bpo-33881: Use NFKC to find duplicate members in make_dataclass

[ValeriyaSinevich]

I measured how it affected the performance by creating a dataclass with 10000 members.
python.bat -m timeit -s "from dataclasses import make_dataclass" -s "arg_list = [chr(k) * i for k in range(97, 123) for i in range(1, 500)]" "make_dataclass('a', arg_list)"
The performance didn't change, both before and after the changes it takes around 14.5s.

https://bugs.python.org/issue33881

[ericvsmith]

This will require some tests. Can you add them?

I haven't had time to give a thorough review, but I will once there are some basic tests.

Thanks!

[bedevere-bot]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[remilapeyre]

Hi @ericvsmith, I wrote some tests for the PR and made a small change: normalize before checking for keywords has there was a bug there too.

commit 27258e9723293178143922d6a503ff6885c3b6a1
Author: Rémi Lapeyre <remi.lapeyre@henki.fr>
Date:   Thu Jan 3 01:17:48 2019 +0100

    Add tests for dataclasses members normalization

diff --git a/Lib/dataclasses.py b/Lib/dataclasses.py
index cdb9bf23ed..64ee5610d7 100644
--- a/Lib/dataclasses.py
+++ b/Lib/dataclasses.py
@@ -1120,9 +1120,9 @@ def make_dataclass(cls_name, fields, *, bases=(), namespace=None, init=True,
 
         if not isinstance(name, str) or not name.isidentifier():
             raise TypeError(f'Field names must be valid identifers: {name!r}')
-        if keyword.iskeyword(name):
-            raise TypeError(f'Field names must not be keywords: {name!r}')
         normalized_name = unicodedata.normalize('NFKC', name)
+        if keyword.iskeyword(normalized_name):
+            raise TypeError(f'Field names must not be keywords: {normalized_name!r}')
         if normalized_name in seen:
             raise TypeError(f'Field name duplicated: {normalized_name!r}')
 
diff --git a/Lib/test/test_dataclasses.py b/Lib/test/test_dataclasses.py
index d9556c7ff9..cb81de80da 100755
--- a/Lib/test/test_dataclasses.py
+++ b/Lib/test/test_dataclasses.py
@@ -2918,6 +2918,15 @@ class TestMakeDataclass(unittest.TestCase):
                 C = make_dataclass(classname, ['a', 'b'])
                 self.assertEqual(C.__name__, classname)
 
+    def test_normalize_members(self):
+        with self.assertRaisesRegexp(TypeError, "Field name duplicated: 'μ'"):
+            make_dataclass('a', ['\u00b5', '\u03bc'])
+
+    def test_normalized_keywords(self):
+        with self.assertRaisesRegexp(TypeError,
+                "Field names must not be keywords: 'assert'"):
+            make_dataclass('a', ['𝚊ssert'])
+
 class TestReplace(unittest.TestCase):
     def test(self):
         @dataclass(frozen=True)

[csabella]

@ValeriyaSinevich, please make the changes requested in the code review and please fix the merge conflict. Thanks!

[ericvsmith]

@ValeriyaSinevich : Any chance you can add the tests to the PR?

[csabella]

I'm going to close this pull request as inactive. It can be reopened if the original author comes back to working on it or someone else can create a new PR. If a new PR is created and the original change is used, please credit the original author as co-author.