app/databus/spl_metadata_decoder.py

"""
SPL Token Metadata Decoder
===========================
Parses raw Solana account data to definitively extract hidden metadata,
mint authorities, and freeze flags, bypassing unreliable third-party APIs.

This is a LOCAL/FREE_API provider that reads directly from public Solana RPCs
and decodes the binary SPL Token mint layout without relying on external indexers.

SPL Token Mint Layout (76 bytes base):
  Byte 0:        mint_authority_option (1 = Some, 0 = None)
  Bytes 1-32:    mint_authority pubkey (if Some, else 32 zeros)
  Bytes 33-40:   supply (u64, little-endian)
  Byte 41:       decimals (u8)
  Byte 42:       is_initialized (bool, 1 byte)
  Byte 43:       freeze_authority_option (1 = Some, 0 = None)
  Bytes 44-75:   freeze_authority pubkey (if Some, else 32 zeros)

Token-2022 Extensions:
  If account data > 76 bytes, parse extension headers to detect:
  - MintCloseAuthority
  - PermanentDelegate
  - TransferFee
  - ConfidentialTransfer
  - DefaultAccountState (frozen by default)
"""

import base64
import hashlib
import logging
from typing import Any

import httpx

logger = logging.getLogger("databus.spl_metadata")

# Free public Solana RPCs (fallback chain)
PUBLIC_RPC_ENDPOINTS = [
    "https://api.mainnet-beta.solana.com",
    "https://solana-rpc.publicnode.com",
    "https://solana.drpc.org",
]

# Metaplex Token Metadata Program ID
METAPLEX_METADATA_PROGRAM = "metaqbxxUerdq28cj1RbAWkYQm3ybzjb6a8bt518x1s"
TOKEN_PROGRAM = "TokenkegQfeZyiNwAJbNbGKPFXCWuBvf9Ss623VQ5DA"
TOKEN_2022_PROGRAM = "TokenzQdBNbLqP5VEhdkAS6EPFLC1PHnBqCXEpPxuEb"

# Known malicious/frozen token flags
HIGH_RISK_FLAGS = [
    "mint_authority_present",
    "freeze_authority_present",
    "default_account_state_frozen",
    "transfer_fee_present",
    "permanent_delegate_present",
]


def decode_pubkey(data: bytes, offset: int) -> str | None:
    """Decode a 32-byte Solana pubkey from bytes."""
    if len(data) < offset + 32:
        return None
    pubkey_bytes = data[offset : offset + 32]
    # Check if it's all zeros (None)
    if all(b == 0 for b in pubkey_bytes):
        return None
    # Encode to base58
    return _bytes_to_base58(pubkey_bytes)


def _bytes_to_base58(data: bytes) -> str:
    """Convert bytes to base58 string (Solana pubkey format)."""
    alphabet = b"123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz"
    num = int.from_bytes(data, "big")
    result = bytearray()
    while num > 0:
        num, mod = divmod(num, 58)
        result.append(alphabet[mod])
    # Add leading '1's for each leading zero byte
    leading_zeros = len(data) - len(data.lstrip(b"\x00"))
    result.extend([alphabet[0]] * leading_zeros)
    return result[::-1].decode("ascii")


def parse_spl_mint_data(raw_data: bytes) -> dict[str, Any]:
    """
    Parse raw SPL Token mint account data.
    Returns decoded metadata including authorities, supply, decimals, and flags.

    SPL Token Mint Layout (82 bytes base):
      Bytes 0-3:   mint_authority_option (u32, 1 = Some, 0 = None)
      Bytes 4-35:  mint_authority pubkey (32 bytes)
      Bytes 36-43: supply (u64, little-endian)
      Byte 44:     decimals (u8)
      Byte 45:     is_initialized (bool)
      Bytes 46-49: freeze_authority_option (u32)
      Bytes 50-81: freeze_authority pubkey (32 bytes)
    """
    result = {
        "is_valid": False,
        "decimals": 0,
        "supply": 0,
        "is_initialized": False,
        "mint_authority": None,
        "mint_authority_revoked": True,
        "freeze_authority": None,
        "freeze_authority_revoked": True,
        "risk_flags": [],
        "extensions": [],
        "raw_size": len(raw_data),
    }

    if len(raw_data) < 82:
        result["error"] = f"Data too short for SPL mint: {len(raw_data)} bytes (expected >= 82)"
        return result

    try:
        # Bytes 0-3: mint_authority_option (u32)
        mint_auth_option = int.from_bytes(raw_data[0:4], "little")
        if mint_auth_option == 1:
            result["mint_authority"] = decode_pubkey(raw_data, 4)
            result["mint_authority_revoked"] = False
            result["risk_flags"].append("mint_authority_present")

        # Bytes 36-43: supply (u64 little-endian)
        supply = int.from_bytes(raw_data[36:44], "little")
        result["supply"] = supply

        # Byte 44: decimals
        result["decimals"] = raw_data[44]

        # Byte 45: is_initialized
        result["is_initialized"] = raw_data[45] == 1

        # Bytes 46-49: freeze_authority_option (u32)
        freeze_auth_option = int.from_bytes(raw_data[46:50], "little")
        if freeze_auth_option == 1:
            result["freeze_authority"] = decode_pubkey(raw_data, 50)
            result["freeze_authority_revoked"] = False
            result["risk_flags"].append("freeze_authority_present")

        result["is_valid"] = result["is_initialized"]

        # Parse Token-2022 extensions if data is larger than base 82 bytes
        if len(raw_data) > 82:
            result["extensions"] = _parse_token_extensions(raw_data[82:], result)

    except Exception as e:
        result["error"] = f"Failed to parse mint data: {e!s}"

    return result


def _parse_token_extensions(extension_data: bytes, base_result: dict[str, Any]) -> list[dict[str, Any]]:
    """
    Parse Token-2022 extension data.
    Extension layout: [extension_type (u16), length (u16), data (variable)]
    """
    extensions = []
    offset = 0

    # Extension type constants (from spl-token-2022)
    EXTENSION_TYPES = {
        1: "Uninitialized",
        2: "TransferFeeConfig",
        3: "TransferFeeAmount",
        4: "MintCloseAuthority",
        5: "ConfidentialTransferMint",
        6: "ConfidentialTransferAccount",
        7: "DefaultAccountState",
        8: "ImmutableOwner",
        9: "MemoTransfer",
        10: "NonTransferable",
        11: "InterestBearingConfig",
        12: "CpiGuard",
        13: "PermanentDelegate",
        14: "NonTransferableAccount",
        15: "TransferHook",
        16: "TransferHookAccount",
        17: "ConfidentialTransferFeeConfig",
        18: "ConfidentialTransferFeeAmount",
        19: "MetadataPointer",
        20: "TokenMetadata",
        21: "GroupPointer",
        22: "GroupMemberPointer",
    }

    while offset + 4 <= len(extension_data):
        try:
            ext_type = int.from_bytes(extension_data[offset : offset + 2], "little")
            ext_length = int.from_bytes(extension_data[offset + 2 : offset + 4], "little")

            ext_name = EXTENSION_TYPES.get(ext_type, f"Unknown({ext_type})")
            ext_info = {"type": ext_type, "name": ext_name, "length": ext_length}

            # Check for high-risk extensions
            if ext_type == 4:  # MintCloseAuthority
                ext_info["risk"] = "high"
                ext_info["description"] = "Mint can be closed by authority"
            elif ext_type == 7:  # DefaultAccountState
                # Check if state is Frozen (2)
                if offset + 4 + ext_length <= len(extension_data):
                    state = extension_data[offset + 4]
                    if state == 2:
                        ext_info["risk"] = "critical"
                        ext_info["description"] = "Accounts are frozen by default"
                        base_result["risk_flags"].append("default_account_state_frozen")
            elif ext_type == 13:  # PermanentDelegate
                ext_info["risk"] = "critical"
                ext_info["description"] = "Permanent delegate can transfer any tokens"
                base_result["risk_flags"].append("permanent_delegate_present")
            elif ext_type == 2:  # TransferFeeConfig
                ext_info["risk"] = "medium"
                ext_info["description"] = "Transfer fees can be modified by authority"
                base_result["risk_flags"].append("transfer_fee_present")

            extensions.append(ext_info)
            offset += 4 + ext_length

        except Exception:
            break  # Stop parsing if extension data is malformed

    return extensions


async def fetch_metaplex_metadata(mint: str, rpc_url: str) -> dict[str, Any] | None:
    """
    Fetch Metaplex Token Metadata for a given mint.
    Uses getProgramAccounts to find the metadata PDA.
    """
    # Metaplex metadata PDA derivation:
    # seeds = ["metadata", METAPLEX_METADATA_PROGRAM, mint]
    # For simplicity, we'll use a known RPC method or derive it

    # Actually, the easiest way is to use the getAccountInfo on the known metadata address
    # But deriving it requires sha256. Let's use a simpler approach:
    # query getProgramAccounts with dataSize filter for the metadata program.

    # Better: use the standard metadata PDA derivation
    # seeds: b"metadata", metaplex_program_bytes, mint_bytes

    metaplex_bytes = bytes.fromhex("".join(f"{ord(c):02x}" for c in METAPLEX_METADATA_PROGRAM))
    mint_bytes = bytes.fromhex("".join(f"{ord(c):02x}" for c in mint))

    # Actually, let's use the standard Solana PDA derivation
    # We'll compute it using hashlib.sha256
    seed1 = b"metadata"

    # Create the seed bytes
    seeds = seed1 + metaplex_bytes + mint_bytes

    # PDA derivation: sha256(seeds + bump_seed)
    # We'll try bump seeds 255 down to 0
    for bump in range(255, -1, -1):
        test_seeds = seeds + bytes([bump])
        hashlib.sha256(test_seeds).digest()
        # Check if it's off the ed25519 curve (valid PDA)
        # For simplicity, we'll just use the first one that doesn't have the high bit set
        # Actually, proper PDA derivation is complex. Let's use a known endpoint or skip if too complex.
        pass

    # Simpler approach: use getProgramAccounts with filters
    payload = {
        "jsonrpc": "2.0",
        "id": 1,
        "method": "getProgramAccounts",
        "params": [
            METAPLEX_METADATA_PROGRAM,
            {
                "encoding": "base64",
                "filters": [
                    {"dataSize": 679},  # Standard metadata size
                    {"memcmp": {"offset": 1, "bytes": mint}},  # Mint address at offset 1
                ],
            },
        ],
    }

    try:
        async with httpx.AsyncClient(timeout=10) as client:
            response = await client.post(rpc_url, json=payload)
            if response.status_code == 200:
                data = response.json()
                accounts = data.get("result", [])
                if accounts:
                    account = accounts[0]
                    pubkey = account.get("pubkey")
                    account_data = account.get("account", {}).get("data", [""])[0]
                    raw_bytes = base64.b64decode(account_data)

                    # Parse basic metadata (name, symbol, uri)
                    # Offset 1: mint (32 bytes)
                    # Offset 33: update_authority (32 bytes)
                    # Offset 65: name length (4 bytes) + name
                    # This is complex, so we'll return raw for now or parse simply

                    return {
                        "address": pubkey,
                        "raw_size": len(raw_bytes),
                        "note": "Full metadata parsing requires borsh deserialization",
                    }
    except Exception as e:
        logger.debug(f"Metaplex metadata fetch failed: {e}")

    return None


async def decode_spl_token_metadata(mint: str, rpc_urls: list[str] | None = None) -> dict[str, Any]:
    """
    Main decoder function. Fetches and parses SPL token metadata from public RPCs.

    Args:
        mint: Solana token mint address (base58)
        rpc_urls: List of RPC URLs to try (defaults to PUBLIC_RPC_ENDPOINTS)

    Returns:
        Dict containing decoded metadata, authorities, and risk flags.
    """
    if rpc_urls is None:
        rpc_urls = PUBLIC_RPC_ENDPOINTS

    result = {
        "mint": mint,
        "success": False,
        "rpc_used": None,
        "account_exists": False,
        "decoded": {},
        "metadata": None,
        "risk_score": 0,
        "risk_flags": [],
        "error": None,
    }

    payload = {
        "jsonrpc": "2.0",
        "id": 1,
        "method": "getAccountInfo",
        "params": [mint, {"encoding": "base64"}],
    }

    for rpc_url in rpc_urls:
        try:
            async with httpx.AsyncClient(timeout=8) as client:
                response = await client.post(rpc_url, json=payload)

                if response.status_code != 200:
                    continue

                data = response.json()
                error = data.get("error")
                if error:
                    if "not found" in str(error.get("message", "")).lower():
                        result["error"] = "Mint account not found"
                        return result
                    continue

                value = data.get("result", {}).get("value")
                if value is None:
                    result["error"] = "Mint account not found"
                    return result

                result["account_exists"] = True
                result["rpc_used"] = rpc_url

                # Get owner program
                owner = value.get("owner", "")
                result["owner_program"] = owner

                # Get raw data
                account_data = value.get("data", [""])[0]
                raw_bytes = base64.b64decode(account_data)

                # Parse the mint data
                decoded = parse_spl_mint_data(raw_bytes)
                result["decoded"] = decoded
                result["risk_flags"] = decoded.get("risk_flags", [])

                # Calculate risk score
                risk_score = 0
                for flag in result["risk_flags"]:
                    if flag == "mint_authority_present":
                        risk_score += 30
                    elif flag == "freeze_authority_present":
                        risk_score += 40
                    elif flag == "default_account_state_frozen":
                        risk_score += 50
                    elif flag == "permanent_delegate_present":
                        risk_score += 60
                    elif flag == "transfer_fee_present":
                        risk_score += 20

                # Check extensions for additional risk
                for ext in decoded.get("extensions", []):
                    if ext.get("risk") == "critical":
                        risk_score += 40
                    elif ext.get("risk") == "high":
                        risk_score += 25
                    elif ext.get("risk") == "medium":
                        risk_score += 15

                result["risk_score"] = min(100, risk_score)
                result["success"] = True

                # Try to fetch Metaplex metadata (non-blocking, best effort)
                try:
                    meta = await fetch_metaplex_metadata(mint, rpc_url)
                    if meta:
                        result["metadata"] = meta
                except Exception:
                    pass  # Metadata is optional

                return result

        except httpx.TimeoutException:
            continue
        except Exception as e:
            logger.debug(f"RPC {rpc_url} failed: {e}")
            continue

    result["error"] = "All RPC endpoints failed or timed out"
    return result


# ── DataBus Provider Integration ───────────────────────────────────


async def _spl_metadata_decoder_provider(mint: str = "", **kwargs) -> dict[str, Any] | None:
    """
    DataBus provider function for SPL Token Metadata Decoder.
    """
    if not mint:
        return {"error": "mint address is required"}

    # Validate base58 format (basic check)
    if len(mint) < 32 or len(mint) > 44:
        return {"error": "Invalid mint address format"}

    result = await decode_spl_token_metadata(mint)

    if result["success"]:
        return {
            "mint": mint,
            "source": "spl_metadata_decoder",
            "tier": "free_api",
            "is_local": False,
            "data": result,
        }

    return {"error": result.get("error", "Failed to decode SPL token metadata")}