sdk-seshat-python 0.3.16__tar.gz → 0.4.2__tar.gz

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sdk-seshat-python
-Version: 0.3.16
+Version: 0.4.2
 Summary: Seshat python SDK is a library to help create ML data pipelines.
 License: Commercial - see LICENSE.txt
 Author: SeshatLabs
@@ -25,7 +25,7 @@ Requires-Dist: loguru (>=0.7.3,<0.8.0)
 Requires-Dist: memory-profiler (>=0.61.0,<0.62.0)
 Requires-Dist: openai (>=1.73.0,<2.0.0)
 Requires-Dist: pandas (>=2.2.1,<3.0.0)
-Requires-Dist: psycopg2 (>=2.9,<3.0) ; extra == "postgres-support"
+Requires-Dist: psycopg2-binary (>=2.9,<3.0) ; extra == "postgres-support"
 Requires-Dist: pyarmor (>=8.5.1,<9.0.0)
 Requires-Dist: pydantic (>=2.7.4,<3.0.0)
 Requires-Dist: pyspark (>=3.5.1,<4.0.0)
@@ -33,6 +33,7 @@ Requires-Dist: python-logstash-async (>=4.0.2,<5.0.0)
 Requires-Dist: requests (==2.32.0)
 Requires-Dist: rich (>=13.9.4,<14.0.0)
 Requires-Dist: scikit-learn (>=1.4.1.post1,<2.0.0)
+Requires-Dist: setuptools (>=80.9.0,<81.0.0)
 Requires-Dist: sqlalchemy (>=2.0.29,<3.0.0)
 Requires-Dist: toml (>=0.10.2,<0.11.0)
 Requires-Dist: typer (>=0.12.3,<0.13.0)

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "sdk-seshat-python"
-version = "0.3.16"
+version = "0.4.2"
 description = "Seshat python SDK is a library to help create ML data pipelines."
 authors =  ["SeshatLabs <info@seshatlabs.xyz>"]
 packages = [{ include = "seshat", from = "." }]
@@ -32,13 +32,14 @@ langchain = "^0.3.23"
 langchain-community = "^0.3.21"
 langchain-openai = "^0.3.12"
 pyarmor = "^8.5.1"
-croniter = "^6.0.0"
-psycopg2 = { version = "^2.9", optional = true }
 python-logstash-async = "^4.0.2"
+croniter = "^6.0.0"
+psycopg2-binary = { version = "^2.9", optional = true }
+setuptools = "^80.9.0"
 
 [tool.poetry.extras]
 flipside_support = ["flipside"]
-postgres_support = ["psycopg2"]
+postgres_support = ["psycopg2-binary"]
 
 [tool.poetry.group.dev.dependencies]
 flake8 = "^7.0.0"

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/__init__.py

@@ -20,6 +20,7 @@ from seshat.general.exceptions import NoConfigSetError, RestClientException
 app = typer.Typer()
 console = Console()
 DEFAULT_DATA_SIZE = 1_000_000  # 1 GB
+DEFAULT_EXPIRATION = 86400
 
 state = {"verbose": False}
 
@@ -141,7 +142,12 @@ def submit_job(
         identifier = manager.store_code(package)
 
         job_response = manager.submit_job(
-            identifier, name, version, executor_image_tag, job_metadata
+            identifier,
+            name,
+            version,
+            executor_image_tag,
+            job_metadata,
+            expiration=config.get("aws", {}).get("expiration", DEFAULT_EXPIRATION),
         )
         job_response_data = job_response.get("data", {})
 

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/data_class/pandas.py

@@ -1,10 +1,9 @@
 from __future__ import annotations
 
-from typing import Iterable, List, Dict
+from typing import Dict, Iterable, List
 
 import pandas as pd
 from pandas import DataFrame
-from pyspark.sql import SparkSession
 
 from seshat.data_class import SFrame
 from seshat.data_class.base import GroupSFrame
@@ -33,7 +32,7 @@ class DFrame(SFrame):
     def to_spf(self) -> SFrame:
         from seshat.data_class import SPFrame
 
-        spark = SparkSession.builder.appName(configs.SPARK_APP_NAME).getOrCreate()
+        spark = SPFrame.get_spark()
         return SPFrame.from_raw(spark.createDataFrame(self.data))
 
     def extend_vertically(self, other: DataFrame):

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/data_class/pyspark.py

@@ -1,9 +1,10 @@
-from typing import Iterable, List, Dict
+from typing import Dict, Iterable, List
 
 from pandas import DataFrame
-from pyspark.sql import DataFrame as PySparkDataFrame, SparkSession
+from pyspark.sql import DataFrame as PySparkDataFrame
+from pyspark.sql import SparkSession
 
-from seshat.data_class import SFrame, DFrame
+from seshat.data_class import DFrame, SFrame
 from seshat.data_class.base import GroupSFrame
 from seshat.general import configs
 
@@ -56,6 +57,8 @@ class SPFrame(SFrame):
     def from_raw(cls, data, *args, **kwargs) -> "SPFrame":
         if isinstance(data, DataFrame):
             data = DFrame.from_raw(data).convert(cls).to_raw()
+        elif not isinstance(data, PySparkDataFrame) and data:
+            data = cls.get_spark().createDataFrame(data or [])
         return cls(data)
 
     @staticmethod

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/feature_view/base.py

@@ -1,5 +1,5 @@
 import logging
-from typing import Dict, Callable
+from typing import Callable, Dict, Optional
 
 from seshat.data_class import SFrame
 from seshat.evaluation.base import Evaluation
@@ -7,6 +7,7 @@ from seshat.profiler import ProfileConfig
 from seshat.profiler.base import profiler
 from seshat.source import Source
 from seshat.source.saver import Saver
+from seshat.transformer.base import Transformer
 from seshat.transformer.pipeline import Pipeline
 from seshat.transformer.splitter import Splitter
 
@@ -36,7 +37,8 @@ class FeatureView:
     saver : Saver, optional
         An optional component responsible for saving the processed data during training
         Required only in offline mode.
-
+    on_save_finished : Transformer, Optional
+        A Transformer to be called after the save operation completes.
     Examples
     --------
     Define feature view:
@@ -81,6 +83,7 @@ class FeatureView:
     saver: Saver = None
     profile_config = ProfileConfig(logging.INFO, default_tracking=True)
     evaluation: Evaluation
+    on_save_finished: Optional[Transformer] = None
 
     def __call__(self, *args, **kwargs):
         source = self._get_source()
@@ -143,6 +146,8 @@ class FeatureView:
                 if hasattr(self, "splitter")
                 else self.saver(self.data)
             )
+        if self.on_save_finished is not None:
+            self.on_save_finished(self.data)
 
     def _split(self, *args, **kwargs):
         self.split_data = self.splitter(self.data, *args, **kwargs)

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/general/command/submit_to_network.py

@@ -313,6 +313,7 @@ class SubmitCommand(BaseTyperCommand):
         version: str,
         executor_image_tag: str,
         metadata: JobMetadata,
+        expiration=86400,
     ) -> dict:
         """Submit job to API after successful upload"""
         if not self.job_config.base_url or not self.job_config.auth_token:
@@ -320,7 +321,7 @@ class SubmitCommand(BaseTyperCommand):
                 "API configuration missing. Please set base_url and auth_token"
             )
 
-        presigned_url = self.backend.generate_presigned_url(s3_key)
+        presigned_url = self.backend.generate_presigned_url(s3_key, expiration)
         executor_label = self.config.get("executor", {}).get("label")
 
         payload = {

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/source/local/base.py

@@ -5,24 +5,29 @@ from seshat.source import Source
 
 class LocalSource(Source):
     """
-    LocalSource is a source that can read data from csv file.
+    LocalSource is a data source that can read from a local file or an in-memory source.
     """
 
     def __init__(
         self,
-        path,
+        data_source,
         query=None,
         schema=None,
         mode=configs.DEFAULT_MODE,
     ):
         super().__init__(query, schema, mode)
-        self.path = path
+        self.data_source = data_source
 
     def convert_data_type(self, data) -> SFrame:
         return self.data_class.from_raw(data)
 
     def fetch(self) -> SFrame:
-        d = self.data_class.read_csv(path=self.path)
+        d = (
+            self.data_class.read_csv(path=self.data_source)
+            if isinstance(self.data_source, str)
+            else self.data_source
+        )
+
         return self.convert_data_type(d)
 
     def calculate_complexity(self):

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/source/saver/database.py

@@ -35,7 +35,13 @@ class SQLDBSaver(SQLMixin, Saver):
             self.create_index(config)
 
             selected_sf = sf.get(config.sf_key)
-            selected_sf = self.drop_nan_ids(selected_sf, config.schema)
+            has_id = False
+            for col in config.schema.cols:
+                if col.is_id:
+                    has_id = True
+                break
+            if has_id:
+                selected_sf = self.drop_nan_ids(selected_sf, config.schema)
 
             if config.clear_table:
                 self.delete(config.table)
@@ -65,6 +71,11 @@ class SQLDBSaver(SQLMixin, Saver):
         table, _ = self.get_table(table_name, autoload=True)
         self.write_on_db(table.delete())
 
+    def drop_table(self, table_name):
+        if table_name in inspect(self.get_engine()).get_table_names():
+            table, _ = self.get_table(table_name, autoload=True)
+            table.drop(self.get_engine())
+
     def insert(self, selected_sf: SFrame, config: SaveConfig):
         values = self.prepare_sf_to_insert(selected_sf, config).to_dict()
         table, _ = self.get_table(config.table, autoload=True)

sdk_seshat_python-0.4.2/seshat/transformer/deriver/__init__.py

@@ -0,0 +1,25 @@
+from .base import (
+    FeatureForAddressDeriver,
+    InteractedSymbolsToSentenceDeriver,
+    OperationOnColsDeriver,
+    PercentileTransactionValueDeriver,
+    SFrameFromColsDeriver,
+    StaticValueColumnAdder,
+    TimeWindowTransformer,
+    FractionDeriver,
+    ProfitLossDeriver,
+    DateTimeTypeDeriver,
+    ComprehensiveFeaturesDeriver,
+    GroupByDeriverCount,
+    GroupByDeriverMeanMax,
+    ChangingOverTimeDeriver,
+    TokenLastPriceDeriver,
+    GroupByTimeWindowDeriver,
+    OneColumnPercentileFilterDeriver,
+    SenderReceiverTokensDeriver,
+    TokenPriceDeriver,
+    TokenSwapTradeDeriver,
+    TokenFeatureTransformationDeriver,
+)
+
+from .from_database import FromSQLDBDeriver

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/transformer/deriver/base.py

@@ -1,12 +1,13 @@
 from datetime import timedelta
-from typing import Dict, Callable, List, Tuple, Any
+from typing import Any, Callable, Dict, List, Literal, Tuple
 
 import pandas as pd
 from pandas import DataFrame
-from pyspark.sql import DataFrame as PySparkDataFrame, Window
+from pyspark.sql import DataFrame as PySparkDataFrame
+from pyspark.sql import Window
 from pyspark.sql import functions as F
-from pyspark.sql.functions import array_distinct, array_union, coalesce, array, lit
-from pyspark.sql.types import IntegerType, StructType, StructField
+from pyspark.sql.functions import array, array_distinct, array_union, coalesce, lit
+from pyspark.sql.types import IntegerType, StructField, StructType
 
 from seshat.data_class import SFrame, SPFrame
 from seshat.data_class.base import GroupSFrame
@@ -15,9 +16,9 @@ from seshat.general import configs
 from seshat.general.exceptions import InvalidArgumentsError
 from seshat.transformer import Transformer
 from seshat.transformer.merger import Merger
-from seshat.transformer.schema import Schema, Col
-from seshat.utils import pandas_func
-from seshat.utils import pyspark_func
+from seshat.transformer.schema import Col, Schema
+from seshat.transformer.trimmer.base import FilterTrimmer
+from seshat.utils import pandas_func, pyspark_func
 from seshat.utils.validation import NumericColumnValidator, TimeStampColumnValidator
 
 SYMBOLS_RECEIVED_COL = "symbols_received"
@@ -1091,54 +1092,32 @@ class GroupByDeriverMeanMax(SFrameDeriver):
 
 class TimeWindowTransformer(SFrameDeriver):
     """
-    Applies transformations to data within specified time windows and merges the results.
+    A class that derives time series-based features from a default SFrame using a transformer and merger strategy.
+    It supports applying transformations over multiple durations and combines the results using a merger.
 
-    This transformer filters data based on time windows defined by durations, applies a
-    specified transformer to each filtered dataset, and merges the results using a provided merger.
-    It's particularly useful for generating time-based features or analyzing data across
-    different time periods.
+    Duration labeling can be handled in two ways:
+    1. By adding a new column with the name specified in `label_col`, where each row is labeled with its corresponding duration.
+    2. By appending the duration label as a suffix to the names of the new columns created by the transformer, excluding those required for merging.
+
+    This behavior is controlled by the `specify_duration_by` parameter.
 
     Parameters
     ----------
-    transformer : Transformer
-        The transformer to apply to each time window's data.
+    transformer : Transformer or Callable
+        An instance of a transformer or a callable function that performs the transformation on the input data.
     merger : Merger
-        The merger used to combine results from different time windows.
-        Will be set to work in-place.
+        An instance of a merger used to combine features generated across different durations.
     time_col : str, optional
-        The column name containing timestamp data.
-        Default is configs.BLOCK_TIMESTAMP_COL.
-    durations : List[Tuple[str, int]], optional
-        List of tuples where each tuple contains a label and a duration in minutes.
-        The label identifies the time window and the duration specifies how far back
-        from the latest timestamp to include data.
+        The name of the timestamp column in the SFrame. Defaults to `configs.BLOCK_TIMESTAMP_COL`.
+    durations : List[Tuple[str, int]]
+        A list of (label, duration) pairs where each label identifies a time window (in minutes) for feature extraction.
     label_col : str, optional
-        The column name to store the time window label.
-        Default is configs.DURATION_LABEL_COL.
-    group_keys : dict, optional
-        Keys used to identify and retrieve data from a grouped SFrame.
-
-    Attributes
-    ----------
-    time_col : str
-        The column name containing timestamp data.
-    transformer : Transformer
-        The transformer applied to each time window's data.
-    label_col : str
-        The column name storing the time window label.
-    merger : Merger
-        The merger used to combine results from different time windows.
-    periods : List[Tuple[str, int]]
-        The list of time windows with their labels and durations.
-
-    Notes
-    -----
-    The transformer processes data by:
-    1. Finding the latest timestamp in the data
-    2. Creating time windows based on the specified durations
-    3. Filtering data for each time window
-    4. Applying the transformer to each filtered dataset
-    5. Merging the results using the provided merger
+        The name of the column that will store the duration label for each record. Defaults to `configs.DURATION_LABEL_COL`.
+    specify_duration_by : {'column', 'suffix'}, default 'column'
+        Specifies how duration information is represented in the output:
+        * 'column' – Adds a new column (named by `label_col`) containing the duration label.
+        * 'suffix' – Appends the duration label as a suffix to the names of columns created by the transformer,
+          excluding those required by the merger.
     """
 
     DEFAULT_GROUP_KEYS = {
@@ -1148,11 +1127,12 @@ class TimeWindowTransformer(SFrameDeriver):
 
     def __init__(
         self,
-        transformer: Transformer,
+        transformer: Transformer | Callable,
         merger: Merger,
+        durations: List[Tuple[str, int]],
         time_col=configs.BLOCK_TIMESTAMP_COL,
-        durations: List[Tuple[str, int]] = None,
         label_col=configs.DURATION_LABEL_COL,
+        specify_duration_by: Literal["column", "suffix"] = "column",
         group_keys=None,
     ):
         super().__init__(group_keys)
@@ -1160,191 +1140,89 @@ class TimeWindowTransformer(SFrameDeriver):
         self.transformer = transformer
         self.label_col = label_col
         self.merger = merger
-        self.periods = durations
+        self.specify_duration_by = specify_duration_by
+        self.durations = durations
 
-        # Ensure that merger work inplace
+        # Ensure that the merge works in place and group keys are set to their default values.
         self.merger.inplace = True
+        self.merger.group_keys = {"default": "default", "other": "other"}
+        if self.specify_duration_by == "column":
+            self.merger.axis = 0
 
     def validate(self, sf: SFrame):
-        """
-        Validate that the input SFrame contains the required columns.
-
-        Parameters
-        ----------
-        sf : SFrame
-            The input SFrame to validate.
-
-        Raises
-        ------
-        ColDoesNotExistError
-            If the required time column is not present in the SFrame.
-        """
         super().validate(sf)
         self._validate_columns(sf, self.group_keys["default"], self.time_col)
 
-    def calculate_metrics_df(self, default: DataFrame, label, start) -> SFrame:
-        """
-        Calculate metrics for a specific time window.
+    def _is_empty(self, sf: SFrame) -> bool:
+        raw = sf.to_raw()
+        if raw is None:
+            return True
+        elif hasattr(raw, "rdd"):
+            return raw.rdd.isEmpty()
+        else:
+            return len(raw) == 0
 
-        Parameters
-        ----------
-        default : DataFrame
-            The input DataFrame containing the data.
-        label : str
-            The label for the time window.
-        start : datetime
-            The start timestamp for the time window.
+    def calculate_metrics(self, sf: SFrame, label, start) -> SFrame:
+        trimmer = FilterTrimmer(op=">=", col=self.time_col, value=start)
+        filtered = trimmer(sf)
 
-        Returns
-        -------
-        SFrame
-            An SFrame containing the calculated metrics for the time window.
-            Returns an empty DFrame if no data is available for the time window.
-        """
-        filtered = default[default[self.time_col] >= start]
+        if self._is_empty(filtered):
+            return filtered
 
-        if len(filtered) == 0:
-            return DFrame(DataFrame())
+        result = self.transformer(sf_input=filtered, label=label, start=start)
+        if self.specify_duration_by == "column":
+            deriver = StaticValueColumnAdder(col_name=self.label_col, value=label)
+            result = deriver(result)
 
-        filtered.loc[:, self.label_col] = label
-        return self.transformer(DFrame(filtered), label=label, start=start)
+        else:
+            # Find columns excluding those used by,
+            # the merger for merging.
+            cols = set(result[self.transformer].get_columns()) - set(self.merger.on)
+            schema = Schema(
+                cols=[Col(col, to=f"{col}_{label}") for col in cols],
+                exclusive=False,
+            )
+            result = schema(result)
+        return result
 
     def derive_df(self, default: DataFrame, *args, **kwargs):
-        """
-        Derive time-based features from the input DataFrame.
-
-        This method:
-        1. Validates that the time column contains timestamp data
-        2. Finds the latest timestamp in the data
-        3. Creates time windows based on the specified durations
-        4. Calculates metrics for each time window
-        5. Merges the results using the provided merger
-
-        Parameters
-        ----------
-        default : DataFrame
-            The input DataFrame containing the data.
-        *args : tuple
-            Variable length argument list.
-        **kwargs : dict
-            Arbitrary keyword arguments.
+        default = TimeStampColumnValidator().validate(default, self.time_col)
+        last_timestamp = default[self.time_col].max()
+        sf_input = DFrame.from_raw(default)
+        result = DFrame()
+        return self.derive(sf_input, result, last_timestamp)
 
-        Returns
-        -------
-        dict
-            A dictionary with two keys:
-            - 'default': The original input DataFrame
-            - 'timeseries_feature': The derived time-based features
-        """
+    def derive_spf(self, default: PySparkDataFrame, *args, **kwargs):
         default = TimeStampColumnValidator().validate(default, self.time_col)
+        last_timestamp = default.select(F.max(self.time_col)).collect()[0][0]
+        sf_input = SPFrame.from_raw(default)
+        result = SPFrame()
+        return self.derive(sf_input, result, last_timestamp)
 
-        last_timestamp = default[self.time_col].max()
+    def derive(self, sf_input: SFrame, result: SFrame, last_timestamp):
         intervals = {
             label: last_timestamp - timedelta(minutes=minutes)
-            for label, minutes in self.periods
+            for label, minutes in self.durations
         }
 
-        result = DFrame(DataFrame())
-        for label, start in intervals.items():
-            metrics = self.calculate_metrics_df(default, label, start)
-            if metrics.data.empty:
+        for idx, (label, start) in enumerate(intervals.items()):
+            metrics = self.calculate_metrics(sf_input, label, start)
+            if self._is_empty(metrics):
                 continue
             metrics = metrics.make_group()
-            metrics[self.merger.group_keys["other"]] = result
+            metrics["other"] = metrics["default"]
+            metrics["default"] = result
             result = (
-                metrics["default"]
-                if result.data.empty
+                metrics["other"]
+                if self._is_empty(result)
                 else self.merger(metrics)["default"]
             )
 
-        return {"default": default, "timeseries_feature": result.to_raw()}
-
-    def calculate_metrics_spf(
-        self, default: PySparkDataFrame, label, start
-    ) -> PySparkDataFrame:
-        """
-        Calculate metrics for a specific time window using PySpark DataFrame.
-
-        Parameters
-        ----------
-        default : PySparkDataFrame
-            The input PySpark DataFrame containing the data.
-        label : str
-            The label for the time window.
-        start : datetime
-            The start timestamp for the time window.
-
-        Returns
-        -------
-        PySparkDataFrame
-            A PySpark DataFrame containing the calculated metrics for the time window.
-            Returns an empty DataFrame if no data is available for the time window.
-        """
-        filtered = default.filter(F.col(self.time_col) >= start)
-
-        if filtered.count() == 0:
-            return SPFrame.get_spark().createDataFrame([], default.schema)
-
-        filtered = filtered.withColumn(self.label_col, F.lit(label))
-        return self.transformer(SPFrame(filtered), label=label, start=start).data
-
-    def derive_spf(self, default: PySparkDataFrame, *args, **kwargs):
-        """
-        Derive time-based features from the input PySpark DataFrame.
-
-        This method:
-        1. Validates that the time column contains timestamp data
-        2. Finds the latest timestamp in the data
-        3. Creates time windows based on the specified durations
-        4. Calculates metrics for each time window
-        5. Merges the results using the provided merger
-
-        Parameters
-        ----------
-        default : PySparkDataFrame
-            The input PySpark DataFrame containing the data.
-        *args : tuple
-            Variable length argument list.
-        **kwargs : dict
-            Arbitrary keyword arguments.
-
-        Returns
-        -------
-        dict
-            A dictionary with two keys:
-            - 'default': The original input DataFrame
-            - 'timeseries_feature': The derived time-based features
-        """
-        default = TimeStampColumnValidator().validate(default, self.time_col)
-
-        last_timestamp_row = default.agg(F.max(self.time_col)).collect()[0][0]
-
-        intervals = {
-            label: last_timestamp_row - timedelta(minutes=minutes)
-            for label, minutes in self.periods
+        return {
+            "default": sf_input[self.default_sf_key].to_raw(),
+            "timeseries_feature": result.to_raw(),
         }
 
-        result = None
-
-        for label, start in intervals.items():
-            metrics = self.calculate_metrics_spf(default, label, start)
-
-            if metrics.count() == 0:
-                continue
-
-            metrics_group = {self.merger.group_keys["default"]: metrics}
-
-            if result is not None:
-                metrics_group[self.merger.group_keys["other"]] = result
-                result = self.merger(SPFrame(metrics_group))["default"]
-            else:
-                result = metrics
-
-        if result is not None:
-            return {"default": default, "timeseries_feature": result}
-        else:
-            return {"default": default}
-
 
 class StaticValueColumnAdder(SFrameDeriver):
     """
@@ -1714,9 +1592,11 @@ class ComprehensiveFeaturesDeriver(SFrameDeriver):
                     last_price=("TOKEN_PRICE", "last"),
                     transaction_count_hourly=(
                         "BLOCK_TIMESTAMP",
-                        lambda x: len(x) / ((x.max() - x.min()).total_seconds() / 3600)
-                        if (x.max() > x.min())
-                        else len(x),
+                        lambda x: (
+                            len(x) / ((x.max() - x.min()).total_seconds() / 3600)
+                            if (x.max() > x.min())
+                            else len(x)
+                        ),
                     ),
                 )
                 .reset_index()

{sdk_seshat_python-0.3.16 → sdk_seshat_python-0.4.2}/seshat/transformer/pseudo/__init__.py

@@ -1 +1,2 @@
+from .action_gate import ActionGate
 from .table_existence import SQLTableExistenceValidator