Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Final py3.9 deprecation #2801

Merged
merged 3 commits into from Sep 16, 2021
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
215 changes: 134 additions & 81 deletions nltk/test/unit/test_util.py
@@ -1,81 +1,134 @@
"""
Unit tests for nltk.util.
"""

import unittest

from nltk.util import everygrams


class TestEverygrams(unittest.TestCase):
def setUp(self):
"""Form test data for tests."""
self.test_data = iter("a b c".split())

def test_everygrams_without_padding(self):
expected_output = [
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("c",),
]
output = everygrams(self.test_data)
self.assertCountEqual(output, expected_output)

def test_everygrams_max_len(self):
expected_output = [
("a",),
("a", "b"),
("b",),
("b", "c"),
("c",),
]
output = everygrams(self.test_data, max_len=2)
self.assertCountEqual(output, expected_output)

def test_everygrams_min_len(self):
expected_output = [
("a", "b"),
("b", "c"),
("a", "b", "c"),
]
output = everygrams(self.test_data, min_len=2)
self.assertCountEqual(output, expected_output)

def test_everygrams_pad_right(self):
expected_output = [
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("b", "c", None),
("c",),
("c", None),
("c", None, None),
(None,),
(None, None),
(None,),
]
output = everygrams(self.test_data, max_len=3, pad_right=True)
self.assertCountEqual(output, expected_output)

def test_everygrams_pad_left(self):
expected_output = [
(None,),
(None, None),
(None, None, "a"),
(None,),
(None, "a"),
(None, "a", "b"),
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("c",),
]
output = everygrams(self.test_data, max_len=3, pad_left=True)
self.assertCountEqual(output, expected_output)
import pytest

from nltk.util import everygrams, usage


def test_usage_with_self(capsys):
class MyClass:
def kwargs(self, a=1):
...

def no_args(self):
...

def pos_args(self, a, b):
...

def pos_args_and_kwargs(self, a, b, c=1):
...

usage(MyClass)

captured = capsys.readouterr()
assert captured.out == (
"MyClass supports the following operations:\n"
" - self.kwargs(a=1)\n"
" - self.no_args()\n"
" - self.pos_args(a, b)\n"
" - self.pos_args_and_kwargs(a, b, c=1)\n"
)


def test_usage_with_cls(capsys):
class MyClass:
@classmethod
def clsmethod(cls):
...

@classmethod
def clsmethod_with_args(cls, a, b, c=1):
...

usage(MyClass)

captured = capsys.readouterr()
assert captured.out == (
"MyClass supports the following operations:\n"
" - cls.clsmethod()\n"
" - cls.clsmethod_with_args(a, b, c=1)\n"
)


def test_usage_on_builtin():
# just check the func passes, since
# builtins change each python version
usage(dict)


@pytest.fixture
def everygram_input():
"""Form test data for tests."""
return iter(["a", "b", "c"])


def test_everygrams_without_padding(everygram_input):
expected_output = [
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("c",),
]
output = list(everygrams(everygram_input))
assert output == expected_output


def test_everygrams_max_len(everygram_input):
expected_output = [
("a",),
("a", "b"),
("b",),
("b", "c"),
("c",),
]
output = list(everygrams(everygram_input, max_len=2))
assert output == expected_output


def test_everygrams_min_len(everygram_input):
expected_output = [
("a", "b"),
("a", "b", "c"),
("b", "c"),
]
output = list(everygrams(everygram_input, min_len=2))
assert output == expected_output


def test_everygrams_pad_right(everygram_input):
expected_output = [
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("b", "c", None),
("c",),
("c", None),
("c", None, None),
(None,),
(None, None),
(None,),
]
output = list(everygrams(everygram_input, max_len=3, pad_right=True))
assert output == expected_output


def test_everygrams_pad_left(everygram_input):
expected_output = [
(None,),
(None, None),
(None, None, "a"),
(None,),
(None, "a"),
(None, "a", "b"),
("a",),
("a", "b"),
("a", "b", "c"),
("b",),
("b", "c"),
("c",),
]
output = list(everygrams(everygram_input, max_len=3, pad_left=True))
assert output == expected_output
32 changes: 19 additions & 13 deletions nltk/util.py
Expand Up @@ -37,32 +37,38 @@
######################################################################


def usage(obj, selfname="self"):
def usage(obj):
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

could alternatively delete this method? it's only used in nltk/test/classify.doctest and that usage feels unnecessary?

Copy link
Member

@tomaarsen tomaarsen Sep 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was unaware of this function, but I could definitely see how it has a nice educatory use. Being able to quickly see the useful methods of a class can be quite useful if you're not entirely aware of the codebase, especially considering that NLTK is definitely not as well documented as I'd like.

There are some classes in NLTK for which users definitely might like to get an overview of the methods that it supports. nltk.tree.Tree, nltk.BigramCollocationFinder and nltk.FreqDist come to mind, but upon trying the last one, I get the following exception:

>>> import nltk
>>> nltk.usage(nltk.FreqDist)
FreqDist supports the following operations:
  - self.B()
  - self.N()
  - self.Nr(r, bins=None)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "[sic]\nltk\util.py", line 53, in usage
    signature = inspect.signature(method)
  File "[sic]\lib\inspect.py", line 3130, in signature    
    return Signature.from_callable(obj, follow_wrapped=follow_wrapped)
  File "[sic]\lib\inspect.py", line 2879, in from_callable
    return _signature_from_callable(obj, sigcls=cls,
  File "[sic]\lib\inspect.py", line 2334, in _signature_from_callable
    return _signature_from_builtin(sigcls, obj,
  File "[sic]\lib\inspect.py", line 2145, in _signature_from_builtin
    raise ValueError("no signature found for builtin {!r}".format(func))
ValueError: no signature found for builtin <method 'clear' of 'dict' objects>

Potentially this is because FreqDist subclasses collections' Counter, which again subclasses dict. Note that I also got a similar exception before this PR.

As far as I'm concerned, this can still be merged, as the bug existed before this already regardless. We can just open an issue for it then. However, if you feel like it, you can also include a fix for this in this PR already.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

good catch!! you're totally right, thanks!

turns out that inspect and some builtin functions don't always work out together. i can't really figure out how to get the failing ones working, my best workaround is to skip them?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, what about Python's built-in help command? It gives you both methods and attributes as well as their docstrings. I use it extensively with all kinds of codebases.

I completely agree that the docs need some love, but that's a separate issue imo.

Copy link
Member

@tomaarsen tomaarsen Sep 16, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's indeed a separate issue I'd say.

I'd even consider usage failing on some builtin functions to be a separate issue. If you'd like to get this merged, then we can just open an issue saying that usage is broken in some cases. That is, assuming that the bug still exists.

Regarding the tests, it might be a good idea to check if the output .startswith("MyClass supports the following operations:") or something, or whether the output is indeed a string with at least length ~35 or something. It's not much, but then at least we have some tests to verify that it prints a string with "... supports the following operations: ...".

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, let's just treat that as a separate issue.

str(obj) # In case it's lazy, this will load it.

if not isinstance(obj, type):
obj = obj.__class__

print("%s supports the following operations:" % obj.__name__)
print(f"{obj.__name__} supports the following operations:")
for (name, method) in sorted(pydoc.allmethods(obj).items()):
if name.startswith("_"):
continue
if getattr(method, "__deprecated__", False):
continue

getargspec = inspect.getfullargspec
args, varargs, varkw, defaults = getargspec(method)[:4]
if (
args
and args[0] == "self"
and (defaults is None or len(args) > len(defaults))
):
args = args[1:]
name = f"{selfname}.{name}"
argspec = inspect.formatargspec(args, varargs, varkw, defaults)
try:
sig = str(inspect.signature(method))
except ValueError as e:
# builtins sometimes don't support introspection
if "builtin" in str(e):
continue
else:
raise

args = sig.lstrip("(").rstrip(")").split(", ")
meth = inspect.getattr_static(obj, name)
if isinstance(meth, (classmethod, staticmethod)):
name = f"cls.{name}"
elif args and args[0] == "self":
name = f"self.{name}"
args.pop(0)
print(
textwrap.fill(
f"{name}{argspec}",
f"{name}({', '.join(args)})",
initial_indent=" - ",
subsequent_indent=" " * (len(name) + 5),
)
Expand Down