defmodule Tarakan.Repo.Migrations.CreateCodePatternRules do
  use Ecto.Migration

  @kinds ~w(code_review threat_model privacy_review business_logic verify_findings write_fix
            diff_review refute_finding reproduce_finding calibrate_severity)
  @with_synthesis @kinds ++ ~w(synthesize_rule)

  def up do
    create table(:code_pattern_rules) do
      # One rule per code cluster. The cluster is a string key rather than a row,
      # so this table is where a cluster first becomes a thing you can own.
      add :code_pattern_key, :string, null: false
      add :engine, :string, null: false, default: "semgrep"
      add :language, :string
      add :rule_yaml, :text, null: false

      # Validation is done by the worker, which has both the rule engine and the
      # code. A rule is only servable if it demonstrably matched the instances it
      # claims to cover: `matched_count` of `checked_count`, with the finding ids
      # it hit recorded so the claim can be re-tested later.
      add :checked_count, :integer, null: false, default: 0
      add :matched_count, :integer, null: false, default: 0
      add :matched_finding_ids, {:array, :binary_id}, null: false, default: []
      add :validated_at, :utc_datetime_usec

      add :author_account_id, references(:accounts, on_delete: :nilify_all)

      timestamps(type: :utc_datetime_usec)
    end

    # Newest validated rule per cluster wins; only one row per author per cluster
    # so a resubmission updates rather than piles up.
    create unique_index(:code_pattern_rules, [:code_pattern_key, :author_account_id])
    create index(:code_pattern_rules, [:code_pattern_key, :validated_at])

    create constraint(:code_pattern_rules, :code_pattern_rules_counts_sane,
             check: "checked_count >= 0 AND matched_count >= 0 AND matched_count <= checked_count"
           )

    # A rule that matched nothing is not evidence of a pattern, it is a guess.
    create constraint(:code_pattern_rules, :code_pattern_rules_validated_needs_a_match,
             check: "validated_at IS NULL OR matched_count > 0"
           )

    alter table(:review_tasks) do
      add :target_code_pattern_key, :string
    end

    create index(:review_tasks, [:target_code_pattern_key])

    drop constraint(:review_tasks, :review_tasks_kind_must_be_valid)

    create constraint(:review_tasks, :review_tasks_kind_must_be_valid,
             check: kind_check(@with_synthesis)
           )
  end

  def down do
    drop constraint(:review_tasks, :review_tasks_kind_must_be_valid)
    create constraint(:review_tasks, :review_tasks_kind_must_be_valid, check: kind_check(@kinds))

    drop index(:review_tasks, [:target_code_pattern_key])

    alter table(:review_tasks) do
      remove :target_code_pattern_key
    end

    drop table(:code_pattern_rules)
  end

  defp kind_check(kinds) do
    "kind IN (" <> Enum.map_join(kinds, ", ", &"'#{&1}'") <> ")"
  end
end