# Author: Simon-Pierre Boucher
# Mail: contact@spboucher.ai
#!/usr/bin/env python3
"""Verify every configured token address on-chain before trusting it.

For each chain: check eth_chainId matches the config; for each token,
eth_call symbol() and decimals() on the address and compare with
tokens.yaml. A wrong or dead address returns empty data and fails loudly.

    python3 scripts/verify_tokens.py
"""

import pathlib
import sys

import yaml

sys.path.insert(0, str(pathlib.Path(__file__).parent.parent))
from indexer.rpc import RpcError, RpcPool  # noqa: E402

ROOT = pathlib.Path(__file__).parent.parent
SEL_SYMBOL = "0x95d89b41"    # keccak4("symbol()")
SEL_DECIMALS = "0x313ce567"  # keccak4("decimals()")


def decode_string(hexdata):
    """ABI string return: word0 = offset, word1 = length, then bytes.
    Some old tokens (MKR-style) return bytes32 instead — handle both."""
    if hexdata in (None, "0x", ""):
        return None
    raw = bytes.fromhex(hexdata[2:])
    if len(raw) >= 64:
        offset = int.from_bytes(raw[:32], "big")
        if offset == 32 and len(raw) >= 64:
            length = int.from_bytes(raw[32:64], "big")
            if 0 < length <= len(raw) - 64:
                return raw[64 : 64 + length].decode("utf-8", "replace")
    return raw.rstrip(b"\x00").decode("utf-8", "replace") or None


def main():
    chains = yaml.safe_load((ROOT / "config" / "chains.yaml").read_text())["chains"]
    tokens = yaml.safe_load((ROOT / "config" / "tokens.yaml").read_text())["tokens"]

    failures = 0
    for chain, toks in tokens.items():
        cfg = chains[chain]
        if cfg.get("family", "evm") != "evm":
            # non-EVM identifiers are verified by their adapters (Deliverable 4)
            print(f"{chain:10s} skipped (family={cfg['family']} — no eth_call)")
            continue
        p = RpcPool(cfg["rpcs"], max_cycles=2)
        try:
            chain_id = int(p.call("eth_chainId"), 16)
        except (RpcError, RuntimeError) as e:
            print(f"{chain:10s} UNREACHABLE: {e}")
            failures += 1
            continue
        cid_ok = "ok" if chain_id == cfg["chain_id"] else f"MISMATCH got {chain_id}"
        if chain_id != cfg["chain_id"]:
            failures += 1
        print(f"{chain:10s} chain_id {cfg['chain_id']} [{cid_ok}]  via {p.current_url}")

        for t in toks:
            addr = t["address"]
            try:
                sym = decode_string(
                    p.call("eth_call", [{"to": addr, "data": SEL_SYMBOL}, "latest"])
                )
                dec_hex = p.call(
                    "eth_call", [{"to": addr, "data": SEL_DECIMALS}, "latest"]
                )
                dec = int(dec_hex, 16) if dec_hex not in (None, "0x") else None
            except (RpcError, RuntimeError) as e:
                print(f"  {t['symbol']:5s} {addr}  ERROR {e}")
                failures += 1
                continue
            # Tether uses '₮' in some deployments (USD₮, USD₮0) — normalize
            def norm(s):
                return (s or "").replace("₮", "T").upper()
            expected = t.get("onchain_symbol", t["symbol"])
            ok = norm(sym) == norm(expected) and dec == t["decimals"]
            mark = "ok" if ok else "MISMATCH"
            if not ok:
                failures += 1
            print(f"  {t['symbol']:5s} {addr}  on-chain: symbol={sym!r} decimals={dec} [{mark}]")

    print()
    if failures:
        print(f"FAILED: {failures} problem(s) — do not index until fixed.")
        sys.exit(1)
    print("All token addresses verified on-chain.")


if __name__ == "__main__":
    main()