repo.query!(
           "SELECT rowid FROM products_search_map WHERE product_id = ?1",
           [product_id]
         ) do
      %{rows: [[rowid]]} ->
        Repo.query!("DELETE FROM products_search WHERE rowid = ?1", [rowid])
        Repo.query!("DELETE FROM products_search_map WHERE rowid = ?1", [rowid])

      _ ->
        :ok
    end
  end

  # Build searchable variant text from enabled variants
  defp build_variant_info(variants) do
    variants
    |> Enum.filter(& &1.is_enabled)
    |> Enum.flat_map(fn v -> [v.title | Map.values(v.options || %{})] end)
    |> Enum.uniq()
    |> Enum.join(" ")
  end

  # Strip HTML tags and decode common entities
  defp strip_html(html) do
    html
    |> String.replace(~r/<[^>]+>/, " ")
    |> String.replace("&amp;", "&")
    |> String.replace("&lt;", "<")
    |> String.replace("&gt;", ">")
    |> String.replace("&quot;", "\"")
    |> String.replace("&#39;", "'")
    |> String.replace("&nbsp;", " ")
    |> String.replace(~r/\s+/, " ")
    |> String.trim()
  end
end