ccxt_client

defmodule CCXT.Signing.Classifier do
  @moduledoc """
  Classifies exchange signing patterns from spec AST data.

  Analyzes `structure.sign_method` AST from each exchange spec to determine
  which of the 9 signing patterns it uses, plus exchange-specific config
  (header names, encoding preferences).

  ## Classification Strategy

  1. **Header-name matching** (primary) — distinctive header strings in the AST
     uniquely identify patterns (e.g., `X-BAPI-*` → Bybit-style headers)
  2. **Hash algorithm fallback** — sha384/sha512/sha256 in AST identifiers
  3. **Parent inheritance** — child exchanges (e.g., binancecoinm) inherit
     from their parent's pattern
  4. **`:custom` fallback** — unclassifiable exchanges get the escape hatch

  ## Usage

      spec = CCXT.Spec.load!("bybit")
      {pattern, config} = CCXT.Signing.Classifier.classify(spec)
      #=> {:hmac_sha256_headers, %{api_key_header: "X-BAPI-API-KEY", ...}}

  """

  @type classification :: {CCXT.Signing.pattern(), config()}
  @type config :: %{optional(atom()) => term()}

  # Child exchanges that inherit signing from a parent.
  # These have no `structure.sign_method` in their spec.
  @parent_map %{
    "binancecoinm" => "binance",
    "binanceus" => "binance",
    "binanceusdm" => "binance",
    "bequant" => "hitbtc",
    "coinbaseadvanced" => "coinbase",
    "exchange_v1" => "hitbtc",
    "fmfwio" => "hitbtc",
    "gateio" => "gate",
    "huobi" => "htx",
    "kucoinfutures" => "kucoin",
    "myokx" => "okx",
    "okxus" => "okx"
  }

  @doc """
  Classifies the signing pattern for an exchange spec.

  Returns `{pattern_atom, config_map}` where config contains
  exchange-specific header names and encoding preferences.

  If the spec has no `sign_method`, falls back to parent inheritance
  or returns `{:custom, %{}}`.
  """
  @spec classify(map()) :: classification()
  def classify(spec) do
    exchange_id = get_in(spec, ["exchange", "id"])
    sign_method = get_in(spec, ["structure", "sign_method"])

    case sign_method do
      nil -> classify_no_sign_method(exchange_id)
      ast -> classify_from_ast(ast)
    end
  end

  @doc """
  Classifies a signing pattern from a raw sign_method AST map.

  Useful when you have the AST directly without a full spec wrapper.
  """
  @spec classify_from_ast(map()) :: classification()
  def classify_from_ast(ast) do
    text = Jason.encode!(ast)
    literals = extract_literals(text)
    idents = extract_idents(text)

    {pattern, base_config} = match_pattern(literals, idents)
    config = extract_config(literals, pattern, base_config)

    {pattern, config}
  end

  @doc """
  Returns the parent exchange ID for a child exchange, or nil.
  """
  @spec parent_for(String.t()) :: String.t() | nil
  def parent_for(exchange_id), do: Map.get(@parent_map, exchange_id)

  @doc """
  Returns all known parent→child mappings.
  """
  @spec parent_map() :: %{String.t() => String.t()}
  def parent_map, do: @parent_map

  # --- Classification for exchanges without sign_method ---

  defp classify_no_sign_method(exchange_id) do
    case Map.get(@parent_map, exchange_id) do
      nil ->
        {:custom, %{}}

      parent_id ->
        parent_spec = CCXT.Spec.load!(parent_id)
        classify(parent_spec)
    end
  end

  # --- Pattern matching (two-tier heuristic) ---

  # Rules are evaluated in order — first match wins.
  # Each rule is {pattern, check_fn}. Separated from cond to keep complexity low.
  defp match_pattern(literals, idents) do
    rules = header_rules() ++ algorithm_rules()

    result =
      Enum.find_value(rules, fn {pattern, check} ->
        if check.(literals, idents), do: {pattern, %{}}
      end)

    result || {:custom, %{}}
  end

  # Tier 1: Header-name patterns (most reliable — unique per exchange family)
  defp header_rules do
    [
      {:deribit, fn lits, _ids -> has_literal_prefix?(lits, "deri-hmac") end},
      {:hmac_sha256_passphrase_signed, fn lits, _ids -> has_literal_prefix?(lits, "KC-API-") end},
      {:hmac_sha256_iso_passphrase, fn lits, _ids -> has_literal_prefix?(lits, "OK-ACCESS-") end},
      {:hmac_sha384_payload, fn lits, _ids -> has_literal?(lits, "bfx-apikey") or has_literal?(lits, "bfx-nonce") end},
      {:hmac_sha512_gate, fn _lits, ids -> has_ident?(ids, "signaturePath") end},
      {:hmac_sha512_nonce, fn lits, _ids -> has_literal?(lits, "API-Key") and has_literal?(lits, "API-Sign") end},
      {:hmac_sha256_query, fn lits, _ids -> has_literal?(lits, "X-MBX-APIKEY") end},
      {:hmac_sha256_headers, fn lits, _ids -> has_literal_prefix?(lits, "X-BAPI-") end},
      {:hmac_sha256_iso_passphrase,
       fn lits, _ids ->
         has_literal_prefix?(lits, "ACCESS-KEY") and has_literal_prefix?(lits, "ACCESS-SIGN") and
           has_literal_prefix?(lits, "ACCESS-PASSPHRASE")
       end}
    ]
  end

  # Tier 2: Hash algorithm + structural cues (fallback)
  defp algorithm_rules do
    [
      {:hmac_sha384_payload, fn _lits, ids -> has_ident?(ids, "sha384") end},
      {:hmac_sha512_nonce, fn _lits, ids -> has_ident?(ids, "sha512") and has_ident?(ids, "sha256") end},
      {:hmac_sha512_nonce, fn _lits, ids -> has_ident?(ids, "sha512") end},
      {:hmac_sha256_iso_passphrase,
       fn lits, ids ->
         (has_ident?(ids, "passphrase") or has_passphrase_literal?(lits)) and has_ident?(ids, "sha256")
       end},
      {:hmac_sha256_query, fn lits, ids -> has_signature_in_query?(lits) and has_ident?(ids, "sha256") end},
      {:hmac_sha256_headers, fn _lits, ids -> has_ident?(ids, "sha256") end}
    ]
  end

  # --- Config extraction ---

  # Extracts exchange-specific signing config (header names) from AST literals.
  # Pattern modules have sensible defaults, so we only need overrides.
  defp extract_config(literals, pattern, base_config) do
    headers = extract_header_literals(literals)
    config = extract_pattern_config(pattern, headers)
    Map.merge(base_config, config)
  end

  defp extract_pattern_config(:hmac_sha256_headers, headers), do: extract_headers_config(headers)
  defp extract_pattern_config(:hmac_sha256_query, headers), do: extract_query_config(headers)
  defp extract_pattern_config(:hmac_sha256_iso_passphrase, h), do: extract_iso_config(h)
  defp extract_pattern_config(:hmac_sha256_passphrase_signed, h), do: extract_kucoin_config(h)
  defp extract_pattern_config(:hmac_sha512_nonce, headers), do: extract_nonce_config(headers)
  defp extract_pattern_config(:hmac_sha512_gate, headers), do: extract_gate_config(headers)
  defp extract_pattern_config(:hmac_sha384_payload, headers), do: extract_payload_config(headers)
  defp extract_pattern_config(_pattern, _headers), do: %{}

  # Extracts config for HMAC-SHA256 headers pattern (Bybit-style)
  defp extract_headers_config(headers) do
    %{}
    |> maybe_put(:api_key_header, find_header(headers, ["apikey", "api-key", "api_key"]))
    |> maybe_put(:timestamp_header, find_header(headers, ["timestamp", "time"]))
    |> maybe_put(:signature_header, find_header(headers, ["sign", "signature", "hmac"]))
    |> maybe_put(:recv_window_header, find_header(headers, ["recv-window", "recvwindow"]))
  end

  # Extracts config for HMAC-SHA256 query pattern (Binance-style)
  defp extract_query_config(headers) do
    maybe_put(%{}, :api_key_header, find_header(headers, ["apikey", "api-key", "api_key"]))
  end

  # Extracts config for ISO passphrase pattern (OKX-style)
  defp extract_iso_config(headers) do
    %{}
    |> maybe_put(:api_key_header, find_header(headers, ["key", "apikey", "api-key"]))
    |> maybe_put(:timestamp_header, find_header(headers, ["timestamp", "time"]))
    |> maybe_put(:signature_header, find_header(headers, ["sign", "signature"]))
    |> maybe_put(:passphrase_header, find_header(headers, ["passphrase"]))
  end

  # KuCoin passphrase-signed pattern uses same header layout as ISO passphrase (OKX)
  defp extract_kucoin_config(headers), do: extract_iso_config(headers)

  # Extracts config for HMAC-SHA512 nonce pattern (Kraken-style)
  defp extract_nonce_config(headers) do
    %{}
    |> maybe_put(:api_key_header, find_header(headers, ["key", "apikey", "api-key"]))
    |> maybe_put(:signature_header, find_header(headers, ["sign", "signature", "authent", "hmac"]))
  end

  # Extracts config for Gate.io pattern
  defp extract_gate_config(headers) do
    %{}
    |> maybe_put(:api_key_header, find_header(headers, ["key"]))
    |> maybe_put(:timestamp_header, find_header(headers, ["timestamp"]))
    |> maybe_put(:signature_header, find_header(headers, ["sign"]))
  end

  # Extracts config for HMAC-SHA384 payload pattern (Bitfinex-style)
  defp extract_payload_config(headers) do
    payload_header = find_header(headers, ["payload"])
    variant = if payload_header, do: :gemini, else: :bitfinex

    base = %{
      variant: variant,
      api_key_header: find_header(headers, ["apikey", "api-key", "key"]),
      signature_header: find_header(headers, ["sign", "signature"])
    }

    variant_specific =
      case variant do
        :gemini -> %{payload_header: payload_header}
        :bitfinex -> %{nonce_header: find_header(headers, ["nonce"])}
      end

    base
    |> Map.merge(variant_specific)
    |> Map.reject(fn {_k, v} -> is_nil(v) end)
  end

  # --- AST text scanning helpers ---

  # Extracts all string literal values from JSON-serialized AST
  defp extract_literals(text) do
    ~r/"value":\s*"([^"]+)"/
    |> Regex.scan(text)
    |> MapSet.new(fn [_, value] -> value end)
  end

  # Extracts all identifier names from JSON-serialized AST
  defp extract_idents(text) do
    ~r/"name":\s*"([^"]+)"/
    |> Regex.scan(text)
    |> MapSet.new(fn [_, name] -> name end)
  end

  # Extracts header-like string literals (contain dash, mixed case, short, not URLs)
  defp extract_header_literals(literals) do
    literals
    |> Enum.filter(fn lit ->
      String.length(lit) < 40 and
        not String.contains?(lit, "http") and
        not String.contains?(lit, "endpoint") and
        not String.contains?(lit, " ") and
        (String.contains?(lit, "-") or lit in ~w(KEY SIGN Timestamp nonce sign timestamp signature))
    end)
    |> MapSet.new()
  end

  defp has_literal?(literals, value), do: MapSet.member?(literals, value)

  defp has_literal_prefix?(literals, prefix) do
    Enum.any?(literals, &String.starts_with?(&1, prefix))
  end

  defp has_ident?(idents, name), do: MapSet.member?(idents, name)

  defp has_signature_in_query?(literals) do
    Enum.any?(literals, fn lit ->
      String.contains?(lit, "signature=") or String.contains?(lit, "&sign=")
    end)
  end

  # Checks if any literal contains "passphrase" (case-insensitive).
  # Catches exchanges like apex (APEX-PASSPHRASE), coinbase (CB-ACCESS-PASSPHRASE)
  # where passphrase appears in header names but not as an AST identifier.
  defp has_passphrase_literal?(literals) do
    Enum.any?(literals, fn lit ->
      String.contains?(String.downcase(lit), "passphrase")
    end)
  end

  # Finds the first header literal matching any of the search terms (case-insensitive).
  # Sorts first for deterministic results — MapSet iteration order is not guaranteed.
  defp find_header(headers, search_terms) do
    headers
    |> Enum.sort()
    |> Enum.find(fn header ->
      header_lower = String.downcase(header)

      Enum.any?(search_terms, fn term ->
        String.contains?(header_lower, term)
      end)
    end)
  end

  defp maybe_put(map, _key, nil), do: map
  defp maybe_put(map, key, value), do: Map.put(map, key, value)
end