MOD: Standardize Python client API with backend

Author: cjdsellers

databento/historical/api/batch.py

@@ -115,10 +115,10 @@ def submit_job(
 
         params: List[Tuple[str, str]] = BentoHttpAPI._timeseries_params(
             dataset=dataset,
-            symbols=symbols,
-            schema=Schema(schema),
             start=start,
             end=end,
+            symbols=symbols,
+            schema=Schema(schema),
             limit=limit,
             stype_in=SType(stype_in),
             stype_out=SType(stype_out),

databento/historical/api/metadata.py

@@ -244,10 +244,10 @@ def list_unit_prices(
     def get_shape(
         self,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         symbols: Optional[Union[List[str], str]] = None,
         schema: Union[Schema, str] = "trades",
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         encoding: Union[Encoding, str] = "dbz",
         stype_in: Optional[Union[SType, str]] = "native",
         limit: Optional[int] = None,
@@ -261,17 +261,17 @@ def get_shape(
         ----------
         dataset : Dataset or str
             The dataset code for the request.
+        start : pd.Timestamp or date or str or int
+            The start datetime for the request range. Assumes UTC as timezone unless otherwise specified.
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
+        end : pd.Timestamp or date or str or int
+            The end datetime for the request range. Assumes UTC as timezone unless otherwise specified.
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         symbols : List[Union[str, int]] or str, optional
             The product symbols to filter for. Takes up to 2,000 symbols per request.
             If `*` or ``None`` then will be for **all** symbols.
         schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
             The data record schema for the request.
-        start : pd.Timestamp or date or str or int, optional
-            The start datetime for the request range. Assumes UTC as timezone unless otherwise specified.
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
-        end : pd.Timestamp or date or str or int, optional
-            The end datetime for the request range. Assumes UTC as timezone unless otherwise specified.
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         encoding : Encoding or str {'dbz', 'csv', 'json'}, optional
             The data encoding.
         stype_in : SType or str, default 'native'
@@ -313,10 +313,10 @@ def get_shape(
     def get_billable_size(
         self,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         symbols: Optional[Union[List[str], str]] = None,
         schema: Union[Schema, str] = "trades",
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         stype_in: Optional[Union[SType, str]] = "native",
         limit: Optional[int] = None,
     ) -> int:
@@ -330,17 +330,17 @@ def get_billable_size(
         ----------
         dataset : Dataset or str
             The dataset code for the request.
-        symbols : List[Union[str, int]] or str, optional
-            The product symbols to filter for. Takes up to 2,000 symbols per request.
-            If `*` or ``None`` then will be for **all** symbols.
-        schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
-            The data record schema for the request.
         start : pd.Timestamp or date or str or int, optional
             The start datetime for the request range. Assumes UTC as timezone unless otherwise specified.
             If an integer is passed, then this represents nanoseconds since UNIX epoch.
         end : pd.Timestamp or date or str or int, optional
             The end datetime for the request range. Assumes UTC as timezone unless otherwise specified.
             If an integer is passed, then this represents nanoseconds since UNIX epoch.
+        symbols : List[Union[str, int]] or str, optional
+            The product symbols to filter for. Takes up to 2,000 symbols per request.
+            If `*` or ``None`` then will be for **all** symbols.
+        schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
+            The data record schema for the request.
         stype_in : SType or str, default 'native'
             The input symbology type to resolve from.
         limit : int, optional
@@ -357,10 +357,10 @@ def get_billable_size(
 
         params: List[Tuple[str, str]] = super()._timeseries_params(
             dataset=dataset,
-            symbols=symbols,
-            schema=Schema(schema),
             start=start,
             end=end,
+            symbols=symbols,
+            schema=Schema(schema),
             stype_in=SType(stype_in),
             limit=limit,
         )
@@ -376,11 +376,11 @@ def get_billable_size(
     def get_cost(
         self,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         mode: Union[FeedMode, str] = "historical-streaming",
         symbols: Optional[Union[List[str], str]] = None,
         schema: Union[Schema, str] = "trades",
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         stype_in: Optional[Union[SType, str]] = "native",
         limit: Optional[int] = None,
     ) -> float:
@@ -394,19 +394,19 @@ def get_cost(
         ----------
         dataset : Dataset or str
             The dataset code for the request.
+        start : pd.Timestamp or date or str or int
+            The start datetime for the request range. Assumes UTC as timezone unless otherwise specified.
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
+        end : pd.Timestamp or date or str or int
+            The end datetime for the request range. Assumes UTC as timezone unless otherwise specified.
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         mode : FeedMode or str {'live', 'historical-streaming', 'historical'}, default 'historical-streaming'
             The data feed mode for the request.
         symbols : List[Union[str, int]] or str, optional
             The product symbols to filter for. Takes up to 2,000 symbols per request.
             If `*` or ``None`` then will be for **all** symbols.
         schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
             The data record schema for the request.
-        start : pd.Timestamp or date or str or int, optional
-            The start datetime for the request range. Assumes UTC as timezone unless otherwise specified.
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
-        end : pd.Timestamp or date or str or int, optional
-            The end datetime for the request range. Assumes UTC as timezone unless otherwise specified.
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         stype_in : SType or str, default 'native'
             The input symbology type to resolve from.
         limit : int, optional
@@ -424,10 +424,10 @@ def get_cost(
 
         params: List[Tuple[str, str]] = super()._timeseries_params(
             dataset=dataset,
-            symbols=symbols,
-            schema=Schema(schema),
             start=start,
             end=end,
+            symbols=symbols,
+            schema=Schema(schema),
             stype_in=SType(stype_in),
             limit=limit,
         )

databento/historical/api/timeseries.py

@@ -22,10 +22,10 @@ def __init__(self, key: str, gateway: str) -> None:
     def stream(
         self,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         symbols: Optional[Union[List[str], str]] = None,
         schema: Union[Schema, str] = "trades",
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         stype_in: Union[SType, str] = "native",
         stype_out: Union[SType, str] = "product_id",
         limit: Optional[int] = None,
@@ -40,18 +40,18 @@ def stream(
         ----------
         dataset : Dataset or str
             The dataset code (string identifier) for the request.
+        start : pd.Timestamp or date or str or int
+            The start datetime (UTC) of the request time range (inclusive).
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
+        end : pd.Timestamp or date or str or int
+            The end datetime (UTC) of the request time range (exclusive).
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         symbols : List[Union[str, int]] or str, optional
             The product symbols to filter for. Takes up to 2,000 symbols per request.
             If more than 1 symbol is specified, the data is merged and sorted by time.
             If `*` or ``None`` then will be for **all** symbols.
         schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
             The data record schema for the request.
-        start : pd.Timestamp or date or str or int, optional
-            The start datetime (UTC) of the request time range (inclusive).
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
-        end : pd.Timestamp or date or str or int, optional
-            The end datetime (UTC) of the request time range (exclusive).
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         stype_in : SType or str, default 'native'
             The input symbology type to resolve from.
         stype_out : SType or str, default 'product_id'
@@ -81,10 +81,10 @@ def stream(
 
         params: List[Tuple[str, str]] = BentoHttpAPI._timeseries_params(
             dataset=dataset,
-            symbols=symbols,
-            schema=Schema(schema),
             start=start,
             end=end,
+            symbols=symbols,
+            schema=Schema(schema),
             stype_in=SType(stype_in),
             stype_out=SType(stype_out),
             limit=limit,
@@ -114,10 +114,10 @@ def stream(
     async def stream_async(
         self,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         symbols: Optional[Union[List[str], str]] = None,
         schema: Union[Schema, str] = "trades",
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         stype_in: Union[SType, str] = "native",
         stype_out: Union[SType, str] = "product_id",
         limit: Optional[int] = None,
@@ -132,18 +132,18 @@ async def stream_async(
         ----------
         dataset : Dataset or str
             The dataset code (string identifier) for the request.
+        start : pd.Timestamp or date or str or int
+            The start datetime (UTC) of the request time range (inclusive).
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
+        end : pd.Timestamp or date or str or int
+            The end datetime (UTC) of the request time range (exclusive).
+            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         symbols : List[Union[str, int]] or str, optional
             The product symbols to filter for. Takes up to 2,000 symbols per request.
             If more than 1 symbol is specified, the data is merged and sorted by time.
             If `*` or ``None`` then will be for **all** symbols.
         schema : Schema or str {'mbo', 'mbp-1', 'mbp-10', 'trades', 'tbbo', 'ohlcv-1s', 'ohlcv-1m', 'ohlcv-1h', 'ohlcv-1d', 'definition', 'statistics', 'status'}, default 'trades'  # noqa
             The data record schema for the request.
-        start : pd.Timestamp or date or str or int, optional
-            The start datetime (UTC) of the request time range (inclusive).
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
-        end : pd.Timestamp or date or str or int, optional
-            The end datetime (UTC) of the request time range (exclusive).
-            If an integer is passed, then this represents nanoseconds since UNIX epoch.
         stype_in : SType or str, default 'native'
             The input symbology type to resolve from.
         stype_out : SType or str, default 'product_id'
@@ -173,10 +173,10 @@ async def stream_async(
 
         params: List[Tuple[str, str]] = BentoHttpAPI._timeseries_params(
             dataset=dataset,
-            symbols=symbols,
-            schema=Schema(schema),
             start=start,
             end=end,
+            symbols=symbols,
+            schema=Schema(schema),
             stype_in=SType(stype_in),
             stype_out=SType(stype_out),
             limit=limit,

databento/historical/http.py

@@ -42,26 +42,26 @@ def __init__(self, key: str, gateway: str):
     def _timeseries_params(
         *,
         dataset: Union[Dataset, str],
+        start: Union[pd.Timestamp, date, str, int],
+        end: Union[pd.Timestamp, date, str, int],
         symbols: Optional[Union[List[str], str]] = None,
         schema: Schema,
-        start: Optional[Union[pd.Timestamp, date, str, int]] = None,
-        end: Optional[Union[pd.Timestamp, date, str, int]] = None,
         limit: Optional[int] = None,
         stype_in: SType,
         stype_out: SType = SType.PRODUCT_ID,
     ) -> List[Tuple[str, str]]:
         # Parse inputs
         dataset = enum_or_str_lowercase(dataset, "dataset")
-        symbols = maybe_symbols_list_to_string(symbols)
         start = maybe_datetime_to_string(start)
         end = maybe_datetime_to_string(end)
+        symbols = maybe_symbols_list_to_string(symbols)
 
         params: List[Tuple[str, Any]] = [
             ("dataset", dataset),
-            ("symbols", symbols),
-            ("schema", schema.value),
             ("start", start),
             ("end", end),
+            ("symbols", symbols),
+            ("schema", schema.value),
             ("stype_in", stype_in.value),
             ("stype_out", stype_out.value),
         ]

tests/test_historical_batch.py

@@ -84,10 +84,10 @@ def test_batch_submit_job_sends_expected_request(
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "trades"),
             ("start", "2020-12-28T12:00:00"),
             ("end", "2020-12-29T00:00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "trades"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("encoding", "csv"),

tests/test_historical_client.py

@@ -119,10 +119,10 @@ def test_request_full_definitions_expected_request(
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "definition"),
             ("start", "2020-12-28T13:00:00+00:00"),
             ("end", "2020-12-29T00:00:00+00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "definition"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("encoding", "dbz"),

tests/test_historical_metadata.py

@@ -279,10 +279,10 @@ def test_get_billable_size_sends_expected_request(
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "mbo"),
             ("start", "2020-12-28T12:00:00"),
             ("end", "2020-12-29T00:00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "mbo"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("limit", "1000000"),
@@ -318,10 +318,10 @@ def test_get_cost_sends_expected_request(self, mocker: MockerFixture) -> None:
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "mbo"),
             ("start", "2020-12-28T12:00:00"),
             ("end", "2020-12-29T00:00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "mbo"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("limit", "1000000"),

tests/test_historical_timeseries.py

@@ -73,10 +73,10 @@ def test_stream_sends_expected_request(self, mocker: MockerFixture) -> None:
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "trades"),
             ("start", "2020-12-28T12:00:00"),
             ("end", "2020-12-29T00:00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "trades"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("encoding", "dbz"),
@@ -114,10 +114,10 @@ def test_stream_with_limit_sends_expected_request(
         )
         assert call["params"] == [
             ("dataset", "glbx.mdp3"),
-            ("symbols", "ESH1"),
-            ("schema", "trades"),
             ("start", "2020-12-28T12:00:00"),
             ("end", "2020-12-29T00:00:00"),
+            ("symbols", "ESH1"),
+            ("schema", "trades"),
             ("stype_in", "native"),
             ("stype_out", "product_id"),
             ("limit", "1000000"),