Replace pydot with pygraphviz. (#308)

Author: tobiasraabe

.github/workflows/main.yml

@@ -65,23 +65,3 @@ jobs:
         if: runner.os == 'Linux' && matrix.python-version == '3.8'
         shell: bash -l {0}
         run: bash <(curl -s https://codecov.io/bash) -F end_to_end -c
-
-
-  docs:
-
-    name: Run documentation.
-    runs-on: ubuntu-latest
-
-    steps:
-      - uses: actions/checkout@v2
-      - uses: conda-incubator/setup-miniconda@v2
-        with:
-          auto-update-conda: true
-
-      - name: Install core dependencies.
-        shell: bash -l {0}
-        run: conda install -c conda-forge tox-conda
-
-      - name: Build docs
-        shell: bash -l {0}
-        run: tox -e sphinx

docs/source/tutorials/visualizing_the_dag.md

@@ -1,11 +1,11 @@
 # Visualizing the DAG
 
 To visualize the {term}`DAG` of the project, first, install
-[pydot](https://github.com/pydot/pydot) and [graphviz](https://graphviz.org/). For
-example, you can both install with conda
+[pygraphviz](https://github.com/pygraphviz/pygraphviz) and
+[graphviz](https://graphviz.org/). For example, you can both install with conda
 
 ```console
-$ conda install -c conda-forge pydot
+$ conda install -c conda-forge pygraphviz
 ```
 
 After that, pytask offers two interfaces to visualize your project's {term}`DAG`.
@@ -27,7 +27,7 @@ There are ways to customize the visualization.
    other layouts, which are listed [here](https://graphviz.org/docs/layouts/).
 1. Using the {option}`pytask dag --output-path` option, you can provide a file name for
    the graph. The file extension changes the output format as supported by
-   [pydot](https://github.com/pydot/pydot).
+   [pygraphviz](https://github.com/pygraphviz/pygraphviz).
 
 ## Programmatic Interface
 
@@ -49,10 +49,10 @@ shape of all nodes to hexagons by adding the property to the node attributes.
 nx.set_node_attributes(dag, "hexagon", "shape")
 ```
 
-For drawing, you better switch to pydot or pygraphviz since the matplotlib backend
-handles shapes with texts poorly. Here we use pydot and store the graph as a `.svg`.
+For drawing, you better switch to pygraphviz since the matplotlib backend handles shapes
+with texts poorly. Here we store the graph as a `.svg`.
 
 ```python
-graph = nx.nx_pydot.to_pydot(dag)
-graph.write_svg(produces)
+graph = nx.nx_agraph.to_agraph(dag)
+graph.draw(path, prog=layout)
 ```

environment.yml

@@ -32,7 +32,7 @@ dependencies:
   - jupyterlab
   - matplotlib
   - pre-commit
-  - pydot
+  - pygraphviz
   - pytest
   - pytest-cov
   - pytest-xdist

src/_pytask/graph.py

@@ -123,7 +123,7 @@ def dag(**config_from_cli: Any) -> NoReturn:
     else:
         try:
             session.hook.pytask_log_session_header(session=session)
-            import_optional_dependency("pydot")
+            import_optional_dependency("pygraphviz")
             check_for_optional_program(
                 session.config["layout"],
                 extra="The layout program is part of the graphviz package which you "
@@ -154,10 +154,11 @@ def build_dag(config_from_cli: dict[str, Any]) -> nx.DiGraph:
     """Build the DAG.
 
     This function is the programmatic interface to ``pytask dag`` and returns a
-    preprocessed :class:`pydot.Dot` which makes plotting easier than with matplotlib.
+    preprocessed :class:`pygraphviz.AGraph` which makes plotting easier than with
+    matplotlib.
 
     To change the style of the graph, it might be easier to convert the graph back to
-    networkx, set attributes, and convert back to pydot or pygraphviz.
+    networkx, set attributes, and convert back to pygraphviz.
 
     Parameters
     ----------
@@ -167,7 +168,7 @@ def build_dag(config_from_cli: dict[str, Any]) -> nx.DiGraph:
 
     Returns
     -------
-    pydot.Dot
+    pygraphviz.AGraph
         A preprocessed graph which can be customized and exported.
 
     """
@@ -190,7 +191,7 @@ def build_dag(config_from_cli: dict[str, Any]) -> nx.DiGraph:
     else:
         try:
             session.hook.pytask_log_session_header(session=session)
-            import_optional_dependency("pydot")
+            import_optional_dependency("pygraphviz")
             check_for_optional_program(
                 session.config["layout"],
                 extra="The layout program is part of the graphviz package which you "
@@ -212,7 +213,6 @@ def _refine_dag(session: Session) -> nx.DiGraph:
     dag = _shorten_node_labels(session.dag, session.config["paths"])
     dag = _clean_dag(dag)
     dag = _style_dag(dag)
-    dag = _escape_node_names_with_colons(dag)
     dag.graph["graph"] = {"rankdir": session.config["rank_direction"].name}
 
     return dag
@@ -239,7 +239,7 @@ def _create_session(config_from_cli: dict[str, Any]) -> nx.DiGraph:
     else:
         try:
             session.hook.pytask_log_session_header(session=session)
-            import_optional_dependency("pydot")
+            import_optional_dependency("pygraphviz")
             check_for_optional_program(session.config["layout"])
             session.hook.pytask_collect(session=session)
             session.hook.pytask_resolve_dependencies(session=session)
@@ -282,19 +282,8 @@ def _style_dag(dag: nx.DiGraph) -> nx.DiGraph:
     return dag
 
 
-def _escape_node_names_with_colons(dag: nx.DiGraph) -> nx.DiGraph:
-    """Escape node names with colons.
-
-    pydot cannot handle colons in node names since it messes up some syntax. Escaping
-    works by wrapping the string in double quotes. See this issue for more information:
-    https://github.com/pydot/pydot/issues/224.
-
-    """
-    return nx.relabel_nodes(dag, {name: f'"{name}"' for name in dag.nodes})
-
-
 def _write_graph(dag: nx.DiGraph, path: Path, layout: str) -> None:
     """Write the graph to disk."""
     path.parent.mkdir(exist_ok=True, parents=True)
-    graph = nx.nx_pydot.to_pydot(dag)
-    graph.write(path, prog=layout, format=path.suffix[1:])
+    graph = nx.nx_agraph.to_agraph(dag)
+    graph.draw(path, prog=layout)

tests/test_graph.py

@@ -10,11 +10,11 @@
 from pytask import ExitCode
 
 try:
-    import pydot  # noqa: F401
+    import pygraphviz  # noqa: F401
 except ImportError:  # pragma: no cover
-    _IS_PYDOT_INSTALLED = False
+    _IS_PYGRAPHVIZ_INSTALLED = False
 else:
-    _IS_PYDOT_INSTALLED = True
+    _IS_PYGRAPHVIZ_INSTALLED = True
 
 _GRAPH_LAYOUTS = ["neato", "dot", "fdp", "sfdp", "twopi", "circo"]
 
@@ -34,7 +34,7 @@
 
 
 @pytest.mark.end_to_end
-@pytest.mark.skipif(not _IS_PYDOT_INSTALLED, reason="pydot is required")
+@pytest.mark.skipif(not _IS_PYGRAPHVIZ_INSTALLED, reason="pygraphviz is required")
 @pytest.mark.parametrize("layout", _PARAMETRIZED_LAYOUTS)
 @pytest.mark.parametrize("format_", _TEST_FORMATS)
 @pytest.mark.parametrize("rankdir", ["LR"])
@@ -70,7 +70,7 @@ def task_example(): pass
 
 
 @pytest.mark.end_to_end
-@pytest.mark.skipif(not _IS_PYDOT_INSTALLED, reason="pydot is required")
+@pytest.mark.skipif(not _IS_PYGRAPHVIZ_INSTALLED, reason="pygraphviz is required")
 @pytest.mark.parametrize("layout", _PARAMETRIZED_LAYOUTS)
 @pytest.mark.parametrize("format_", _TEST_FORMATS)
 @pytest.mark.parametrize("rankdir", [_RankDirection.LR.value, _RankDirection.TB])
@@ -91,9 +91,9 @@ def task_example(): pass
     def task_create_graph():
         dag = pytask.build_dag({{"paths": Path(__file__).parent}})
         dag.graph = {{"rankdir": "{rankdir_str}"}}
-        graph = nx.nx_pydot.to_pydot(dag)
+        graph = nx.nx_agraph.to_agraph(dag)
         path = Path(__file__).parent.joinpath("dag.{format_}")
-        graph.write(path, prog="{layout}", format=path.suffix[1:])
+        graph.draw(path, prog="{layout}")
     """
 
     tmp_path.joinpath("task_example.py").write_text(textwrap.dedent(source))
@@ -124,7 +124,7 @@ def task_example(): pass
 
     monkeypatch.setattr(
         "_pytask.compat.importlib.import_module",
-        lambda x: _raise_exc(ImportError("pydot not found")),  # noqa: U100
+        lambda x: _raise_exc(ImportError("pygraphviz not found")),  # noqa: U100
     )
 
     result = runner.invoke(
@@ -133,8 +133,9 @@ def task_example(): pass
     )
 
     assert result.exit_code == ExitCode.FAILED
-    assert "pytask requires the optional dependency 'pydot'." in result.output
-    assert "pip or conda" in result.output
+    assert "pytask requires the optional dependency 'pygraphviz'." in result.output
+    assert "pip" in result.output
+    assert "conda" in result.output
     assert "Traceback" not in result.output
     assert not tmp_path.joinpath("dag.png").exists()
 
@@ -150,22 +151,23 @@ def test_raise_error_with_graph_via_task_missing_optional_dependency(
 
     def task_create_graph():
         dag = pytask.build_dag({"paths": Path(__file__).parent})
-        graph = nx.nx_pydot.to_pydot(dag)
+        graph = nx.nx_agraph.to_agraph(dag)
         path = Path(__file__).parent.joinpath("dag.png")
-        graph.write(path, prog="dot", format=path.suffix[1:])
+        graph.draw(path, prog="dot")
     """
     tmp_path.joinpath("task_example.py").write_text(textwrap.dedent(source))
 
     monkeypatch.setattr(
         "_pytask.compat.importlib.import_module",
-        lambda x: _raise_exc(ImportError("pydot not found")),  # noqa: U100
+        lambda x: _raise_exc(ImportError("pygraphviz not found")),  # noqa: U100
     )
 
     result = runner.invoke(cli, [tmp_path.as_posix()])
 
     assert result.exit_code == ExitCode.FAILED
-    assert "pytask requires the optional dependency 'pydot'." in result.output
-    assert "pip or conda" in result.output
+    assert "pytask requires the optional dependency 'pygraphviz'." in result.output
+    assert "pip" in result.output
+    assert "conda" in result.output
     assert "Traceback" in result.output
     assert not tmp_path.joinpath("dag.png").exists()
 
@@ -211,9 +213,9 @@ def test_raise_error_with_graph_via_task_missing_optional_program(
 
     def task_create_graph():
         dag = pytask.build_dag({"paths": Path(__file__).parent})
-        graph = nx.nx_pydot.to_pydot(dag)
+        graph = nx.nx_agraph.to_agraph(dag)
         path = Path(__file__).parent.joinpath("dag.png")
-        graph.write(path, prog="dot", format=path.suffix[1:])
+        graph.draw(path, prog="dot")
     """
     tmp_path.joinpath("task_example.py").write_text(textwrap.dedent(source))
 

tox.ini

@@ -30,7 +30,7 @@ conda_deps =
     # Optional and test dependencies
     graphviz
     pexpect
-    pydot
+    pygraphviz
 
 commands =
     pytest {posargs}
@@ -42,12 +42,6 @@ commands =
     sphinx-build -T -b html -d {envtmpdir}/doctrees . {envtmpdir}/html
     - sphinx-build -T -b linkcheck -d {envtmpdir}/doctrees . {envtmpdir}/linkcheck
 
-[doc8]
-ignore = D002, D004
-ignore-path =
-    docs/build
-max-line-length = 88
-
 [flake8]
 docstring-convention = numpy
 exclude =
@@ -75,7 +69,6 @@ filterwarnings =
     ignore: the imp module is deprecated in favour of importlib
     ignore: Using or importing the ABCs from 'collections' instead of from
     ignore: The (parser|symbol) module is deprecated and will be removed in future
-    ignore: nx\.nx_pydot\.to_pydot depends on the pydot package
 markers =
     wip: Tests that are work-in-progress.
     unit: Flag for unit tests which target mainly a single function.