Merge pull request #2605 from opentensor/feat/roman/brigh-chain-metagraph

[RAO] Add methods to fetch metagraph data from the chain

Author: basfroman

bittensor/core/async_subtensor.py

@@ -14,10 +14,10 @@
 from scalecodec.base import RuntimeConfiguration
 from scalecodec.type_registry import load_type_registry_preset
 
-from bittensor.core.types import SubtensorMixin
 from bittensor.core.chain_data import (
     DelegateInfo,
     StakeInfo,
+    MetagraphInfo,
     NeuronInfoLite,
     NeuronInfo,
     ProposalVoteData,
@@ -60,6 +60,7 @@
 from bittensor.core.metagraph import AsyncMetagraph
 from bittensor.core.settings import version_as_int, TYPE_REGISTRY, DELEGATES_DETAILS_URL
 from bittensor.core.types import ParamWithTypes
+from bittensor.core.types import SubtensorMixin
 from bittensor.utils import (
     decode_hex_identity_dict,
     format_error_message,
@@ -1282,6 +1283,33 @@ async def get_minimum_required_stake(self):
 
         return Balance.from_rao(getattr(result, "value", 0))
 
+    async def get_metagraph_info(
+        self, netuid: int, block: Optional[int] = None
+    ) -> Optional[MetagraphInfo]:
+        block_hash = await self.get_block_hash(block)
+
+        query = await self.substrate.runtime_call(
+            "SubnetInfoRuntimeApi",
+            "get_metagraph",
+            params=[netuid],
+            block_hash=block_hash,
+        )
+        metagraph_bytes = bytes.fromhex(query.decode()[2:])
+        return MetagraphInfo.from_vec_u8(metagraph_bytes)
+
+    async def get_all_metagraphs_info(
+        self, block: Optional[int] = None
+    ) -> list[MetagraphInfo]:
+        block_hash = await self.get_block_hash(block)
+
+        query = await self.substrate.runtime_call(
+            "SubnetInfoRuntimeApi",
+            "get_all_metagraphs",
+            block_hash=block_hash,
+        )
+        metagraphs_bytes = bytes.fromhex(query.decode()[2:])
+        return MetagraphInfo.list_from_vec_u8(metagraphs_bytes)
+
     async def get_netuids_for_hotkey(
         self,
         hotkey_ss58: str,

bittensor/core/chain_data/__init__.py

@@ -6,22 +6,50 @@
 from scalecodec.types import GenericCall
 
 from .axon_info import AxonInfo
+from .chain_identity import ChainIdentity
 from .delegate_info import DelegateInfo
 from .delegate_info_lite import DelegateInfoLite
+from .dynamic_info import DynamicInfo
 from .ip_info import IPInfo
+from .metagraph_info import MetagraphInfo
 from .neuron_info import NeuronInfo
 from .neuron_info_lite import NeuronInfoLite
 from .neuron_certificate import NeuronCertificate
 from .prometheus_info import PrometheusInfo
 from .proposal_vote_data import ProposalVoteData
 from .scheduled_coldkey_swap_info import ScheduledColdkeySwapInfo
-from .subnet_state import SubnetState
 from .stake_info import StakeInfo
 from .subnet_hyperparameters import SubnetHyperparameters
-from .subnet_info import SubnetInfo
-from .dynamic_info import DynamicInfo
 from .subnet_identity import SubnetIdentity
+from .subnet_info import SubnetInfo
+from .subnet_state import SubnetState
 from .weight_commit_info import WeightCommitInfo
 from .utils import custom_rpc_type_registry, decode_account_id, process_stake_data
 
 ProposalCallData = GenericCall
+
+__all__ = [
+    AxonInfo,
+    ChainIdentity,
+    DelegateInfo,
+    DelegateInfoLite,
+    DynamicInfo,
+    IPInfo,
+    MetagraphInfo,
+    NeuronInfo,
+    NeuronInfoLite,
+    NeuronCertificate,
+    PrometheusInfo,
+    ProposalCallData,
+    ProposalVoteData,
+    ScheduledColdkeySwapInfo,
+    StakeInfo,
+    SubnetHyperparameters,
+    SubnetIdentity,
+    SubnetInfo,
+    SubnetState,
+    WeightCommitInfo,
+    custom_rpc_type_registry,
+    decode_account_id,
+    process_stake_data,
+]

bittensor/core/chain_data/chain_identity.py

@@ -0,0 +1,15 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class ChainIdentity:
+    """Dataclass for chain identity information."""
+
+    # In `bittensor.core.chain_data.utils.custom_rpc_type_registry` represents as `ChainIdentityOf` structure.
+
+    name: str
+    url: str
+    image: str
+    discord: str
+    description: str
+    additional: str

bittensor/core/chain_data/metagraph_info.py

@@ -0,0 +1,225 @@
+from dataclasses import dataclass
+from typing import Optional
+
+from bittensor.core.chain_data.axon_info import AxonInfo
+from bittensor.core.chain_data.chain_identity import ChainIdentity
+from bittensor.core.chain_data.subnet_identity import SubnetIdentity
+from bittensor.core.chain_data.utils import (
+    ChainDataType,
+    from_scale_encoding,
+)
+from bittensor.utils import u64_normalized_float as u64tf, u16_normalized_float as u16tf
+from bittensor.utils.balance import Balance
+from scalecodec.utils.ss58 import ss58_encode
+
+
+# to balance with unit (just shortcut)
+def _tbwu(val: int, netuid: Optional[int] = 0) -> Balance:
+    """Returns a Balance object from a value and unit."""
+    return Balance.from_tao(val, netuid)
+
+
+@dataclass
+class MetagraphInfo:
+    # Subnet index
+    netuid: int
+
+    # Name and symbol
+    name: str
+    symbol: str
+    identity: Optional[SubnetIdentity]
+    network_registered_at: int
+
+    # Keys for owner.
+    owner_hotkey: str  # hotkey
+    owner_coldkey: str  # coldkey
+
+    # Tempo terms.
+    block: int  # block at call.
+    tempo: int  # epoch tempo
+    last_step: int
+    blocks_since_last_step: int
+
+    # Subnet emission terms
+    subnet_emission: Balance  # subnet emission via tao
+    alpha_in: Balance  # amount of alpha in reserve
+    alpha_out: Balance  # amount of alpha outstanding
+    tao_in: Balance  # amount of tao injected per block
+    alpha_out_emission: Balance  # amount injected in alpha reserves per block
+    alpha_in_emission: Balance  # amount injected outstanding per block
+    tao_in_emission: Balance  # amount of tao injected per block
+    pending_alpha_emission: Balance  # pending alpha to be distributed
+    pending_root_emission: Balance  # pending tao for root divs to be distributed
+
+    # Hparams for epoch
+    rho: int  # subnet rho param
+    kappa: float  # subnet kappa param
+
+    # Validator params
+    min_allowed_weights: float  # min allowed weights per val
+    max_weights_limit: float  # max allowed weights per val
+    weights_version: int  # allowed weights version
+    weights_rate_limit: int  # rate limit on weights.
+    activity_cutoff: int  # validator weights cut off period in blocks
+    max_validators: int  # max allowed validators.
+
+    # Registration
+    num_uids: int
+    max_uids: int
+    burn: Balance  # current burn cost.
+    difficulty: float  # current difficulty.
+    registration_allowed: bool  # allows registrations.
+    pow_registration_allowed: bool  # pow registration enabled.
+    immunity_period: int  # subnet miner immunity period
+    min_difficulty: float  # min pow difficulty
+    max_difficulty: float  # max pow difficulty
+    min_burn: Balance  # min tao burn
+    max_burn: Balance  # max tao burn
+    adjustment_alpha: float  # adjustment speed for registration params.
+    adjustment_interval: int  # pow and burn adjustment interval
+    target_regs_per_interval: int  # target registrations per interval
+    max_regs_per_block: int  # max registrations per block.
+    serving_rate_limit: int  # axon serving rate limit
+
+    # CR
+    commit_reveal_weights_enabled: bool  # Is CR enabled.
+    commit_reveal_period: int  # Commit reveal interval
+
+    # Bonds
+    liquid_alpha_enabled: bool  # Bonds liquid enabled.
+    alpha_high: float  # Alpha param high
+    alpha_low: float  # Alpha param low
+    bonds_moving_avg: float  # Bonds moving avg
+
+    # Metagraph info.
+    hotkeys: list[str]  # hotkey per UID
+    coldkeys: list[str]  # coldkey per UID
+    identities: list[Optional[ChainIdentity]]  # coldkeys identities
+    axons: list[AxonInfo]  # UID axons.
+    active: list[bool]  # Active per UID
+    validator_permit: list[bool]  # Val permit per UID
+    pruning_score: list[float]  # Pruning per UID
+    last_update: list[int]  # Last update per UID
+    emission: list[Balance]  # Emission per UID
+    dividends: list[float]  # Dividends per UID
+    incentives: list[float]  # Mining incentives per UID
+    consensus: list[float]  # Consensus per UID
+    trust: list[float]  # Trust per UID
+    rank: list[float]  # Rank per UID
+    block_at_registration: list[int]  # Reg block per UID
+    alpha_stake: list[Balance]  # Alpha staked per UID
+    tao_stake: list[Balance]  # TAO staked per UID
+    total_stake: list[Balance]  # Total stake per UID
+
+    # Dividend break down.
+    tao_dividends_per_hotkey: list[
+        tuple[str, Balance]
+    ]  # List of dividend payouts in tao via root.
+    alpha_dividends_per_hotkey: list[
+        tuple[str, Balance]
+    ]  # List of dividend payout in alpha via subnet.
+
+    @classmethod
+    def from_vec_u8(cls, vec_u8: bytes) -> Optional["MetagraphInfo"]:
+        """Returns a Metagraph object from encoded MetagraphInfo vector."""
+        if len(vec_u8) == 0:
+            return None
+        decoded = from_scale_encoding(vec_u8, ChainDataType.MetagraphInfo)
+        if decoded is None:
+            return None
+
+        return MetagraphInfo.fix_decoded_values(decoded)
+
+    @classmethod
+    def list_from_vec_u8(cls, vec_u8: bytes) -> list["MetagraphInfo"]:
+        """Returns a list of Metagraph objects from a list of encoded MetagraphInfo vectors."""
+        decoded = from_scale_encoding(
+            vec_u8, ChainDataType.MetagraphInfo, is_vec=True, is_option=True
+        )
+        if decoded is None:
+            return []
+
+        decoded = [
+            MetagraphInfo.fix_decoded_values(meta)
+            for meta in decoded
+            if meta is not None
+        ]
+        return decoded
+
+    @classmethod
+    def fix_decoded_values(cls, decoded: dict) -> "MetagraphInfo":
+        """Returns a Metagraph object from a decoded MetagraphInfo dictionary."""
+        # Subnet index
+        _netuid = decoded["netuid"]
+
+        # Name and symbol
+        decoded.update({"name": bytes(decoded.get("name")).decode()})
+        decoded.update({"symbol": bytes(decoded.get("symbol")).decode()})
+        decoded.update({"identity": decoded.get("identity", {})})
+
+        # Keys for owner.
+        decoded["owner_hotkey"] = ss58_encode(decoded["owner_hotkey"])
+        decoded["owner_coldkey"] = ss58_encode(decoded["owner_coldkey"])
+
+        # Subnet emission terms
+        decoded["subnet_emission"] = _tbwu(decoded["subnet_emission"])
+        decoded["alpha_in"] = _tbwu(decoded["alpha_in"], _netuid)
+        decoded["alpha_out"] = _tbwu(decoded["alpha_out"], _netuid)
+        decoded["tao_in"] = _tbwu(decoded["tao_in"])
+        decoded["alpha_out_emission"] = _tbwu(decoded["alpha_out_emission"], _netuid)
+        decoded["alpha_in_emission"] = _tbwu(decoded["alpha_in_emission"], _netuid)
+        decoded["tao_in_emission"] = _tbwu(decoded["tao_in_emission"])
+        decoded["pending_alpha_emission"] = _tbwu(
+            decoded["pending_alpha_emission"], _netuid
+        )
+        decoded["pending_root_emission"] = _tbwu(decoded["pending_root_emission"])
+
+        # Hparams for epoch
+        decoded["kappa"] = u16tf(decoded["kappa"])
+
+        # Validator params
+        decoded["min_allowed_weights"] = u16tf(decoded["min_allowed_weights"])
+        decoded["max_weights_limit"] = u16tf(decoded["max_weights_limit"])
+
+        # Registration
+        decoded["burn"] = _tbwu(decoded["burn"])
+        decoded["difficulty"] = u64tf(decoded["difficulty"])
+        decoded["min_difficulty"] = u64tf(decoded["min_difficulty"])
+        decoded["max_difficulty"] = u64tf(decoded["max_difficulty"])
+        decoded["min_burn"] = _tbwu(decoded["min_burn"])
+        decoded["max_burn"] = _tbwu(decoded["max_burn"])
+        decoded["adjustment_alpha"] = u64tf(decoded["adjustment_alpha"])
+
+        # Bonds
+        decoded["alpha_high"] = u16tf(decoded["alpha_high"])
+        decoded["alpha_low"] = u16tf(decoded["alpha_low"])
+        decoded["bonds_moving_avg"] = u64tf(decoded["bonds_moving_avg"])
+
+        # Metagraph info.
+        decoded["hotkeys"] = [ss58_encode(ck) for ck in decoded.get("hotkeys", [])]
+        decoded["coldkeys"] = [ss58_encode(hk) for hk in decoded.get("coldkeys", [])]
+        decoded["axons"] = decoded.get("axons", [])
+        decoded["pruning_score"] = [
+            u16tf(ps) for ps in decoded.get("pruning_score", [])
+        ]
+        decoded["emission"] = [_tbwu(em, _netuid) for em in decoded.get("emission", [])]
+        decoded["dividends"] = [u16tf(dv) for dv in decoded.get("dividends", [])]
+        decoded["incentives"] = [u16tf(ic) for ic in decoded.get("incentives", [])]
+        decoded["consensus"] = [u16tf(cs) for cs in decoded.get("consensus", [])]
+        decoded["trust"] = [u16tf(tr) for tr in decoded.get("trust", [])]
+        decoded["rank"] = [u16tf(rk) for rk in decoded.get("trust", [])]
+        decoded["alpha_stake"] = [_tbwu(ast, _netuid) for ast in decoded["alpha_stake"]]
+        decoded["tao_stake"] = [_tbwu(ts) for ts in decoded["tao_stake"]]
+        decoded["total_stake"] = [_tbwu(ts, _netuid) for ts in decoded["total_stake"]]
+
+        # Dividend break down
+        decoded["tao_dividends_per_hotkey"] = [
+            (ss58_encode(alpha[0]), _tbwu(alpha[1]))
+            for alpha in decoded["tao_dividends_per_hotkey"]
+        ]
+        decoded["alpha_dividends_per_hotkey"] = [
+            (ss58_encode(adphk[0]), _tbwu(adphk[1], _netuid))
+            for adphk in decoded["alpha_dividends_per_hotkey"]
+        ]
+
+        return MetagraphInfo(**decoded)