Preserve inputs passed to depends_on and produces decorator. (#42)

Author: tobiasraabe

docs/changes.rst

@@ -21,6 +21,9 @@ all releases are available on `Anaconda.org <https://anaconda.org/pytask/pytask>
 - :gh:`39` releases v0.0.9.
 - :gh:`40` cleans up the capture manager and other parts of pytask.
 - :gh:`41` shortens the task ids in the error reports for better readability.
+- :gh:`42` ensures that lists with one element and dictionaries with only a zero key as
+  input for ``@pytask.mark.depends_on`` and ``@pytask.mark.produces`` are preserved as a
+  dictionary inside the function.
 
 
 0.0.8 - 2020-10-04

src/_pytask/nodes.py

@@ -7,8 +7,10 @@
 from abc import abstractmethod
 from pathlib import Path
 from typing import Any
+from typing import Dict
 from typing import Iterable
 from typing import List
+from typing import Tuple
 from typing import Union
 
 import attr
@@ -82,17 +84,22 @@ class PythonFunctionTask(MetaTask):
     """List[MetaNode]: A list of products of task."""
     markers = attr.ib(factory=list)
     """Optional[List[Mark]]: A list of markers attached to the task function."""
+    keep_dict = attr.ib(factory=dict)
     _report_sections = attr.ib(factory=list)
 
     @classmethod
     def from_path_name_function_session(cls, path, name, function, session):
         """Create a task from a path, name, function, and session."""
+        keep_dictionary = {}
+
         objects = _extract_nodes_from_function_markers(function, depends_on)
-        nodes = _convert_objects_to_node_dictionary(objects, "depends_on")
+        nodes, keep_dict = _convert_objects_to_node_dictionary(objects, "depends_on")
+        keep_dictionary["depends_on"] = keep_dict
         dependencies = _collect_nodes(session, path, name, nodes)
 
         objects = _extract_nodes_from_function_markers(function, produces)
-        nodes = _convert_objects_to_node_dictionary(objects, "produces")
+        nodes, keep_dict = _convert_objects_to_node_dictionary(objects, "produces")
+        keep_dictionary["produces"] = keep_dict
         products = _collect_nodes(session, path, name, nodes)
 
         markers = [
@@ -109,6 +116,7 @@ def from_path_name_function_session(cls, path, name, function, session):
             depends_on=dependencies,
             produces=products,
             markers=markers,
+            keep_dict=keep_dictionary,
         )
 
     def execute(self):
@@ -124,15 +132,15 @@ def _get_kwargs_from_task_for_function(self):
         """Process dependencies and products to pass them as kwargs to the function."""
         func_arg_names = set(inspect.signature(self.function).parameters)
         kwargs = {}
-        for name in ["depends_on", "produces"]:
-            if name in func_arg_names:
-                attribute = getattr(self, name)
-                kwargs[name] = (
+        for arg_name in ["depends_on", "produces"]:
+            if arg_name in func_arg_names:
+                attribute = getattr(self, arg_name)
+                kwargs[arg_name] = (
                     attribute[0].value
-                    if len(attribute) == 1 and 0 in attribute
-                    else {
-                        node_name: node.value for node_name, node in attribute.items()
-                    }
+                    if len(attribute) == 1
+                    and 0 in attribute
+                    and not self.keep_dict[arg_name]
+                    else {name: node.value for name, node in attribute.items()}
                 )
 
         return kwargs
@@ -208,32 +216,49 @@ def _extract_nodes_from_function_markers(function, parser):
 
 
 def _convert_objects_to_node_dictionary(objects, when):
-    list_of_tuples = _convert_objects_to_list_of_tuples(objects)
+    """Convert objects to node dictionary."""
+    list_of_tuples, keep_dict = _convert_objects_to_list_of_tuples(objects)
     _check_that_names_are_not_used_multiple_times(list_of_tuples, when)
     nodes = _convert_nodes_to_dictionary(list_of_tuples)
-    return nodes
+    return nodes, keep_dict
 
 
 def _convert_objects_to_list_of_tuples(objects):
+    """Convert objects to list of tuples.
+
+    Examples
+    --------
+    _convert_objects_to_list_of_tuples([{0: 0}, [4, (3, 2)], ((1, 4),))
+    [(0, 0), (4,), (3, 2), (1, 4)], False
+
+    """
+    keep_dict = False
+
     out = []
     for obj in objects:
         if isinstance(obj, dict):
             obj = obj.items()
 
         if isinstance(obj, Iterable) and not isinstance(obj, str):
+            keep_dict = True
             for x in obj:
                 if isinstance(x, Iterable) and not isinstance(x, str):
                     tuple_x = tuple(x)
                     if len(tuple_x) in [1, 2]:
                         out.append(tuple_x)
                     else:
-                        raise ValueError("ERROR")
+                        raise ValueError(
+                            f"Element {x} can only have two elements at most."
+                        )
                 else:
                     out.append((x,))
         else:
             out.append((obj,))
 
-    return out
+    if len(out) > 1:
+        keep_dict = False
+
+    return out, keep_dict
 
 
 def _check_that_names_are_not_used_multiple_times(list_of_tuples, when):
@@ -263,7 +288,19 @@ def _check_that_names_are_not_used_multiple_times(list_of_tuples, when):
         )
 
 
-def _convert_nodes_to_dictionary(list_of_tuples):
+def _convert_nodes_to_dictionary(
+    list_of_tuples: List[Tuple[str]],
+) -> Dict[str, Union[str, Path]]:
+    """Convert nodes to dictionaries.
+
+    Examples
+    --------
+    >>> _convert_nodes_to_dictionary([(0,), (1,)])
+    {0: 0, 1: 1}
+    >>> _convert_nodes_to_dictionary([(1, 0), (1,)])
+    {1: 0, 0: 1}
+
+    """
     nodes = {}
     counter = itertools.count()
     names = [x[0] for x in list_of_tuples if len(x) == 2]

tests/test_execute.py

@@ -169,3 +169,32 @@ def task_dummy(depends_on, produces):
     result = runner.invoke(cli, [tmp_path.as_posix()])
 
     assert result.exit_code == 0
+
+
+@pytest.mark.parametrize("input_type", ["list", "dict"])
+def test_preserve_input_for_dependencies_and_products(tmp_path, input_type):
+    """Input type for dependencies and products is preserved."""
+    path = tmp_path.joinpath("in.txt")
+    input_ = {0: path.as_posix()} if input_type == "dict" else [path.as_posix()]
+    path.touch()
+
+    path = tmp_path.joinpath("out.txt")
+    output = {0: path.as_posix()} if input_type == "dict" else [path.as_posix()]
+
+    source = f"""
+    import pytask
+    from pathlib import Path
+
+    @pytask.mark.depends_on({input_})
+    @pytask.mark.produces({output})
+    def task_dummy(depends_on, produces):
+        for nodes in [depends_on, produces]:
+            assert isinstance(nodes, dict)
+            assert len(nodes) == 1
+            assert 0 in nodes
+        produces[0].touch()
+    """
+    tmp_path.joinpath("task_dummy.py").write_text(textwrap.dedent(source))
+
+    session = main({"paths": tmp_path})
+    assert session.exit_code == 0

tests/test_nodes.py

@@ -7,6 +7,7 @@
 from _pytask.nodes import _check_that_names_are_not_used_multiple_times
 from _pytask.nodes import _convert_nodes_to_dictionary
 from _pytask.nodes import _convert_objects_to_list_of_tuples
+from _pytask.nodes import _convert_objects_to_node_dictionary
 from _pytask.nodes import _create_task_name
 from _pytask.nodes import _extract_nodes_from_function_markers
 from _pytask.nodes import _find_closest_ancestor
@@ -108,21 +109,25 @@ def state(self):
 
 @pytest.mark.unit
 @pytest.mark.parametrize(
-    ("x", "expected"),
+    ("x", "expected_lot", "expected_kd"),
     [
-        (["string"], [("string",)]),
-        (("string",), [("string",)]),
-        (range(2), [(0,), (1,)]),
-        ([{"a": 0, "b": 1}], [("a", 0), ("b", 1)]),
+        (["string"], [("string",)], False),
+        (("string",), [("string",)], False),
+        (range(2), [(0,), (1,)], False),
+        ([{"a": 0, "b": 1}], [("a", 0), ("b", 1)], False),
         (
             ["a", ("b", "c"), {"d": 1, "e": 1}],
             [("a",), ("b",), ("c",), ("d", 1), ("e", 1)],
+            False,
         ),
+        ([["string"]], [("string",)], True),
+        ([{0: "string"}], [(0, "string")], True),
     ],
 )
-def test_convert_objects_to_list_of_tuples(x, expected):
-    result = _convert_objects_to_list_of_tuples(x)
-    assert result == expected
+def test_convert_objects_to_list_of_tuples(x, expected_lot, expected_kd):
+    list_of_tuples, keep_dict = _convert_objects_to_list_of_tuples(x)
+    assert list_of_tuples == expected_lot
+    assert keep_dict is expected_kd
 
 
 ERROR = "'@pytask.mark.depends_on' has nodes with the same name:"
@@ -253,3 +258,30 @@ def test_shorten_node_name(node, paths, expectation, expected):
     with expectation:
         result = shorten_node_name(node, paths)
         assert result == expected
+
+
+@pytest.mark.integration
+@pytest.mark.parametrize("when", ["depends_on", "produces"])
+@pytest.mark.parametrize(
+    "objects, expectation, expected_dict, expected_kd",
+    [
+        ([0, 1], does_not_raise, {0: 0, 1: 1}, False),
+        ([{0: 0}, {1: 1}], does_not_raise, {0: 0, 1: 1}, False),
+        ([{0: 0}], does_not_raise, {0: 0}, True),
+        ([[0]], does_not_raise, {0: 0}, True),
+        ([((0, 0),), ((0, 1),)], ValueError, None, None),
+        ([{0: 0}, {0: 1}], ValueError, None, None),
+    ],
+)
+def test_convert_objects_to_node_dictionary(
+    objects, when, expectation, expected_dict, expected_kd
+):
+    expectation = (
+        pytest.raises(expectation, match=f"'@pytask.mark.{when}' has nodes")
+        if expectation == ValueError
+        else expectation()
+    )
+    with expectation:
+        node_dict, keep_dict = _convert_objects_to_node_dictionary(objects, when)
+        assert node_dict == expected_dict
+        assert keep_dict is expected_kd