add nbqa_ipynb to temporary files (#870)

allows config globs to apply to nbqa-generated files.

e.g.

    default_magicgr4dg5bw_nbqa_ipynb.py

Added to suffix, so e.g. all generated .py files match *nbqa_ipynb.py

docs/known-limitations.rst

@@ -11,7 +11,19 @@ then the following will still not be processed:
 - cells with code which ``IPython`` would transform magics into (e.g. ``get_ipython().system('ls')``).
 
 Because ``nbQA`` converts the code cells in Jupyter notebooks to temporary Python files for linting, certain flags like ``flake8``'s
-``--per-file-ignores`` don't work. The temporary Python files will not match the specified file patterns and ignored error codes will still
+``--per-file-ignores`` don't work perfectly.
+The temporary Python files will not match the specified file patterns and ignored error codes will still
 surface (`GH issue <https://github.com/nbQA-dev/nbQA/issues/730>`_).
+nbqa-generated temporary files will contain the string ``nbqa_ipynb``,
+so you can still apply per-file-ignores if you add an additional pattern:
+
+.. sourcecode:: ini
+
+    [flake8]
+    per-file-ignores =
+      examples/*.ipynb: E402
+      examples/*nbqa_ipynb.py: E402
+
+The directory and the stem of the filename are preserved, so e.g. ``path/to/mynotebook.ipynb`` will be ``path/to/mynotebook{randomstring}_nbqa_ipynb.py`` when nbqa passes it to the linter.
 
 Any other limitation is likely unintentional - if you run into any, please do report an issue.

nbqa/__main__.py

@@ -526,7 +526,7 @@ def _get_nb_to_tmp_mapping(
                 prefix=remove_suffix(
                     os.path.basename(notebook), os.path.splitext(notebook)[-1]
                 ),
-                suffix=SUFFIX[md],
+                suffix="_nbqa_ipynb" + SUFFIX[md],
             )
         )
         relative_path, _ = get_relative_and_absolute_paths(

tests/tools/test_flake8_works.py

@@ -1,6 +1,7 @@
 """Check :code:`flake8` works as intended."""
 
 import os
+from pathlib import Path
 from textwrap import dedent
 from typing import TYPE_CHECKING
 
@@ -91,3 +92,49 @@ def test_cell_with_all_magics(capsys: "CaptureFixture") -> None:
     out, err = capsys.readouterr()
     assert out == ""
     assert err == ""
+
+
+def test_per_file_ignores(
+    tmp_notebook_for_testing: Path, capsys: "CaptureFixture"
+) -> None:
+    """
+    Check flake8 per-file-ignore patterns work.
+
+    Parameters
+    ----------
+    tmp_notebook_for_testing
+        notebook Path to test
+    capsys
+        Pytest fixture to capture stdout and stderr.
+    """
+    # enable per-file ignores with nbqa glob
+    flake8_ini = Path(".flake8")
+    flake8_ini.write_text(
+        dedent(
+            """
+        [flake8]
+        per-file-ignores =
+          **/*.ipynb: E402
+          **/*nbqa_ipynb.py: E402
+        """
+        ),
+        encoding="utf-8",
+    )
+
+    main(["flake8", str(tmp_notebook_for_testing)])
+    flake8_ini.unlink()
+
+    expected_path_0 = os.path.join("tests", "data", "notebook_for_testing.ipynb")
+
+    out, err = capsys.readouterr()
+    expected_out = dedent(
+        f"""\
+        {expected_path_0}:cell_1:1:1: F401 'os' imported but unused
+        {expected_path_0}:cell_1:3:1: F401 'glob' imported but unused
+        {expected_path_0}:cell_1:5:1: F401 'nbqa' imported but unused
+        {expected_path_0}:cell_2:19:9: W291 trailing whitespace
+        {expected_path_0}:cell_4:1:1: F401 'random.randint' imported but unused
+        """
+    )
+    assert err == ""
+    assert sorted(out.splitlines()) == sorted(expected_out.splitlines())