WIP4

curvesim/exceptions/__init__.py

@@ -27,6 +27,10 @@ def __repr__(self):
         return f"HttpClientError({self.status}, {self.message}, url={self.url})"
 
 
+class CurvesimTypeError(CurvesimException, TypeError):
+    """Raised when an argument is the wrong type."""
+
+
 class CurvesimValueError(CurvesimException, ValueError):
     """Raised when an argument has an inappropriate value (but the right type)."""
 

curvesim/iterators/price_samplers/price_volume.py

@@ -1,7 +1,6 @@
 from typing import Iterator
 
 from curvesim.logging import get_logger
-from curvesim.price_data import get_price_data
 from curvesim.templates.price_samplers import PriceSample, PriceSampler
 from curvesim.utils import dataclass, override
 
@@ -29,15 +28,7 @@ class PriceVolume(PriceSampler):
     An iterator that retrieves price/volume and iterates over timepoints in the data.
     """
 
-    def __init__(
-        self,
-        assets,
-        *,
-        days=60,
-        data_dir="data",
-        src="coingecko",
-        end=None,
-    ):
+    def __init__(self, data):
         """
         Retrieves price/volume data and prepares it for iteration.
 
@@ -57,17 +48,7 @@ def __init__(
             Identifies pricing source: coingecko or local.
 
         """
-        prices, volumes, _ = get_price_data(
-            assets.addresses,
-            chain=assets.chain,
-            days=days,
-            data_dir=data_dir,
-            src=src,
-            end=end,
-        )
-
-        self.prices = prices.set_axis(assets.symbol_pairs, axis="columns")
-        self.volumes = volumes.set_axis(assets.symbol_pairs, axis="columns")
+        self.data = data
 
     @override
     def __iter__(self) -> Iterator[PriceVolumeSample]:
@@ -76,16 +57,18 @@ def __iter__(self) -> Iterator[PriceVolumeSample]:
         -------
         :class:`PriceVolumeSample`
         """
-        for price_row, volume_row in zip(
-            self.prices.iterrows(), self.volumes.iterrows()
-        ):
-            price_timestamp, prices = price_row
-            volume_timestamp, volumes = volume_row
-            assert (
-                price_timestamp == volume_timestamp
-            ), "Price/volume timestamps don't match"
-
-            prices = prices.to_dict()
-            volumes = volumes.to_dict()
-
-            yield PriceVolumeSample(price_timestamp, prices, volumes)  # type:ignore
+        for row in self.data.iterrows():
+            timestamp, row_data = row
+
+            prices = row_data["price"].to_dict()
+            volumes = row_data["volume"].to_dict()
+
+            yield PriceVolumeSample(timestamp, prices, volumes)  # type:ignore
+
+    @property
+    def prices(self):
+        return self.data["price"]
+
+    @property
+    def volumes(self):
+        return self.data["volume"]

curvesim/metrics/base.py

@@ -312,8 +312,6 @@ def __init__(self, coin_names, **kwargs):
             Symbols for the coins used in a simulation. A numeraire is selected from
             the specified coins.
         """
-
-        self.coin_names = coin_names
         self.numeraire = get_numeraire(coin_names)
         super().__init__(**kwargs)
 
@@ -348,9 +346,6 @@ def get_market_price(self, base, quote, prices):
         return prices[(base, quote)]
 
 
-pandas_coin_pair_attr = {DataFrame: "columns", Series: "index"}
-
-
 def get_coin_pairs(prices):
     """
     Returns the coin pairs available in the price data.
@@ -418,4 +413,4 @@ def __init__(self, pool, **kwargs):
             :func:`pool_config` and stored as :python:`self._pool` for access during
             metric computations. Number and names of coins derived from pool metadata.
         """
-        super().__init__(pool.assets.symbols, pool=pool)
+        super().__init__(pool.asset_names, pool=pool)

curvesim/metrics/metrics.py

@@ -88,7 +88,7 @@ def config(self):
         }
 
     def __init__(self, pool, **kwargs):
-        super().__init__(pool.assets.symbols)
+        super().__init__(pool.asset_names)
 
     def compute_arb_metrics(self, **kwargs):
         """Computes all metrics for each timestamp in an individual run."""
@@ -218,7 +218,7 @@ def get_stableswap_metapool_volume(self, **kwargs):
         """
         trade_data = kwargs["trade_data"]
 
-        meta_asset = self._pool.assets.symbols[0]
+        meta_asset = self._pool.asset_names[0]
 
         def per_timestamp_function(trade_data):
             volume = 0

curvesim/pipelines/common/__init__.py

@@ -1,13 +1,17 @@
 """
 Contains variables and functions common to the arbitrage pipelines.
 """
+__all__ = ["DEFAULT_METRICS", "get_arb_trades", "get_asset_data", "get_pool_data"]
 
 from scipy.optimize import root_scalar
 
 from curvesim.logging import get_logger
 from curvesim.metrics import metrics as Metrics
 from curvesim.templates.trader import ArbTrade
 
+from .get_asset_data import get_asset_data
+from .get_pool_data import get_pool_data
+
 logger = get_logger(__name__)
 DEFAULT_METRICS = [
     Metrics.Timestamp,

curvesim/pipelines/common/get_asset_data.py

@@ -0,0 +1,22 @@
+from datetime import datetime, timedelta, timezone
+
+from curvesim.pool_data import get_pool_assets
+from curvesim.price_data import get_price_data
+from curvesim.templates import DateTimeSequence
+
+
+def get_asset_data(pool_metadata, time_sequence, data_source):
+    sim_assets = get_pool_assets(pool_metadata)
+    time_sequence = time_sequence or _make_default_time_sequence()
+    asset_data = get_price_data(
+        sim_assets, time_sequence, data_source=data_source
+    )  # allow DataSource to be passed
+    return asset_data, time_sequence
+
+
+def _make_default_time_sequence():
+    t_end = datetime.now(timezone.utc) - timedelta(days=1)
+    t_end = t_end.replace(hour=23, minute=0, second=0, microsecond=0)
+    t_start = t_end - timedelta(days=60) + timedelta(hours=1)
+    time_sequence = DateTimeSequence.from_range(start=t_start, end=t_end, freq="1h")
+    return time_sequence

curvesim/pipelines/common/get_pool_data.py

@@ -0,0 +1,44 @@
+from datetime import datetime
+
+from curvesim.exceptions import CurvesimTypeError
+from curvesim.pool import get_sim_pool
+from curvesim.pool_data import get_metadata
+from curvesim.pool_data.metadata import PoolMetaDataInterface
+
+
+def get_pool_data(metadata_or_address, chain, env, pool_ts):
+    pool_metadata = _parse_metadata_or_address(metadata_or_address, chain)
+    end_ts = _parse_timestamp(pool_ts)
+    pool = get_sim_pool(pool_metadata, env=env, end_ts=end_ts)
+
+    return pool, pool_metadata
+
+
+def _parse_timestamp(timestamp) -> int:
+    if not timestamp:
+        return timestamp
+
+    if isinstance(timestamp, datetime):
+        timestamp = int(timestamp.timestamp())
+
+    if not isinstance(timestamp, int):
+        _type = type(timestamp).__name__
+        raise CurvesimTypeError(f"'Pool_ts' must be 'int' or 'timestamp', not {_type}.")
+
+    return timestamp
+
+
+def _parse_metadata_or_address(metadata_or_address, chain):
+    if isinstance(metadata_or_address, str):
+        pool_metadata: PoolMetaDataInterface = get_metadata(metadata_or_address, chain)
+
+    elif isinstance(metadata_or_address, PoolMetaDataInterface):
+        pool_metadata = metadata_or_address
+
+    else:
+        _type = type(metadata_or_address).__name__
+        raise CurvesimTypeError(
+            f"'Metadata_or_address' must be 'PoolMetaDataInterface' or 'str', not {_type}."
+        )
+
+    return pool_metadata

curvesim/pipelines/simple/__init__.py

@@ -11,21 +11,19 @@
 from curvesim.metrics.results import make_results
 from curvesim.pipelines import run_pipeline
 from curvesim.pipelines.simple.strategy import SimpleStrategy
-from curvesim.pool import get_sim_pool
 
-from ..common import DEFAULT_METRICS
+from ..common import DEFAULT_METRICS, get_asset_data, get_pool_data
 
 
 def pipeline(  # pylint: disable=too-many-locals
-    pool_address,
-    chain,
+    metadata_or_address,
     *,
+    chain="mainnet",
     variable_params=None,
     fixed_params=None,
-    end_ts=None,
-    days=60,
     src="coingecko",
-    data_dir="data",
+    time_sequence=None,
+    pool_ts=None,
     ncpu=None,
     env="prod",
 ):
@@ -84,15 +82,12 @@ def pipeline(  # pylint: disable=too-many-locals
     """
     ncpu = ncpu or os.cpu_count()
 
-    pool = get_sim_pool(pool_address, chain, env=env, end_ts=end_ts)
-
-    sim_assets = pool.assets
-    price_sampler = PriceVolume(
-        sim_assets, days=days, end=end_ts, data_dir=data_dir, src=src
-    )
+    pool, pool_metadata = get_pool_data(metadata_or_address, chain, env, pool_ts)
+    asset_data, _ = get_asset_data(pool_metadata, time_sequence, src)
 
     # pylint: disable-next=abstract-class-instantiated
     param_sampler = ParameterizedPoolIterator(pool, variable_params, fixed_params)
+    price_sampler = PriceVolume(asset_data)
 
     _metrics = init_metrics(DEFAULT_METRICS, pool=pool)
     strategy = SimpleStrategy(_metrics)

curvesim/pipelines/simple/__main__.py

@@ -3,4 +3,4 @@
 if __name__ == "__main__":
     pool_address = "0xbebc44782c7db0a1a60cb6fe97d0b483032ff1c7"
     chain = "mainnet"
-    results = pipeline(pool_address, chain, ncpu=1)
+    results = pipeline(pool_address, chain=chain, ncpu=1)

curvesim/pipelines/vol_limited_arb/__init__.py

@@ -8,29 +8,29 @@
 from curvesim.iterators.price_samplers import PriceVolume
 from curvesim.logging import get_logger
 from curvesim.metrics import init_metrics, make_results
-from curvesim.pool import get_sim_pool
 from curvesim.pool_data import get_pool_volume
 
 from .. import run_pipeline
-from ..common import DEFAULT_METRICS
+from ..common import DEFAULT_METRICS, get_asset_data, get_pool_data
 from .strategy import VolumeLimitedStrategy
 
 logger = get_logger(__name__)
 
 
 # pylint: disable-next=too-many-locals
 def pipeline(
-    pool_metadata,
+    metadata_or_address,
     *,
+    chain="mainnet",
     variable_params=None,
     fixed_params=None,
     metrics=None,
-    days=60,
     src="coingecko",
-    data_dir="data",
+    time_sequence=None,
     vol_mult=None,
+    pool_ts=None,
     ncpu=None,
-    end=None,
+    env="prod",
 ):
     """
     Implements the volume-limited arbitrage pipeline.
@@ -91,16 +91,17 @@ def pipeline(
         cpu_count = os.cpu_count()
         ncpu = cpu_count if cpu_count is not None else 1
 
-    pool = get_sim_pool(pool_metadata)
+    pool, pool_metadata = get_pool_data(metadata_or_address, chain, env, pool_ts)
+    asset_data, time_sequence = get_asset_data(pool_metadata, time_sequence, src)
 
     # pylint: disable-next=abstract-class-instantiated
     param_sampler = ParameterizedPoolIterator(pool, variable_params, fixed_params)
-    price_sampler = PriceVolume(
-        pool.assets, days=days, data_dir=data_dir, src=src, end=end
-    )
+    price_sampler = PriceVolume(asset_data)
 
     if vol_mult is None:
-        pool_volume = get_pool_volume(pool_metadata, days=days, end=end)
+        pool_volume = get_pool_volume(
+            pool_metadata, time_sequence[0], time_sequence[-1]
+        )
         vol_mult = pool_volume.sum() / price_sampler.volumes.sum()
         logger.info("Volume Multipliers:\n%s", vol_mult.to_string())
         vol_mult = vol_mult.to_dict()

curvesim/pool_data/queries/pool_assets.py

@@ -3,7 +3,7 @@
 from curvesim.constants import Chain
 from curvesim.pool_data.metadata import PoolMetaDataInterface
 from curvesim.pool_data.queries.metadata import get_metadata
-from curvesim.templates.sim_asset import OnChainAsset, OnChainAssetPair
+from curvesim.templates import OnChainAsset, OnChainAssetPair
 from curvesim.utils import get_pairs