FIX-modin-project#7240: Allow 'doc_checker.py' works with 'functools.…

…cached_property'

Signed-off-by: Anatoly Myachev <[email protected]>

.github/workflows/ci-required.yml

@@ -42,7 +42,7 @@ jobs:
       - uses: actions/checkout@v4
       - uses: ./.github/actions/python-only
       # The `numpydoc` version here MUST match the versions in the dev requirements files.
-      - run: pip install pytest pytest-cov pydocstyle numpydoc==1.1.0
+      - run: pip install pytest pytest-cov pydocstyle numpydoc==1.6.0
       - run: python -m pytest scripts/test
       - run: pip install -e ".[all]"
       - run: |

environment-dev.yml

@@ -69,4 +69,4 @@ dependencies:
       # Fixes breaking ipywidgets changes, but didn't release yet.
       - git+https://github.com/modin-project/modin-spreadsheet.git@49ffd89f683f54c311867d602c55443fb11bf2a5
       # The `numpydoc` version should match the version installed in the `lint-pydocstyle` job of the CI.
-      - numpydoc==1.1.0
+      - numpydoc==1.6.0

requirements/env_hdk.yml

@@ -45,4 +45,4 @@ dependencies:
   - pip:
       - dataframe-api-compat>=0.2.7
       # The `numpydoc` version should match the version installed in the `lint-pydocstyle` job of the CI.
-      - numpydoc==1.1.0
+      - numpydoc==1.6.0

requirements/env_unidist_linux.yml

@@ -58,4 +58,4 @@ dependencies:
       - git+https://github.com/modin-project/modin-spreadsheet.git@49ffd89f683f54c311867d602c55443fb11bf2a5
       - connectorx>=0.2.6a4
       # The `numpydoc` version should match the version installed in the `lint-pydocstyle` job of the CI.
-      - numpydoc==1.1.0
+      - numpydoc==1.6.0

requirements/env_unidist_win.yml

@@ -59,4 +59,4 @@ dependencies:
       - git+https://github.com/modin-project/modin-spreadsheet.git@49ffd89f683f54c311867d602c55443fb11bf2a5
       - connectorx>=0.2.6a4
       # The `numpydoc` version should match the version installed in the `lint-pydocstyle` job of the CI.
-      - numpydoc==1.1.0
+      - numpydoc==1.6.0

requirements/requirements-no-engine.yml

@@ -52,4 +52,4 @@ dependencies:
       # Fixes breaking ipywidgets changes, but didn't release yet.
       - git+https://github.com/modin-project/modin-spreadsheet.git@49ffd89f683f54c311867d602c55443fb11bf2a5
       # The `numpydoc` version should match the version installed in the `lint-pydocstyle` job of the CI.
-      - numpydoc==1.1.0
+      - numpydoc==1.6.0

scripts/doc_checker.py

@@ -32,8 +32,8 @@
 import types
 from typing import List
 
-from numpydoc.docscrape import NumpyDocString
-from numpydoc.validate import Docstring
+from numpydoc.docscrape import NumpyDocString, get_doc_object
+from numpydoc.validate import Validator
 
 # Let the other modules to know that the doc checker is running.
 os.environ["_MODIN_DOC_CHECKER_"] = "1"
@@ -73,14 +73,14 @@
 }
 
 
-def get_optional_args(doc: Docstring) -> dict:
+def get_optional_args(doc: Validator) -> dict:
     """
     Get optional parameters for the object for which the docstring is checked.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
 
     Returns
     -------
@@ -98,13 +98,13 @@ def get_optional_args(doc: Docstring) -> dict:
     }
 
 
-def check_optional_args(doc: Docstring) -> list:
+def check_optional_args(doc: Validator) -> list:
     """
     Check type description of optional arguments.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
+    doc : numpydoc.validate.Validator
 
     Returns
     -------
@@ -139,14 +139,14 @@ def check_optional_args(doc: Docstring) -> list:
     return errors
 
 
-def check_spelling_words(doc: Docstring) -> list:
+def check_spelling_words(doc: Validator) -> list:
     """
     Check spelling of chosen words in doc.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
 
     Returns
     -------
@@ -188,9 +188,18 @@ def check_spelling_words(doc: Docstring) -> list:
     ]
 
     docstring_start_line = None
-    for idx, line in enumerate(inspect.getsourcelines(doc.code_obj)[0]):
+    code_obj = doc.code_obj
+    if isinstance(code_obj, property):
+        # inspect.getsourcelines doesn't work with property
+        code_obj = code_obj.fget
+    elif isinstance(code_obj, functools.cached_property):
+        # inspect.getsourcelines doesn't work with cached_property
+        code_obj = code_obj.func
+
+    rows, starting_line = inspect.getsourcelines(code_obj)
+    for idx, line in enumerate(rows):
         if '"""' in line or "'''" in line:
-            docstring_start_line = doc.source_file_def_line + idx
+            docstring_start_line = starting_line + idx
             break
 
     errors = []
@@ -210,14 +219,14 @@ def check_spelling_words(doc: Docstring) -> list:
     return errors
 
 
-def check_docstring_indention(doc: Docstring) -> list:
+def check_docstring_indention(doc: Validator) -> list:
     """
     Check indention of docstring since numpydoc reports weird results.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
 
     Returns
     -------
@@ -240,14 +249,14 @@ def check_docstring_indention(doc: Docstring) -> list:
     return errors
 
 
-def validate_modin_error(doc: Docstring, results: dict) -> list:
+def validate_modin_error(doc: Validator, results: dict) -> list:
     """
     Validate custom Modin errors.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
     results : dict
         Dictionary that numpydoc.validate.validate return.
 
@@ -263,14 +272,14 @@ def validate_modin_error(doc: Docstring, results: dict) -> list:
     return results
 
 
-def skip_check_if_noqa(doc: Docstring, err_code: str, noqa_checks: list) -> bool:
+def skip_check_if_noqa(doc: Validator, err_code: str, noqa_checks: list) -> bool:
     """
     Skip the check that matches `err_code` if `err_code` found in noqa string.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
     err_code : str
         Error code found by numpydoc.
     noqa_checks : list
@@ -286,22 +295,31 @@ def skip_check_if_noqa(doc: Docstring, err_code: str, noqa_checks: list) -> bool
 
     # GL08 - missing docstring in an arbitary object; numpydoc code
     if err_code == "GL08":
-        name = doc.name.split(".")[-1]
-        # Numpydoc recommends to add docstrings of __init__ method in class docstring.
-        # So there is no error if docstring is missing in __init__
-        if name == "__init__":
-            return True
+        try:
+            name = doc.name.split(".")[-1]
+            # Numpydoc recommends to add docstrings of __init__ method in class docstring.
+            # So there is no error if docstring is missing in __init__
+            if name == "__init__":
+                return True
+        except AttributeError as exc:
+            # Cashed properties are not used for constructors, so we can safely ignore this exception
+            if (
+                not len(exc.args) > 0
+                and exc.args[0]
+                == "'cached_property' object has no attribute '__name__'"
+            ):
+                raise exc
     return err_code in noqa_checks
 
 
-def get_noqa_checks(doc: Docstring) -> list:
+def get_noqa_checks(doc: Validator) -> list:
     """
     Get codes after `# noqa`.
 
     Parameters
     ----------
-    doc : numpydoc.validate.Docstring
-        Docstring handler.
+    doc : numpydoc.validate.Validator
+        Validator handler.
 
     Returns
     -------
@@ -343,6 +361,11 @@ def get_noqa_checks(doc: Docstring) -> list:
     return [check.strip() for check in noqa_checks]
 
 
+def construct_validator(import_path: str) -> Validator:  # GL08
+    # helper function
+    return Validator(get_doc_object(Validator._load_obj(import_path)))
+
+
 # code snippet from numpydoc
 def validate_object(import_path: str) -> list:
     """
@@ -361,7 +384,7 @@ def validate_object(import_path: str) -> list:
     from numpydoc.validate import validate
 
     errors = []
-    doc = Docstring(import_path)
+    doc = construct_validator(import_path)
     if getattr(doc.obj, "__doc_inherited__", False) or (
         isinstance(doc.obj, property)
         and getattr(doc.obj.fget, "__doc_inherited__", False)
@@ -521,15 +544,6 @@ def monkeypatch(*args, **kwargs):
 
     modin.utils.instancer = functools.wraps(modin.utils.instancer)(lambda cls: cls)
 
-    # monkey-patch numpydoc for working correctly with properties
-    def load_obj(name, old_load_obj=Docstring._load_obj):
-        obj = old_load_obj(name)
-        if isinstance(obj, property):
-            obj = obj.fget
-        return obj
-
-    Docstring._load_obj = staticmethod(load_obj)
-
     # for testing hdk-engine docs without `pyhdk` installation
     sys.modules["pyhdk"] = Mock()
     sys.modules["pyhdk"].__version__ = "999"

scripts/test/test_doc_checker.py

@@ -12,12 +12,12 @@
 # governing permissions and limitations under the License.
 
 import pytest
-from numpydoc.validate import Docstring
 
 from scripts.doc_checker import (
     MODIN_ERROR_CODES,
     check_optional_args,
     check_spelling_words,
+    construct_validator,
     get_noqa_checks,
     get_optional_args,
 )
@@ -34,7 +34,7 @@
     ],
 )
 def test_get_optional_args(import_path, result):
-    optional_args = get_optional_args(Docstring(import_path))
+    optional_args = get_optional_args(construct_validator(import_path))
     assert optional_args == result
 
 
@@ -57,7 +57,7 @@ def test_get_optional_args(import_path, result):
     ],
 )
 def test_check_optional_args(import_path, result):
-    errors = check_optional_args(Docstring(import_path))
+    errors = check_optional_args(construct_validator(import_path))
     assert errors == result
 
 
@@ -88,7 +88,7 @@ def test_check_spelling_words(import_path, result):
                 ),
             )
         )
-    errors = check_spelling_words(Docstring(import_path))
+    errors = check_spelling_words(construct_validator(import_path))
     # the order of incorrect words found on the same line is not guaranteed
     for error in errors:
         assert error in result_errors
@@ -105,5 +105,5 @@ def test_check_spelling_words(import_path, result):
     ],
 )
 def test_get_noqa_checks(import_path, result):
-    noqa_checks = get_noqa_checks(Docstring(import_path))
+    noqa_checks = get_noqa_checks(construct_validator(import_path))
     assert noqa_checks == result