tp.begin and tp.end operators

docs/src/tutorials/m5_competition.py

@@ -319,8 +319,9 @@
 print("Export to csv file")
 print("==================")
 
-tabular_dataset_data.to_pandas().to_csv(
-    os.path.join(work_directory, "tabular_dataset.csv"), index=False
+tp.to_csv(
+    tabular_dataset_data,
+    path=os.path.join(work_directory, "tabular_dataset.csv"),
 )
 
 print("The artefacts are available in:", work_directory)

temporian/core/operators/BUILD

@@ -25,6 +25,8 @@ py_library(
         ":since_last",
         ":unary",
         ":unique_timestamps",
+        ":begin",
+        ":end",
         "//temporian/core/operators/binary",
         "//temporian/core/operators/calendar:day_of_month",
         "//temporian/core/operators/calendar:day_of_week",
@@ -250,3 +252,29 @@ py_library(
         "//temporian/proto:core_py_proto",
     ],
 )
+
+py_library(
+    name = "begin",
+    srcs = ["begin.py"],
+    srcs_version = "PY3",
+    deps = [
+        ":base",
+        "//temporian/core:operator_lib",
+        "//temporian/core/data:node",
+        "//temporian/core/data:schema",
+        "//temporian/proto:core_py_proto",
+    ],
+)
+
+py_library(
+    name = "end",
+    srcs = ["end.py"],
+    srcs_version = "PY3",
+    deps = [
+        ":base",
+        "//temporian/core:operator_lib",
+        "//temporian/core/data:node",
+        "//temporian/core/data:schema",
+        "//temporian/proto:core_py_proto",
+    ],
+)

temporian/core/operators/all_operators.py

@@ -72,3 +72,5 @@
 from temporian.core.operators.window.moving_max import moving_max
 from temporian.core.operators.unique_timestamps import unique_timestamps
 from temporian.core.operators.since_last import since_last
+from temporian.core.operators.begin import begin
+from temporian.core.operators.end import end

temporian/core/operators/begin.py

@@ -0,0 +1,70 @@
+# Copyright 2021 Google LLC.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+"""Begin operator class and public API function definitions."""
+
+from temporian.core import operator_lib
+from temporian.core.data.node import Node, create_node_new_features_new_sampling
+from temporian.core.operators.base import Operator
+from temporian.proto import core_pb2 as pb
+
+
+class Begin(Operator):
+    def __init__(self, input: Node):
+        super().__init__()
+
+        self.add_input("input", input)
+
+        self.add_output(
+            "output",
+            create_node_new_features_new_sampling(
+                features=[],
+                indexes=input.schema.indexes,
+                is_unix_timestamp=input.schema.is_unix_timestamp,
+                creator=self,
+            ),
+        )
+        self.check()
+
+    @classmethod
+    def build_op_definition(cls) -> pb.OperatorDef:
+        return pb.OperatorDef(
+            key="BEGIN",
+            attributes=[],
+            inputs=[pb.OperatorDef.Input(key="input")],
+            outputs=[pb.OperatorDef.Output(key="output")],
+        )
+
+
+operator_lib.register_operator(Begin)
+
+
+def begin(input: Node) -> Node:
+    """Generates a single timestamp at the beginning of the input.
+
+    Args:
+        input: Guide input
+
+    Example:
+        Input
+            timestamps: [1, 5, 10]
+        Output
+            timestamps: [1]
+
+    Returns:
+        A feature-less node with a single timestamp.
+    """
+
+    return Begin(input=input).outputs["output"]

temporian/core/operators/end.py

@@ -0,0 +1,70 @@
+# Copyright 2021 Google LLC.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+"""End operator class and public API function definitions."""
+
+from temporian.core import operator_lib
+from temporian.core.data.node import Node, create_node_new_features_new_sampling
+from temporian.core.operators.base import Operator
+from temporian.proto import core_pb2 as pb
+
+
+class End(Operator):
+    def __init__(self, input: Node):
+        super().__init__()
+
+        self.add_input("input", input)
+
+        self.add_output(
+            "output",
+            create_node_new_features_new_sampling(
+                features=[],
+                indexes=input.schema.indexes,
+                is_unix_timestamp=input.schema.is_unix_timestamp,
+                creator=self,
+            ),
+        )
+        self.check()
+
+    @classmethod
+    def build_op_definition(cls) -> pb.OperatorDef:
+        return pb.OperatorDef(
+            key="END",
+            attributes=[],
+            inputs=[pb.OperatorDef.Input(key="input")],
+            outputs=[pb.OperatorDef.Output(key="output")],
+        )
+
+
+operator_lib.register_operator(End)
+
+
+def end(input: Node) -> Node:
+    """Generates a single timestamp at the end of the input.
+
+    Args:
+        input: Guide input
+
+    Example:
+        Input
+            timestamps: [1, 5, 10]
+        Output
+            timestamps: [10]
+
+    Returns:
+        A feature-less node with a single timestamp.
+    """
+
+    return End(input=input).outputs["output"]

temporian/core/test/registered_operators_test.py

@@ -29,6 +29,7 @@ def test_base(self):
             "ADDITION_SCALAR",
             "ADD_INDEX",
             "AND",
+            "BEGIN",
             "CALENDAR_DAY_OF_MONTH",
             "CALENDAR_DAY_OF_WEEK",
             "CALENDAR_DAY_OF_YEAR",
@@ -42,6 +43,7 @@ def test_base(self):
             "DIVISION",
             "DIVISION_SCALAR",
             "DROP_INDEX",
+            "END",
             "EQUAL",
             "EQUAL_SCALAR",
             "FILTER",

temporian/implementation/numpy/data/plotter.py

@@ -227,6 +227,9 @@ def get_num_plots(
 def auto_style(uniform: bool, xs, ys) -> Style:
     """Finds the best plotting style."""
 
+    if len(ys) <= 1:
+        return Style.marker
+
     if len(ys) == 0:
         all_ys_are_equal = True
     else:

temporian/implementation/numpy/operators/BUILD

@@ -26,6 +26,8 @@ py_library(
         ":since_last",
         ":unary",
         ":unique_timestamps",
+        ":begin",
+        ":end",
         "//temporian/implementation/numpy/operators/binary:arithmetic",
         "//temporian/implementation/numpy/operators/binary:logical",
         "//temporian/implementation/numpy/operators/binary:relational",
@@ -250,3 +252,34 @@ py_library(
         "//temporian/implementation/numpy/data:event_set",
     ],
 )
+
+py_library(
+    name = "begin",
+    srcs = ["begin.py"],
+    srcs_version = "PY3",
+    deps = [
+        # already_there/numpy
+        ":base",
+        "//temporian/core/data:duration",
+        "//temporian/core/operators:begin",
+        "//temporian/implementation/numpy:implementation_lib",
+        "//temporian/implementation/numpy:utils",
+        "//temporian/implementation/numpy/data:event_set",
+    ],
+)
+
+
+py_library(
+    name = "end",
+    srcs = ["end.py"],
+    srcs_version = "PY3",
+    deps = [
+        # already_there/numpy
+        ":base",
+        "//temporian/core/data:duration",
+        "//temporian/core/operators:end",
+        "//temporian/implementation/numpy:implementation_lib",
+        "//temporian/implementation/numpy:utils",
+        "//temporian/implementation/numpy/data:event_set",
+    ],
+)

temporian/implementation/numpy/operators/all_operators.py

@@ -42,3 +42,5 @@
 from temporian.implementation.numpy.operators.calendar import second
 from temporian.implementation.numpy.operators import since_last
 from temporian.implementation.numpy.operators import unique_timestamps
+from temporian.implementation.numpy.operators import begin
+from temporian.implementation.numpy.operators import end

temporian/implementation/numpy/operators/begin.py

@@ -0,0 +1,59 @@
+# Copyright 2021 Google LLC.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+"""Implementation for the Begin operator."""
+
+
+from typing import Dict
+import numpy as np
+
+from temporian.implementation.numpy.data.event_set import IndexData, EventSet
+from temporian.core.operators.begin import Begin
+from temporian.implementation.numpy import implementation_lib
+from temporian.implementation.numpy.operators.base import OperatorImplementation
+
+
+class BeginNumpyImplementation(OperatorImplementation):
+    def __init__(self, operator: Begin) -> None:
+        assert isinstance(operator, Begin)
+        super().__init__(operator)
+
+    def __call__(self, input: EventSet) -> Dict[str, EventSet]:
+        assert isinstance(self.operator, Begin)
+        output_schema = self.output_schema("output")
+
+        # create output event set
+        output_evset = EventSet(data={}, schema=output_schema)
+
+        # fill output event set data
+        for index_key, index_data in input.data.items():
+            if len(index_data.timestamps) == 0:
+                dst_timestamps = np.array([], dtype=np.float64)
+            else:
+                dst_timestamps = np.array(
+                    [index_data.timestamps[0]], dtype=np.float64
+                )
+            output_evset[index_key] = IndexData(
+                [],
+                dst_timestamps,
+                schema=output_schema,
+            )
+
+        return {"output": output_evset}
+
+
+implementation_lib.register_operator_implementation(
+    Begin, BeginNumpyImplementation
+)

temporian/implementation/numpy/operators/end.py

@@ -0,0 +1,57 @@
+# Copyright 2021 Google LLC.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+"""Implementation for the End operator."""
+
+
+from typing import Dict
+import numpy as np
+
+from temporian.implementation.numpy.data.event_set import IndexData, EventSet
+from temporian.core.operators.end import End
+from temporian.implementation.numpy import implementation_lib
+from temporian.implementation.numpy.operators.base import OperatorImplementation
+
+
+class EndNumpyImplementation(OperatorImplementation):
+    def __init__(self, operator: End) -> None:
+        assert isinstance(operator, End)
+        super().__init__(operator)
+
+    def __call__(self, input: EventSet) -> Dict[str, EventSet]:
+        assert isinstance(self.operator, End)
+        output_schema = self.output_schema("output")
+
+        # create output event set
+        output_evset = EventSet(data={}, schema=output_schema)
+
+        # fill output event set data
+        for index_key, index_data in input.data.items():
+            if len(index_data.timestamps) == 0:
+                dst_timestamps = np.array([], dtype=np.float64)
+            else:
+                dst_timestamps = np.array(
+                    [index_data.timestamps[-1]], dtype=np.float64
+                )
+            output_evset[index_key] = IndexData(
+                [],
+                dst_timestamps,
+                schema=output_schema,
+            )
+
+        return {"output": output_evset}
+
+
+implementation_lib.register_operator_implementation(End, EndNumpyImplementation)